Workshop Proceedings

EuropeanWorkshop on photonic

solutions for wireless, access, and in house networks

May 18-20, 2009 inHaus Innovation Centre University Campus Duisburg GERMANY

Workshop programme at a glance

Monday, May 18 Tuesday, May 19 Wednesday, May 209:0010:00

Session #5 Home networks

11:00 12:00

Session #2

European project presentations

Session #6 Poster session

13:00 14:00 Session #3

Components15:00 Registration Session #7

Access networks

Welcome talks 16:00 Sessions #4 Radio over Fiber

techniques17:00

Session #1 Optical Wireless Communications

Session 1 Monday May 18th, 2009 16:40 – 18:00 h

Optical Wireless Communications

Session Chair: D. JägerUniversity of Duisburg-Essen, Duisburg (Germany)

Visible light communications (invited) S. Haruyama Keio University, Yokohama (Japan)

High-Speed Information Broadcast using Phosphorescent White-Light LEDs – Advances and Prospects (invited)K.-D. Langer1, J. Vu i 1, C. Kottke1, L. Fernández del Rosal1, S. Nerreter2,J. Walewski31 Fraunhofer Institute for Telecommunications, Berlin, (Germany) 2 Siemens AG, Berlin (Germany) 3 Siemens AG, Munich (Germany)

Choice of the Modulation for Gbps Wireless Infrared Systems M. Wolf, L. Grobe, J. LiIlmenau University of Technology, Ilmenau (Germany)

Line-of-sight Infrared Wireless Path Loss Simulation in an Aircraft CabinS. Dimitrov1,2, R. Mesleh1, H. Haas1,3, M. Cappitelli2, M. Olbert2, E. Bassow4

1 Jacobs University Bremen (Germany) 2 EADS Deutschland GmbH, Hamburg (Germany) 3 Institute for Digital Communications, University of Edinburgh (United Kingdom)4 Airbus Deutschland GmbH, Hamburg (Germany)

European workshop on photonic solutions for wireless, access, and in-house networks Duisburg, May 18-20, 2009

- 1 -


- 2 -

Kei

o U

nive

rsity

SD

M1

Vis

ible

Lig

ht C

omm

unic

atio

n an

d Ja

pane

se V

isib

le L

ight

Com

mun

icat

ions

Con

sort

ium

Shin

ichi

ro H

aruy

ama,

Pro

fess

or

The

Gra

duat

e Sc

hool

of S

yste

m D

esig

n an

d M

anag

emen

tK

eio

Uni

vers

ity, Y

okoh

ama,

Jap

anE

mai

l:ha

ruya

ma@

sdm

.kei

o.ac

.jp

Cha

irm

an, V

isib

le L

ight

Com

mun

icat

ions

Con

sort

ium

,T

okyo

, Jap

an

Dui

sbur

g, G

erm

any

May

200

9

Kei

o U

nive

rsity

SD

M2

1.V

isib

le L

ight

Com

mun

icat

ion

2.D

evic

es u

sed

for

Vis

ible

Lig

ht C

omm

unic

atio

n3.

Com

mun

icat

ion

usin

g Im

age

Sens

ors

4.A

pplic

atio

ns o

f Vis

ible

Lig

ht C

omm

unic

atio

n5.

Vis

ible

Lig

ht C

omm

unic

atio

ns C

onso

rtiu

m6.

Stan

dard

izat

ion

of V

isib

le L

ight

Com

mun

icat

ion

Con

tent

s

Kei

o U

nive

rsity

SD

M3

Vis

ible

Lig

ht c

omm

unic

atio

n is

a w

irel

ess c

omm

unic

atio

n te

chno

logy

that

use

s lig

ht th

at is

vis

ible

to h

uman

s.

An

exam

ple

of v

isib

le li

ght c

omm

unic

atio

n: th

e us

e of

LE

D il

lum

inat

ion

as a

tr

ansm

itter

1.V

isib

le L

ight

Com

mun

icat

ion

Kei

o U

nive

rsity

SD

M4

A. L

ED

ligh

ts w

ill b

e us

ed e

very

whe

reB

. Inf

rast

ruct

ure

loca

ted

at id

eal l

ocat

ions

C. E

asy

iden

tific

atio

n of

pla

ces o

r th

ings

D. T

here

is n

o re

gula

tion

for

visi

ble

light

com

mun

icat

ion

so fa

r

Cha

ract

eris

tics o

f Vis

ible

Lig

ht C

omm

unic

atio

n

Kei

o U

nive

rsity

SD

M5

A: L

ED

ligh

ts w

ill b

e us

ed e

very

whe

re

Man

y lig

ht so

urce

s suc

h as

elec

tric

ligh

ts, a

utom

obile

hea

d/re

ar

lam

ps, t

raff

ic li

ghts

, ind

icat

or la

mps

, etc

. are

now

bei

ng r

epla

ced

by L

ED

ligh

ts.

If L

ED

ligh

ts c

an tr

ansm

it da

ta in

add

ition

to

emitt

ing

visi

ble

light

s, th

ey c

an b

e us

ed a

s per

vasi

ve

com

mun

icat

ion

equi

pmen

ts.

B:I

nfra

stru

ctur

e lo

cate

d at

idea

l loc

atio

nsIn

gen

eral

, ele

ctri

c lig

hts a

re a

ttac

hed

toth

e ce

iling

, whi

ch is

an

idea

l pos

ition

for

send

ing

data

to u

sers

.V

isib

le li

ght c

anno

t sen

d da

ta if

ther

e is

an

obst

acle

bet

wee

n a

tran

smitt

er a

nd a

rec

eive

r. If

vis

ible

ligh

t tra

nsm

itter

is a

ttac

hed

to th

e ce

iling

, the

cha

nces

of o

bstr

uctio

n is

smal

ler

than

oth

erlig

htin

g po

sitio

ns.

Kei

o U

nive

rsity

SD

M6

C: E

asy

iden

tific

atio

n of

pla

ces o

r th

ings

If v

isib

le li

ght c

omm

unic

atio

n is

don

e us

ing

LE

D li

ght,

hum

an c

an e

asily

iden

tify

it as

a d

ata

tran

smitt

er.

If a

flas

hlig

ht L

ED

is u

sed

as a

dat

a tr

ansm

itter

, hum

an

can

accu

rate

ly p

oint

the

light

to a

vis

ible

ligh

t rec

eive

r.

D: T

here

is n

o re

gula

tion

for

visi

ble

light

com

mun

icat

ion

so fa

r

For

radi

o w

aves

, a li

cens

e is

nec

essa

ry fo

r its

use

and

de

taile

d fr

eque

ncy

allo

catio

n is

stri

ctly

reg

ulat

ed in

eac

h co

untr

y. H

owev

er, c

omm

unic

atio

n us

ing

visi

ble

light

is n

ot

regu

late

d so

far.

Kei

o U

nive

rsity

SD

M7

Tra

nsm

itter

devi

ce o

f vis

ible

light

com

mun

icat

ion

-Vis

ible

ligh

t LE

DL

ED

ligh

t int

ensi

ty is

mod

ulat

ed b

y co

ntro

lling

its

cur

rent

.da

ta r

ate:

low

spee

d to

ver

y hi

gh sp

eed

(up

to se

vera

l hun

dred

Mbp

s)-F

luor

esce

nt la

mp

FSK

mod

ulat

ion

of h

igh

freq

uenc

yflu

ores

cent

ligh

tda

ta r

ate:

up

to se

vera

l kilo

bps

2. D

evic

es u

sed

for

Vis

ible

Lig

ht C

omm

unic

atio

n

Kei

o U

nive

rsity

SD

M8

Rec

eive

r de

vice

of v

isib

le li

ght c

omm

unic

atio

n

-PIN

pho

to d

iode

-hig

h sp

eed

rece

ptio

n up

to 1

Gbp

s-A

vala

nche

pho

to d

iode

-ver

y se

nsiti

ve r

ecep

tion

-Im

age

sens

or-s

imul

tane

ous i

mag

e ac

quis

ition

and

dat

a re

cept

ion

Kei

o U

nive

rsity

SD

M9

Dat

a ca

n be

rece

ived

with

CC

D o

rC

MO

S im

age

sens

or.

3. C

omm

unic

atio

n us

ing

Imag

e Se

nsor

s


- 3 -

Kei

o U

nive

rsity

SD

M10

LED

ligh

t is m

odul

ated

lens

time

Rec

eive

d da

ta:1

Rec

eive

d da

ta:0

Rec

eive

d da

ta:1

Prin

cipl

esof

com

mun

icat

ion

usin

g im

age

sens

or

Cam

era

(rec

eive

r) c

ontin

uous

ly ta

kes i

mag

es o

f a sc

ene

with

an

LE

D li

ght a

nd a

rec

eive

r de

tect

s the

opt

ical

inte

nsity

at a

pix

el

whe

re th

e L

ED

ligh

t is f

ocus

ed o

n.

reco

rded

imag

es b

y an

imag

e se

nsor

Kei

o U

nive

rsity

SD

M11

Purp

ose

of c

omm

unic

atio

n w

ith im

age

sens

or

Purp

ose

1:

Sim

ulta

neou

s dis

play

of c

onte

nts a

nd d

irec

tion

of

visi

ble

light

sour

ce

Purp

ose

2:A

ccur

ate

posi

tion

dete

ctio

n of

a tr

ansm

itter

or

a re

ceiv

er

Kei

o U

nive

rsity

SD

M12

Exa

mpl

es:

ID-C

AM

by

Sony

Kih

ara

Lab

orat

ory

Lig

htho

use

visi

ble

light

com

mun

icat

ion

by V

isib

le L

ight

C

omm

unic

atio

ns C

onso

rtiu

m (V

LC

C)

Purp

ose1

: Sim

ulta

neou

s dis

play

of c

onte

nts a

nddi

rect

ion

of v

isib

le li

ght s

ourc

e

Kei

o U

nive

rsity

SD

M13

Imag

e se

nsor

Com

putin

g ar

ea

Act

ual s

cene

Rep

rodu

ced

imag

eon

a c

ell p

hone

dis

play

Exa

mpl

e of

serv

ices

•Adv

ertis

emen

ts/D

isco

unts

•Nav

igat

ion/

Lan

dmar

ks

ID-C

AM

by

Son

y K

ihar

a L

abor

ator

y

“ID

CA

M: a

smar

t cam

era

for

scen

e ca

ptur

ing

and

ID r

ecog

nitio

n”, M

atsu

shita

, N. ,

Hih

ara,

D.,

Ush

iro,

T.,

Yos

him

ura,

S.,

R

ekim

oto,

J.,

Yam

amot

o, Y

., T

he S

econ

d IE

EE

and

AC

M In

tern

atio

nal S

ympo

sium

on

Mix

ed a

nd A

ugm

ente

d R

ealit

y, 7

-10

Oct

. 20

03, p

p. 2

27 -

236

Sony

Kih

ara

Lab

orat

ory

Kei

o U

nive

rsity

SD

M14

Lig

htho

use

visi

ble

light

com

mun

icat

ion

by V

isib

le L

ight

Com

mun

icat

ions

Con

sort

ium

(VL

CC

)

Mar

itim

e Sa

fety

Age

ncy

Res

earc

h C

ente

r o

f the

Jap

an

Coa

st G

uard

req

uest

ed V

LC

C (V

isib

le L

ight

C

omm

unic

atio

ns C

onso

rtiu

m) t

o do

res

earc

h ab

out v

isib

le

light

com

mun

icat

ion

usin

g lig

htho

use

or b

uoy

light

s in

2008

, and

VL

CC

mem

ber

com

pani

es (C

asio

Com

pute

r C

o.,

Ltd

., N

EC

, and

Tos

hiba

) are

doi

ng e

xper

imen

ts.

We

succ

eede

din

the

expe

rim

ent o

f2km

com

mun

icat

ion

usin

g an

imag

e se

nsor

in S

epte

mbe

r, 2

008.

Kei

o U

nive

rsity

SD

M15

Cas

io C

ompu

ter

Co.

, Ltd

.

Lig

htho

use

visi

ble

light

com

mun

icat

ion

1300

kilo

bits

per

seco

nd27

80 fr

ames

of p

ictu

res p

er se

cond

Rec

eive

d im

age

and

data

LE

D li

ght i

n a

phot

ogra

ph im

age

LE

D li

ght f

or a

bu

oyL

ED

ligh

t at

tach

ed to

a b

uoy

The

pur

pose

of t

his p

roje

ct is

to d

evel

op a

new

tech

nolo

gy th

at

enab

les t

he v

isib

le li

ght c

omm

unic

atio

n us

ing

visi

ble

light

from

light

hous

es a

nd b

uoys

. An

imag

e se

nsor

of a

cam

era

on a

boa

t de

mod

ulat

es th

e in

com

ing

data

from

ligh

thou

ses a

nd b

uoys

and

di

spla

ys it

s con

tent

on

a di

spla

y m

onito

r.

Kei

o U

nive

rsity

SD

M16

Purp

ose

2:A

ccur

ate

posi

tion

dete

ctio

n of

a tr

ansm

itter

or

a re

ceiv

er

Obj

ects

can

be m

easu

red

by r

ecei

ving

and

det

ectin

g th

e di

rect

ion

of v

isib

le li

ghts

igna

l w

ith a

nim

age

sens

or.

The

err

or o

f pos

ition

for

a 10

0 m

eter

obj

ect s

uch

as a

bri

dge

in th

e ph

oto

is a

bout

5m

m.

Thr

ee d

imen

sion

al p

ositi

on m

easu

ring

syst

em u

sing

vis

ible

ligh

t com

mun

icat

ion

by K

eio

Uni

vers

ity, S

umito

mo

Mits

ui C

onst

ruct

ion

Co.

, Ltd

., an

d N

akag

awa

Lab

orat

ory

LE

D li

ghts

att

ache

d to

a

brid

ge u

nder

con

stru

ctio

n

Kei

o U

nive

rsity

SD

M17

4. A

pplic

atio

ns o

f Vis

ible

Lig

ht C

omm

unic

atio

n

4.1.

App

licat

ion

of v

isib

le li

ght c

omm

unic

atio

n to

ele

ctri

c lig

hts

4.1.

1. C

omm

unic

atio

n us

ing

spot

light

4.1.

2. G

loba

l loc

atio

n se

rvic

e th

at u

ses v

isib

le li

ght I

D sy

stem

4.1.

3. V

isib

le li

ght i

nfor

mat

ion

broa

dcas

t sys

tem

4.1.

4. F

low

pla

nnin

g su

rvey

syst

em fo

r a

stor

e4.

2. A

pplic

atio

n of

vis

ible

ligh

t com

mun

icat

ion

to IT

S (I

ntel

ligen

t Tra

nspo

rt S

yste

m)

4.3.

App

licat

ion

of v

isib

le li

ght c

omm

unic

atio

n to

am

usem

ent

Kei

o U

nive

rsity

SD

M18

4.1.

1.C

omm

unic

atio

n us

ing

spot

light

Spot

ligh

ts se

nds d

ata

such

as p

ictu

re in

form

atio

n in

a m

useu

m a

s sh

own

abov

e an

d a

rece

iver

gui

des g

uest

s by

voic

e so

und.

Hik

ari G

uide

by

Shim

izu

Cor

pora

tion,

Jap

an


- 4 -

Kei

o U

nive

rsity

SD

M19

Vis

ible

Lig

ht ID

Inte

rnet

4.1.

2. G

loba

l loc

atio

n se

rvic

e th

at u

ses v

isib

le li

ght I

D sy

stem

It a

cces

ses t

he In

tern

et b

y fir

st o

btai

ning

cod

e fr

om a

vis

ible

ligh

tso

urce

such

as L

ED

ligh

ts. I

tthe

n ac

cess

es th

e lo

catio

n se

rver

from

th

e ce

llula

r ph

one

in o

rder

to o

btai

n lo

catio

n-re

late

d in

form

atio

n.

Serv

er o

f loc

atio

n in

form

atio

n

Vis

ible

Lig

ht ID

Vis

ible

Lig

ht ID

Vis

ible

Lig

ht ID

Prot

otyp

e m

ade

by N

EC

and

Mat

sush

ita

Ele

ctri

c W

orks

, mem

bers

of V

LC

C

Kei

o U

nive

rsity

SD

M20

4.1.

3. V

isib

le li

ght i

nfor

mat

ion

broa

dcas

t sys

tem

Info

rmat

ion

of p

rodu

cts a

t a su

perm

arke

tis o

btai

ned

by a

vi

sibl

e lig

htre

ceiv

er th

at is

inst

alle

d in

a sh

oppi

ng c

art.

Prot

otyp

e m

ade

by N

EC

and

Mat

sush

ita E

lect

ric

Wor

ks, m

embe

rs o

f VL

CC

Kei

o U

nive

rsity

SD

M21

Vis

ible

ligh

t tra

nsm

itter

usi

ng L

ED

ligh

t

4.1.

4. F

low

pla

nnin

g su

rvey

syst

em fo

r a

stor

eV

isib

le li

ght I

D fr

om L

ED

ligh

ts is

rec

eive

d by

a r

ecei

ver

atta

ched

at t

he b

otto

m o

f a sh

oppi

ng c

art.

LE

D li

ght t

rans

mitt

ers

are

inst

alle

d in

eac

h pa

ssag

e in

a su

perm

arke

t. Sh

oppe

rs’

mov

emen

t is a

naly

zed

base

d on

the

rece

ived

ID h

isto

ry.

Vis

ible

ligh

t rec

eive

r at

tach

ed a

t th

e bo

ttom

of a

shop

ping

car

t

Nak

agaw

a La

bora

tori

es, I

nc.

Nak

agaw

a La

bora

tori

es, I

nc.

Kei

o U

nive

rsity

SD

M22

Exa

mpl

e of

flow

pla

nnin

g of

shop

pers

in a

supe

rmar

ket

Ave

rage

spee

d

slow

fast

Thi

s sys

tem

is a

ble

to a

naly

ze h

ow m

any

shop

pers

pas

sed

in

each

pas

sage

and

how

fast

they

wal

ked.

4.1.

4. F

low

pla

nnin

g su

rvey

syst

em fo

r a

stor

e(c

ontin

ued)

The

thic

knes

s of l

ines

indi

cate

the

traf

fic a

mou

nt in

eac

h pa

ssag

e, a

ndth

e co

lor

indi

cate

s how

fast

shop

pers

wal

ked

on th

e av

erag

e.

Kei

o U

nive

rsity

SD

M23

4.2.

App

licat

ion

of v

isib

le li

ght c

omm

unic

atio

n to

ITS

(Int

ellig

ent T

rans

port

Sys

tem

)

Info

rmat

ion

broa

dcas

t sys

tem

usi

ng L

ED

traf

fic li

ghts

The

tran

smis

sion

met

hod

is b

ased

on

JEIT

A C

P-12

22 st

anda

rd.

The

Nip

pon

Sign

al C

o., L

td. a

nd V

LCC

, at

Jap

an S

hop

2006

Kei

o U

nive

rsity

SD

M24

Dem

onst

ratio

n of

mus

ic b

road

cast

syst

em

4.3.

App

licat

ion

of v

isib

le li

ght c

omm

unic

atio

n to

am

usem

ent

By

Sony

and

Agi

lent

at C

EA

TE

C 2

004,

Jap

anSo

undS

pot

Kei

o U

nive

rsity

SD

M25

Hon

orar

y C

hair

man

: Pr

ofes

sor

Mas

ao N

akag

awa,

Kei

o U

nive

rsity

, Jap

anC

hair

man

:Pr

ofes

sor

Shin

ichi

ro H

aruy

ama,

Kei

o U

nive

rsity

, Jap

anV

ice-

Cha

irm

an:

Prof

esso

r K

en S

akam

ura,

Uni

vers

ity o

f Tok

yo,

Japa

nPr

ofes

sor

Mits

uji M

atsu

mot

o, W

ased

a U

nive

rsity

, Jap

anD

ate

of F

ound

ing:

N

ovem

ber

2003

Purp

ose

of V

isib

le L

ight

Com

mun

icat

ions

Con

sort

ium

:V

LC

C w

as e

stab

lishe

d in

ord

er to

rea

lize

safe

, ubi

quito

uste

leco

mm

unic

atio

n sy

stem

usi

ng v

isib

le li

ght t

hrou

gh th

e ac

tiviti

es o

f mar

ket r

esea

rch,

pro

mot

ion,

and

stan

dard

izat

ion.

Hom

e Pa

ge: h

ttp:

//ww

w.v

lcc.

net/?

ml_

lang

=en

5. V

isib

le L

ight

Com

mun

icat

ions

Con

sort

ium

Kei

o U

nive

rsity

SD

M26

VL

CC

mem

ber

com

pani

es

Part

icip

atio

n fr

om v

ario

us in

dust

ries

such

as t

elec

omm

unic

atio

nsco

mpa

nies

, lig

htin

g co

mpa

nies

, LE

D m

aker

, ele

ctri

c po

wer

com

pani

es, e

lect

roni

cs m

aker

s, et

c.

The

Tok

yo E

lect

ric

Pow

er C

o., I

nc.

NE

C C

orpo

ratio

nT

oshi

ba C

orpo

ratio

n So

ny C

orpo

ratio

n T

he N

ippo

n Si

gnal

Co.

, Ltd

. T

oyod

a G

osei

Co.

, Ltd

. Sa

msu

ng E

lect

roni

cs C

o. L

td.

Pana

soni

c E

lect

ric

Wor

ks, L

td.

KD

DI R

&D

Lab

orat

orie

s N

TT

DoC

oMo,

Inc.

C

asio

Com

pute

r C

o., L

td.

JAPA

N R

UR

AL

INFO

RM

AT

ION

SY

SYT

EM

ASS

OC

IAT

ION

Info

rmat

ion

Syst

em

Res

earc

h In

stitu

teN

akag

awa

Lab

orat

orie

s, In

c.Fu

ji T

elev

isio

nO

i Ele

ctri

c C

o., L

td.

Sum

itom

o M

itsui

Con

stru

ctio

n C

o., L

td.

MoM

oAlli

ance

Co.

, Ltd

.T

amur

a C

orpo

ratio

nN

itto

Den

ko C

orpo

ratio

nSh

arp

Cor

pora

tion

Com

tech

200

0O

utst

andi

ng T

echn

olog

yJa

pan

Coa

st G

uard

-C

oast

Gua

rd R

esea

rch

Cen

ter

Ris

e C

o., L

td.

NH

K (J

apan

Bro

adca

stin

g C

orpo

ratio

n )

Lis

t of m

embe

rs:

Kei

o U

nive

rsity

SD

M27

It is

nec

essa

ry to

stan

dard

ize

visi

ble

light

com

mun

icat

ion

to c

omm

unic

ate

betw

een

equi

pmen

tsm

ade

by d

iffer

ent c

ompa

nies

.

VL

CC

is d

oing

vis

ible

ligh

t com

mun

icat

ion

stan

dard

izat

ion

activ

ities

in J

apan

.

VL

CC

mem

bers

are

als

o pa

rtic

ipat

ing

in I

EE

E 8

02.1

5 W

PAN

Tas

k G

roup

7 (T

G7)

Vis

ible

Lig

ht

Com

mun

icat

ion

(802

.15.

7)

6. S

tand

ardi

zatio

n of

Vis

ible

Lig

ht C

omm

unic

atio

n


- 5 -

Kei

o U

nive

rsity

SD

M28

Stan

dard

izat

ion

effo

rt o

f vis

ible

ligh

tcom

mun

icat

ion

at V

LC

C

2006

–20

07 :

Stan

dard

izat

ion

of v

isib

le li

ght c

omm

unic

atio

n w

ith J

EIT

A

2008

-pre

sent

Thr

ee g

roup

s (V

LC

C,I

rDA

, IC

SA) r

elat

ed to

free

-spa

ce o

ptic

alco

mm

unic

atio

n ge

t tog

ethe

r fo

r th

e jo

int d

evel

opm

ento

f vis

ible

lig

ht c

omm

unic

atio

n st

anda

rds

Kei

o U

nive

rsity

SD

M29

In 2

007,

VL

CC

pro

pose

d tw

o vi

sibl

e lig

ht

stan

dard

s to

JEIT

A (J

apan

Ele

ctro

nics

and

In

form

atio

n T

echn

olog

y In

dust

ries

Ass

ocia

tion)

an

d th

e tw

o pr

opos

als b

ecam

e JE

ITA

stan

dard

s in

Jun

e 20

07.

2006

–20

07 :

Stan

dard

izat

ion

of v

isib

le li

ght c

omm

unic

atio

n w

ith J

EIT

A

Kei

o U

nive

rsity

SD

M30

JEIT

ASt

anda

rds o

f vis

ible

ligh

tcom

mun

icat

ion

JEIT

A C

P-12

22:

Vis

ible

Lig

ht ID

Sys

tem

Stan

dard

JEIT

A C

P-12

21:

Vis

ible

Lig

ht C

omm

unic

atio

nSy

stem

Stan

dard

The

se st

anda

rds d

ocum

ents

can

be

purc

hase

d at

http

://w

ww

.jeita

.or.

jp/ja

pane

se/s

tand

ard/

list/l

ist.a

sp?c

atei

d=1&

subc

atei

d=50

2006

–20

07 :

Stan

dard

izat

ion

of v

isib

le li

ght c

omm

unic

atio

n w

ith J

EIT

A

Kei

o U

nive

rsity

SD

M31

JEIT

A C

P-12

21: V

isib

le L

ight

Com

mun

icat

ion

Syst

emSt

anda

rd

The

vis

ible

ligh

tcom

mun

icat

ion

syst

em st

anda

rd p

ropo

sed

atJE

ITA

is th

e m

ost b

asic

in th

e vi

sibl

e lig

htco

mm

unic

atio

n sy

stem

.T

he p

urpo

se o

f thi

s sta

ndar

d:(a

)Pre

sent

an

indi

cato

r m

inim

um in

ord

er to

pre

vent

the

inte

rfer

ence

bet

wee

n di

ffer

ent o

ptic

al c

omm

unic

atio

n eq

uipm

ents

.(b

)Def

ine

a m

inim

um n

eces

sary

req

uire

men

t in

vari

ous v

isib

le

light

com

mun

icat

ion

appl

icat

ions

.(c

) Pro

posa

l of t

he u

se o

fsub

carr

ier

met

hod

by m

odul

atin

gth

e lig

ht in

tens

ity b

y a

spec

ific

freq

uenc

y. B

y us

ing

a di

ffer

ent

subc

arri

er fr

eque

ncie

s, in

terf

eren

ce a

mon

g di

ffer

ent v

isib

le

light

app

licat

ions

can

be

avoi

ded.

Kei

o U

nive

rsity

SD

M32

JEIT

A C

P-12

22: J

EIT

A V

isib

le L

ight

ID S

yste

mSt

anda

rd

Subc

arri

er fr

eque

ncy:

28.

8kH

zT

rans

mis

sion

rat

e: 4

.8kb

psM

odul

atio

n: S

ubca

rrie

r 4-

PPM

The

mod

ulat

ion

was

cho

sen

not t

o ca

use

flick

erin

g.E

rror

con

trol

met

hod:

Err

or d

etec

tion

by C

RC

Tra

nsm

issi

on c

onte

nts:

ID

(fix

ed d

ata)

and

arb

itrar

y da

ta(n

on-f

ixed

)

Kei

o U

nive

rsity

SD

M33

2008

-pre

sent

: T

hree

gro

ups (

VL

CC

,IrD

A,I

CSA

) rel

ated

to a

n op

tical

wir

eles

s com

mun

icat

ion

for

the

join

t dev

elop

men

tof v

isib

le

light

com

mun

icat

ion

The

Infr

ared

Dat

a A

ssoc

iatio

n (I

rDA

) and

Vis

ible

Lig

ht

Com

mun

icat

ions

Con

sort

ium

(VL

CC

) beg

an

com

plim

enta

ry r

esea

rch

and

deve

lopm

ent e

ffor

t in

Sept

embe

r, 2

008.

The

Infr

ared

Dat

a A

ssoc

iatio

n (I

rDA

) w

hich

est

ablis

hes i

nfra

red

stan

dard

s tha

t pro

vide

co

nven

ient

wir

eles

s con

nect

ivity

and

the

Vis

ible

Lig

ht

Com

mun

icat

ions

Con

sort

ium

(VL

CC

) ann

ounc

ed a

co

oper

ativ

e ag

reem

ent b

etw

een

the

two

orga

niza

tions

.

VL

CC

and

ICSA

in J

apan

als

o ag

reed

to d

evel

op

stan

dard

s for

vis

ible

ligh

t LA

N.

Kei

o U

nive

rsity

SD

M34

Dat

a tr

ansm

issi

on fr

om L

ED

ligh

ts

Dat

a tr

ansm

issi

on

Con

cept

of v

isib

le li

ght c

omm

unic

atio

n en

visi

oned

by

VL

CC

and

IrD

AV

isib

le li

ght t

rans

mitt

er

LE

D li

ght

In M

arch

200

9, V

LC

C r

elea

sed

thei

r fir

st S

peci

ficat

ion

Stan

dard

ado

ptin

g an

d ex

pand

ing

the

IrD

A P

hysi

cal L

ayer

to th

e ne

wly

rel

ease

d V

LC

C S

peci

ficat

ion

Stan

dard

. By

adap

ting

IrD

A sp

ecifi

catio

n, th

e ex

istin

g Ir

DA

opt

ical

mod

ules

can

be

use

d fo

r V

LC

C d

ata

tran

smis

sion

with

slig

ht m

odifi

catio

n.

Kei

o U

nive

rsity

SD

M35

Dat

a co

mm

unic

atio

n

Con

cept

of v

isib

le li

ght c

omm

unic

atio

n en

visi

oned

by

VL

CC

and

ICSA

Vis

ible

ligh

t L

AN

ena

bles

sim

ulta

neou

s com

mun

icat

ion

for

mul

tiple

term

inal

s.

Opt

ical

vis

ible

ligh

t LA

N sy

stem

indo

or

Dat

a co

mm

unic

atio

n

Kei

o U

nive

rsity

SD

M36

Con

clus

ion

Vis

ible

ligh

t com

mun

icat

ion

may

bec

ome

com

mon

com

mun

icat

ion

tech

nolo

gy w

hen

LE

D

light

s bec

ome

wid

ely

avai

labl

e.


- 6 -

Choice of the Modulation for Gbps Wireless Infrared Systems

Mike Wolf∗, Liane Grobe and Jianhui Li

Ilmenau Univ. of Technology, CRL, PO Box 100565 D-98684 Ilmenau, Germany∗Corresponding author: Phone: +493677692619, Fax: +493677691195, E-Mail: [email protected]

Abstract – Optical Transmission can be affected both by white and f2 noise, where for large area PIN-

photodetectors the latter dominates at high transmission speeds. We extend the results with respect to the

required optical power for on-off keying and pulse-position modulation to f2 noise limited channels.

Introduction – As a part of the EU SeventhFramework R&D programme (FP7), the hOME Gi-gabit Access (OMEGA) project aims at bridging thegap between mobile broadband terminals and thewired backbone at home. To provide Gbps connec-tivity, three main technologies are considered. Oneof these is wireless infrared (IR).Whereas coherent and differential detection can beconsidered as serious alternatives for fibre optics orfor free space optics (like extremely focused outdooror inter satellite links), Gbps indoor communicationdemands for the simplest solution, namely direct de-tection. This implies intensity modulation.The most popular modulation schemes with respectto wireless IR are by far OOK (on-off keying) andPPM (pulse-position modulation) — including M -PPM with M orthogonal waveforms each of the sameduration, differential PPM or Multipulse PPM. Thepopularity of PPM has primarily two reasons. Firstly,compared to uncoded OOK, it may provide an ad-vantage with respect to the required average opticalpower. This is important from the power consump-tion and eye-safety point of view. Secondly, PPMhas a favourable spectral characteristic. The contin-uous part of the spectrum vanishes at and near DCwhich makes the signals robust to highpass filtering.A large highpass cut-on frequency corresponds to agood suppression of the harmonic interference causedby fluorescent lighting and ensures a short durationof the transient baseline wander at the beginning ofa received packet, cf. [1].Both advantages with respect to OOK are boughtby an increased bandwidth. This may be a seriousissue for Gbps transmission, since the device speedsare limited, especially if low capacitance, large areasilicon photodiodes are intended to be used. Further-more, the power advantage of PPM is usually provenfor white noise only, where it is well known that op-tical receivers are also affected by f2 noise, whosepower spectral density increases with f2. If f2 noisedominates, the photo current’s noise power σ2

n in-creases with the third power of the bandwidth. Thismeans that σ2

n increases by a factor of 1000 if thedata rate Rb is changed from 100 Mbps to 1 Gbps.In this paper, we compare the power requirementof various modulation schemes in white noise andin f2 noise. At 1 Gbps, our internal analysis showthat a PIN-photodiode based large area detector is

truly limited by f2 noise. However, we will notprove this statement here, and we will “only” con-sider non return-to-zero (NRZ) OOK, return-to-zero(RZ) OOK, and M -PPM. All these schemes exhibitonly two levels with respect to the instantaneous op-tical power making the laser diode driver much easierto build and much more power efficient than a lineardriver required for subcarrier modulations or pulseamplitude modulation.

Noise Modelling – We assume that the current atthe output of the photodiode consists of a signal partand a noise1 part n(t), where n(t) has a two sidedpower spectral density Φnn(f) = N0

2 + N2

2 f2. N0 and

N2 are constants. After lowpass filtering, σ2n is given

byσ2

n = N0I2Rb + N2I3R3b ,

where I2 and I3 are the corresponding Personick-integrals [2]. With G(f) being the lowpass filter’s

transfer function and f̃ = f/Rb, I2 and I3 are de-fined as

I2 =

∫∞

0

G(f̃)df̃ and I3 =

∫∞

0

f̃2G(f̃)df̃ .

Transmission in White Noise – As mentionedabove, we assume only two power levels and thus arectangular pulse shape. In the case of OOK witha duty cycle γ, 0 < γ ≤ 1, the pulse width isTrect = γTb. For PPM, Trect is assumed to be equal tothe chip interval, i.e., Trect = log2(M)Tb/M . Withrespect to the noise rejection filter, a pulse matchedfilter with the transfer function G(f) = sinc(fTrect)is supposed, which is associated with the Personickintegral I2 = 1

21γ

for OOK and I2 = 12

Mlog2(M) for

PPM.For signal space considerations, which are used toderive the required optical powers Preq here, it is ad-vantageously to normalise the transfer function G(f)of the “truth” filter by a factor

k =√

2RbI2.

This leads to a unit energy filter and those to a fixednoise variance of N0

2 at the sampling times. Since theM orthogonal base functions of M -PPM differ onlyin a delay, only one single pulse matched filter, whoseoutput is sampled with the chip rate, is required foroptimal PPM detection. Each receive vector consists

1input referred noise

1


- 7 -- 7 -

10 log10Preq

Preq,ookdB — white noise 10 log10

Preq

Preq,ookdB — f2 noise Tb

Trect

50% RZ-OOK -1.5 +1.5 225% RZ-OOK -3 +3 4

2-PPM, soft Dec. 0 +3 24-PPM, soft Dec. -3 0 28-PPM, soft Dec. -5.4 -1.13 2.66

Table 1: Normalised required power. Hard detected M -PPM requires additionally 1.5 dB.

of M subsequent samples of the filter output signal.Without noise, the samples can only take the values 0and RMP/k — just as the components of the signalvectors in an orthonormal signal space. R denotesthe photodiode responsivity, MP the optical peakpower. For soft detected PPM, the Euclidian distancedE between the orthogonal signal vectors is therefore

dE =

√2RMP

k= 2RP ·

√M log2(M)

2· 1

Rb

.

If hard decision is applied, the detection basicallytakes place on the chip level by means of a thresh-old which reduces dE by a factor

√2.

In the case of OOK with an optical peak power 2P/γ,we get

dE = 2RP/(kγ) = 2RP/√

γRb,

which shows that the required power depends on thepulse shape, too, cf. also [3]. The required power Preq

relative to NRZ-OOK can be easily estimated by

Preq

Preq,ook=

dE,ook

dE,

where dE,ook = 2RP/√

Rb.

Transmission in f2 noise – Here we assume thatN0I2Rb � N2I3R

3b . The results can be easily ob-

tained, if the transfer function of the filter is nownormalised by

k =√

2R3bI3,

which leads to the fixed noise variance of N2

2 atthe sampling times. A major issue with respectto the modulation scheme performance is the selec-tion of an appropriate noise rejection filter. A shorttime integrator can not be applied, since the enve-lope of sinc2(fTrect) drops only with f2. Fifth or-der Bessel lowpass filters are proven to perform well,since they provide a good tradeoff between linear dis-tortions and noise rejection. For NRZ-OOK, thatis for γ = 1, we set the 3 dB cut-off frequency ofthis filter to fg = Rb/2, which leads to the Person-ick integral I3 = I3,OOK = 0.084 and a (vertical)eye-opening penalty of 0.5 dB. Similarly, we choosefg = 1/(2Trect) for PPM and RZ-OOK, which yields

to I3 = I3,OOK ·(

Mlog2(M)

)3

and I3 = I3,OOK/γ3 for

PPM and RZ-OOK, respectively at a similar eye-opening penalty. By ignoring this penalty, we getfor NRZ-OOK

dE = dE,ook = 2RP/k =2RP√

2R3b

· 1√I3,OOK

and for soft detected PPM

dE =

√2RMP

k=

2RP√2R3

b

· 1√I3,OOK

·√

log32(M)

2M.

The relative required optical power for soft detectedPPM and for RZ-OOK is therefore

Preq

Preq,ook=

√2M

log32(M)

andPreq

Preq,ook=

1√γ

,

respectively.

Results – Table 1 shows the relative required op-tical power of RZ-OOK and M -PPM both for whitenoise and for f2 noise, where NRZ-OOK acts as thereference. Whereas RZ-OOK promises a gain in awhite noise regime, just the opposite occurs if f2

noise dominates. In f2 noise, even soft detected PPMcan only achieve a gain for M > 4. However, at1 Gbps, hard decision is more likely. In this case,M -PPM performs worse than OOK even for M=8.Clearly, with respect to OOK, the table does not showthe whole story, since OOK has to be combined witha line code [1] in a practical system. An 8B10B linecode will increase the required bandwidth by a fac-tor 10/8, which increases the required power by 0.5dB and 1.5 dB in white and f2 noise, respectively.However, if the limited speed of the components isalso taken into account, OOK can be considered as agood choice at 1 Gbps.

Acknowledgement – This work was carried outwithin the framework of the European integratedproject OMEGA.

References

[1] J. Li, M. Wolf, and M. Haardt, “Investigationof the Baseline Wander Effect on Gbps WirelessInfrared System Employing 8B10B Coding,” inInternational Conference on Telecommunications

2009, 2009, accepted paper.

[2] T. van Muoi, “Receiver Design for High-SpeedOptical-Fiber Systems,” Journal of Lightwave

Technology, vol. LT-2, pp. 243–267, June 1984.

[3] M. Wolf and M. Haardt, “Coding limits for shortrange wireless infrared transmission,” in IEEE

PIMRC ’05, (Berlin, Germany), Sept. 2005.

2


- 8 -- 8 -

Line-of-sight Infrared Wireless Path Loss Simulation in an Aircraft Cabin

Svilen Dimitrov1,2, Raed Mesleh1, Harald Haas1,3, Mario Cappitelli2, Michael Olbert2, Erhard Bassow4

1 Jacobs University Bremen, Campus Ring 1, 28759 Bremen, Germany, e-mail: s.dimitrov & [email protected] 2 EADS Deutschland GmbH, Nesspriel 1, 21129 Hamburg, Germany, e-mail: mario.cappitelli & [email protected] 3 Institute for Digital Communications, The University of Edinburgh, Edinburgh EH9 3JL, UK, e-mail: [email protected] 4 Airbus Deutschland GmbH, Lueneburger Schanze 30, 21614 Hamburg, Germany, e-mail: [email protected]

Abstract – In this paper, the line-of-sight (LOS) infrared wireless path loss in an aircraft cabin is estimated via a Monte Carlo ray-tracing (MCRT) simulation, performed on a geometric CAD model with defined transmitters and receivers.

Introduction – With the advent of non-coherent high-power light emitting diodes (LEDs) and highly sensitive photodiodes (PDs), optical wireless communication has become a viable candidate for medium range data transmission [1,2]. Compared to radio frequency (RF) communication, it offers an almost unlimited bandwidth, license-free operation, low-cost front ends and expected high data rates. Since there is no interference with RF-based technology, an optical wireless system can be deployed in an aircraft cabin to enhance inter-system communication and multimedia acquisition. One particular implementation regards the infrared optical communication [3]. To optimize the performance, the infrared wireless channel requires characterization. In a LOS case, the infrared signal suffers mainly the attenuation of the surrounding objects due to their reflection / absorption properties [4].

Fig. 1.: Generalized model of the setup, including three paths for path loss estimation, the transmitter and some of the receivers.

In this paper, a MCRT simulative approach is presented, which is used to compute the infrared signal power distribution in an aircraft cabin along chosen paths for a predefined light source. Considering the reflective properties of the environment, particular scenarios of LOS cases are studied, in which the azimuth (AZ) and elevation (EL) properties of a selected of-the-shelf LED are varied to find the appropriate orientation towards a static set of PDs positioned along the chosen paths. Simulation Modeling – The infrared irradiation simulation is performed with the software tool Specter by Integra Inc. A MCRT algorithm is utilized as a primary lighting simulation method. The following inputs are required: geometric model of the simulation setup, materials reflection characteristic, simulation color mode, and definition of key properties of light sources and observers. The geometric 3D model of the aircraft cabin is constructed with the CAD tool Rhinoceros 3D. A generalized model of the setup is presented in Fig. 1. The reflection characteristic of the different objects in

the geometry is defined through measured diffuse and specular reflection coefficients. The simulation focuses on monochromatic light at 940nm in a black and white simulation color mode. A single transmitter LED fixture is placed in the setup as shown on Fig. 1. The light source is modeled according to the specifications of the OSRAM high power infrared emitter "Golden Dragon" SFH 4231. Input parameters are the ball-formed LED's radiation characteristic, field of view (FOV) of 120o, and a total radiant flux of 500mW/nm. The irradiance of the setup volume is sampled by means of a planar irradiance observer tool, which is a collection of receiver points. A receiver in the scene is modeled according to the specifications of the OSRAM PD BP 104 F with a FOV of 120o and a radiant sensitive area of 4.84mm2. With the help of equidistantly separated irradiance observer planes, a three dimensional array of 100x100x100 receiver points is defined and sketched in Fig. 1. The limitations of the simulation are related to the MCRT simulation stopping criterion of stochastic


- 9 -

accuracy and the accuracy of definition of the simulation input parameters [4]. Simulation Results and Discussion – To estimate the LOS path loss in the cabin, three paths are chosen as shown in Fig. 1. The result in Fig. 2 shows that the path loss curve is almost linear in the log domain, after an alignment of the transmitter’s and the receivers’ FOVs is achieved. Thus, the generic RF path loss model [5] can be employed:

(1)

where is the path loss at reference distance d0, n is the path loss exponent, and d is the distance between transmitter and receiver. To avoid the FOV misalignment between the transmitter and the receivers along the paths, the receiver planes at distances greater than 1m away

from the transmitter are considered. Thus, in this shortened spatial region, the path loss curves along the shortened paths exhibit the linear behavior in log domain shown in Fig. 2. Hence, in the rest of the paper the path loss along the considered parts of the paths is described by its average value in dB and its path loss exponent. The results for the different LED orientation scenarios are presented in Fig. 3. A better orientation is defined as the one, which provides less average path loss along the shortened path. Along the CENTER path the best irradiation is given for AZ=0o

and EL=30o, because of the high reflection from the upper objects. Along the LEFT path the best orientation is AZ=-30o and EL=0o. The UP path is best irradiated for AZ=0o and EL=60o.

Fig. 2.: Path loss along paths LEFT and UP.

Fig. 3.: Average path loss and path loss exponent in the linear region along paths CENTER, LEFT and UP.

Conclusion – In this paper, an approach for the estimation of the infrared wireless LOS path loss is proposed. An MCRT irradiation simulation provides a comprehensive set of measurements of the signal power distribution in an aircraft cabin. The results are consistent with the RF LOS indoor propagation

model and can be used to calculate the link budget and optimize the radiation pattern of the transmitter.

References[1] T. Komine, M. Nakagawa, “Fundamental

Analysis for Visible-light Communication System Using LED Lights”, IEEE Transactions on Consumer Electronics, vol.50, no.1, Feb.2004

[2] M. Afgani, H. Haas, H. Elgala D. Knipp,“Visible Light Communication Using OFDM”, in Proc. of TridentCom2006, Barcelona, Spain, Mar. 2006

[3] F. Gfeller, U. Bapst, “Wireless In-House Data Communication via Diffuse Infrared Radiation“, in Proc. of IEEE, vol.67, no.11, Nov. 1979

[4] M. Cappitelli, M. Olbert, C. Mussmann and R. Greule, “GUM in Simulation und Messung”, in Proc. of Licht2008, Ilmenau, Germany, Sep.2008

[5] T.S. Rappaport, “Wireless Communications: Principle and Practice”, 2nd ed., 2002


- 10 -

Session 2 Tuesday May 19th, 2009 9:30 – 11:00 h

European project presentations (Part 1)

Session Chair: P-Y. FonjallazKista Photonic Research Centre, Kista (Sweden) A. Seeds University College London, London (United Kingdom)

Photonic in FP7-ICT - Workprogramme 2009-2010 F. Gillessen Deutsches Zentrum für Luft und Raumfahrt, DLR, ICT Strategies and EC Synergies, Köln (Germany)

Photonics21 – The European Technology Platform K. Flaig VDI Technologiezentrum GmbH, Düsseldorf (Germany)

ALPHA FP7-IP http://www.ict-alpha.eu/M. Popov ACREO AB, Kista (Sweden)

OMEGA FP7-IP http://www.ict-omega.eu/J.-P. Javaudin France Telecom Research & Development, Cesson-Sévigné (France)

FUTON FP7-IP http://www.ict-futon.eu/N.J. GomesUniversity of Kent, Canterbury (United Kingdom)

UROOF FP6-STReP http://www.ist-uroof.org/M. Ran Holon Institute of Technology, Holon (Israel)


- 11 -


European project presentations (Part 2)

Session Chair: A. Seeds University College London, London (United Kingdom) P-Y. FonjallazKista Photonic Research Centre, Kista (Sweden)

IPHOBAC FP6-IP http://www.ist-iphobac.org/A. Stöhr University of Duisburg-Essen, Duisburg (Germany)

GIBON FP6-STReP http://www.ist-gibon.eu/A. Scavennec Alcatel-Thalès III-VLab, Marcoussis (France)

HECTO FP6-STReP http://www.hecto.eu/U. Westergren Kungliga Tekniska Högskolan, Stockholm (Sweden)

ISIS FP6-NoE http://www.isis.minatec.grenoble-inp.fr/J-P. Vilcot Institut d'Electronique, de Microelectronique et de Nanotechnologie, Villeneuve d'Ascq (France)

BONE FP7-NoE http://www.ict-bone.eu/P. van Daele Ghent University, Ghent (Belgium)

EUROFOS FP7-NoE http://www.euro-fos.eu/E. Kehayas National Technical University of Athens, Athens (Greece)


- 12 -


Components for Wireless Photonics Applications

Session Chair: C. GonzalezAlcatel-Thalès III-VLab, Palaiseau (France)

Large Bandwidth Detectors and Receivers for Telecom and Wireless (invited)H.-G. Bach Fraunhofer Institute for Telecommunications, Berlin (Germany)

Generation of 54.8 GHz signal using a modelocked quantum dash semiconductor laser F. van Dijk, A. Enard, A. Akrout, G.-H. Duan, F. Lelarge Alcatel-Thalès III-VLab, Palaiseau (France)

Injection-locked Integrated Twin DBR Lasers for mm-wave Generation L. Ponnampalam1, C.C. Renaud1, I.F. Lealman2, L. Rivers2, P. Cannard2,M.J. Robertson2, D. Moodie2, F. van Dijk3, A. Enard3, F. Blache3, M. Goix3,F. Mallécot3, A.J. Seeds1

1 University College London, London (United Kingdom) 2 CIP Technologies, Ipswich (United Kingdom) 3 Alcatel-Thalès III-V Lab, Palaiseau (France)

Photodiodes and Reflective Electroabsorption Modulators for mm-wave and UWB Applications D.G. Moodie1, D.C. Rogers1, P.J. Cannard1, A. Borghesani1, C.W. Ford1,R. Firth1, R. Cronin1, M.J. Robertson1, D.W. Smithy, L. Ponnampalam2,C. Renaud2, A.J. Seeds2, M. Thakur3, T. Quinlan3, S. Walker3

1 CIP Technologies, Ipswich (United Kingdom) 2 University College London (United Kingdom) 3 University of Essex, Colchester(United Kingdom)

Demultiplexing Photoreceivers Comprising pin and pinTWA Frontends for 107 Gbit/s ETDM H.-G. Bach1, G.G. Mekonnen1, R. Kunkel1, C. Schubert1, D. Pech1, T. Rosin1,A. Konczykowska2, F. Jorge2, A. Scavennec2, M. Riet21 Fraunhofer Institute for Telecommunications, Berlin (Germany) 2 Alcatel Thalès III-V Lab, Marcoussis (France)


- 13 -


- 14 -

Large Bandwidth Detectors and Receivers for Telecom and Wireless

H.-G. Bach Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, Einsteinufer 37, D-10587 Berlin, Germany

Phone. ++49 30 31002-503, FAX ++49 30 31002-558, e-mail: [email protected]

Abstract: A family of ultrafast (160 Gbit/s) photodetectors based on waveguide-integrated evanescently coupled photodiodes, is described. The detectors are monolithically integrated with bias-Ts, coplanar transmission lines and MMI-couplers. Examples of miniaturized ultra-broadband-, travelling wave-, and narrowband photodetectors, operating up to -3 dB frequencies of 150 GHz are given. Novel self-biased detectors are demonstrated and integrated with antennas.

Introduction

Fast photodetectors with large saturation photocurrent are key components in high-bitrate fibre networks and photonic microwave applications incorporating optical pre-amplification. For high-speed operation, the photodetector has to be designed for low capacitance and small carrier transit times, cf. to chapter 1. The need for -3dB cut-off frequencies beyond 100 GHz leads to a further reduced size of the photodiode (PD) with the consequence of earlier limited output saturation current, but now with a 150 GHz response (chapter 2). A way to overcome the trade-off between speed and saturation photocurrent is to distribute symmetrically the optical signal to several photodiodes and combine their photocurrents by means of a transmission line [1, 2]. Now, due to the uniform optical power distribution, the unsaturated output photocurrent scales directly with the number of photodiodes. By embedding the discrete PDs within a transmission line, a travelling wave photodetector (TWPD) can be built. A 85 GHz periodic travelling wave photodetector with parallel optical feed [3], based on evanescently coupled low-capacitance waveguide-integrated micro-pin photodiodes, is presented in chapter 3. The generation of mm-wave signals by photonic technologies is advantageous for a variety of applications like fibre radio wireless communication systems, remotely controlled optically steered antenna systems, clock recovery in photonic-based telecommunication networks, and high-speed measurement systems. We report in chapter 4 on the design, fabrication and characterization of an efficient InP-based narrowband photodetector family, covering the frequency range from 38 to 100 GHz. The detectors comprise a waveguide-integrated photodiode and a resonating short stub circuit CPW-based transformer network [4, 7]. Chapter 5 presents novel self-biased photodetectors, suitable for remote mm-wave generation in EMI sensitive environments

or remote antenna driving [8]. Chapter 6 describes monolithic pin-antenna chips suitable for generation of THz radiation [6]. The self-biasing scheme [8] was recently combined with the pin-antenna integration, forming a self-biased pin-antenna chip, which radiates simply from the tip of a fiber into free space.

Chapter 1: Photodiodes up to 160 Gbit/s Side-illuminated photodetectors show an improved high-power behaviour, because the absorption is distributed laterally into a larger length of a thinner absorption layer in a controlled manner. The fabricated photodetector chips are based on InP and comprise an evanescently coupled mesa photodiode of 5x20 m2 size, a spot-size converter for increased fiber alignment tolerances, a biasing network and a 50 matching resistor. An optimized impedance line connecting the p-mesa to the electrical output line of the detector leads to an increase of the cut-off frequency up to 110 GHz [5].

The chip is assembled into a housing equipped with a 1 mm coaxial output connector and a fiber pigtail (Fig. 1, inset), exhibiting a responsivity of 0.55 A/W.

Fig. 1: Relative frequency response of the PD module (+2.3 dBm optical input power) measured with hp437B

(black) and PM3 (red) , inset: PD module photo.


- 15 -

Fig. 1 shows the calibrated frequency response of the photodetector module at 2 V reverse bias (Vbias).A -3 dB bandwidth of 100 GHz is obtained. The module converts excellently eyepattern from RZ PRBS modulation at 80/100/107 Gbit/s into the electrical domain. Even at 160 Gbit/s well opened

eyes are obtained, see Fig. 2, which now are converted to NRZ due to bandwidth limitations of the sampling head (Agilent 86118A) and the module.

Chapter 2: Micro-pin Photodiodes up to 150 GHz Micro-pin photodiodes are further miniaturized devices from a standard size 5x20 μm2 [3, 6] with a relatively broad absorber region to a much more reduced size of 4x7 μm2 with a reduced depth absorber region, without loosing too much of the responsivity (> 0,35 A/W), and to extend the

bandwidth far over 100 GHz. Therefore an optimization of the evanescent coupling scheme has been performed [6], c.f. Fig. 3. Here the contact and matching layer with the protrusion length L serves as low-resistance n-type contact layer, controls the optical intensity distribution in the device, and provides refractive index match between the waveguide and absorber.

Fig. 4: Measured frequency responses of lumped element (micro-pin) PDs and a TWPD with d = 110 μm (c.f. ch. 3). All devices stem from the same wafer with dabs =200 nm.

The measurements were done at a dc photocurrent of 2 mA and Vbias = -2 V.

In Fig. 4 the relative frequency responses of three novel PDs from the same wafer with dabs = 200 nm are shown. All devices contain an internal 50 matching resistor. In case of the photodetector with an active area of 5 x 20 μm2 a 3 dB bandwidth of 85 GHz is achieved, which is primarily limited due to the RC limit. The responsivity amounts to 0.52 A/W. By minimizing the active area to 4 x 7 μm2, the bandwidth is drastically increased to 150 GHz. The calculated 3 dB bandwidth due to pure carrier transit effects amounts to 150 GHz, this implies that this PD is mainly limited by the transit time. Due to the optimized matching layer, a notable responsivity of 0.35 A/W with a PDL of 0.22 dB can be maintained for this shortened photodiode.

Chapter 3: Travelling-Wave Photodetectors

In contrast to the lumped photodetectors described in the previous section, the travelling wave photodetector (TWPD) is a completely distributed structure. The inherent trade-off between maximized RC-limited and carrier-transit-time-limited bandwidth can be overcome. Our fabricated travelling wave photodetector with parallel feed comprises a spot-size converter for effective fibre-chip coupling and a 1 x 4 multimode interference (MMI) power splitter of which output waveguides

4 ps

0.5 V

Fig. 2 Detected eye pattern under 160 Gbit/s RZ excitation, Vbias = -2.5 V.

Fig. 3: Schematic cross-sectional view of the 1.55 μm wavelength photodiode . The light is injected from the

left into the spot-size converter (not shown) and couples evanescently from the semi-insulating

waveguide into the pin-mesa. The insets depict the optical 2D field intensity profiles in the singlemode waveguide (left), multimode matching layer (center)

and multimode PD mesa (right).


- 16 -

feed four miniaturized p-i-n waveguide-integrated photodiodes with active areas of 4 x 7 μm2 [3, 6].

Fig 5: Partial view of the fabricated TWPD chip. The input signal is fed from the left via the spot-size converter and rib

waveguide (not shown) into the MMI splitter. The PD distance is dpd = 90 μm.

The bandwidth of such travelling-wave detector amounts to ~85 GHz, which could be optimised further.

Fig. 6: Detected output power vs. the DC photocurrent (both devices without R50).

Fig. 6 gives the dependence of the detected RF output power delivered to a 50 load at 10 GHz on the DC photocurrent for both devices. In case of the TWPD a maximum electrical output power of +10.3 dBm has been achieved. Compared to the single PD this is an improvement of 10 dB available power. The 1 dB RF compression point amounts to IPD: 8 mA and 27 mA for the single PD and the TWPD, respectively.

Chapter 4: Resonant Narrowband Photodetectors

The electrical output signal of the photodiode is coupled to a resonant transformer network (L1, L2),which transforms the output load impedance of 50 Ohm to a considerably higher impedance Zeff

(200-250 Ohm) at the diode’s anode feeding point, see Fig. 7. Consequently the available RF power (~i2*Zeff) is increased, due to the constant current output of the detector. The transformer network feeds this increased power via L3 to the output load with only minor losses [7].

Fig. 7: Circuit scheme of the resonant narrowband photodetector.

Thus an effective gain in the range of ~ 7dB, compared to 50 Ohm internally terminated broadband detectors, is achieved at resonance.

The frequency response of a narrowband detector module for 80 Gbit/s clock recovery purposes and its reverse bias tuneable O/E conversion behaviour is presented in Fig. 8. By increasing the reverse bias from 0.5V to 4 V, the resonance peak is shifted from 76 GHz to 80 GHz. The resonance position for a fixed bias value was proven to be stable when the optical input power level was varied from -3 dBm to + 12 dBm, while the output return loss was better than 10 dB over the full range of optical input power levels. Typical Q values of the narrowband detectors range between 6-8.

Fig. 8: Opto-electronic frequency response of the module (inset) for 80 Gbit/s clock recovery purposes, showing slight resonance tuning with reverse bias of 0.5…4 V.

Chips with resonances at 40/60/80/85/90/100 GHz and responsivities up to 1 A/W are available, higher frequencies can be achieved further.


- 17 -

Chapter 5: Self-biased Photodetectors

The principle circuit scheme of a self-biased photodetector is depicted in Fig. 9.

Fig. 9. Basic circuit scheme of a self-biased photodetector, and packaged into a module (no! bias pins needed).

The incoming (modulated) light is split by an e.g. 1:4 MMI coupler into four waveguides. One waveguide feeds the small high-speed signal photodiode (5x20 m2), the other three waveguides supply light to three longer biasing photodiodes (10x40 m2), each of which produces a bias up to the bandgap of GaInAs. The series connection of these diodes cascades the bias voltages up to approx. 1.5 V, which now serves as the reverse biasing of the signal photodiode [8]. Fig. 10 shows 107 Gbit/s operation.

Figure 10: 107 Gbit/s eye patterns at +6 dBm (left) and +15 dBm (right) optical input power; opt. input pulses RZ:

2 ps, OTDM multiplexed, PRBS 27-1.

Chapter 6: Pin-Antenna Chips (opt. self-biased)

Micro-pin photodiodes have been integrated with log.-periodic antennas. Such chips radiate (sub) Thz radiation at IPD = 5 mA into free space, see Fig. 11.

Fig. 11: Heterodyne measured radiated RF output power vs. frequency of chip at top-right, top curve: via silicon lens at substrate side (inset), bottom curve: radiation at antenna

side; chip at lower right: novel self-biased pin-antenna chip.

The pin-antenna chips were further furnished by the self-biasing scheme, forming a self-biased pin-antenna chip, which radiates > W simply from the tip of a fiber into free space, see Fig. 11 (lower right).

Conclusion A family of broadband (150 GHz) photodetectors based on waveguide-integrated evanescently coupled photodiodes, is described. The detectors are monolithically integrated with bias-Ts, coplanar transmission lines and MMI-couplers. Examples of detectors, capable for 160 Gbit/s RZ conversion, miniaturized ultra-broadband-, travelling wave detectors, operating up to -3 dB frequencies of 150 GHz and narrowband photodetectors, comprising resonant output networks for increased responsivity up to 1 A/W, are given. Novel self-biased detectors are demonstrated at 107 Gbit/s. Pin-antenna chips for THz wave generation were shown and further combined with a self-biasing power cascade for free space radiation just from the fiber tip.

Acknowledgements The contributing work from G.G. Mekonnen, R. Kunkel, D. Schmidt, W. Ebert, A. Seeger, T. Gärtner, D. Pech, C. Sakkas, and R. Zhang at the FhI HHI is gratefully acknowledged. This work was financed partly by the MultiTeraNet program of the German Federal Ministry of Education and Research and supported by the GIBON and HECTO projects of the EC. References[1] C. L. Goldsmith, G. A. Magel, R. J. Baca, "Principles and Performance of Travelling-Wave Photodetector Arrays," IEEE Trans. Microwave Theory Tech., vol. 45, no. 8, August 1997, pp. 1342-1349. [2] S. Murthy, M. C.Wu, D. Sivco and A. Y. Cho, "Parallel feed traveling wave distributed pin photodetectors with integrated MMI couplers," Elect. Lett., vol. 38, no.2, 17th Jan. 2002, pages: 78-80. [3] A. Beling et al.: “High-Speed Miniaturized Photodiodes and Parallel-fed Traveling Wave Photodetector based on InP”, IEEE J. of Selected Topics in Quantum Electronics, special issue on High-Speed Photonic Integration, Vol. 13, No. 1 (2007), pp. 15-21. [4] H. Ito et al.: “High-Power photonic millimetre wave generation at 100 GHz using matching-circuit-integrated uni-traveling-carrier photodiodes” IEE Proc.-Optoelectron., vol. 150, no. 2, pp. 138-142, April 2000. [5] H.-G. Bach et al, IEEE JSTQE, Vol. 10, no.4, pp. 668-672, 2004. [6] H.-G. Bach, Ultrafast Waveguide-integrated pin-Photodiodes and Photonic Mixers from GHz to THz Range (invited), ECOC 2007, Sept. 16.-20, 2007, Berlin, Germany, 05.5.1_1569030753. [7] H.-G. Bach, Ultrafast Efficient Photodiodes exceeding 100 GHz Bandwidth (invited), Proc. of 19th Intern. Conf. on InP and Related Materials (IPRM 2007), May 14-18, 2007, Matsue, Japan, paper TuB3-1, pp. 71-76. [8] H.-G. Bach, High-Speed Photodetectors: Self-Biasing and High-Power Behaviour (invited), 214th Meeting of the Electro-chemical Society (ECS) 2008, October 12-17, 2008, Honolulu, Hawaii, USA, Abs. E5 2030, and in ECST 2009, Vol. 16, HI-E5.

40 50 60 70 80 90 100 110 120-60

-55

-50

-45

-40

-35

-30

-25

-20

-15

-10

Pel(d

Bm)

Freq.(GHz)

P-Log-2 TWPD2#0655 antenna-side P-Log-2 TWPD2#0655 Si-lensed substrate-side


- 18 -

Generation of 54.8 GHz signal using a mode-locked quantum dash semiconductor laser

Frédéric van Dijk*, Alain Enard, Akram Akrout, Guang-Hua Duan, François Lelarge Alcatel-Thalès III-V Lab, joint lab of « Bell labs » and « Thales Research & Technology », 1, avenue Augustin

Fresnel, 91767 Palaiseau, France

* Corresponding author: Phone: +33 1 69 41 57 35, Fax: +33 1 69 41 57 38, E-Mail: [email protected]

Abstract – In this paper we present how a 1.55µm quantum dash mode-locked laser can be used in order to generate a 54.8 GHz modulated optical signal with a phase noise compatible with wireless signal transmission.

Introduction –Mode-locked laser sources are very attractive solutions for various applications such as, pulse generation, clock extraction from digital data and optical microwave signal generation and processing [1]. Previously published work showed how FP QDs mode-locked lasers could be used to get low phase noise oscillators at 39.8 GHz [2]. In this paper, we will report on the generation of 54.8 GHzsignal at 1.55 µm using a QDs laser with dispersioncompensation using standard single-mode fibre.

I. FABRY-PEROT QUANTUM DASH LASER

DEVICE PRESENTATION AND PASSIVE MODE-LOCKING

CHARACTERISTICS

The studied semiconductor lasers are made of a buried ridge structure, and contain an active layerbased on QDs on InP substrate. The vertical structure was described in a previously published work [2]. Both facets are cleaved, forming a 774 µm-long FP cavity. The QDs-FP Laser was mounted on an AlN carrier with a GSG coplanar guide in for biasing and direct modulation.

-70-65-60

-55-50

-45-40

-35

-30-25

-20

54.8162 54.8164 54.8166 54.8168 54.817

Frequency (GHz)

Pho

tode

tect

ed e

lect

rica

l pow

er (d

Bm

) -

RBW = 3 kHzspan = 1 MHz

Figure 1: Self pulsation electrical spectrum at 370 mA

Passive mode-locking with these devices has been obtained without the use of any specific saturable absorber. Figure 1 shows an example of the beating

spectrum observed at a DC biased current of 370 mA,with the resolution bandwidth of the electrical spectrum analyser (ESA) set to 3 kHz. One can observe a self pulsation frequency close to 54.8 GHz. A nearly lorentzian lineshape is obtained, exhibiting a -3 dB linewidth narrower than 18 kHz. A lorentzian fit is shown on the figure. The extremely narrow linewidth for QDs lasers is believed to be a consequence of reduced spontaneous emission rate coupled to the lasing mode, and sufficient four wave mixing in these QDs structures [3].

II. MODE-LOCKED LASER DISPERSION

COMPENSATION

The electrical power that is obtained from a mode-locked laser source is obtained thanks to the beating of the couples of optical modes during the photodetection process. The highest photodetected RF power will be obtained if the relative phase difference between the adjacent modes is the same. There are two main contributions to the relative phase: the dispersion in the laser itself and the dispersion associated with the optical fibre used for the transport of the optical signal. Optimizing the transmission efficiency will require to minimize these two contributions. It has been shown that it could be done using standard single-mode fibre and enabled to getsub-picosecond pulses [4], [5], [6].

We have measured the electrical power after photodetection from the mode-locked laser. The laser was biased at 370 mA. At this bias level the coupled optical power was 11 mW. Detection was made using a 50 GHz photodiode from U²T. The electrical power at 58.4 GHz was measured using an Agilent E4448A electrical spectrum analyzer (ESA) coupled to a V band (50-75 GHz) H-P 11974V preselected harmonic mixer. The power was measured after propagation trough different lengths of standard single mode fibre (diamonds in Figure 2). A correction was applied toremove the contribution of the coupling losses associated to the numerous sections that had to be used by comparison of the DC photocurrents. Withoutcorrection, an electrical power of -6.77 dBm was measured for a length of fibre of 50 m and -15 dBm for a length of fibre of 2370 m. The signal has also


- 19 -

been measured using an auto-correlator and had a FWHM of 722 fs.

-50

-45

-40

-35

-30

-25

-20

-15

-10

-5

0

0 1000 2000 3000 4000 5000

fibre length (m)

corr

ecte

d el

ectr

ical

pow

er a

fter

ph

otod

etec

tion

(dB

m)

modelmeasurements

Figure 2: corrected photodetected electrical power as a function of fibre length

From these measurements it can be observed that an improvement of the electrical power of more the 15 dB compared to the laser alone can be obtained just by using 65 m of standard single mode optical fibre.

III. CONCLUSIONS

A study of the dispersion effects on transmission efficiency of the signal of a 54.8 GHz mode-locked laser has been performed. From this study it was found that a significant part of dispersion of the laser could be minimized using a standard single-mode fibre. For the optimum fibre length, auto-correlator traces with a FWHM of 722fs were obtained. The different optimum fibre lengths for the signal transmission were identified. These lasers have been successfully used for 60 GHz wireless signal transmission with data rates of 3 Gb/s when modulating the data by directly modulating the laser and more than 10 GB/s using an external modulator [7].

Acknowledgement - This work was carried out within in the framework of the European integrated project IPHOBAC

References[1] J. Renaudier, B. Lavigne, F. Lelarge, M.

Jourdran, B. Dagens, O. Legouezigou, P. Gallion, G.-H. Duan, “Standard-compliant jitter transfer function of all-optical clock recovery at 40 GHz based on a quantum-dot self-pulsating semiconductor laser”, IEEE Photonics Technology Letters, vol. 18, Issue 11, June 2006 Page(s):1249 - 1251Quantum Elec

[2] F. van Dijk, A. Enard, X.Buet, F. Lelarge, G.-H. Duan; “Quantum Dash mode-locked laser for millimeter-wave Coupled Opto-Electronic Oscillator”, IEEE International Topical Meeting on Microwave Photonics, 3-5 Oct. 2007 Page(s):66 – 69

[3] F. Lelarge, B. Dagens, J. Renaudier, R. Brenot, A. Accard, F. van Dijk, D. Make, O. Le Gouezigou, J.-G. Provost, F. Poingt, J. Landreau, O. Drisse, E. Derouin, B. Rousseau, F. Pommereau, G.-H. Duan, “Recent Advances on InAs/InP Quantum Dash Based Semiconductor Lasers and Optical Amplifiers Operating at 1.55 µm”, Invited paper, IEEE Journal of Selected Topics in Quantum Electronics, vol. 13, Issue 1, Jan.-feb. 2007 Page(s):111 – 124

[4] Arahira, S.; Kutsuzawa, S.; Matsui, Y.; Kunimatsu, D.; Ogawa, Y.; “Repetition-frequency multiplication of mode-locked pulses using fiber dispersion”, Journal of Lightwave Technology, Vol. 16, Iss. 3, March 1998 Page(s):405 – 410

[5] Arahira, S.; Kutsuzawa, S.; Matsui, Y.; Ogawa, Y.; “Higher order chirp compensation of femtosecond mode-locked semiconductor lasers using optical fibers with different group-velocity dispersions”, IEEE JSTQE Volume 2, Issue 3, Sept. 1996 Page(s):480 – 486

[6] Sato, K.; “Optical pulse generation using fabry-Pe/spl acute/rot lasers under continuous-wave operation”, IEEE JSTQE, Vol. 9, Issue 5, Sept.-Oct. 2003 Page(s):1288 - 1293

[7] A. Stoehr, A. Akrout, R. Buss, B. Charbonnier, F. van Dijk, A. Enard, S. Fedderwitz, D. Jaeger, M. Huchard, F. Lecoche, J. Marti, R. Sambaraju, A. Steffan, A. Umbach, and M. Weiss, 60 GHz Radio-over-Fiber Technologies for Broadband Wireless Services, Journal of Optical Networking, vol. 8, no. 5, 2009, (accepted)


- 20 -

Injection-locked Integrated Twin DBR Lasers for mm-wave Generation

L. Ponnampalam1*, C. C. Renaud1, I. F. Lealman2, L. Rivers2, P. Cannard2, M. J. Robertson2, D. Moodie2, F. van Dijk3, A. Enard3, F. Blache3, M. Goix3, F. Mallécot3 and A. J. Seeds1

1 Department of Electronic and Electrical Engineering, University College London, Torrington Place, London WC1E 7JE, United Kingdom

2 CIP Technologies, Adastral Park, Martlesham Heath, Ipswich, Suffolk, IP5 3RE, United Kingdom 3 Alcatel-Thalès III-V Lab, joint lab of « Bell labs » and « Thales Research & Technology », 1, avenue Augustin

Fresnel, 91767 Palaiseau, France

* Phone: +44 207 679 4464, Fax: +44 207 388 9325, E-Mail: [email protected] Abstract – A compact highly coherent mm-wave source is described, based on optical injection-locking of integrated twin-tunable lasers with a quantum-dash mode-locked laser. Output power was 0dBm at 98GHz and phase noise was -70dBc/Hz at 10kHz offset. Introduction – Many applications, such as high bit rate communications, security, radar and instrumentation require highly coherent mm-wave sources. Generation of mm-wave signals using photonic technologies [1] such as optical heterodyning has attracted remarkable interest in the recent years due to the availability of low loss fibres and cost effective optical devices. For compact low cost systems it is advantageous to integrate all components onto a single chip. In this paper, we describe the performance of a compact tunable mm-wave synthesizer that employs a quantum dash semiconductor mode-locked laser as an optical frequency comb generator (OFCG), injection locked integrated twin distributed Bragg reflector (DBR) lasers as highly selective coherent filters and a uni-travelling carrier photodiode (UTC-PD) as the mm-wave emitter. Experimental Arrangement and Results – The experimental arrangement is shown in Fig.1. The OFCG is a mode-locked Fabry-Perot semiconductor laser with a quantum dash gain medium. The comb lines are locked in phase and are separated by 24.5GHz over a 1.6THz span (Fig.2). The twin DBR lasers are buried heterostructure lasers fabricated 30μm apart on a single InP/InGaAsP chip, each stripe having 4 sections: front and rear grating sections of lengths 150μm and 450μm respectively, a 400μm long gain section and a 100μm long phase section. The waveguide of each stripe was of different width thus changing the centre frequency of the Bragg section to achieve the necessary wavelength offset between them. The outputs are combined by an MMI coupler which is further integrated with an angled tapered semiconductor optical amplifier (SOA) to boost the output power to 20 – 25mW with minimum facet reflections. The output from each of the stripes

had SMSR greater than 35dB (Fig.2) and a tuning range of 7 – 8nm with an offset of 6nm between them, giving an heterodyne tuning range of 0Hz to 1.8THz. A further advantage of the twin laser integration is that a 50C change in temperature causes the heterodyne frequency to change only by 2.5GHz, even though each of the stripes tunes by 60GHz. In order to achieve high spectral purity both the outputs from the twin laser must be locked in phase which is achieved by injection locking each of the outputs to two different lines of the OFCG. The combined output is amplified and then applied to high speed photodetector to generate the mm-wave signal. To achieve high power widely tunable mm-wave synthesizer high power high speed photodiodes are essential. This can be accomplished by using a UTC-PD where the light is absorbed in the p-region and hence the electrons are the only active carriers which can be made to travel at the overshoot velocity, which is an order of magnitude greater than the hole velocity. The response of UTC-PD is very fast and a 3dB bandwidth of 310GHz has been reported [2]. In addition, the higher electron velocity in the depletion region leads to much reduced space charge effect which enables the UTC-PD to have very high saturation power [3]. The UTC-PD used in this experiment had a 3dB bandwidth of 90GHz and a responsivity of 0.15A/W at a bias voltage of -4V. This was designed in a waveguide configuration in order to further improve the saturation power, and it was not saturated at an optical input power of 140mW. The mm-wave output was collected using a coplanar probe, and an external mixer was added for frequencies above 50GHz. Fig.3 shows the measured spectrum with a centre frequency of 98GHz with an input optical power of 11dBm (photocurrent of 2mA). The generated power increased to 0dBm at a photocurrent of 20mA (optical power of 140mW).


- 21 -

The measured linewidth of the generated mm-wave signal is less than 1kHz (limited by the resolution bandwidth of the spectrum analyzer) at any heterodyne frequency within the tuning range of the source. Many applications such as instrumentation require phase noise less than -90dBc/Hz at an offset frequency of 10kHz. However, the requirements can be less stringent for some areas such as high speed short reach indoor telecommunication where it is acceptable to have phase noise <-70dBc/Hz at an offset frequency of 100kHz. Fig.4 shows the measured phase noise spectra of the signal at 98GHz compared to that of the mode-locked laser. The OFCG is driven with a 15dBm signal at 24.5GHz to improve its phase noise performance. The resulting phase noise was -70dBc/Hz for offset frequencies less than 100kHz. The phase noise of the master OFCG was the limitation for the phase noise of the heterodyne signal at 98GHz. Incorporation of self injection loops can further improve the phase noise of the OFCG as reported in [4] which should reduce the phase noise level of the synthesized signal. Summary – We have demonstrated a compact tunable high spectral purity mm-wave synthesizer that has the potential of generating signals into the terahertz band. The generated power was measured to be up to 0dBm at 98GHz and the phase noise was

-72dBc/Hz at an offset frequency of 100kHz. This can be further improved by employing self injection loops in the OFCG. The tunability could be improved if integrated optical phase lock loops are used to replace the optical injection locking mechanism for the twin lasers.

Acknowledgement - This work is supported by the European Union within the framework of the Integrated project IPHOBAC.

References [1] A. J. Seeds and K. J. Williams, “Microwave

Photonics”, J. of Lightwave Technology, vol. 24, pp. 4628-4641, 2006.

[2] H. Ito, T. Furuta, S. Kodama and T. Ishibashi, “InP/InGaAs uni-traveling-carrier photodiode with 310GHz bandwidth”, Electron. Lett., vol. 36, pp. 1809-1810, 2000.

[3] C. C. Renaud, D. Moodie, M. Robertson and A. J. Seeds, “High Output Power at 110GHz with a Waveguide Uni-Travelling Carrier photodiode”, LEOS, ThM3, pp. 782-783, 2007

[4] F. van Dijk, A.Enard, X. Buet, F. Lelarge and G. Duan, “Phase noise reduction of a Quantum dash mode-locked laser in a millimetre-wave Coupled Opto-Electronic Oscillator”, J. of Lightwave Technology, vol. 26, no. 15, pp. 2789-2794, 2008.

Fig. 1.: Experimental Arrangement

-60

-50

-40

-30

-20

-10

0

1560 1565 1570 1575 1580Wavelength (nm)

Pow

er (d

Bm

)

OFCGTwin DBR Laser

Fig. 2.: Optical spectrum of OFCG and the twin DBR lasers

-80

-70

-60

-50

-40

-30

-20

-80000 -60000 -40000 -20000 0 20000 40000 60000 80000Offset Frequency from 98GHz (Hz)

Pow

er (d

Bm

)

Fig. 3.: Spectrum of the generated mm-wave signal at 98GHz

-120

-110

-100

-90

-80

-70

-60

-50

-40

1000 10000 100000 1000000 10000000 100000000Frequency offset (Hz)

Phas

e N

oise

(dBc

/Hz)

OFCG:No electrical injection

OFCG:With electrical injection

Beat signal at 98GHz

Fig. 4.: Phase noise spectrum of the mm-wave signal at 98GHz compared to that of OFCG.


- 22 -

Photodiodes and Reflective Electroabsorption Modulators for mm-wave and UWB Applications

D. G. Moodie1,*, D. C. Rogers1, P. J. Cannard1, A. Borghesani1, C. W. Ford1, R. Firth1, R. Cronin1, M. J. Robertson1, D. W. Smith1, L. Ponnampalam2, C. Renaud2, A. J. Seeds2, M. Thakur3,

T. Quinlan3, S. Walker3

1 CIP Technologies, B55, Adastral Park, Martlesham Heath, Ipswich, Suffolk, IP5 3RE, UK. 2 Dept. of Electronic and Electrical Engineering, University College London, Torrington Place, WC1E 7JE, UK.

3 School of Computer Science and Electronic Engineering, University of Essex, Colchester, C04 3SQ, UK. * Phone: +44 (0)1473 663223, Fax: +44 (0)1473 663295, E-Mail: [email protected]

Abstract – This paper describes recent progress made on photodiodes for mm-wave applications and EAMs designed for low cost sub 11 GHz antenna remoting applications.

Introduction – Photodiodes and electroabsorption modulators (EAMs) are key building blocks of many photonic systems. This paper will describe progress on high bandwidth photodiodes developed on the IPHOBAC project and lower bandwidth reflective EAMs developed on the UROOF project. Both devices were designed to have large optical modes at their front facets for low fibre-chip coupling losses.

Mode Expanded Photodiodes – Edge-coupled photodiodes with coplanar electrode configurations for operation at up to 110 GHz and some with integrated antenna electrode patterns for operation between 30 and 300 GHz were designed. Some earlier edge-coupled photodiodes, made without the mode expansion at the facet, had shown promising frequency response and power handling characteristics [1]. A similar active design was used in the mode expanded devices, although they were designed to have a partially p-doped absorber layer rather than an UTC structure. The mode expanded photodiodes were fabricated to a multi-level ridge waveguide design, shown schematically in Fig. 1 (a). A 1 mm long double stage taper was employed. Its purpose was to match the mode at the facet to that of a cleaved fibre or that of the passive silica waveguides used on the silicon motherboards at CIP [2], while delivering relatively tightly confined light into the active region to maximise the responsivity of a short low capacitance active region. Fabricated chips also incorporate precision cleave and scribe features, Fig. 1(b). The variation in measured responsivity with photodiode active section length is shown in Fig. 2. The polarisation sensitivity was typically 0.5 dB. A lens ended input fibre with a 10 degree FWHM far-field in intensity beam profile was used as it gave better coupling efficiencies than a cleaved fibre suggesting that the mode at the facet was less dilute than design simulations had predicted.

Chip capacitances comprised the parasitic capacitance of the coplanar electrode structure ~11 fF and the active capacitance. Chips with intended active region dimensions of 20 m x 2.8 m had an average active region capacitance measured at 1 MHz and -2 V bias of 33 fF. This would give an idealised RC limited bandwidth of ~96 GHz with no matching circuit when R=50 . The series resistance estimated from Smith chart traces was ~ 19 . Measured 3 dBe bandwidths of chips without matching circuits were ~80 GHz, Fig, 3, for reverse biases of 1.5 to 4 V.

(a)

(b)Fig. 1.: Mode expanded photodiode chip (a) schematic and (b) SEM micrograph.

First stage taper

Second stage taper

Active region


- 23 -

Two photodiode chips from a previous fabrication run with higher series resistances were packaged in a ‘W’ connectorised module, Fig. 4(a), and showed 3 dBe bandwidths of ~ 60 GHz.

00.05

0.10.15

0.20.25

0.30.35

0.40.45

0.5

0 5 10 15 20 25 30 35

Active length (um)

Res

pons

ivity

(A/W

)

Fig. 2. Responsivity versus active length of photodiodes with double stage tapers.

Fig. 3. Frequency response of photodiode chips measured using a coplanar probe.

(a) (b) Fig. 4. Photographs of (a) High speed photodiode module and (b) Reflective EAM module.

Mode Expanded Reflective Electroabsorption Modulators – For ultra-wide band radio (UWB) a maximum electrical frequency of only 10.6 GHz is required, however, a key requirement of the remote optoelectronic interface is that it needs to be low cost. Edge-coupled reflective EAMs have been developed for applications up to 10.6 GHz with very low insertion losses and polarization sensitivities. The 3.4 dB insertion loss was achieved by low coupling losses to a lens ended input/output fibre with a 10 degree FWHM far-field in intensity beam profile, Fig. 5. Using a package with only a single fibre, Fig. 4(b) helps reduce costs, while chips have been designed for compatibility with other low cost packaging techniques such as flip chip bonding and

passive optical alignment. Reflective EAMs fabricated on the UROOF project have been used as modulators and photodiodes in UWB radio over fibre systems demonstrators [3, 4].

0

0.1

0.2

0.3

0.4

0.5

-1 0 1 2 3 4

Reverse bias (V)

Tran

smiss

ion

0

0.2

0.4

0.6

0.8

1

Res

pons

ivity

(A/W

)

0

0.1

0.2

0.3

0.4

0.5

-1 0 1 2 3 4

Reverse bias (V)

Tran

smiss

ion

0

0.2

0.4

0.6

0.8

1

Res

pons

ivity

(A/W

)

(a)

-12-9-6-30369

12

0 2 4 6 8 10 12 14

Frequency (GHz)

Rel

ativ

e re

spon

se (d

Be)

(b)Fig. 5. Measured characteristics of packaged reflective EAM (a) DC characteristics measured at 1550 nm and 20 degrees C; solid lines - TM, dashed lines - TE, (b) Modulation bandwidth at -2 V bias.

Acknowledgement - This work was supported by the European projects IPHOBAC and UROOF.

References[1] C. C. Renaud, D.G. Moodie, D.C.Rogers, P.J.Cannard,

R.Firth, A.Borghesani, M. Robertson, A. J. Seeds, “High Output Power at 110 GHz with a Waveguide Uni-Travelling Carrier photodiode”, LEOS 2007, Paper ThM3.

[2] G. Maxwell et al, “Very low coupling loss, hybrid-integrated all-optical regenerator with passive assembly” ECOC, Post Deadline Paper, 2002.

[3] M. P. Thakur, T. Quinlan, S. Dudley, M. Toycan, C. Bock, S. D. Walker, D. W. Smith, A. Borghesani, D. G. Moodie, R. Llorente, M. Ran, Y. Ben-Ezra, “Bi-directional, 480Mbps, Ultra-Wideband, Radio-over-Fibre Transmission Using a 1310/1564nm Reflective Electro-absorption Transducer and Commercially-Available Components”, . ECOC’08, vol. 2, pp. 201-202, 2008.

[4] M. Thakur, T. Quinlan, S. B. A. Anas, D. Hunter, S. Walker, D. W. Smith, A. Borghesani, D. G. Moodie, “Triple-Format, UWB-WiFi-WiMax, Radio-over-Fibre Co-Existence Demonstration Featuring Low-Cost 1308/1564 nm VCSELs and a Reflective Electro-Absorption Transceiver”, OFC’09, Paper OTuJ2, 2009.


- 24 -

Demultiplexing Photoreceivers Comprising pin- and pinTWA Frontends for 107 Gbit/s ETDM

H.-G. Bach (1*), G.G. Mekonnen (1), R. Kunkel (1), C. Schubert (1), D. Pech (1), T. Rosin (1) A. Konczykowska (2), F. Jorge (2), A. Scavennec (2), and M. Riet (2)

1: Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut, Einsteinufer 37, D-10587 Berlin, GermanyPhone. ++49 30 31002-503, FAX ++49 30 31002-558, e-mail: [email protected]

2: Alcatel Thales III-V Lab(*), Route de Nozay, F-91461 Marcoussis Cedex Phone:++ 33 1 3077 6867, Fax:++33 1 3077 6786, e-mail: [email protected]

(*) joint lab: Bell Labs and Thales Research and Technology

Abstract: Demultiplexing photoreceivers, composed of either pin or pinTWA packaged frontends and subsequent packaged InP-HBT-based demultiplexers are reported for 107 Gbit/s operation, paving the way to ultra compact co-packaged pin/pinTWA-DEMUX receivers.

Introduction

For short range communications over a single-wavelength, serial 100 G OOK (on-off keying), applying ETDM transmitter and receivers, gains interest when components will be available at affordable prices. 107 Gbit/s 1:2 demultiplexing photoreceivers are key components [1, 2]. Within the European projects GIBON and HECTO InP-based components for 100 G serial transmission are developed; this work focuses on the receiver part.

Chapter 1: Waveguide-integrated pin photodiode and pinTWA-OEIC photoreceiver modules

Waveguide-integrated pin photodiodes allow ultra-high bandwidths exceeding 100 GHz, while exhibiting responsivities of > 0.7 A/W and polarization dependent losses (PDL) < 0.4 A/W [3], see Fig. 1(a).

Fig. 1(a) BCB-passivated Photodetector chip with 110 GHz bandwidth, (b) packaged PD module capable of 107 Gbit/s data rates, (c) pinTWA photoreceiver OEIC with 72 GHz bandwidth, (d) packaged pinTWA OEIC.

The final module for 107 Gbit/s data rate is shown in Fig. 1(b). An integrated TWA [4] behind the photodiode allows to save a 100 GHz interconnection, which otherwise is accomplished by tedious wire bonding or costly 1mm-connectors. Further on its on-chip-gain may spare an otherwise needed expensive EDFA in short range communications.

The HEMTs with 0.18 μm gate length, exhibit fT/fmax frequencies of typically 140/300 GHz. The pinTWA OEIC bandwidth amounts to 72 GHz. This bandwidth is completely preserved also in the module see Fig. 1(d).

Fig. 2(a) Eye pattern of photodetector module Fig. 1 (b) at 107 Gbit/s data rate (PRBS7), bias 2V, and +9 dBm opt. input power, 2(b) eye pattern of pinTWA photoreceiver module Fig. 1 (d) at 107 Gbit/s data rate (PRBS7) and +10 dBm opt. input power.

The eye patterns at 107 Gbit/s data are given in Fig. 2 (a) for the detector module (conversion gain 15 V/W into 50 ) and in Fig. 2(b) for the pinTWA receiver module, respectively. The conversion gain of the pinTWA receiver module was 20 V/W, the output eye is clearly opened at 107 Gbit/s, a first of all for a packaged pinTWA-OEIC.

Chapter 2: Demultiplexers based on InP HBTs for up to 107 Gbit/s data rates Demultiplexing is performed with a decision circuit developed at ATL, started with 2 μm InP DHBTtechnology (see Fig. 3 (left)), followed by 2nd

generation circuits from a novel 0.7μm HBT process.The circuit architecture is similar to the one presented in [5], but designed for higher operating bitrates (up to 100 Gb/s). Biasing employs two bias values at -2 V and -4 V, to be provided at the top of the chip. The data input on the left hand side is single-ended. The data input can provide the DC return path for the photodiode current, if needed.

(c)

(b)

(a) x: 5 ps/div y: 50 mV/div

(a)

(d)(c)

(b) x: 5 ps/div y: 80 mV/div


- 25 -

Fig. 3 (left) Photo of the decision circuit in 2 m HBT technology; (middle) decision circuit in 0.7 m HBT technology, (right) packaged DEMUX from (b).

The demultiplexing IC with ~500 mVpp output voltage swing comprises an input for the clock signal and two differential data outputs at the right side. The improved circuit in Fig. 3 (middle), packaged in Fig. 3 (right), assuring 107 Gbit/s demultiplexing operation can operate with unipolar input signals.

Chapter 3: The pin ½-DEMUX 1:2 107 Gbit/s demultiplexing photoreceiver

The pin-DEMUX receiver was combined from the modules in Fig. 1(b) and Fig. 3 (right) by direct dc-coupling using 1mm connectors.

Fig. 4 (left) Demultiplexed eye of pin-DEMUX at 53.5 Gbit/s and PRBS7; (middle) same as (left) but with PRBS31; (right) BER vs. OSNR.

Fig. 4 (left, middle) show for photocurrents of 4 mA the demultiplexed eye patterns with 53.5 Gbit/s data rate for PRBS7 and PRBS31, respectively. Output amplitudes of 500 mVpp are achieved. The BER measurement vs. the OSNR is given in Fig. 4 (right) with comparison to a reference 53.5 Gbit/s BER measurement using the same receiver setup. It delivers for PRBS7 20.5 dB at BER 10-3 (28 dB at BER 10-9), which compares favorably to results 21 dB (30 dB at BER 10-9) in [1], respectively.

Chapter 4: The pinTWA ½-DEMUX 1:2 107 Gbit/s demultiplexing photoreceiver The pinTWA module was directly dc-coupled to the input of the DEMUX module, thanks to the zero-bias dc-output of the TWA and its bipolar bias supply (-2V, +4V), forming the 107 Gbit/s pinTWA-DEMUX photoreceiver.

Fig. 5 depict the demultiplexed eyes at 53.5 Gbit/s and the BER vs. OSNR, respectively. The additional penalty of 7.7 dB is caused by the ripple of the TWA transfer characteristics, which will be improved in subsequent wafer runs to fully exploit the additional on-chip gain advancement.

Fig. 5 (left) Demultiplexed eye of pinTWA-DEMUX at 53.5 Gbit/s and PRBS7, Popt.: 9 dBm, IPD: 3 mA; (right) BER vs. OSNR.

Conclusion and outlook Packaged waveguide-integrated 110 GHz pin photodetector chips and monolithic pinTWA OEICs were hybridly combined with 1:2 HBT-based half-rate DFF demultiplexing modules as sensitive 107 Gbit/s receivers. Very clean demultiplexed output signals with ~500 mV amplitudes have been obtained in both cases. On-going work faces co-packaging of pin/pinTWA and DEMUX ICs, as was recently demonstrated with an 80 Gbit/s demultiplexing single-housing receiver [2], where DEMUX chips of Fig. 3(a) have been used.

Acknowledgements The authors gratefully acknowledge contributions from their colleagues D. Schmidt, W. Ebert, A. Seeger, and T. Gärtner at the FhI HHI. Further they acknowledge the cooperation with G. Veith and E. Lach at Alcatel-Lucent Germany and P. Berdaguer and S. Vuye at Alcatel-Thales III-V Lab France. This work was financed partly by the MultiTeraNet program of the German Federal Ministry of Education and Research and supported by the GIBON and HECTO projects of the EC.

References [1] J.H. Sinsky et al., “107-Gbit/s Opto-Electronic Receiver with Hybrid Integrated Photodetector and Demultiplexer”, OFC 2007, PDP-30.[2] G.G. Mekonnen et al., “InP Waveguide-integrated pin-Photodiode Hybrid Packaged with an HBT-DEMUX-Chip for Receiver Modules of 80-100 Gb/s Data Rates”, ECOC 2008, Vol. 3-33, We.1.C.5. [3] A. Beling, H.-G. Bach, G.G. Mekonnen, R. Kunkel, and D. Schmidt: “High-Speed Miniaturized Photodiodes and Parallel-fed Traveling Wave Photodetector based on InP”, IEEE Journal of Selected Topics in Quantum Electronics, special issue on High-Speed Photonic Integration, Vol. 13, No. 1 (2007), pp. 15-21. [4] Bach H.-G., A. Beling, G. G. Mekonnen, “Development Roadmap towards 100 GHz Photodetectors and Receivers and beyond”, 11th European Conference on Networks and Optical Communications (NOC 2006), Berlin, Germany, pp. 87-96. [5] A. Konczykowska et al., IEEE Trans on MTT, Vol. 53, No. 4, pp. 1228-1234.

16 18 20 22 24 26 28 30 32 3413121110987

6

5

4

3

-log(

BER

)

OSNR (dB)

PRBS 31

error-floor

PRBS 7

+2.7dB

53.5 Gb/s 107 Gb/s

16 18 20 22 24 26 28 30 32 34 3613121110987

6

5

4

3

-log(

BER

)

OSNR (dB)

53.5 Gb/s

+7.7dB

PRBS 7

107 Gb/s


- 26 -


Radio-over-Fibre techniques

Session Chair: C.G. SchäfferTechnical University Dresden, Dresden (Germany)

In-Building Distributed Antenna Systems using Radio over Fibre (invited)D. Wake University of Kent, Canterbury (United Kingdom)

Full-Duplex remodulation of optical microwave signals by Feed-Forward Current Injection in Reflective SOAs M. Presi1, A. Chiuchiarelli1, G. Contestabile1, L. Giorgi2, E. Ciaramella1

1 Scuola Superiore Sant’Anna, Pisa (Italy) 2 Ericsson, Pisa (Italy)

Wireless-Photonics-Wireless Interfaces Based on Resonant Tunneling Diode OptoelectronicOscillators J.M.L. Figueiredo1, C.N. Ironside2, B. Romeira1, T.S. Slight2, L. Wang2

E. Wasige2

1 Universidade do Algarve, Faro (Portugal) 2 University of Glasgow, Glasgow (United Kingdom)

A multi-hop UWB Radio over Polymer fibre system for 60-GHz hybrid NetworksC. Lethien, C. Loyez, J-P. Vilcot, N. Rolland Université des Sciences et Technologies de Lille, Villeneuve d’Ascq (France)

Optical Routing of Millimeter-Wave Signals with a New Optical Frequency Multiplication Scheme H.-D. Jung, C. Okonkwo, E. Tangdiongga, T.Koonen COBRA Research Institute, Eindhoven University of Technology, Eindhoven (The Netherlands)


- 27 -


- 28 -

In-Building Distributed Antenna Systemsusing Radio over Fibre

D. WakeBroadband and Wireless Communications Group, University of Kent, Canterbury, UK.

Phone: +44 1229 823244, E-Mail: [email protected]

Abstract Radio over fibre is being used increasingly for in-building distributed antenna systems to provide enhanced coverage and capacity of wireless networks. This paper looks at the architectures, requirements andcomponents used for these systems.

Introduction In-building wireless coverage and capacity is an important and growing market inenvironments such as office buildings, shopping malls and transport hubs. The most effective and efficient way of providing this coverage with good service quality is to place one or more radio base stations (BSs) at a central location inside the building and use a distributed antenna system (DAS) to distribute the wireless signals from the BSs to the various antenna locations around the building.Although DASs can be constructed using coaxial cable, the preferred option for larger installations is optical fibre cable using the radio over fibre (RoF)technique. This is because optical fibre provides a low loss, high frequency transmission medium. This paper looks at the architectures, requirements and components for these RoF-DAS systems.

Distributed Antenna Systems A typical RoF DAS architecture is illustrated in Fig. 1, which represents a large multi-storey building. The BSs are co-located in an equipment room, in this case in the basement of the building. A central hub places the radio signals from the BSs onto optical carriers for distribution over optical fibre to remote hubs, placed in services closets around the building. The remote hubs remove

the radio signals from the optical carriers for onward transmission over coaxial copper cables to antennas.The reverse sequence applies to the uplink direction. The optical fibre span can be a few km in large buildings, whereas the coaxial cable length is generally of the order of 20m or less. The DASarchitecture provides several important advantagesover alternative architectures in which the base stations are distributed around the building:

The remote hubs have small size, light weight and low power consumption.Co-location of the BSs allows capacity gains through increased efficiency and better utilization of radio resources. Centralized base stations are easier and consequently cheaper to operate, maintain and upgrade.

Radio over Fibre There are four main RoF typesused in current in-building wireless deployments. These are illustrated in Fig. 2.

RF transmission over single mode fibre (SMF)directly at the radio carrier frequency (usually in the range 800 2700MHz). RF transmission over multimode fibre (MMF).The ability to use existing MMF cabling is important for low installation cost. However, the bandwidth of MMF is not sufficiently high to support RF transmission over the required link lengths using traditional launch arrangements forthe optical input signal. However, successful transmission of RF signals over MMF can be achieved using restricted launch conditions [1]. IF transmission over MMF or SMF. The RF signal from the BS is downconverted to IF and transmitted to the remote hubs where it is upconverted back to RF. This allows MMF cables to be used, although at the expense of additional cost and complexity.

Fig. 1.: Typical architecture for RoF-DAS installation.


- 29 -

Digitised IF over MMF or SMF. This approach uses downconversion to IF and then digitises the signal for transmission over optical fibre. The analogue signal is then re-constructed and converted back to RF at the remote hub. This has the advantages of digital transmission (no impairments due to noise or distortion), but at the expense of high complexity.

Link Requirements RoF links suffer from noise and distortion and these impairments affect both downlink and uplink directions. Link noise affects the downlink because it adds to transmitted noise, which increases interference with other mobiles. It affects the uplink because it reduces the sensitivity of the base station receiver. Distortion in both downlink and uplink creates additional interference with other channels. A convenient metric that includes both noise and distortion is the spur-free dynamic range (SFDR). The SFDR requirement depends on the radio system and the environment, and can be determined from four inter-related radio system specifications (downlink power, downlink noise,uplink noise figure and uplink blocking level) for any particular deployment scenario. There are twocritical blocking mechanisms which influence the four radio system specifications mentioned above;

Blocking of the DAS resulting from approach of an uncoordinated handset (i.e. one being served by a BS outside the DAS) to a DAS antenna. Thehandset is likely to be transmitting at high power to try to overcome high attenuation losses. The

DAS uplink needs to be sufficiently linear to uplink overload.

Broadband noise transmitted by the DAS which blocks handsets near DAS antennas. This primarily affects uncoordinated handsets operating at low signal level.

Similar considerations apply in the reverse direction.For the case of a GSM system deployed in an officebuilding, it has been shown previously that the downlink SFDR requirement is 103dB.Hz2/3 and the uplink requirement is 95dB.Hz2/3 [2].

Link Performance All link types use direct modulation of laser diodes for low cost andperformance depends critically on the choice of laser. There are three main types of laser that areconsidered for this type of application;

VCSEL (vertical cavity surface emitting laser). These lasers usually operate at 850nm withmultiple transverse modes. Cost is very low(~$20) because they are produced in high volume for datacomms applications. VCSELs typically have SFDR values of around 90dB.Hz2/3.FP (Fabry-Perot laser). These lasers are edge-emitters and predominantly operate at longer wavelength (1310 or 1550nm windows) with multiple longitudinal modes. Cost is typically $50 - $100. Low cost uncooled FPs typically have SFDR values of around 110dB.Hz2/3.DFB (distributed feedback laser). These lasers are edge-emitters and predominantly operate at longer wavelength (1310 or 1550nm windows) with a single longitudinal mode. Cost ranges from $150 - >$500 depending on the specification. Low cost uncooled DFBs typically have SFDR values of around 110dB.Hz2/3.

Both FP and DFB lasers have adequate performance for RoF DAS systems.

Conclusions In-building wireless is an important and growing market for cellular network operators. The use of low cost optical components is leading to more widespread use of radio over fibre for theseapplications.

References[1] D. Wake et al -QAM radio transmission over

IEEE Int. Topical Mtg on Microwave Photonics,2001.

[2] D. Wake et al, communications IEEE Int. Topical Mtg on Microwave Photonics, 2004.

Fig. 2.: RoF link types used for DAS.

BTSSMF coax

CENTRAL HUB REMOTE HUBANTENNA UNIT

PDLRF RF

IF

BTSMMF coax


PDLRF RF

BTSSMF / MMF


PDLRF RF

UTP

IFIF

IFBTSSMF / MMF

CENTRAL HUBANTENNA UNIT

PDLRF RFIFIF

LO LO

LOLO

A/D D/A


- 30 -

Full-Duplex remodulation of optical microwave signals by Feed-Forward Current Injection in Reflective SOAs

M. Presi1,*, A. Chiuchiarelli1, G. Contestabile1, L. Giorgi2 and E. Ciaramella1�� #

�� #%�� ?��\� ?�^`{�|�|��{��{��}��?�^`{�|�|��{��{`��?��

�� !�� "##$� �� %�� &';�<=�">�?�$=�>J�?�$K� ��Q!J�YZ[��

�� #�� #�� }�� #�� #��\� � �\ � �� \�� \��} � �� \�� \ ��#� � �� \�� \�� \��\�#�� \ � �� \�� }� �� #� �� #�� \��\�� \�� \ � �#�� \�� \� ��\��\�� \ � �} � �� \�� \�� \� �#��\�� \ � �� ¡��¢�� ¢��

�� }��}�� }}�� ¡�� \�� \��\ ��}}�� \�� \�� \��\� �� \�� \�� #��\�� \�� \��\�� \��\�� \��\�� \��\�� \��\�� #� \ �� }��

��?��

}�� \ � ��\�� # � �� \�� \ � �� \��\�� \��\�� \ � �� } � �� ¡� � ��¢�� ¢�� #�� #�� |��# �� \ � �� £#�� \�¤\ �� ¥ � ¦ � ��{ ��\�� \ ��\��\��}��\�� £§��\� ��#��¨|�`|©��`|©��\�� \�� ª��\��\�� \��\�� \�� #�� |��# � � � �� \��?�� \��\�� \�� # � �� \�� \�� #�� \ � �� # � �� `� � �� \�� \��#��\��?�� \� �� |� � � � �|�� |� � � �� # �\�� \�� #�� #�� «��


- 31 -

�� ?� «��? � �� \��}}�� \�}}��`��ª��\�?��}�� ¢��

�� # � �� \�� |�� \�� \�� ?�� \�� \�� \�� \ � �� \ � ��\ ��\�� #�� #��\�� \ ��\�}}�� \��\�� \ ��\�}}�� \�� \ � �� # � �� `�� \��}�� ¢�� \ � �� \ � �\�� ?�� \��\��\�� \ ��\�}}�� \�� }�� \�� \ ��\�}}�� \� � �� \ � }}�� \��\ � � �� \ � ��¡��¢�� \�� \ � � �� \�� \�� \�� ?�� \�� \�� #�� \��\�}}�� ª � �\�� \�� \�� \��\�� \�� `��\�� \�`��

�� ?� �� \ � ��

� ��\��\��\�� \ � �� ¬��\�¤\ �� \ � �� # � �� }}�� ¬}}��£��}�� \�� \�� \��\��\ �� \ � �\ � �� ¬ � �� \�� #�®��©�� \�� \�� \�� ?�©��\��¢�� ¢�� ©�� \� ��¡�� \�� \��\�� \�� ? � � �� \ � �� \��\��©��\��

�� #��\�� \�� \��\�� \�� #�� }��}�� £��\�� \� � �� \��\�}}��\��\�� \�� \ � �� \ � ��¡� � � �� ¢�� ¢�� #��{¯|��\��\�� # � �� \�� # � �� \�� \ � �� \��\�� \ �� \# � �� #�� \�� #��

�� «��\��°��«��¡��°� ��

� � � � � ¡�� ¬�� #� � � � �� \ � }��}�� }��£}��||¨

�� #��\�� \�� ¬��# � �� |�� ¢��¡ � ��§��£��\�� }��£}��||¨


- 32 -

Wireless-Photonics-Wireless Interfaces Based on Resonant Tunneling Diode Optoelectronic Oscillators

J. M. L. Figueiredo1,*, C. N. Ironside2, B. Romeira1, T. S. Slight2, L. Wang2, E. Wasige2

1 Universidade do Algarve, CEOT, Departamento de Física, Campus de Gambelas, 8005-139 Faro, Portugal 2 University of Glasgow, Department of Electronics and Electrical Engineering, Glasgow G12 8LT, UK

* Phone: +351 289 800 987/905, Fax: +351 289 800 066, E-Mail: [email protected]

Abstract – Resonant tunneling diode voltage controlled oscillators integrated with laser diodes and photo-detectors can work as wireless to optical and optical to wireless converters, respectively, with high-potential to act as low-cost wireless/photonic interfaces in future picocellular networks.

Introduction – The rapid growth of wireless network users demanding higher coverage and higher data rates has motivated an increasing attention into hybrid wireless-optical communication systems to take advantage of the large bandwidth of optical communications networks [1]. Recent work on an Optoelectronic integrated circuit (OEIC) based on the integration of a resonant tunneling diode (RTD) oscillator with a laser diode (LD), the RTD-LD, has demonstrated it can act as an optoelectronic voltage controlled oscillator (OVCO) [2,3]. Resonant tunneling diodes are semiconductor nanostructures with strong nonlinear current-voltage (I-V)characteristic showing wide-bandwidth negative differential conductance (NDC) at room temperature [4]. Since circuits showing NDC can act as oscillators with built-in amplifiers, simply dc biasing the RTD-LD circuit in the NDC region produces relaxation oscillations at frequencies determined by both the circuit components and the dc voltage value, with the relaxation oscillations modulating the laser optical output [3]. Both monolithic and hybrid (separate RTD and LD chips) have been investigated. A hybrid RTD-LD circuit showing oscillation frequency as high as 2.2 GHz, tunable from 1.8 GHz to 2.2 GHz as the dc voltage cross the NDC region (a voltage range around 0.5 V) was already demonstrated [3]. This is a simply way of converting a RF signal into an optical subcarrier. Here we describe the work on RTD OVCOs that are capable to phase-lock to wireless signals with significant noise reduction, and produce broadband chaotic signals. We also present preliminary results on optical to electrical conversion using another OEIC consisting of a RTD VCO incorporating a photoconductive region. Synchronized and chaos wireless-photonic conversion – The RTD based wireless microwave-photonic circuit converter consists of a RTD-LD OVCO which incorporates a patch antenna for wireless detection, Fig. 1(a). In the presence of a wireless signal with frequency close to circuit free

running oscillation frequency, the RTD-LD phase-locks to broadcasted signals with the laser output being modulated by the broadcasted signal [5]. We have observed locking with considerable noise reduction for broadcasted powers as lower as -40 dBm. Figure 1(b) shows the RF spectra of the detected laser optical output at the second harmonic (1.8 GHz) of the circuit relaxation oscillations at 600MHz and when phase-locked to a 1.8 GHz injected wireless signal. When the 1.8 GHz wireless signal is present, stable frequency locking with significant noise reduction occurs, as shown in Fig. 1(b). Phase-locking was also observed for broadcasted signals with frequencies close to other harmonics of the circuit’s free running frequency. In all cases, the frequency locking range was adjusted using either the DC bias or the wireless power. The circuit operation mode applications includes phase modulated wireless to phase modulated optical sub-carrier conversion [5]. We propose that these characteristics could be employed in a digital wireless access network employing phase shift keying (PSK) modulation. Outside the above phase-locked regions and under appropriate wireless injection conditions the RTD nonlinear characteristics can induce the RTD-LD OVCO to operate on other nonlinear regimes including period-adding and intermittence routes to chaos, producing broadband (GHz-wide) chaotic current oscillations that modulate the laser diode optical output, generating optical sub-carriers with the same broadband chaotic features, Fig. 1(c). The numerical model of the RTD OVCO based on the circuit differential equations in the form of a Liénard’s system and the laser diode rate equations accurately predicts the observed circuit optoelectronic operating regimes including synchronization, quasi-periodicity and chaos generation [6]. Photonics-wireless conversion – A RTD oscillator containing a photo-conductive region can be used to extract a RF carrier from an optical signal, using the synchronization between a modulated optical signal and RTD oscillations [7]. Embedding a RTD within


- 33 -

an optical waveguide core (as described in [4]) we have implemented a photo-detector with a built-in amplifier that can act as a photonics-wireless converter, Fig. 2(a). Figure 2(b) presents the RF injection locking capture level using light from a tuneable laser diode modulated by a sinusoidal signal at 1 GHz, showing the RTD-PD responsitivity-gainincreases with the transition from RTD peak to valley by more than 15 dB. Figure 2(c) illustrates photo-detection capabilities at 5 GHz. We are currently investigating the synchronization between optical sub-carriers and RTD oscillations to take advantages of the RTD NDC to transfer and amplify the information bearing signal from the optical to the RF wireless domain. Circuit applications include optical sub-carrier to wireless conversion and optical control of microwave circuits. We believe the conversion characteristics can also be employed for phase shift keying (PSK) modulation. Conclusion – We have described the current activity on design, fabrication, characterization and modeling of a new class of wireless-photonic-wireless interfaces based on integration of RTD with optoelectronic devices such as laser diodes and photo-detectors to convert phase modulated wireless/optical signals into phase modulated optical sub-carriers/wireless signals. Phase-locking of a RTD-LD OEIC relaxation oscillator by direct injection of low power wireless signals has been demonstrated. The frequency locking range is tunable by adjusting the circuit natural frequency using the dc bias or the wireless power. We have also shown a RTD with a photoconductive region that provides a

simple way to convert an optical signal sub-carrier onto an RF signal. The RTD based OEICs applications can include single chip platforms with reduced size and low cost microwave/photonics interfaces for Radio-over-Fiber communication networks. Acknowledgement - This work was carried out with financial support of Fundação para a Ciência e Tecnologia, and Research Networks - Treaty of Windsor Programme – Acções Integradas Luso-Britânicas 2008/09 - U32, Portugal.

References[1] M. Sauer and A. Kobyakov, Radio over fiber for

picocellular network architectures, J. Lightw. Technol. 25, 3301 (2007).

[2] J. M. L. Figueiredo, et al., Self-oscillation and period adding from a resonant tunnelling diode – laser diode circuit, Electron. Lett. 44, 876 (2008).

[3] T. J. Slight, et al., A Liénard Oscillator Resonant Tunnelling-Laser Diode Hybrid Integrated Circuit: Model and Experiment, IEEE J. Quant. Electron. 44, 1158 (2008).

[4] J. M. L. Figueiredo, et al., Electric field switching in a resonant tunneling diode electroabsorption modulator,IEEE J. Quant Electron., 37, 1547 (2001).

[5] B. Romeira, et al., Wireless/Photonics Interfaces Based on Resonant Tunneling Diode Optoelectronic Oscillators, CLEO 2009, paper CTuT4.

[6] B. Romeira, et al., Synchronization and Chaos in a Laser Diode Driven by a Resonant Tunneling Diode,IET Optoelectronics 2, 211 (2008).

[7] T. M. Ramond, et al, Low-noise optical injection locking of a resonant tunneling diode to a stable optical frequency comb, Appl. Phys. Lett. 90, 171124 (2007)

(a) (b) (c) Fig. 1.: (a) Wireless microwave-photonics circuit converter. (b) Phase-locking at 1.8 GHz induced by a broadcast signal of -40 dBm at the RTD-LD receiving antenna. A phase noise reduction of ~25 dB at 10 kHz offset is observed (resolution and video bandwidths of 1 kHz). (c) Generation of chaotic signals induced by a 3 GHz injected wireless signal.

(a) (b) (c)Fig. 2.: (a) RTD-PD photonic-wireless converter. (b) RTD-PD I-V characteristic and RF injection locking capture level of an optical signal 1 mW@1550 nm modulated at 1 GHz. (c) Photo-detection of a 5 GHz modulated optical signal at the valley region.


- 34 -

A multi-hop UWB Radio over Polymer fibre system

for 60-GHz hybrid Networks

C. Lethien1, 2*, C. Loyez1, 2, J-P. Vilcot2 and N. Rolland1, 2

1 Institut de Recherche sur les Composants logiciels et matériels pour l'Information et la Communication Avancée

(IRCICA), Université des Sciences et Technologies de Lille, CNRS FR 3024, Parc Scientifique de la Haute Borne, 50 avenue Halley, 59650 Villeneuve d'Ascq – France

2 Institut d'Electronique, de Microélectronique et de Nanotechnologie (IEMN), Université des Sciences et

Technologies de Lille, CNRS UMR 8520 Avenue Poincaré - BP 60069 - 59652 Villeneuve d'Ascq cedex – France * Corresponding author: C. Loyez; �: +33 36253 1620; E-mail address: [email protected]

Abstract – A specific impulse optical fibre system has been achieved to enhance the connectivity of 60-GHz wireless networks. Properties of the multimode plastic optical fibre (POF) based on a perfluorinated material are exploited to transmit such signals..

I – Introduction This study proposes an efficient alternative to 60-

GHz ROF systems by reducing the architecture complexity: we have achieved a specific impulse fibre-radio system based on a simple RF architecture using optical COTS (commercial off the shelf) components. The RF transmission is assumed by a monolithically integrated RF system using sub-nanosecond pulses up-converted in the 60-GHz frequency band. The optical transmission is performed by modulating the intensity of a Laser: only the envelope of sub-nanosecond pulses is transmitted through the optical link by using the intensity modulation (IM) of a distributed feedbacklaser (DFB) operating at 1300nm (cutoff frequency: 3GHz) and the direct detection (DD) technique in a large area PIN photodiode. Most of conventional 60-GHz ROF systems use the glass single mode fibre (SMF) with frequency conversion and studies have already been led on IF-over-glass multimode fibre (MMF) for 60-GHz wireless networks [1]. In this study, we deliberately aim getting it compliant with a low cost application: it has been so achieved thanks to the simultaneous use of a perfluorinated graded index polymer optical fibre (PF GIPOF) and the IMDD technique. Furthermore, this system potentially enables a temporal multiplexing of the communications. In this paper, both RF and optical topologies are described in details. The main measurements of RF modules are depicted as well as the experimental results which concern the signal quality of a typical multi-hop transmission performed by this system.

II - RF Topology – The overall topology is depicted in the figure 1. The RF emitter and receiver include MMICs realized with different pHEMTs from Ommic foundry (ft up to 120 GHz). The main component of the emitter is a pulse generator. Thisgenerator consists of a high-speed NOR logic gate associated to a varactor diode which enables to delay the relative position of generated pulses. By this way, NRZ data are transmitted using a pulse position modulation (PPM) scheme. The 60-GHz carrier signal is generated by a 30-GHz VCO associated to afrequency doubler. Based on a pulse position modulation (PPM) technique, this system not requires local oscillators with strict phase-noise properties. The frequency up-conversion of the sub-nanosecond pulses is obtained by modulating the amplitude of the 60-GHz OL signal. This amplitude modulation (AM) is performed by a switch having a SPDT (single-poledouble-throw) topology and ensuring an isolation ofmore than 23 dB. The RF modulated signal is then amplified using a medium power amplifier (MPA) which provides an output power of 16 dBm. The receiver includes a low noise amplifier (LNA), a RFdetector and a correlator. The LNA has a noise figure equal to 6.5 dB and a power gain of 42 dB at 60-GHz. The RF detector performs the envelope detection of 60-GHz up-converted pulses. Next, the correlator mainly consists of a matched filter and a fast sampling and hold amplifier (SHA). Another pulses generator is used to trig this SHA. This second pulses generator is synchronised with the incoming signal by using a specific synchronisation technique which renders output data. No channel coding is applied within this first demonstrator:


- 35 -

II - Experimental Set-up and Results As depicted by the figure1, the NRZ data stream is provided by a data timing generator Tektronix DTG 5334. These data modulate the pulse position of a first 60-GHz transmitter. The pulse width is close to 300 ps and corresponds to a good trade-off between channel capacity and spectral band filling: the narrowness of the pulse width enables to decrease the probability of collisions between users and offers a sufficient spread spectrum to limit signal fading due to the multipath propagation. The radiated signal has a power of 16 dBm and is transmitted over a distance of 4 meters. The emitter and receiver antennas have a 3dB aperture beam-width equal to 60 degrees. The radiofrequency waves are circularly polarized and the cross-polarization rejection is equal to 23 dB. Thepropagation environment is a large room (6 m x 15 m) with metallic and wood furniture. The RF part ofthe electro-optic transceiver is only composed by aLNA and a detector enabling to recover the pulse envelope. Then, theses pulses modulate the intensity of a DFB operating at 1300nm. The modulated optical signal is transmitted through the fibre characterized above. The coupled optical power through this fibre is -4dBm. We choose to transmit the sub-nanosecond pulses through 100m of PF GIPOF due to the trade-off between fibre optical bandwidth and pulse spectrum. Nevertheless, this length is compatible with indoor applications [4]. A low cost PIN photodiode followed by a DC-6GHz amplifier delivers the sub-nanosecond pulses to theRF part of the optoelectronic transceiver. Recovered pulses modulate a second 60-GHz LO signal. The

modulated signal is amplified in the same manner asfor the first RF hop and is transmitted over a distance of 4 meters by using similar antennas. The 60-GHz receiver, as the first one, is only composed by a LNA and a detector which enables to recover the pulse envelope. The eye diagram of the received pulse stream can be observed in Fig. 2. The eye aperture shows the quality of the multi-hop 100 Mbps transmission. To our knowledge, frequency selectivefading is the main degradation factor of this system and prevent it from reaching performance such as inan additive white gaussian noise (AWGN) channel. By the way, a large amount of measurements is required to qualify statistically the system performance in similar configurations. Nevertheless, we give further information about the signal quality thanks to BER measurements: for data rates up to 200 Mbps, we obtained BER results reaching values greater than 10-6 for the configuration described previously.

Acknowledgement - This work was supported by the ERDF (European Regional Development Fund) and by the Nord-Pas-de-Calais Region (France).

References

[1] Loyez, C.; Lethien, C.; Kassi, R.; Vilcot, J.P.; Decoster, D.; Rolland, N.; Rolland, P.A, “Subcarrier radio signal transmission over multimode fibre for 60 GHz WLAN using a phase noise cancellation technique”, IEE Electronics Letters, Vol. 41, no. 2, pp.91 – 92, 20Jan. 2005

PIN photodiode

DC

60-GHz emitter

X 2

pulse generator (PPM)

SPDT switch (AM)

MPA

electo-optic transceiver

LNADFB

1300nm

RF detector

DC

PF GIPOF 62.5µm (100 m)

X 2

VCO

SPDT switch (AM)

MPASHA

SYNC

correlator

4 m

4 m RF

detector

LNA

60-GHz receiver

opto-electronic transceiver

data

NRZ

VCO

Fig. 1 - multi-hops opto-RF transmission

Fig. 2 - Eye diagram of received PPM signal

data NRZ


- 36 -

Optical Routing of Millimeter-Wave Signals with a New Optical Frequency Multiplication Scheme

Hyun-Do Jung,*, Chigo Okonkwo, Eduward Tangdiongga, and Ton Koonen COBRA Research Institute, Eindhoven University of Technology, 5600MB, the Netherlands

* Corresponding author: Phone: +31 40 247 5627, Fax: +31 40 245 5197, E-Mail: [email protected]

Abstract – We demonstrate a new OFM configuration using XGM in SOA for simultaneous all-optical generation and routing of mm-wave signals for in-building networks. After routing, EVM of 4.5% is achieved for 20MS/s, 64-QAM data at 39.6GHz.

1. Introduction –The millimeter-wave (mm-wave) frequency band (26~70GHz) has raised much interest in broadband wireless access applications as it offers a large bandwidth, avoids the spectral congestion occurring in the lower microwave bands, and reducesinterference by creating picocells, but the difficulties in millimeter-wave generation, transmission, and processing limit its wide usage. However, with the development of fiber-optic technologies, Radio-over-Fiber (RoF) techniques can offer valuable solutionsfor these problems [1],[2]. Moreover, the combination of mm-wave technology with Radio-over-Fiber (mm-RoF) has emerged as a key technology for in-building networks. By means of mm-RoF systems, picocells in a building can be extended to several rooms instead of having a single large wireless network, which covers the whole building and causes interference problems. Recent proposals have suggested that radio signals are broadcasted through the optical fiber link and the channel selection is done at each destination. Withbroadcast-and-select configurations, however, thereare concerns regarding high power consumption and security. In this paper, we demonstrate a new optical frequency multiplication (OFM) configuration [3] for in-building networks shown in Fig. 1. Key functionalities of the proposed architecture are simultaneous all-optical mm-wave generation and selective routing of the mm-wave signals

to individual rooms based on encoded header information [4]. This centralized optical routing can dynamically adjust the optical connectivity and RF frequency allocation. Hence the radio-cells can be dynamically adjusted in size and capacity, which improve the traffic handling capacities and networkoperational efficiency. As a proof of concept, we successfully demonstrated optical up-conversion of 3.6GHz radio signal carrying 20MSymbols/s 64-QAM data to 39.6GHz mm-wave frequency, and selective optical routing of mm-wave signals to thedestination. At the destination, the detected signal showed error vector magnitude (EVM) of 4.5% at 39.6GHz.

2. Operational principle of the proposed system –Figure 1 illustrates the in-building network scenario. An optical transparent gateway routes wired radio data signals from the central station (CS) to each room based on address information [4]. The proposedall-optical routing system for mm-wave signals consists of a tunable source, a phase modulator, a 3-dB coupler, an SOA, a Mach-Zehnder Interferometer (MZI), and an optical router based on an arrayed waveguide grating (AWG). The continuous wave (CW) optical signal (λCW) from the tunable source, which are selected based on the address informationextracted from the downstream optical signal,

AWG

Fig. 1 Concept and experimental setup (Radio signal from the central station: 20Msymbols/s, 64-QAM at 3.6GHz RF carrier)


- 37 -

are phase-modulated (PM) by the RF sweep signal (fSW) to generate optical harmonics (the OFM technique [3]). These PM optical signal (λCW) is injected into the SOA together with the intensity-modulated (IM) optical signal (λMOD) carrying the radio signal from the CS. By cross-gain modulation (XGM) in the SOA, the radio signal is duplicated onto the PM optical signal (λCW). In the MZI, PM-IM conversion allows the mm-wave carriers at the multiples of the RF sweep frequency (fSW) to appear at the optical wavelength (λCW) as illustrated in Fig. 1. Then, the converted optical signal (λCW) are routed by means of the AWG to the destination (room), where it is detected and the mm-wave radio signal is selected by a bandpass filter (BPF).

3. Experimental results and discussion –To generate mm-wave carrier signal, the CW optical signal (λCW) from the tunable source was phase-modulated with a 6GHz RF sweep signal (fSW). The CW wavelength (λCW) was selected by header processing based on the address information extracted from the modulated optical signal (λMOD) [4]. Figure 2 (a) shows the RF spectra of the multiple harmonics generated by the OFM technique. The proper adjustment of both the phase modulation index (β) and the center-wavelength (λCW) of the tunable source allows the amplitude of each harmonic signal to be tuned. In the experiment, the 6th order harmonic (36GHz) was optimized at 6.8 (β), 1550.735 nm (λCW). This PM optical signal was inserted into the SOA with the IM optical signal carrying the radio data signal (fRF = 3.6GHz) shown in Fig. 2 (b). Then, the radio signal was duplicated (wavelength-converted) on the PM optical signal andoptically up-converted along with the harmonics of fSW to fUP = n�fSW ± fRF (where n is the order of harmonics) by XGM of the SOA; Fig. 2 (c) depicts the radio signal up-converted to fUP = 39.6GHz (6th

harmonic of fSW). As shown in the figure, the SNR of the mm-wave signal is reduced by around 16dB and

there is an EVM penalty of 2.5%, compared to the input RF signal. In addition, a nonlinear skirt slope appears at the edge of the signal band. This degradation comes from the ASE noise of the SOA, the wavelength-conversion penalty, and the nonlinearity of the SOA gain profile. Nevertheless,the performance of the routed signal at the destination meets the EVM requirements for the wireless standards.

4. Conclusions –In this paper, we proposed a new configuration for all-optical generation and routing of mm-wave signals using the OFM technique for in-building networks. By using XGM in an SOA, we can optically up-convert radio signals at low frequency to mm-wave frequency region and convert radio signals to different wavelength signal at thesame time. In the experiment, we successfully demonstrated optical up-conversion from a 3.6GHz radio signal carrying 20MS/s 64-QAM data to a 39.6GHz mm-wave frequency and optical routing of the mm-wave signal to different destination. At thereceiver side, the routed signal showed the EVM performance of 4.5% for 20MS/s 64-QAM data at 39.6GHz.

Acknowledgement - This work was carried out within in the framework of the European integrated project ALPHA.

References[1] A.J. Seeds, “Microwave phtonics,” IEEE Trans.

Microw. Theory Tech., 50, 877-887 (2002) [2] L. Noel, D. Wake, D.G. Moodie, D.D. Marcenac, L.D.

Westbrook, and D. Nesset, “Novel techniques for high-capacity 60GHz fiber-radio transmission systems,” IEEE Trans. Microw. Theory Tech., 45, 1416-1423 (1997).

[3] A.M.J. Koonen and M. Garcia Larrode, “Radio-over-MMF techniques – Part II: microwave to millimeter –wave systems,” J. Lightw. Technol., 26, 2396-2408 (2008).

[4] Hyun-Do Jung and et. al., “All-Optical Routing Architecture of Radio Signals using Label Processing Technique for In-building Optical Network”, in Proc. ECOC 2008, Brussels, paper Tu.4.F.2.

Fig. 1 RF Spectra of (a) multiple harmonics generated with RF sweep frequency (fSW = 6GHz), (b) input 64-QAM signal (20MS/s) at 3.6GHz, (c) received 64-QAM signal (20MS/s) at 39.6GHz. (Inset : Constellation diagram)


- 38 -

Session 5 Wednesday May 20th, 2009 9:15 – 11:00 h

Home networks

Session Chair: N. GomesUniversity of Kent, Canterbury (United Kingdom)

Future Home Network Requirements (invited) B. Charbonnier1, H. Wessing2, B. Lannoo3, M. Popov4

1 France Telecom R&D, Lannion (France) 2 Technical University of Denmark, Lyngby (Denmark) 3 INTEC, Ghent University-IBBT, Gent (Belgium)4 ACREO AB, Kista (Sweden)

Comparison of two types of 60 GHz photonic millimeter-wave generation and distribution of a 3 Gb/s OFDM signal F. Lecoche1, B. Charbonnier1, E, Tanguy2, H. Li2, F. van Dijk3, A. Enard3

F. Blache3, M. Goix3, F. Mallecot31 Orange Labs, Lannion (France) 2 Université de Nantes (France) 3 Alcatel-Thalès III-V Lab, Palaiseau (France)

Bidirectional Multi-Standard RoMMF Transmission Using a Reflective Electro-Optic Transceiver I. Möllers, D. Jäger University of Duisburg-Essen, Duisburg (Germany)

High modal bandwidth glass multimode fibers used for the simultaneous transmission of 10GbE and band group 5 MB-OFDM Ultra-Wide Band signals C. Lethien, C. Loyez, J-P. Vilcot, P.-A. Rolland Université des Sciences et Technologies de Lille, Villeneuve d’Ascq (France)

Experimental implementation of real-time optical OFDM modems for optical access networks (invited) R.P. Giddings, X.Q. Jin, H.H. Lee, X.L. Yang, J.M. Tang Bangor University (United Kingdom)


- 39 -


- 40 -

Future Home Network Requirements

B. Charbonnier1,*, H. Wessing2, B. Lannoo3, M. Popov4

1 France Telecom Research and Development, 2 AV. P. Marzin, 22307 LANNION Cédex, France 2 Technical University of Denmark., DTU Fotonik, Bldg. 343, Oersteds Plads, 2800 Kgs. Lyngby, Denmark

3 Dep. of Information Technology (INTEC), Ghent University-IBBT, G. Crommenlaan, 8/201, 9050 Gent, Belgium 4 ACREO AB, SE 164 40 Kista, Stockholm, Sweden

* Corresponding author: Phone: +33 2 96052176, Fax: +33 2 96051723, Mail: [email protected]

Abstract – This paper presents the requirements for future Home Area Networks (HAN). Firstly, we discuss the applications and services as well as their requirements. Then, usage scenarios are devised to establish a first specification for the HAN. The main requirements are an increased bandwidth (towards 1 Gbps) and a reduced delay (< 10-20 ms).

In this paper, an overview of the requirements for future Home Area Networks (HAN) is given. First, the applications that will be required to run over this network are defined and described and then, by a suitable choice of usage scenarios, the network requirements themselves are specified.

Introduction – Residential multimedia content usage has evolved tremendously in the last few years starting from the single home PC with dial-up connection to what is seen today with multiple connected end-devices all exchanging information within or to/from the network via a broadband connection to the outside [1]. This has induced as well some changes in the way we consume information, becoming more and more content producers ourselves (e.g. digital photos and videos). Pressure on requirements for a HAN capable of handling large data rates and high Quality of Service (QoS) is increasing. This factor is particularly important and is supported by a number of industrial initiatives or pre-standardization work such as WWRF [2], DLNA [3] or HGI [4] aiming at offering access to remote services by operator's networks and high speed connectivity between end devices such asTVs, media centres, home cinema or PCs. In this paper, we will review the different applications available to the end-users today and tomorrow. We will also try to establish the networkrequirements for each application. Further, throughusage scenarios we will aggregate those applications and derive an overall target for the requirements put on the HAN. We will extensively base this study on the work performed in the framework of the ICT-FP7 ALPHA project deliverable D1.1p [5] but as well, toa lesser extent, on other projects working on the subject such as the ICT-FP7 OMEGA project [6] and the IST-FP6 MUSE project [7]. Applications – First of all, the basic applications that are used today by residential private users are mainly issued from the triple/quadruple play offers i.e. broadband Internet access, mobile and fixed

telephony and television will certainly be used in the future with improved Quality of Experience (QoE) and user friendliness. As a first step we can forecast the evolution of those services. For television, the emerging video standards (e.g. Ultra High Definition TV – UHDTV, immersive TV) lead to large bandwidth requirements (for instance 640 Mbps for compressed UHDTV) but relaxed delay constraints. This bandwidth demand is reinforced by the multiplication in homes of multimedia connected devices (Media servers and TV sets). For telephony over IP, the strictest requirements come from delay(<200 ms) but it has to be expected that this service will evolve towards video-phony or even video-conferencing adding also bandwidth requirements depending on the image quality targeted. Lastly, Internet cannot be considered as a single application anymore because many services are being offered viaInternet such as email, web browsing, gaming, videoconferencing, peer-to-peer, messaging etc. All these applications have to be considered separately to define precisely what the network requirements are.As a second step, we can consider new applications which are emerging to allow different usages to develop such as social networking, tele-working, remote services and/or e-health. Each of these new applications will have again a different set of requirements which are analysed in [5]. Usage Scenarios – The two main keywords for HANs in the future are Simplicity and Mobility: Theend-devices must connect to the HAN without human intervention and recognize the other connected elements. A protocol like UPnP allows devices to connect seamlessly and to simplify network implementation in the home [8]. In addition, the data which are stored locally or remotely must be accessible to the user wherever he is. Technological candidates here are wireless access to the end-user, e.g., WiFi in terms of flexibility and user-friendliness [9]. This wireless connectivity shall comply of course with the bandwidth and delay requirements for Home


- 41 -

Networking and could be achieved via IEEE802.11n or HSDPA (FemtoNode) in the short to medium term while other radio standards may be provided in the future such as ECMA-387 [10], IEEE802.15.3c [11], IEEE802.11ac [12], IEEE802.11ad [12] or LTE [13]. The HANs will have to be able support a plurality of services dealing mainly with Multimedia content, Online Entertainment, Tele-Working, Home Management and Health. Many usage scenarios can be derived from the different combinations of service types [5]. We will detail here a Home Office scenario as it is one which has a large potential for growth in the near future [14]. To really facilitate the trend toward Tele-Working it is necessary to establish an environment that does not isolate the worker. A variety of communication tools should coexist with different non-compatible requirements. In addition, in order for the enterprise to keep control of the work-related applications used by the tele-worker as well as to secure the sensitive data handled remotely, it is anticipated that the tele-worker's computer will act as a thin client to the enterprise local servers. Some remote PC management applications as well as the standard access to telephony and email are also provided. However, their impact on the HAN is low compared to the main applications, i.e. videoconferencing and thin client. Aggregating the different requirementsarising from these services as they stand today (Basic quality Video Conference requires 4Mbps bandwidth and less than 200 ms delay, Thin Client ranging from 100 kbps to 6 Mbps and 150ms), it is anticipated that the HAN will have to handle in the medium term, data flows of the order of less than 30 Mbps but with extremely tight constraints on delay as the higher limit of 150 ms applies for the overall Home, Access plus eventually Metro/Core network over which the thin client application will have to run. The HAN itself shall not contribute to a noticeable delay increase and thus we limit the tolerable HAN delay to less than a tenth of the overall delay requirement. In the longer term, as the video standards improve,the applications requirement will increase to 640 Mbps and 150 ms for Video Conferencing (UHD/Immersive Type) and Thin Client will require 6 Mbps and even higher and 80ms to allow video streaming. This will lead to HAN requirements in the region of 1Gbps for bandwidth and less than around 10 ms for delay. The coexistence of the high bandwidth consuming applications together with the delay sensitive ones, requires that complex though scalable QoS measures are provided to isolate each flow or at least provide the necessary prioritization. Conclusion: Home Network Specification – Today the main constraint on the Home Network is given by

delay as applications necessitating real time interaction develop and are more and more widely adopted by end users. Delay requirements are evaluated to be in the region of 15 ms max in the medium term, decreasing to <10 ms in the long term.If Bandwidth or Data rate is not really a challengetoday, it is anticipated that this requirement willsignificantly grow in the future due the development of high quality video applications leading to requirements of around up to 1 Gbps in a five yearstime frame. Other requirements must be considered as well such as user-friendliness (plug and play, wireless connectivity), Quality of Experience (flowmanagement) but as well low electrical consumption.Acknowledgement - The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7) under project 212 352 ALPHA “Architectures for fLexible Photonic Home and Access networks”. All authors who contributed to ALPHA Deliverables D1.1 [5] and D1.2 which material has been used in this paper are gratefully acknowledged. References

[1] Philip Marshall, "Surviving the Digital Home", Yankee Group, April 2008.

[2] WWRF SIG 4 "Home and enterprise networks" http://www.wireless-world-research.org/?id=92

[3] Digital Living Network Alliance http://www.dlna.org/home

[4] Home Gateway Initiative http://www.homegatewayinitiative.org/

[5] ALPHA Project Deliverable D1.1p available at http://www.ict-alpha.eu/

[6] OMEGA Project Deliverables D1.1 and D1.2 available at http://www.ict-omega.eu/

[7] MUSE Project Deliverables D3.3 and D1.7 available at http://www.ist-muse.org/

[8] UPnP Forum, http://www.upnp.org

[9] "Home Networking Forecast, 2006 to 2011", Jupiter Research, March 2007

[10] Standard ECMA-387, "High Rate 60GHz PHY, MAC and HDMI PAL", December 2008.

[11] IEEE802.15 WPAN Task Group 3c, http://www.ieee802.org/15/pub/TG3c.html.

[12] IEEE802.11Task Group AD, "Very High Throughput in 60 GHz" and AC "Very High Throughput <6GHz"

[13] 3GPP Long Term Evolution (LTE), http://www.3gpp.org/article/lte

[14] Jeremy Green "Homeworkers and the Enterprise", OVUM 7th May 2008.


- 42 -

Comparison of two types of 60 GHz photonic millimeter-wave generation and distribution of a 3 Gb/s OFDM signal

F. Lecoche1,*, B. Charbonnier1, E, Tanguy2, Hongwu. Li2, F. van Dijk3, A. Enard3, F. Blache3, M. Goix3, F. Mallecot3

1 Orange Labs, 2 Avenue Pierre Marzin, 22307, Lannion, France 2 Université de Nantes, IREENA, UFR Sciences et Techniques, 2 rue de la Houssinière, 44322, Nantes, France

3 Alcatel-Thalès III-V Lab, 1, avenue Augustin Fresnel, 91767 Palaiseau, France * F. Lecoche : Phone: +33 296 051 042, E-Mail: [email protected]

Abstract – We demonstrate and compare experimentally two set-ups achieving very high data rate (3 Gbps) wireless transmission in the 60 GHz window, both using Radio-over-Fiber (RoF) for reach extension with OFDM signal compliant to the IEEE 802.15.3.c pre-standard.

Introduction – Multiplication of connected devices and services (computers, media center, videophony, TVoIP etc…) [1] lead to a novel home network architecture and technology to enable wireless coverage offering a data rate above 1 Gbps (Figure 1) [2]. This Home Network architecture is based on a wired backbone network distributing local very highspeed wireless connectivity in the different rooms of the house/building because the only radio technology whose capacity approaches some Gb/s uses the 60 GHz frequency window as for instance ECMA-387 [3] or upcoming standards like IEEE 802.15.3c [4] or IEEE 802.11ad [5]. Such radio interfaces areinherently short reach (<15 m) and the radio signal is confined to within one single room. The radio home network then becomes a multicellular network where the cell interference and management issues are similar to that of larger scale mobile/radio networks. In this context, again, the use of the optical infrastructure with RoF, to link the different remote antennas to provide a cost effective and flexible solution, must be considered. In this paper, we present and compare the performance of two types ofphotonic system for 60 GHz millimeter-wave generation and distribution of a 3 Gb/s OFDM signal. The first one uses low cost well known components and the second one a new generation of optoelec-tronic components designed for 60 GHz applications.

I. RADIO OVER FIBRE SET-UPS

In this part, we show two RoF set-ups generating a 60 GHz millimeter-wave radio signal. For both set-ups,we use an OFDM QPSK signal compliant to IEEE802.15.3.c pre standard [4] (3 Gb/s).

A. Multi-Mode Fibre Set-up (Fig 2)

The OFDM signal is generated first on an intermediate frequency (IF=4.5 GHz) and is used to modulate directly a VCSEL (850 nm) which converts

Fig 1: RoF home architecture and solution for very high data rate wireless next generation

the electrical signal into an optical one. After transport over 300 meters of multimode fiber (OM3 with 4000 MHz.km), an 8.5 GHz bandwidth photodiode converts the optical signal into an electrical one. After, the signal is sent into a low noise amplifier (LNA) and a mixer fed with a 54.5 GHz local oscillator (LO) to reach the 60 GHz radio frequency window. A high power amplifier (HPA) and a 20 dBi horn antenna are used for the subsequent wireless transmission (10 m). In the radio receiver, two LNA amplify the signal before frequency down-conversion and analysis by a Real Time Oscilloscope (RTO).

Fig 2: VCSEL set-up with distribution of OFDM QPSK signal before up-conversion to 60 GHz

B. Single Mode Fibre Set-up (Fig3)

This set-up up-converts signal to 60 GHz before transmission over fiber using a 54.8 GHz Mode Locked Fabry Perrot Laser (ML-FPL). The IF OFDM signal modulates directly the ML-FPL. The laser pulses with a repetition rate of 54.8 GHz. Its


- 43 -

modulation produces a mixing between the pulsating frequency and the IF carrier leading to an optical frequency up-conversion of the original signal to 59.3 GHz. The optical radio signal is then transmitted through 50 m of standard Single Mode Fibre (SMF) up to a commercial 70 GHz photodetector followed by a LNA, a band-pass filter and a HPA. The transmit antenna and the receiver radio front end are identical to the ones used in the Multi-Mode Fibre Set-up.

Fig 3: ML-FPL set-up with OFDM QPSK signal up-conversion to 60 GHz before distribution

II. RESULTS

In this part, we show the results obtained for an OFDM modulation at 3 Gb/s and for each set-up.

A. Multi-Mode Fibre Set-up

The electrical power at the input of VCSEL is set to -10 dBm and the power at the output of the 20 dBi horn antenna is also +10 dBm. In figure 4 a) and b), we present respectively the spectrum of the received OFDM signal and the associated QPSK constellation diagram obtained after demodulation. In spite of a strong disturbance of the spectrum, the mean computed EVM is 18.7% for a signal to noise ratio (SNR) of 23 dB. From the calculated EVM, a BER can be estimated to be around 10-10 [6]. Theoretically, the measured SNR should provide an EVM of 7% [6]. The difference between the measured and theoretical values of EVM is attributed to the residual non-linearities of the system.

a) b) Fig 4: Multi-mode fibre set-up received OFDM

Spectrum (a) and respective constellation (b) (dots around coordinates [1,0] are pilot tones used for equalization).

B. Single-Mode Fibre Set-up

The level of radiated power is similar to the previous set-up at the output of the transmit horn antenna. In figure 5 a) and b), we exhibit, respectively, the

spectrum of the received signal and the QPSK constellation diagram. The computed EVM is 19% for a SNR of 21.5 dB. The value of BER can be estimated at 10-10 [6]. Again, the measured SNR should provide a theoretical EVM of 8.4% [6]. Distortions and non linearity do affect the calculated EVM but results are still acceptable.

a) b) Fig 6: Multi-mode fibre set-up received OFDM Spectrum (a) and respective constellation (b).

III. CONCLUSION

In this paper, we presented two different set-ups for very high data rate wireless transmission using radio-over-fibre at 60 GHz. The first one uses low cost commercial components (VCSEL emitting at 850 nm and 8.5 GHz photodiode) as well as Multi Mode Fibre and the up-conversion to 60 GHz is performed at the remote antenna site after the signal opticaldistribution. The second one uses a ML-FPL at 54.8 GHz which is able to up-convert directly the signal to 60 GHz before distribution over Single Mode Fibre. For both set-ups, we present the performance evaluations with an OFDM QPSK signal carrying 3 Gb/s. At 10 meters of wireless radio transmission, we report a computed EVM of 18.7% for the VCSEL set-up and 19% for ML-FPL. These values should lead to a BER around 10-10 and show that both set-ups have quite similar performances. On one hand, the Multi-Mode Fibre Set-up has proven commercial availability but requires a more complexremote antenna station. On the other hand, the Single Mode Fibre Set-up allows the remote antenna stationto be simplified but the components maturity is notyet achieved.

References [1] European project ICT-FP7-ALPHA deliverable 1.1 available

online at www.ict-alpha.eu [2] B. Charbonnier, H. Le Bras, P. Urvoas, Q.T. N'Guyen, M.

Huchard, A. Pizzinat, "Upcoming perspectives and future challenges for ROF", MWP 2007, Pages: 21 – 23.

[3] Standard ECMA-387, "High Rate 60GHz PHY, MAC and HDMI PAL", December 2008, /www.ecma-international.org/publications/files/ECMA-ST/Ecma-387.pdf.

[4] IEEE802.15 WPAN Task Group 3c, http://www.ieee802.org/15/pub/TG3c.html.

[5] IEEE802.11Task group AD, "Very High Throughput in 60 GHz".

[6] V.J.Urick et al, " Wide-Band QAM-Over-Fibre Using Phase Modulation and Interferometric Demodulation", IEEE PTL vol. 16, No 10, 2004


- 44 -

Bidirectional Multi-Standard RoMMF Transmission Using a Reflective Electro-Optic Transceiver

I. Möllers*, D. Jäger Universität Duisburg-Essen, Zentrum für Halbleitertechnik und Optoelektronik, Lotharstr. 55, Duisburg, Germany

* Phone: +49 203 379 4635, Fax: +49 203 379 2409, E-Mail: [email protected]

Abstract – A characterization of a Radio-over-Multimode fiber link for picocellular architectures using a novel reflective electro-optic transceiver (REOT) for bidirectional transmission over a single fiber is presented. Spurious-free-dynamic-range (SFDR) analysis and error-vector-magnitude (EVM) measurements for multi-standard and subcarrier-multiplexed (SCM) wireless signals are carried out.

Introduction – The ever increasing demand on high data rates in conjunction with mobility for different services leads to a convergence of fixed and wireless access systems supporting picocellular networks [1,2]. Especially for short-range and in-building systems highly integrated but low-cost solutions are required. Radio-over-Multimode Fiber (RoMMF) or Radio-over-Polymer Optical Fiber (RoPOF) systems can address exactly this scenario by providing highbandwidth and reliability coexisting with relatively low installation costs. Distributed Antenna Systems(DAS) based on RoMMF techniques together with one or more central units and several fiber linked distributed base stations, both containing active devices, have been reported [2-4]. We propose a DASby coevally reusing frequency channels in separate picocells. This enables using separate frequency channels for full-duplex transmission for up- and downlink in one picocell. New developments show that self-sustaining or even passive base stations, e.g. by power over fiber techniques [5] or using fully passive eo/oe-components, could be a solution for DAS in the future. Addressing these systems we present the characterization of intensity modulation-direct detection (IM-DD) RoMMF WDM link using a passive full-duplex vertically integrated reflective electro-optic transceiver as key element for DAS base stations for the first time. No bias voltage or current

is needed for the system due to a 0 V biased modulator and photodiode operation. Multi-standard signal transmissions like WLAN, GSM, UMTS and DPRS are demonstrated.

Transceiver Design – The transceiver is a monolithically integrated epitaxially grown GaAs/AlAs/AlGaAs heterostructure with pinip-configuration resulting in two pin-diodes. A bragg with an intrinsic resonator layer composes a vertical modulator changing the reflected intensity of a continuous incident beam at 790 nm wavelength utilizing the well known Franz-Keldysh-Effect. The modulator is quasi transparent for a modulated incident beam at a wavelength of 850 nm. This signal is detected by the subjacent nip-photodiode structure for the downlink [6]. Fig. 1 shows a block diagram of the RoMMF link containing the passive bidirectionalfull-duplex REOT as key oe/eo-element for low-power consuming base stations in DAS systems.

Central Station Base Station

RX TXREOT

f2

f1

Antennas

Central Station Base Station

RX TXREOT

f2

f1

Antennas

Fig. 1.: Block diagram of bidirectional RoMMF system using separate channels (f1, f2) for up- and downlink

REOT

VSGWLAN (fc=2.44GHz)GSM (fc=0.9GHz,

1.8GHz)UMTS (fc=2.0GHz)DPRS (1.88GHz)

LDλ1,CW

ROSA(PD+TIA)

VSA

Y-coupler50/50

MMF/POF(10m ≤ lMMF ≤ 510m)

P

ffλ1ffc

P

f

fc

fλ1

P

ffc

DCSupply

REOT

VSGWLAN (fc=2.44GHz)GSM (fc=0.9GHz,

1.8GHz)UMTS (fc=2.0GHz)DPRS (1.88GHz)

LDλ1,CW

ROSA(PD+TIA)

VSA

Y-coupler50/50

MMF/POF(10m ≤ lMMF ≤ 510m)

P

ffλ1

P

ffλ1ffc ffc

P

f

fc

fλ1

P

f

fc

fλ1

P

ffc

P

ffc

DCSupply

Fig. 2.: System block diagram for RoMMF uplink measurements using the modulator function of the REOT

REOT TX RX

Signal Processing


- 45 -

Experimental Setup – In the proposed system (fig. 1) the uplink turns out to be the more critical transmission; hence we focus on the uplink experiment in this paper. A RoMMF link with the REOT device was set up for SFDR and EVM uplink measurements of multi-standard wireless access signals, shown in fig. 2. All fibers used were 62.5µm core diameter with graded-index (GI) profile. A continuous wave signal, provided by a fiber pigtailed edge emitting LD (λ1 = 790 nm, Popt = 5.8 dBm @ 110 mA), is guided through an optical isolator and y-coupler, multimode glass optical fiber (MM-GOF), over different lengths and types of MMF/POF fibers.A MM-GOF pigtail bare end is free-space coupled to the REOT device. An Agilent vector signal generator(VSG) E4438C is directly connected to the REOT modulator with 0 V bias using microprobe contact equipment. The modulated signal is received by the same fiber and guided through the y-coupler to a ROSA package (f3dB,ROSA = 9 GHz). The demodulation was provided by an Agilent MXA N9020A vector signal analyzer (VSA).

Link Analysis – The SFDR of a transmission system is used to analyze the uplink performance with respect to intermodulation distortion (IMD) of the link. A dynamic range measurement according to [7] has been done with the proposed optical link for input powers ranging from -4 dBm to +12 dBm at a frequency of 2.45 GHz. Fig. 3 displays the SFDR analysis of the system. A SFDR value of 77.1 dB/Hz2/3 and a link gain of -38 dB was observed considering a measured noise level of -122 dBm (1 Hz). The IP3 value was found to be -17 dBm. An uplink with this value achieves the requirements ofWLAN 802.11b/g transmission systems of 75 dB/Hz2/3 [8].

In addition to the analog analysis RoMMF data transmission measurements were carried out for different wireless standards, lengths and types of fibers. Table 1 summarizes the EVM measurement results for different wireless standards such as GSM (GSM900 and DCS1800), DPRS (DECT), UMTS, and WLAN 802.11 b/g. It can be seen that the EVM requirements for all GSM, UMTS, DPRS and WLAN 802.11b transmissions can, to some extend, be highly exceeded with the proposed system. Selected constellation and eye diagrams are displayed in fig. 4. Clearly open eye diagrams were observed for GSM (DCS1800) and UMTS transmission experiments.

In order to relate the link performance to carrier frequency up to 5 GHz EVM/SNR vs. frequency measurements were carried out applying a 16QAM 24 Mbps (6 MHz modulation bandwidth) signal (fig. 5). By adjusting the applied modulator input power for each measurement point separately EVM values

-80 -60 -40 -20 0 20-140

-120

-100

-80

-60

-40

-20

0

20

IP3 = -17 dBm

SFDR = 77.1dB/Hz2/3

Fundamental IMD3 Meas. Noise Level (-122dBm)

Link Gain G = -38 dBFrequency f = 2.45 GHz

Out

put P

ower

(dB

m)

Input Power (dBm)

Fig. 3.: SFDR analysis (RBW = 1 Hz) with input vs. output power of fundamental and third order intermodulation signal (IMD3)

Fig. 4.: Selected constellation (1) and eye diagrams (2) for transmission experiments listed in table 1: (a) DCS 1800 EVM: 1.74 %rms, (b) UMTS QPSK 2 GHz EVM: 2.61%rms, (c) WLAN 802.11b DSSS EVM: 6.82 %rms, (d) WLAN 802.11g OFDM EVM: 9.79 % rms

Standard (Technique)

Modulation Format

Carrier Frequency

Chip / Data Rate

Filter Required

EVM (%rms)

Measured EVM

(%rms)

GSM 900(TDMA)

GMSK 900 MHz 270.833 kbps BT= 0.3 < 7.0 1.19

DCS1800(TDMA)

GMSK 1800 MHz 270.833 kbps BT= 0.3 < 7.0 1.74

DPRS (TDMA)

64QAM 1.88 GHz 1.152 Msps α = 0.5 < 2.6 2.51

UMTS (WCDMA)

QPSK 2 GHz 3.84 Mcps α = 0.22 < 12.5 2.61

WLAN 802.11b(DSSS)

QPSK 2.45 GHz 11 Mbps α = 0.3 < 35 6.82

WLAN 802.11g(OFDM)

64QAM 2.45 GHz 54 Mbps α = 0.3 < 5.6 9.79

Table 1.: EVM requirements [4] and measured results of IEEE and ETSI standard signal transmission using the proposed uplink with a GOF length of 25 m (fig.2)

a1)

c) d)

a2)

b1) b2)


- 46 -

below 5 %rms were recorded up to carrier frequencies of 3.5 GHz. This allows IEEE 802.16e WiMAX WCDMA transmission (Germany: 3.4 GHz - 3.6 GHz) with medium data rate performance. For in-building scenarios different types of fibers andlengths will be of interest for such RoMMF systems.Therefore we investigated the influence of fiber types of 62.5µm GI-GOF and 62.5µm perfluorinated (PF)-GI-POF and lengths for the system, respectively. A QPSK signal at a carrier frequency of 2.44 GHz witha data rate of 2 Mbps was used for EVM analysis. The results of the measurement are summarized in fig. 6. For both fibers we obtain a somehow linear correlation of fiber length and EVM value for this uplink system. The comparison of both slope values for the different fibers shows ten times higher slope in PF-GI-POF compared to GI-GOF. The difference of NA between both fibers (PF-GI-POF and pigtail GI-GOF) results in coupling losses at both, the y-coupler and to the pigtail fiber of the REOT causing signal loss and EVM rising. For all measurements a broadband circulator for a wavelength range from

790 nm to 850 nm could increase link performance significantly due to optical loss reduction by y-coupler and isolator. Coupling efficiency of the REOT to the fiber and modulator matching or matched driving circuitry could further improve measurement results.

Conclusion – In this paper we have proposed a RoMMF transmission system using a passive bidirectional full-duplex transceiver for eo-conversion at the base station for the first time. A SFDR value of 77.1 dB/Hz(2/3) was found for 2.45 GHz by achieving a uplink gain of -38 dB. EVM analysis show that multiple standard wireless access signals such as GSM, UMTS, WLAN 802.11b and WiMAX can be transmitted with values partially far below the required ones according to the standards.Different scenarios on fiber type and length were tested in order to demonstrate their dependence forin-building installations.

Acknowledgement – The work described in this paper was carried out with the support of the BONE-project (“Building the Future Optical Network in Europe”), a Network of Excellence funded by the European Commission through the 7th ICT-Framework Programme, and the FP6 IST ePIXnet-project (“European Network of Excellence on Photonic Integrated Components and Circuits”). We thank E. Arslangiray from Agilent Technologies Deutschland GmbH for his support.

References[1] R. Gaudino, A. Pizzinat, I. Möllers et al., “Future

Internet in Home Networks: Towards Optical Solutions?“, Future of the Internet Conference, Prague, May 2009, accepted for publication

[2] M. Sauer et al., “Radio Over Fiber for Picocellular Network Architectures”, JLT, Vol. 25, No. 11, 2007

[3] H. Al-Raweshidy et al., “Radio over Fiber Technologies for Mobile Communication Networks”, Artech House, 2002

[4] C. Lenthien et al., “Potentials of Radio over Multimode Fiber Systems for the In-Building Coverage of Mobile and Wireless LAN Applications”, PTL, Vol. 17, No. 12, 2005

[5] D. Wake et al., “Optically Powered Remote Units for Radio-Over-Fiber Systems,” JLT, Vol. 26, No. 15, 2008

[6] I. Möllers, M. Bülters et al., “High-Speed Transceiver for Radio-over-POF Applications”, ICPOF, pp. 48-51, Turin, Italy, 2007

[7] J. Capmany, “Measurement issues in Microwave Photonics”, Symposium on Optical Fiber Measurements, Tech. Digest, 2004

[8] K. Hagedorn et al., “Heterogeneous wireless/ wireline optical access networks with R-EAT as backend component, SPIE, Vol. 5466, pp.27-33, 2004

0.1 1 100

2

4

6

8

10

20

22

24

26

28

30

Conditions16QAM, 24Mbps (6MHz mod.bw) 10m MM-GOF 62.5µm, 0V BIASOptimal Modualtion Input Power adjusted

EV

M (%

rms)

Frequency (GHz)

SN

R (dB

)

Fig. 5.: EVM/SNR vs. carrier frequency, 16 QAM, 24Mbps, 0V Bias at Modulator, Modulation Input Power adjusted (range: 3 dBm to 13 dBm)

0 100 200 300 400 5000

5

10

15

20

PF-GI-POF (62.5µm) GI-GOF (62.5µm)

EV

M (%

rms)

Fiber Length (m)

Fig. 6.: EVM measurement results for different lengths (excluding pigtail fiber of 10m MM-GOF) of MM-GOF and MM-POF with a QPSK 2Mbps at 2.44GHz carrier frequency


- 47 -


- 48 -

High modal bandwidth glass multimode fibers used for the simultaneous transmission of 10GbE and band group 5 MB-

OFDM Ultra-Wide Band signals

C. Lethien1, 2*, C. Loyez1, 2, J-P. Vilcot2 and P-A. Rolland1, 2

1 Institut de Recherche sur les Composants logiciels et matériels pour l'Information et la Communication Avancée

(IRCICA), Université des Sciences et Technologies de Lille, CNRS FR 3024, Parc Scientifique de la Haute Borne, 50 avenue Halley, 59650 Villeneuve d'Ascq – France

2 Institut d'Electronique, de Microélectronique et de Nanotechnologie (IEMN), Université des Sciences et

Technologies de Lille, CNRS UMR 8520 Avenue Poincaré - BP 60069 - 59652 Villeneuve d'Ascq cedex – France * Corresponding author: C. Lethien; �: +33 36253 1617; E-mail address: [email protected]

Abstract – This paper deals with the potential of high modal bandwidth glass multimode fibers used for the simultaneous transmission of baseband 10GbE and MB-OFDM Ultra-wide band signals.

Introduction – In this paper, we describe the potential of the OM4 (or A1a.3) MaxCap 550 glass multimode fiber (GMMF) from Draka Comteq combined with wavelength division multiplexing devices in order to carry out the simultaneous transmission of high data rate 10GbE signals with high throughput radio signal.

The idea is to use the indoor fiber Ethernet network for the enhancement of the coverage of wireless signals such as developed by the Wimedia and the Multiband OFDM alliances (under the acronym ECMA-368 [1]). Due to the large amount of exchanged data, the bandwidth requirement of the next generation of wired optical Local Area Networkinduces the deployment of the 10GbE standard. Moreover, most of the existing fiber network is based on short haul glass multimode fiber (GMMF) topology. In this study, we focus firstly on the 10GBase-SR declination of the 10GbE standard (IEEE802.3ae) dealing with an 850 nm communication over GMMF. Then, the optical multiplexing of the ECMA-368 WPAN Standard (by using a radio over fiber technique at 1300 nm) withthe baseband one is realized over the GMMF. This radio standard uses the frequency range from 3.1GHzup to 10.6GHz (divided into 14 unlicensed bands) and supports data rates of up to 480 Mbps. Regarding to the ECMA standard proposal, we focus on the band group presenting the highest carrier frequency to exhibit the high potentialities of the OM4 fiber. Reference transmitter and receiver have been used in order not to induce penalties owing to the E/O and the O/E conversion. The properties of the GMMF under test are summarized in the fig. 1a.

Test bench – The test setup used during these experiments is described in fig. 1b (downlink topology). A Pseudo Random Bit Sequence (PRBS) of 223-1 bits is provided by the pattern generator with a 10.3125Gbps data rate and feeds the baseband input of a commercially available XFP transceiver; the baseband signal directly modulates the bias current of an 850nm VCSEL (TX part of a 10G XFP). The combined uses of an arbitrary waveform generator (IQ generation) and a vector signal generator (RF generation) allow to generate the bandgroup 5 (TimeFrequency Code 5 – carrier frequency close to 9.768GHz) of the MB-OFDM Wimedia [1] radio signal. In order to develop an easy architecture, the intensity modulation/direct detection technique is performed for the transmission of the radio signal over the OM4 fiber at 1300nm (DFB laser). The two signals are then optically combined and divided after propagation over a single OM4 fiber thanks to the use of wavelength division multiplexing components specially designed for multimode communication. Results – Concerning the 10GbE analysis, we have measured the bit error rate as a function of the received optical power in order to exhibit the power dispersion penalties of the fibers under test in the back-to-back case and for the 5 fiber lengths undertest. An optical variable attenuator has been inserted just before the photodetector in order to reduce the received optical power. The results are presented in the fig. 1c. The 5 lengths of high bandwidth fiber exhibit power dispersion penalties from 0dB (100m) to 2.4dB (1100m). These penalties are inherent to the modal bandwidth of the fiber. The mask test compliance of the 10GbE signal transmitted over theGMMF under test allows to investigate the bandwidth


- 49 -

limitations, the intersymbol interference (ISI) inherent to the overall system (TX, RX and optical fiber) and the signal jitter. Concerning the radio UWB MB-OFDM analysis, we have performed the measurement of the relative constellation error (RCE) as a function of the RF input power. The RCE (max RCE requirements: 10.6% for 480Mbps data rate) measurement has been led firstly over the GMMF. An additional radio propagation of 1m length is realized secondly to transmit the radio signal till the mobile receiver. The test setup composed of the MB-OFDM Wimedia generation and analysis equipments linked by a short length of coaxial cable provides a RCE up to 4%. All the measured curves (fig. 1d) present the same evolution. At a specific RF power level, the RCE increases quickly with the growth of the RF power mainly due to the non linearity effect occurring in the laser and the peak to average power ratio (PAPR) ofthe MB-OFDM UWB signal: this increase is inherent to the high level of RF power used to modulate the bias current of the laser [2]. Further results dealing with RCE variation as a function of the fiber attenuation and the fiber length could be presentedduring the workshop. A RCE value up to 7.2 % has been obtained for an 1100m OM4 fiber link and 1m wireless radio path. As expected, the OM4 fiber could be considered as a high performance transmission radio path over a large RF power range for the bandgroup 5 (BG5) centered to 9.768GHz and thus, for all

the band groups dedicated to the MB OFDM UWB Wimedia deployment.

Conclusion – Regarding to the obtained results, we can conclude that the OM4 fiber is a promising candidate to be used for multi-standard transmission both for wired and wireless high data rate signals such as tested in this paper. Owing to its high modal bandwidth at 850nm and 1300nm, the 5 lengths of OM4 GMMF under test are suitable for 10GbE applications. Thanks to the wavelength multiplexingtechnique, no influence between the two signals hasbeen quantified.

Acknowledgement - This work was supported by the ERDF (European Regional Development Fund) and by the Nord-Pas-de-Calais Region (France). The authors want to thank DrakaComteq for the supply ofthe OM4 fibers.

References

[1] Standard ECMA-368, “High Rate UltraWideband PHY and MAC Standard,” 2nd edition, (December 2007) http://www.ecma-international.org/publications/standards/Ecma- 368.htm

[2] C. Lethien et al, “Review of glass and polymer multimode fibers used in a Wimedia UltraWide Band MB-OFDM radio over fiber system”, IEEE/OSA Journal of Lightwave Technology, accepted for publication since January 2009

� ��

�� !�"��#�$��!�"��!��%��&�� '()*$"��#�+��)*$"��

,�� &� ��-��-��'��.�� $��

�

/,!�� /,!��

��&�� &��

��0�123��0�123

�4/��4/�

�4/��4/�

5�

6�(�

�,

!��

��7�� (�2

8,�� 90�� 1��(��

��&�� &��

0�� 1��7��

Access point

:;�� 9�� 74�<�� (��

I Q

10GbE

TX

/RX

m

ountedin an

evaluationboard

-18 -17 -16 -15 -14 -13 -12 -11 -10 -9 -81E-161E-151E-141E-131E-121E-111E-10

1E-91E-81E-71E-61E-51E-41E-3

Short length 100m 200m 300m 600m 1100m

Bit

Err

or R

ate

Received Optical Power (dBm)

-32 -28 -24 -20 -16 -12 -8 -43

4

5

6

7

8

9

10

11

12 Short length 100m 200m 300m 600m 1100m

Rel

ativ

e C

onst

ella

tion

Err

or (

% r

ms)

PRF-In

(dBm)

Fig. 1.: (a) Summary of the OM4 fiber properties; (b) test setup of the simultaneous transmission (10 GbE and Wimedia BG5); (c) Bit Error Rate as a function of the received optical power ; (d) RCE as a function of the input power level and the fiber length


- 50 -

Experimental Implementation of Real-Time Optical OFDM Modems for Optical Access Networks

R.P. Giddings, X.Q. Jin, H.H. Kee, X.L. Yang and J.M. Tang School of Electronic Engineering, Bangor University, Bangor, LL57 1UT, UK

AbstractReal-time optical OFDM transmitters and receivers are, for the first time, successfully demonstrated for 1.5Gb/s transmission over 500m 62.5/125 m multimode fibers in an intensity-modulation and direct-detection system involving a directly modulated DFB laser. A BER of less than 1.0×10-9 was observed in the corresponding optical back-to-back system. The implemented modems only use standard, commercially available components including FPGAs and DACs/ADCs.

IntroductionThe concept of optical orthogonal frequency division multiplexing (OOFDM) was first proposed in 2005 [1], soon after, opportunities of employing OOFDM signals converted by directly modulated DFB lasers (DMLs) were theoretically explored over multimode fiber (MMF)-based LANs [2] and single mode fiber (SMF)-based MANs [3]. Since then, extensive investigations of OOFDM transceivers of various configurations have been reported in long-haul [4,5], MANs [6] and LANs [7]. However, all the experimental works published so far have been undertaken using non-real-time signal processing approaches, which do not consider the limitations imposed by the precision and speed of practical digital signal processing (DSP) hardware. In addition, the non-real time approach is just able to utilise relatively short data transmission sequenecs in analysing the transmission performance of various systems. The experimental demonstration of real-time OOFDM transceivers is critical for not only rigorously validating the OOFDM technique but also establishing a solid platform for evaluating its feasibility for practical implementation. In an optical back-to-back system, a real-time coherent OOFDM receiver has been reported recently [4], which is, however, not able to perform real-time data transmission due to the absence of a corresponding real-time transmitter.

In this paper, real-time OOFDM transmitters and receivers based on standard commercially available components such as FPGAs and ADC/DACs are, for the first time, demonstrated experimentally, whose transmission performance is investigated over a DML-based MMF IMDD transmission system. This work indicates the great potential of OOFDM as an important technology for practical implementation in next generation high capacity optical networks of various architectures.

Real-time OOFDM modems and experimental setup Fig.1a shows the real-time experimental system configuration. The transmitter consists of an Altera Stratix II GX FPGA, which performs the real-time digital signal processing (DSP) on the generated data source and outputs four 8-bit samples in parallel at a rate of 500MHz. These samples are fed to an 8-bit DAC running at 2GS/s. The analog electrical signal with a 1GHz bandwidth is attenuated by a variable attenuator to adjust the modulating current injected into a 1550nm DML having a modulation bandwidth of 10GHz. The optical signal emerging from the DML is coupled into a 500m OM1 MMF link having a 3-dB optical bandwidth of 1200MHz·km. At the receiver, a 12GHz bandwidth PIN with a TIA converts the transmitted optical signal to the electrical domain. The electrical signal is amplified by a 2.5GHz, 20dB RF amplifier, attenuated as needed to adjust the signal amplitude. The low-pass-filtered, single ended electrical signal is converted via a balun to a differential signal to feed an 8-bit ADC operating at 2GS/s, whose digital interface format is identical to the DAC input in the transmitter. Finally, the digital samples are fed to a second Altera Stratix II GX


- 51 -

FPGA, which performs the real-time DSP on the received symbols and determines the BER measurement. Clock synthesizers based on a common reference clock are used to generate the system clocks for the transmitter and the receiver.

Fig.1b shows the detailed transmitter (top) and receiver (bottom) architectures. The digital logic is entirely implemented with self-designed logic blocks. The core functions required for real-time OOFDM transceivers are the Inverse Fast Fourier Transform (IFFT) in the transmitter, and FFT in the receiver. The developed real-time IFFT/FFT logic function is a 32 point, decimation in time, pipelined architecture, whose key parameters can be fully adjusted and optimized to minimise the finite computational error inherent in physical DSP hardware.

FPGAAltera

Stratix IIGX

DAC2Gsps

4x 8 bit

DFB

modulatingcurrent

bias current

MMF Fiber

ADC2GSps

4x 8 bit

Balu

n

LPF

attenuator 2

RFamplifier(a)

Para

llel D

ata

Gen

erat

or

DQ

PSK

M

odul

ator

s

IFFT

Clip

ping

&

Qua

ntis

atio

n

Add

Cyc

lic P

refix

Bit &

Sam

ple

Org

anis

atio

n

Sign

ed to

U

nsig

ned

Hig

h sp

eed

IO(P

aral

lel t

o Se

rial)

Logic Array Dedicated HW

32 b

it Pa

ralle

l LV

DS

Inte

rface

(b)

Com

plex

C

onju

gate

Transmitter

Receiver

Hig

h sp

eed

IO(S

eria

l to

Para

llel)

Dedicated HW

32 b

it Pa

ralle

l LV

DS In

terfa

ce

Logic Array

Rem

ove

Cyc

lic P

refix

Bit &

Sam

ple

Org

anis

atio

n

Uns

igne

d to

Si

gned

Symbol Position

FFT

DQ

PSK

Dem

odul

ator

s

BER

Ana

lyse

r

Signal Tap II

Para

llel D

ata

Gen

erat

or

Sync

opticalattenuator

A

B C

D

500mV

636mV

+20dBPIN & TIA

V to I

FPGAAltera

Stratix IIGX

attenuator 1

Signal Tap II

Figure 1: a) Experimental transmission system, b) Real-time OOFDM transceiver architectures

In the transmitter, except for the digital back-to-back case, DQPSK is considered. 32 subcarriers are used, of which 15 carry data. A 30-bit parallel data sequence feeds 15 DQPSK modulators which generate the complex data for the data-carrying subcarriers. To achieve a real-valued IFFT output, the data-carrying subcarriers are arranged to satisfy the Hermitian symmetry with respect to their complex conjugate counterparts [2]. The signed, 32 real-valued IFFT outputs are clipped at a clipping ratio of 11.6 dB and quantized to 8 bits. A cyclic prefix of 8 samples is added to each symbol, producing 40 samples per symbol. The DAC sampling rate of 2GHz gives a symbol rate of 50MHz. The signed samples are converted to unsigned values as the DAC requires positive values only. Sample reordering and bit arrangement are performed to present the symbol data to the 32 high-speed, 10:1, serialisers in the required order and to ensure that the serialisers feed the four sample interface to the DAC in the correct sequence. In the receiver, an inverse DSP procedure, compared to that described above, is used to recover the received data, as shown in Fig.1b.

To achieve symbol alignment, a test symbol of a fixed pattern is sent repeatedly by the transmitter, the received symbol is detected and analyzed using the Signal Tap II, embedded logic analyzer and the Altera Quartus II software, the measured sample offset is compensated for by adding an appropriate time delay. As shown in the Fig.1b, a BER analyser block continuously detects and counts errors occurring within one million symbols.


- 52 -

The error count is viewed via the Signal Tap II embedded analyser and an average BER is obtained over a large number of readings.

-100 -50 0 50 100-100

-50

0

50

100

I

Q

(a) Digital back-to-back (One FPGA) 9,375 Gbps, 16QAM, subcarrier 1

-100 -50 0 50 100-100

-50

0

50

100

I

Q

(b) Analog back-to-back1.5Gbps, DQPSK, subcarrier 3

-100 -50 0 50 100-100

-50

0

50

100

(c) Optical back-to-back1.5Gb/s, DQPSK, subcarrier 3

I

Q

-100 -50 0 50 100-100

-50

0

50

100

I

Q

(d) 500m, 62.5um, MMF1.5Gb/s, DQPSK, subcarrier 3

Figure 2: Subcarrier constellations for different system configurations

ResultsTo evaluate the developed real-time DSP function only, investigations are first undertaken of the performance a digital back-to-back transmitter and receiver configuration implemented within a single FPGA without involving the DAC/ADC. Based on a symbol rate of 156.25 MHz (less than half of the maximum FPGA clock speed), 9.375Gb/s is achieved at a BER of zero by using a 16QAM modulation format on all subcarriers. Fig.2a shows a representative constellation of subcarrier 1. This confirms the capability of the developed real-time DSP function for supporting high speed transmission.

By including the DAC/ADC, experimental measurements are also conducted in an analog back-to-back transmitter and receiver configuration, in which points A and D, as shown in Fig.1a, are connected with attenuator 1 being set to 3dB. With the sampling rate of 2GS/s and DQPSK, 1.5Gb/s transmission at a BER of zero is achieved with the constellation of subcarrier 3 being shown in Fig.2b. Further measurements are also performed in an optical back-to-back configuration, in which points B and C, as shown in Fig.1a, are connected. For this case, attenuator 1, the optical attenuator and the receiver’s electrical gain are taken to be 5dB, 6dB and 3dB, respectively, also a DFB bias current of 38mA is adopted. 1.5Gb/s transmission at a BER of <<1.0×10-9 is measured, as shown in Fig.3, and the constellation of subcarrier 3 is shown in Fig.2c.

Finally, experimental measurements are undertaken of 1.5Gb/s transmission over a 500m 62.5/125 m MMF IMDD link illustrated in Fig.1a. The measured BER as a function of optical launch power is plotted in Fig 3. For an optical launch power of -3dBm, a BER of <1.0× 10-6

is observed with a corresponding constellation of subcarrier 3 being shown in Fig.2d. The constellation of the same subcarrier for a BER of 1.0x10-4 is also inserted in Fig.3. A power penalty of 5.8dB at a BER of 1.0x10-3 is observed in Fig.3. Experimental measurements also show that the transmission performance is insensitive to different launch conditions.


- 53 -

Figure 3: Measured BER performance for 1.5Gb/s transmission over a 500m MMF

ConclusionsFirst real-time OOFDM modems have been demonstrated successfully for 1.5Gb/s transmission over a 500m MMF IMDD system. Experimental measurements also indicate that the demonstrated transceivers are potentially capable of supporting much higher data rates, when use is made of higher modulation formats and faster DAC/ADC sampling rates. Results indicate that the accuracy and speed of the developed real-time DSP hardware are not limiting factors for achieving 10Gb/s real-time OOFDM data transmission.

References 1. N.E. Jolley, H. Kee, R. Rickard, J. Tang, K. and Cordina, OFC/NFOEC, Anaheim,

California USA, paper OFP3, March, 2005. 2. J.M. Tang, P.M. Lane, and K.A. Shore, J. Lightwave Technol., 2006, 24, (1), pp.429-

4413. J.M. Tang, P.M. Lane, and K.A. Shore, OFC/NFOEC, Anaheim, California USA,

paper JThB8, March, 2006 4. Q. Yang, N. Kaneda, X. Liu, S. Chandrasekhar, W. Shieh, and Y.K. Chen,

OFC/NFOEC, San Diego, California USA, paper PDPC5, March, 2009 5. B.J.C. Schmidt, Z. Zan, L.B. Du, and A.J. Lowery, OFC/NFOEC, San Diego,

California USA, paper PDPC3, March, 2009 6. W. Wei, M.D. Feuer, D. Qian, P.N. Ji, N. Cvijetic, C. Wang, T. Wang, OFC/NFOEC,

San Diego, California USA, paper PDPA7, March, 2009 7. H. Yang, S.C.J. Lee, E. Tangdiongga, F. Breyer, S. Randel, and T. Koonen,

OFC/NFOEC, San Diego, California USA, paper PDPD8, March, 2009

Acknowledgement

This work was partly supported by the European Community's Seventh Framework Programme (FP7/2007-2013) within the project ICT ALPHA under grant agreement n° 212 352, in part by the U.K. Engineering and Physics Sciences Research Council under Grant EP/D036976, and in part by The Royal Society Brian Mercer Feasibility Award. The work of X.Q. Jin was also supported by the School of Electronic Engineering and the Bangor University.

I

Q

BER 1.0E-4500m MMF


- 54 -


POSTER Session

[P1] System Concept for 60 GHz DWDM-Radio-over-Fibre R. Herschel, N. Neumann, C.G. Schäffer Technische Universität Dresden, Dresden (Germany)

[P2] Colourless Radio over Fibre Access Network architecture using New RSOA devices for high performances G. de Valicourt1, M.A. Violas2, F. van Dijk1, D. Maké1, R. Brenot1 1 Alcatel-Thalès III-Vlab, Palaiseau (France) 2 Universidade de Aveiro, Aveiro (Portugal)

[P3] Improvements in the Radio-over-Fiber Transmission for Broadband Wireless Networks E. Udvary, T. Berceli Budapest University of Technology and Economics, Budapest (Hungary)

[P4] Bidirectional Multi-UMTS FDD Carrier Distribution over an Extended-Reach PON Architecture using a shared SOA F. Frank1, B. Charbonnier1, A. Pizzinat1, Ph. Chanclou1, C. Algani2 1 France Telecom, Lannion (France) 2 CNAM, Paris (France)

[P5] Radio over Fibre for the Support of 4th Generation Mobile/Wireless Communications N.J. Gomes1, S. Pato2, P. Monteiro2, A. Gameiro3

1 University of Kent, Canterbury (United Kingdom) 2 Nokia Siemens Networks S.A., Amadora (Portugal) 3 University of Aveiro (Portugal)

[P6] Research and development of wireline and wireless physical layer components and subsystems J. Bauwelinck, C. Mélange, X. Yin, B. Baekelandt, T. De Ridder, X.Z. Qiu, J. Vandewege

IMEC/INTEC, Ghent University, Gent (Belgium)


- 55 -


POSTER Session (cont'd)

[P7] All-Optical Mixer Based on an 850nm Emitting VCSEL Y. Ben-Ezra, M. Ran, B.I. Lembrikov, M. Haridim, A. Leibowitch Holon Institute of Technology, Holon (Israel)

[P8] The Possibility of UWB Signal Detection in a Thin SiGe Layer B.I. Lembrikov, Y. Ben Ezra Holon Institute of Technology, Holon (Israel)

[P9] Advanced phase detection for electro-optical phase-locked loop (EO-PLL) L.Naglic, L.Pavlovic, M.Vidmar University of Ljubljana, Ljubljana (Slovenija)

[P10] An Antenna-Integrated Photonic Millimeter-Wave Transmitter V. Rymanov1, M. Weiß1, A. Steffan2, S. Fedderwitz1, A. Stöhr1, D. Jäger1

1 University of Duisburg-Essen, Duisburg (Germany) 2 u2t Photonics AG, Berlin (Germany)

[P11] Millimeter-wave frequency generation with dual-wavelength DFB laser S. Ginestar1, F. van Dijk2, A. Accard2, O. Legouezigou2, F. Poingt2 F. Pommereau2, L. Legouezigou2, F. Lelarge2, B. Rousseau2, J. Landreau2, J-P. Vilcot1, G.H. Duan2

1 Institut d'Electronique, de Microelectronique et de Nanotechnologie, Villeneuve d'Ascq (France) 2 Alcatel-Thalès IIIVLab, Marcoussis (France)

[P12] Performance of 802.11g signals over a multimode fibre-fed distributed antenna system incorporating optical splitting L.C. Vieira, A. Nkansah, P. Assimakopoulos, N.J. Gomes University of Kent, Canterbury (United Kingdom)


- 56 -


POSTER Session (cont'd)

[P13] Static nonlinear distortion modelling of radioover-fibre systems L.C. Vieira1, N.J. Gomes1, A. Nkansah1, F. van Dijk2, A. Enard2, F. Blache2, M. Goix2

1 University of Kent,Canterbury (United Kingdom) 2 Alcatel-Thalès III-V Lab, Palaiseau (France)

[P14] Dynamic Binary Interleaving Codes for ECMA-368 standard and future Multi-Gigabit Wireless Systems I. Siaud, A.M. Ulmer-Moll Orange-Labs, Cesson-Sévigné (France)

[P15] Line-of-sight Infrared Wireless Path Loss Simulation in an Aircraft Cabin S. Dimitrov1,2, R. Mesleh1, H. Haas1,3, M. Cappitelli2, M. Olbert2, E. Bassow4

1 Jacobs University Bremen (Germany) 2 EADS Deutschland GmbH, Hamburg (Germany) 3 Institute for Digital Communications, University of Edinburgh (United Kingdom) 4 Airbus Deutschland GmbH, Hamburg (Germany)

[P16] Nonlinear optical Ti:PPLN wavelength conversion modules for free-space communication at 3.8 μm

K.-D. Büchter, C. Langrock, H. Herrmann, M.M. Fejer, W. SohlerUniversität Paderborn, Germany, Stanford University, USA


- 57 -


- 58 -

System Concept for 60 GHz DWDM-Radio-over-Fibre R. Herschel1,*, N. Neumann1, C. G. Schäffer1,

, …. 1 Technische Universität Dresden, Communications Laboratory, Chair of Microwave Technology/Photonics,

Helmholtzstraße 18, 01062 Dresden, Germany * Corresponding author: Phone: +49 (0)351 463 38759, Fax: +49 (0) 351 463 37163,

E-Mail: [email protected]

Abstract – The paper presents a concept for a DWDM Radio-over- fibre system based on an optical frequency comb source and periodic optical filtering, including phase correlated uplink carriers for optical downconversion reducing system complexity and enabling the step beyond 100GHz.

Introduction – A lot of research concerning radio over fibre systems in the 60 GHz range was done in the last years [4]. Due to low losses and reduced dispersion penalty these fibre optic systems show a far better performance for transmitting RF signals in the higher GHz range than classical copper wire systems. Most systems suffer from high costs for generating RF carriers and optical processing at the base station, especially as the systems have to be designed for multi channel functionality. Therefore, the aim of the presented system is to make use of the spectral periodicity of the used components in order to realize a simultaneous carrier generation as well as optical processing for all channels in a DWDM RoF system at the central unit. The paper provides a general description of the entire system in the following paragraph while each part such as filtering at the central unit, the structure of the RF base stations as well as the optical downconversion of the uplink signal at the central unit are described in further detail in separate paragraphs.General system design – As shown in figure 1 an optical frequency comb [1], [2] is used to generate the carriers with the appropriate spectral distance. By using a periodic filter with a high finesse the frequency lines with the desired spectral distance are selected out of the spectrum. Every second carrier is then modulated with a data signal and transmitted over SMF. At each base station an Add-Drop-Multiplexer (ADM) is used to select one pair of the frequency component modulated with the desired data signal and an unmodulated carrier signal as phase reference for optical heterodyning as already shown for single carrier systems [3]. A fraction of the reference carrier is used as optical carrier for the up-link signal at every base station. At the central unit the uplink signal can then be down-converted optically due to the phase correlation between the up-link carrier and the frequency components of the optical frequency comb.

Figure 1: Block diagram of DWDM system concept

Optical filtering and modulation – One key aspect is taking advantage of the spectral periodicity of the used filters. As shown in figure 2 the spectrum generated by the optical frequency comb generator is filtered in a first step by a 60 GHz FSR high finesse fabry-perot filter to achieve the 60 GHz grid for the downlink carriers. The resulting comb is then divided into two interleaving 120 GHz spectra where one is split by an AWG and the channels modulated separately whereas the second spectrum remains unmodulated as phase reference for the optical heterodyning [3]. An optical N-to-1 combiner couples the modulated as well as the reference carriers into one SMF.

Optical comb generation

Optical filters and carrier modulation

RF Wireless Base Station

RF Wireless Base Station

Optical downconversion ADM ADM

60 GHz

Central Unit


- 59 -

Figure 2: Block diagram of optical filtering and modulation

Base Station design – In order to reduce the base station’s complexity a fraction of the reference carrier needed for optical heterodyning in the photodiode is split off by a Fibre Bragg Grating (FBG) and reused as carrier for the optical uplink. This approach replaces the optical source at the base station and ensures phase correlation between the frequency comb at the central unit and the uplink signal needed for optical downconversion.

Figure 3: Scheme of the Wireless Base Station

Optical downconversion – Especially in systems operating in the higher GHz range the down-conversion to baseband needs high bandwidth components adding much complexity to the system. One of the main advantages of the system concept is the fixed phase correlation between the frequency comb source and the uplink carriers. This feature enables direct mixing with the appropriate comb lines in the photodiode for RF frequencies matching the grid of the optical frequency comb. Depending on the comb grid spacing this type of down-conversion promises flexibility in terms of the uplink carrier frequency.

Figure 4: Block diagram of optical down-conversion

Conclusion – The paper presents a concept for a DWDM Radio-over-fibre system working at 60 GHz carrier frequency. This system was designed for cost-effective realisation of a multi-channel system by utilisation of the periodicity of a frequency comb source as well as the optical filtering. The reuse of the unmodulated phase reference carriers for the optical downlink lowers the complexity of every base station by replacing the optical source needed for classical uplink realisations. Due to the phase coherence of the optical source at the central unit and the uplink carriers the down-conversion to base band can be shifted to the optical domain in order to lower the bandwidth requirements for the electrical components for receiving the uplink data signal. First experiments on frequency comb generation with a comb grid spacing of 10 GHz as well as carrier filtering by a fabry-perot filter with 40 GHz FSR have been performed successfully while further experimental work has to be invested on phase stabilisation of the optical frequency comb. Also the concept was validated by simulation results from VPIphotonics™. This concept is meant to be a promising option for high data rate systems enabling the use of complex data formats and being suitable for next generation RoF-networks in the 120 GHz domain and above.

References [1] P. Shen, N. J. Gomes, P. A. Davies, “Generation of 2 THz

Span Optical Comb in a Tunable Fiber Ring Based Optical Frequency Comb Generator”, IEEE International Topical Meeting on Microwave Photonics, pp .45-49, 2007

[2] Keang-Po Ho, J. M. Kahn, “Optical Frequency Comb Generator Using Phase Modulation in Amplified Circulating Loop”, IEEE Photon. Technol. Lett., vol. 5, no. 6, pp. 721-725, June 1993

[3] I. González Insua, K. Kojucharow, C. G. Schäffer: "MultiGbit/s transmission over a fiber optic mm-wave link", International Microwave Symposium 2008 in Atlanta, USA,paper no. WE3D-02

[4] J. J. Vegas Olmos, Takahiro Sono, Kazunori Tamura, Hiroyuki Toda, Ken-ichi Kitayama, “Reconfigurable 2.5- Gb/s Baseband and 60-GHz (155-Mb/s) Millimeter- Waveband Radio-Over-Fiber (Interleaving) Access Network”, J. Lightw.Technol., vol. 26, no. 15, pp. 2506– 2512, Aug. 2008

FP-Filter FSR = 60 GHz

MZI FSR = 120 GHz

Downlink Datachannels

ADM

EOM

FBG (3dB)


- 60 -

Colourless Radio over Fibre Access Network architecture using New RSOA devices for high performances

G. de Valicourt1,*, M. A. Violas 2, F. van Dijk 1, D. Maké1 and R. Brenot1

1Alcatel-Thalès III-V labs, A joint Laboratory of "Alcatel Lucent Bell Labs" and "Thales Research & Technology"

Campus Polytechnique, 1, Avenue A. Fresnel, 91767 Palaiseau cedex, France2Instituto de Telecomunicações, Universidade de Aveiro, Campus Universitário, Aveiro 3810-193 , PORTUGAL

*Phone: +33 169 41 60 83, E-Mail: [email protected]

Abstract - We demonstrated that the use of a high gain reflective semiconductor optical amplifier as remote modulator antenna used at the Base Station (BS) for colourless operation.

1.Introduction – The development of higher capacity networks is required in order to satisfy the demandfrom the different communication technologies. Internet, wireless and mobile network are being widely used and present customers want to enjoy thesame quality for mobile multimedia service as fixednetworks. In this context, WLAN hot-spots or third generation of wireless networks (3G) are already able to establish a wireless personal area network. In order to obtain a wide service coverage area, many BS should be connected to a Central Station (CS) via an optical fiber network. Efficient architecture has been proposed in [1] using WDM techniques and allowing a RoFnet-Reconfigurable Radio over Fiber network. RSOA devices were used as a low cost solution with wide bandwidth at the BS in order to perform modulation and amplification functions [1-3]. RSOA devices with electrical bandwidth up to 1.5 GHz have been used [1] and Error-free transmissions of 1.25 Gbit/s have been demonstrated [2-4].

In this paper, we demonstrated the use of colourless RSOA device in a Radio over fiber AccessNetwork architecture driven by standard wifi 802.11g signal and also modulated at 5 Gbit/s with digital modulation. Uplink configuration exploits the capabilities of using RSOA in order to perform modulation and amplification for a OFDM modulation.

2. Network description and experimental set up – The schematic of radio over fiber network concept is represented in Fig. 1. N base stations are used to provide wireless connection with mobile users and BS are connected via optical fiber to the CS. WDM is used in order to provide high capacity network by allocating different wavelength to individual BS [5].

RSOA are perfect devices for low cost uplink configuration. They replace high cost WDM source at the BS and high speed external modulator.

Fig. 1.: Schematic diagram of radio over fiber Network architectureDue to their broad bandwidth characteristics, same devices can be used over a wide wavelength range, therefore they allow Colorless Radio-on-Fiber network with dynamic wavelength allocation.

In order to experimentally simulate the behaviour of a BS, a setup represented by figure 1 was implemented. It consists of an optical network which uses an AWG at the BS to separate the wavelength assigned to a particular BS. A vector signal generator (SMU 200A) provides the standard wifi signal at an IF carrier of 1 GHz, which modulates the RSOA through a linear power amplifier. An ECL (External Cavity Laser) is used to launch the required wavelength into the system. Thereceiver is composed of a photodiode and a low noise Mini-circuits amplifier connected to the Vector Signal Analyzer. All the other parts are a standardoptical circulator and optical attenuator.

3. Device characteristics – The InGaAsP active zone of the RSOA was grown using molecular beam epitaxy. AR (HR) coatings were deposed on the angled front (real) facet. The RSOA structure has been designed to achieve a small optical confinement (Γ~20%) which ensures lower noise figure, less gain ripple and wider optical bandwidth compared to


- 61 -

larger optical confinements. Polarization independent optical gain values up to 30 dB were observed in the small signal regime for devices with a length of 500 µm and 700 µm respectively, and the optical gain bandwidth is 50 nm typically. Dynamic measurements performed at 1558.9 nm show a modulation bandwidth up to 3GHz for both RSOA (Fig. 2.).

Fig. 2.: RSOA direct modulation bandwidth

4. Experimental measurements – A 700 µm cavity length RSOA is used in the experimental setup explained in part 2. The RSOA was driven by a 80 mA DC current and standard IEEE802.11wifi signal with a 64QAM modulation format OFDM multiplexing at 54 MB/s. CW input optical power of –14 dBm and –7 dBm were used. In Figure 3, EVM measurements are presented at 1530 nm and frequency subcarrier of 1 GHz as a function of the input electrical power.

Fig. 3.: EVM values as a function of the electrical RFFigure 4 displays the eye-diagram at the output of the RSOA for a direct digital modulation (pseudo aleatory scheme 231-1) at 5 Gbit/s.

Fig. 4.: 5 Gbps eye-diagram at the output of the RSOA.

EVM measurements show the feasibility of this configuration using Reflective SOA and on a wide range of input power (from –22 dB to17 dB).

5. Conclusion – A cost effective WDM Radio over fiber network is presented. The large gain of the RSOA and also its high output power has also permitted to obtain a suitable transmission over fiber from a RF signal. This upstream colourless result allows investigating this solution to achieve in the trunk line a wavelength multiplex of several next generation radio over fiber solution solutions. Acknowledgement – This work was supported by ICT-FUTON projects.

References

[1] M. C. R. Medeiros, R. Avó, P; Laurêncio, N. S. Correia, A. Barradas, H. J. A. da Silva, I. Darwazeh, J. E. Mitchell and P. M. N. Monteiro, ICTON Mediterranean Winter Conference, pp. 1-5 (2007).

[2] Yong-Yuk Won, Hyuk-Choon Kwon, Sang-KooK Han, “ 1.25 Gib/s Wavelength Colorless Radio-on-Fiber Systems Using Reflective Semiconductor Optical Amplifier”, journal of lightwave tevhnology, Vol. 25, No. 11 (2007).

[3] Maria C.R. Medeiros, Ricardo Avo, Paula Laurencio, Noelia S. Correia, Alvaro Barradas, Henrique J.A. daSilva, Izzat Darwazeh, John E. Mitchell and Paulo M. N. Monteiro,“RoFnet-Reconfigurable Radio over Fiber Network Architecture Overview”

[4] N.Calabretta, G. Carvalho Kassar and I. Tafur Monroy, Electronics Letters, Vol. 43, No. 22 (2007).

[5] H. Bolcskei, A. J Paulraj, K. V. S. Hari, and R. U. Nabar, “Fixed broadband wireless access: State of the art, challenges, and future directions”, IEEE Commun. Mag., Vol. 39, No. 1, pp. 100-108 (2001)


- 62 -

Improvements in the Radio-over-Fiber Transmission for Broadband Wireless Networks

Eszter Udvary, Tibor Berceli Budapest University of Technology and Economics, Budapest, Hungary

[email protected], [email protected]

Abstract - An approach is presented to reduce the chromatic dispersion-induced distortions in radio-over-fiber links carrying combined baseband and millimeter wave signals. This new method is based on the interplay of intensity dependent phase modulation in SOAs and fiber nonlinearities.

Introduction - The microwave/mm-wave optical trans-missions are severely deteriorated by the chromatic dispersion of the optical link using Standard Single Mode Fiber (SSMF) near 1550nm [1, 2]. This limitation is mainly caused by the RF carrier suppression effect due to dispersion-induced sideband cancellations at certain combinations of microwave/mm-wave frequencies and propagation distances. Several techniques have been proposed to overcome this effect like optical single sideband (SSB) modulation [3], chirped fiber gratings [4], fiber self-phase modulation [5], dual mode lasers [6], etc. A new approach will be presented to overcome dispersion-induced effects and adjusting the chirp and phase modulation generated by saturated semiconductor optical amplifiers.

Fiber nonlinearities - With the increase of input optical power the fiber non-linearity can no longer be neglected. At a specific photon flux density, self-phase modulation (SPM) can become significant resulting in distortions. High input intensity modifies the transfer function of the fiber and can compensate modulation suppression caused by dispersion [7]. As the signal intensity increases the notches caused by dispersion shift to higher modulation frequencies (see Fig. 1). The measurement results show that the fiber-induced SPM hardly influences the frequency notches even at relatively low powers.

Fig. 1 Measurement results over 30 km fiber at four different average input intensities SOA based dispersion compensation - When the incoming power of the semiconductor optical amplifier (SOA) is intensity modulated, the optical gain is affected in both magnitude and phase via the modulation of the complex refractive index caused by the electron density changes in a

SOA device. Consequently, in SOA the optical signal becomes amplitude modulated (AM) and phase modulated (PM). It is fundamental to know the behavior of the refractive index within the active region. That can be modeled using the Linewidth Enhancement Factor (LEF = Henry factor = α factor) approximation. LEF is a function of bias current, wavelength and input optical power [8]. As the optical input power (Pin) increases, carrier depletion occurs in SOA and this induces gain saturation. In optical amplifiers under saturation conditions, an increasing input intensity causes a decrease in the amplifier gain.

The chirping parameter which is positive for light sources and unsaturated optical amplifiers, is negative for saturated amplifiers [9]. Fig. 2 represents the optical gain and the LEF dependence on the input optical power. When the input power becomes larger, the chirp parameter falls to a negative value.

Fig. 2 Optical gain saturation and the calculated chirp

The negative chirp of saturated SOA cancels the positive chirp-parameter of the optical modulator, in such way enhances the transmission distance and operating frequency. Furthermore the optical amplification causes RF signal gain, too [10]. However the SOA adds significant noise to the system. The negative chirp affects both sidebands and then causes the asymmetrical optical power between the sidebands. As a result, the RF carrier suppression effect is reduced.

Experimental work was performed in the laboratory over different lengths of single mode fibers. Fig. 3 shows the simplified measurement setup. The SOA under test worked with different bias (DC) currents. The polarization state of the incoming optical power was set by a polarization controller.

0

5

10

15

20

25

30

-10 -5 0 5 10 15 20 25 30

Input Optical Power [dBm]

Opt

ical

Gai

n [d

B]

-20

-15

-10

-5

0

5

10

LEF

(Hen

ry fa

ctor

) [aU

]

3dB

Psat

-60

-50

-40

-30

-20

-10

0

10

20

5 7 9 11 13 15Modulation Frequency [GHz]

Det

ecte

d Su

bcar

rier [

dBm

]

16 dBm11 dBm 6 dBm 0 dBm

Fiber length=30 kmWavelength=1550nm

Input Optical power


- 63 -

Fig. 3 Simplified experimental setup

Fig. 4 Measured RF response with SOA, normalized to back-to-back optical link parameters

The required optical power and wavelength were produced by a tunable laser source. The intensity modulated optical signal was detected by a photo-detector. The harmful effect of the optical reflection was eliminated by optical isolators. The setup was controlled by a computer program, hence the measurement parameters were carefully set by the program and the measurement results were processed and stored.

The RF response was measured with different parameters (see Fig. 4). As the SOA bias current (optical gain) increases, the frequency notches of the RF response are reduced and shifted to higher modulation frequencies.

Fig. 5 Simulated eye diagram, without and with SOA compensator, 400km optical link, subcarrier frequency 3.2GHz, modulation bandwidth 512MHz

Based on the results, we conclude that the interplay of chirp generated by the saturated SOA and the chromatic dispersion of the fiber enables a significant reduction in the dispersion-induced effects.

In radio over fiber systems the radio frequency carrier is modulated by digital information. The above presented dispersion compensation technique affects the eye diagram and the BER of the modulation signal. If the subcarrier frequency is

near to a frequency notch caused by the dispersion, the eye diagram closes and the communication deteriorates or lost. Applying the optimized SOA compensator the eye diagram opens and the BER is improved (see Fig. 5).

Conclusions - The optical fiber is more and more employed all the way to the home to enable broadband connections. The microwave and millimeter-wave optical transmissions are limited by the fiber chromatic dispersion. To overcome the RF carrier suppression effect an approach has been proposed based on the joint effect of SOA chirp, chromatic dispersion and fiber nonlinearities. The results show that the fading of the RF-to-RF system response can be significantly alleviated. Hence the transmitted digital signal performance can be improved. Acknowledgement - The authors acknowledge the EU Network of Excellence project called ISIS (IST-FP6-26592) for the support to their research work.

References [1] W. van Etten, J. van der Plaats, “Fundamentals of Optical

Fiber Communications”, Prentice Hall Int. Series in Optoelectronics, pp. 62-68, UK, 1991.

[2] H. Schmuck, “Comparison of optical millimetre-wave system concepts with regard to chromatic dispersion”, Electronics Letters, Vol. 31, No. 21, pp. 1848-1849, October 1995.

[3] G.H. Smith, D. Novak, “Broad-band millimeter-wave (38 Ghz) fiber-wireless transmission system using electrical SSB modulation to overcome dispersion”, Photonic Techn. Letters, Vol. 10, No. 1, pp. 141-143, Jan. 1998.

[4] J. Marti et al.: “Experimental reduction of chromatic dispersion effects in lightwave microwave/millimeter-wave transmission using tapered linearly chirped fiber gratings”, Electron. Lett., Vol. 33, No. 13, 1997

[5] V. Polo, J. Marti, F. Ramos: “Mitigation of chromatic dispersion effects employing electroabsorption modulator-based transmitters”, IEEE Photon. Technol. Lett., Vol. 11, pp. 883–885, July 1999.

[6] D. Wake, C.R. Lima, P.A. Davies : “Optical Generation of Millimeter-Wave Signals for Fiber-Radio Systems Using a Dual-Mode DFB Semiconductor Laser”, Tans. MTT, Vol. 43, No. 9, pp. 2270-2276, September 1995.

[7] Z. Várallyay, I. Frigyes, O. Schwelb, E. Udvary, L. Jakab, P. Richter, “Soliton Propagation of Microwave Modulated Signal through Single-Mode Optical Fiber”, Acta Physica Hungarica B), Quantum Electronics, Akadémiai Kiadó (Academic Press), Vol. 23, No. 3-4, pp. 175-186, 2005

[8] L. Occhi, L. Schares, G. Guekos: “Phase Modeling Based on the α Factor in Bulk Semiconductor Optical Amplifiers”, IEEE Journal of Selected Topics In Quantum Electronics, pp. 788-797, 2003.

[9] T. Watanabe, N. Sakaida, H. Yasaka, F. Kano, M. Koga, “Transmission performance of chirp-controlled signal by using semiconductor optical amplifier”, IEEE Journal of Lightwave Technology, pp. 1069-1077, August 2000.

[10] Sang-Yun Lee, Bon-Jo Koo, Hyun-Do Jung, and Sang-Kook Han, ”Reduction of chromatic dispersion effects and linearization of dual-drive Mach-Zehnder Modulator by using semiconductor optical amplifier in analog optical links” in Proc. ECOC 2002, 28th European Conf. Optical Comm., Copenhagen, Denmark, September 8-12, 2002.

-80-70-60-50-40-30-20-10

0102030

2 3 4 5 6 7 8 91st order of modulation frequency, fRF [GHz]

Nor

mal

ised

Tra

nsm

issi

on [d

B]

SOA bias current

without SOA

SOA bias current=400mA, 200mA, 125mA, 100mAOptical Gain=15dB, 13.5dB, 4.5dB, -9.5dB

Fiber length=50kmWavelength=1550nm Reference fiber=4m

MZ SSMF

O/E Spectrum analyzer LD SOA

RF carrier

Microwave sweep

generator

Pol. contr.

Computer

Current source

Bias

Temperature controller


- 64 -

Bidirectional Multi-UMTS FDD Carrier Distribution over an Extended-Reach PON Architecture using a shared SOA

F. Frank1,*, B. Charbonnier1, A. Pizzinat1, Ph. Chanclou1, C. Algani2

1 France Telecom, Orange Labs, RD/RESA/ANA/ASHA, 2 av. P. Marzin, 22307 Lannion, France 2 CNAM, ESYCOM (EA2552), Communications Systems, 292 rue St-Martin, 75003 Paris, France

* Corresponding author: Phone: +33 296 051 175, Fax: +33 296051723, E-Mail: [email protected]

Abstract - We demonstrate the distribution of a multiplex of UMTS RF carriers over extended reach PON architectures for overall optical budgets up to 47 dB. A SOA and an APD are respectively used for the reach extension and the photodetection.

Introduction – Today, there are more than 1.3 million FTTH/B customers throughout Europe [1] and this figure is expected to rise to 14.5 millions by 2013 [2]. Similarly to what is currently seen is Asia FTTH/B will gradually replace DSL connections. However, FTTH connection costs are estimated to be around 1500 in most densely populated areas (i.e. Paris) and rising to 2500 for average density areas [3]. This is the most expensive connectivity solution among all the others (WiFi Hot spots, xDSL, WiMAX, etc…) and 70% of this sum is engineering work for the actual fiber installation. Nevertheless, it is the technology of choice by all operators because of the large bandwidth provided and its ability to meet all potential future bandwidth demands. For instance, IDATE has identified in 2007, 201 optical fibre current deployments in Europe [4]. In parallel, optical infrastructures are also deployed mainly for upgrading the mobile telephony network to 3G (UMTS) and 3.5G as there are now more than 116 million 3G users in Europe [1]. It is then key for operators to study ways of sharing the optical infrastructure investments between mobile telephony networks and access networks.

The goal of this paper is to demonstrate experimentally the possibility of transporting UMTSFDD (Frequency Division Duplexing) signals over current and more importantly over future GPON architecture (Extended reach) so as to validate this infrastructure sharing concept. Hence, we describe in this paper a system for distributing and collecting

simultaneously three UMTS Band I RF signals over an Extended Reach Passive Optical Network (ER-PON) Architecture using radio-over-fiber techniques. Experimental results are reported. The reach extension is realized by an in-line Semiconductor Optical Amplifier (SOA) which is shared between the uplink and downlink radio signals each carried by a dedicated wavelength.

Radio Set-up – The used UMTS signal is a Wideband–Code Division Multiplex Access (W-CDMA) QPSK signal at 3.84MChip/s composed of the five Physical Channels composing the Test Model4 of [5] which is specified by 3GPP for EVM (Error Vector Magnitude) measurement purposes.

The above mentioned UMTS baseband signal is processed offline on a computer using Matlab® in order to be generated on a dual output Arbitrary Waveform Generator (AWG). Each output of the AWG represent the I and Q components of a signal composed of a multiplex of three 10 MHz spaced UMTS carriers. After frequency up-conversion using a Vector Signal Generator (VSG#1, 2) the three carriers are centered around 1940 MHz for the uplink (UL) and 2140 MHz for the downlink (DL). Finally a Low Noise Amplifier (LNA) is used as a Laser Driver and is followed by a UMTS Band I Diplexer (DL and UL BPF on Figure 1) to filter out the noiseand unwanted mixing products. All three carriers are always maintained with the same RF power.

For the experiment, each UMTS RF signal power can

AWG

VSG #1

VSA VSG #2

VSADL BPF{I, Q}

{/I, /Q}

VOA #1

VOA #2 SOA

20km FO

APD + TIADFB #1

LNA

Remote Antenna SiteAccess SectionTransport Section

Reach Extension

Site

Circ.#1 Circ.#2

Central Office containing Radio Base Band Unit

UL BPF

DL BPF

DFB #2

UL BPF

Towards Antenna

LNA

Figure 1 : Radio & Optical Set-up


- 65 -

be swept from -40 to +15 dBm/carrier. Performance evaluation is done by a Vector Signal Analyzer (VSA) through measurement of the Error Vector Magnitude (EVM).

Optical Set-up – The optical path of an ER-PON can be divided into a Transport Section (starting at the Central Office) and an Access Section (finishing atthe Remote Antenna Site), the Reach Extension Site – made of a single SOA – lying in between. The Transport Section is represented by pure optical attenuation corresponding to a 17dB budget, whereasthe latter is composed by 20km of optical fiber andoptical attenuation representing a typical PON Class B+ budget of 28dB. Here we used cooled DFB lasers emitting each ~ +9dBm of optical power at 1551nm and 1546nm for the downlink and the uplink respectively. The different wavelengths are separated by optical circulators each having an insertion loss of ~0.8dBand which are not included in the PON budget calculations. The SOA has a gain of 14dB. For practical purposes the APD and the VSA (shown dotted on Figure 1) are permuted for realizing the up- and downlink measurements, however optical signals remain unchanged, thus simultaneity is conserved.

Results – For an overall optical budget of 47dB, the EVM is plotted against the received RF power (in a 5 MHz band) for each one of the three multiplexed carriers. The EVM measurements are referenced to electrical back to back evaluations in order to differentiate between the distortions originating from the optoelectrical conversions and the optical amplification from the ones originating from our test equipment.

In the downlink direction, Figure 2 shows that an EVM power margin (calculated as the difference between the maximum and minimum RF power to achieve the UMTS EVM limit value of 17.5%) of

25dB can be achieved. However this limit of 17.5% is specified to be the maximum EVM to be received at the end of the air link. Thus in our case considering an EVM target of 8% at the output of the radio-over-fiber-system is more realistic since the RF signalsstill have to propagate through the air link. In this case we still have a power margin of 18dB.

In the uplink direction, an EVM power margin - for a limit at 17.5% - of 26dB can be observed on Figure 3.

In this case we do not consider the 8% threshold aswe measure the signal after is would have propagated in the air.

Conclusion – We prove the capability of transporting transparently and simultaneously a Multiplex of three UMTS RF FDD carriers per direction over an Extended Reach PON architecture with an optical budget of 47 dB relying on the use of an SOA for reach extension and an APD for photodetection.

This work validates that current and next generation PON infrastructures can be shared with the UMTS FDD mobile networks' backhaul infrastructure leading to potentially large investment savings.

Acknowledgement – The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7) under project 212 352 ALPHA “Architectures for fLexible Photonic Home and Access networks”

References[1] Didier Pouillot, "World Market for Telecom Services",

IDATE, December 2008.

[2] FTTX Watch Service, IDATE July 2008

[3] Roland Montagne, "FFTH deployments: what solutions to lower the costs?", IDATE, August 2008.

[4] Roland Montagne, "FFTx Market Watch 2008", IDATE, March 2008.

[5] TS 25.141 – Base Station (BS) Conformance Testing (FDD)

[6] TS 25.104 – BS radio transmission and reception (FDD)

[7] TS 25.101 – UE radio transmission and reception (FDD)

Fig. 2: UMTS RF carrier multiplex on Downlink Wavelength

Fig. 3: UMTS RF carrier multiplex on Uplink Wavelength

0

2

4

6

8

10

12

14

16

18

20

-55 -53 -51 -49 -47 -45 -43 -41 -39 -37 -35 -33 -31 -29 -27 -25 -23 -21

Received RF Power per UMTS carrier (5MHz)[dBm]

Co

mp

osi

te E

VM

%

EVM @ 2130 MHzEVM @ 2140 MHzEVM @ 2150 MHzEVM @2130MHz at launchEVM @2140MHz at launchEVM @2150MHz at launch3GPP EVM Spec. Limit%

0

2

4

6

8

10

12

14

16

18

20

-48.00 -44.00 -40.00 -36.00 -32.00 -28.00 -24.00 -20.00 -16.00Received RF Power per UMTS carrier (5MHz)[dBm]

Co

mp

osi

te E

M (%

)

EVM @1930 MHzEVM @ 1940 MHzEVM @ 1950 MHzEVM @1930MHz at launchEVM @1940MHz at launchEVM @1950MHz at launch3GPP EVM Spec. Limit


- 66 -

Radio over Fibre for the Support of 4th Generation Mobile/Wireless Communications

N.J. Gomes1,*, S.Pato2, P. Monteiro2,, A. Gameiro3

1 University of Kent, Dept. of Electronics, Canterbury, United Kingdom 2 Nokia Siemens Networks S.A., Amadora, Portugal

3 University of Aveiro, Instituto de Telecommunicações, Aveiro, Portugal * Corresponding author: Phone: +44 1227 823719, Fax: +44 1227 456084, E-Mail: [email protected]

Abstract – The main aims of the FUTON EU project are presented. Progress in the radio over fibre work package during the first year of the project is outlined and the key achievements are summarised.

Introduction – It is expected that 4th generation (4G) mobile/wireless systems will be specified with bit-rates up to 1 Gb/s. To cope with such increased bandwidth demands in a cost-effective manner requires changes in system architecture. In this context, the FUTON project [1] proposes the development of a hybrid fibre-wireless infrastructure for the transparent interconnection of multiple remote access units (RAUs) to a central unit (CU), which is responsible for the joint processing of all radio signals, as depicted in Fig.1. This architecture allows the development of virtual MIMO concepts to achieve broadband wireless transmission, and inter-cell interference cancellation, fulfilling the objectives specified for 4G systems.

The general aims and objectives of the FUTON project have been presented elsewhere [2]. In this paper, the discussion will focus on the radio over fibre infrastructure (RoF) being proposed within the FUTON project. Previously, radio over fibre has been proposed for the interconnection of simplified RAUs in micro-cellular environments [3], or for antenna remounting to eliminate dead-spots. In FUTON, however, the RoF infrastructure is a key enabler of the wireless technologies being proposed, such that the vision shifts from a remoting technique to an aggregating one.

Fig.1: FUTON joint processing of radio signals.

To meet these objectives, the radio over fibre infrastructure needs to support the transmission of multiple 4G signals to/from the RAUs, with each 4G signal occupying a bandwidth of 100 MHz, operating at around 3.5 GHz and employing high-level QAM for the highest bit-rates (up to 256-QAM assumed), and large numbers of OFDM subcarriers (up to 2048 assumed). These specifications follow those typically proposed for 4G systems. For wireless transmission over ranges of several hundred metres, these are very demanding specifications [4]. High dynamic range analogue radio over fibre links, relying on high performance but low-cost optoelectronic components will be necessary. In addition, to minimize costs, it is desirable to share deployments with fixed (wireline) infrastructures, such as passive optical networks (PONs). The work in the radio over fibre work package in FUTON covers studies over the full range of activity areas from component development, to link analysis and design, through to access network architecture and topology proposals.

Progress and main achievements – The work in the radio over fibre work package has been divided into several tasks covering the definition of the optimum hybrid optical-radio infrastructure (including the topology/architecture studies and the link performance analysis), the development of optoelectronic components (with specifications derived from the link analysis), subsystem design and integration for final testing. Work on the latter tasks has only just commenced, and in this paper we concentrate on the progress and achievements in the former areas.

The FUTON radio-optical infrastructure uses a combination of subcarrier multiplexing (SCM) with frequency translation of the signals (necessary as several signals to/from RAUs occupying the same RF spectrum must be transported) and coarse wavelength division multiplexing (CWDM), most likely for the


- 67 -

addressing of individual RAUs. The use of CWDM enables the use of less expensive and uncooled components. A bidimensional signal space is defined with RAUs addressed by wavelength and individual signals for the RAUs separated in the electrical domain by SCM. By defining the appropriate granularity, both new RAUs and additional wireless systems can be easily added. The current proposal for the support of the 4G wireless system only, uses 16 CWDM channels to support 8 RAUs for each joint processing unit, as shown in Fig.2. Space division multiplexing or more dense forms of WDM may be necessary for the support of additional systems or RAUs.

Fig. 2: Proposed topology using 16 CWDM channels to support up to 8 RAUs.

For the radio over fibre link analysis, the work has investigated the requirements for digital as well as analogue link types. This is very important as specifications for “digital” base stations with remote radio heads have already been made for 3rd

generation (3G) mobile and WiMAX systems, and are currently being extended for LTE. The work has shown that very high bit-rate transmission and time-division multiplexing equipment will be necessary to cater for the 4G systems with digital links, with the costs for the FUTON system being an order of magnitude more expensive than the use of analogue links [4]. The analogue links are dependant on directly modulated laser performance close to the state-of-the-art in terms of slope efficiency, linearity and noise and requiring some effort to obtain such performance in a single device, uncooled, stably over a range of operating temperatures, and at 1550 nm. Other analogue link types have been examined, in particular using reflective semiconductor optical amplifiers (RSOAs) for the uplink. The work on using these devices for analogue radio over fibre in their saturated region has shown improved performance over that previously reported world-wide [5]. Compensation techniques are also being investigated which may alleviate some of the more stringent demands placed on the device linearity [6]. Initial performance comparisons of analogue link types show that Mach-Zehnder modulator and

directly modulated laser links do not significantly affect performance compared to wireless transmission with no supporting radio over fibre link, and work is now progressing on the design of the RAUs and central unit radio-optical interfaces [7].

Conclusion – The work conducted during the first year of the FUTON project in the work package devoted to radio over fibre has been briefly reviewed. The work has resulted in architecture/topology proposals compatible with PON deployments, and link budget and cost analyses which demonstrate the advantages of the analogue radio over fibre approach compared to digital approaches, and that the radio over fibre transmission does not degrade the wireless ranges possible compared to the radio transmission on its own.

Acknowledgement - This work was carried out within the framework of the European Union Integrated Project FUTON (FP7 ICT-2007-215533). The authors are grateful to all of their collaborators in FUTON, particularly D. Wake, A. Nkansah (Univ. Kent), J. Pedro, J. Santos (NSN), M. Violas (Univ. Aveiro), F. van Dijk, R. Brenot, G. de Valicourt, C. Gonzalez (AT 3-5 Labs), E. Lopez and M. Lobeira (Acorde Technologies).

References[1] FUTON project website: www.ict-futon.eu[2] see, e.g., S. Pato, P. Monteiro, N.J. Gomes, A.

Gameiro, T. Kawanishi, “Next-generation distributed and heterogeneous radio architectures: the FUTON project”, accepted for publication at APMP 2009, Beijing, April 2009.

[3] see, e.g., A.J. Cooper, “Fibre/radio for the provision of cordless/mobile telephony services in the access network”, Electron. Lett., 26, pp. 2054-2056, 1990

[4] D. Wake, S. Pato, J. Pedro, E. Lopez, N.J. Gomes, P. Monteiro, “A comparison of remote radio head optical transmission technologies for next generation wireless systems”, submitted to ECOC 2009

[5] G. de Valicourt, R, Brenot, M. Violas, “Colourless radio over fibre access network architecture using new RSOA devices for high performances”, submitted to ECOC 2009

[6] A. Hekkala, M. Lasanen, L.C. Vieira, A. Nkansah, N.J. Gomes, “Architecture for joint compensation of RoF and PA with nonideal feedback”, submitted to PIMRC 2009

[7] see, e.g., D.Wake, A. Nkansah, N.J. Gomes, “Optical transmission link design for a broadband distributed wireless system”, this workshop.


- 68 -

Research and development of wireline and wireless physical layer components and subsystems

J. Bauwelinck*, C. Mélange, X. Yin, B. Baekelandt, T. De Ridder, X.Z. Qiu, J. Vandewege1 IMEC/INTEC, Ghent University, Sint-Pietersnieuwstraat 41, Gent, Belgium


Abstract – This paper presents an overview of the various research activities at INTEC_design. In this lab, high-frequency electronic components and subsystems are designed for the physical layer of emerging wireline and wireless applications.

Introduction – INTEC_design is one of the research groups of the Department of Information Technology(INTEC) at Ghent University and an associated laboratory of IMEC [1].The INTEC_design lab is specialized in the design of high-frequency and high-speed electronic and opto-electronic circuits and systems. It has gained extensive experience by the study and development of innovative physical layer components and subsystems, and associated instrumentation for wireless, wired and broadband optical accessnetworks.The development of such prototypes and demonstrators requires the ability to trade off system level requirements with implementation details and a combination of various skills in a multidisciplinary team:

System modeling, architecture partitioning and subsystem specification.Board level design of complex mixed analog/digital systems using commercial off-the-shelf components, taking care ofsignal integrity challenges.Transistor level circuit design on chip in CMOS and BiCMOS technologies. INTEC_design is one of the few university labs certified by ST microelectronics.Package and interconnect modeling,including 3D electromagnetic simulation.Experimental characterization of high-frequency, low-noise, and high-linearity circuits and systems.

This broad expertise is the basis for the successful development of various applications, beyond the state-of-the art.

Ranging-enabled RF transceivers – Well-established technologies for position determination are mostly addressing outdoor scenarios, such as global positioning system (GPS). Indoor, a high number of reflections and often obstructed lines-of-

sight, complicate the time-of-arrival estimation of the transmitted signal.A proof-of-concept ranging transceiver, shown in Fig. 1, was developed comprising both the analog front-end and the digital signal processing in FPGA.In a number of field trials in warehouse-like environments and harsh industrial environments this prototype showed a significant improvement in precision (<1m) compared to existing technologies.

Fig. 1.: PCB demonstrator of a 2.4GHz ranging transceiverOn-going research

Automotive optical networks – Fiber optic networks will bring many advantages into the automotive environment, such as lower weight, higher data rates and improved electromagnetic compatibility. A number of transmitters (VCSEL drivers) and receivers that comply with the MOST standard were developed for POF and PCS fibers. Ongoing research targets higher data rates (150Mbps, 1Gbps). An important challenge in this application is the extreme low cost requirement. For this reason, the photodiode was integrated into the silicon receiver chip, as shown in Fig 2(a).

Broadband over power line –Broadband over power lines (BPL) aims to offer low-cost broadband communications over the ubiquitous power grid. Within the IST project POWERNET, INTEC_design was mainly responsible for the specification and development of the analog front-end transceiver chip, shown in Fig. 2(b) [2].


- 69 -

(a) (b)Fig. 2.: Die micrographs (a) a Silicon photodiode

monolithically integrated with a transimpedance amplifier (b) BPL analog front-end transceiver chip

This novel design pushed the BPL frequency range from 30MHz up to 60MHz and achieved the highest dynamic range (99.5dB) available to date. The chips were integrated in modems, and evaluated in field trials. These new modems outperformed state-of-the-art commercial modems, and the technologycomplies with the emerging IEEE P1901 standard.

Passive optical networks (PONs) – PONs are one of the most promising solutions for fiber-to-the-premises (FTTP) applications, as feeder and central office costs can be shared by multiple customers. Within the framework of the EU-funded FP6 IST PIEMAN (Photonic Integrated Extended Metro and Access Network) project an optically amplified, large-split, long-reach PON is being developed [3].

This PON offers a high capacity hybrid wavelengthdivision multiplexing/time division multiple access (WDM/TDMA) physical layer, integrating access and metro networks into one system. Within thisframework three 10Gbps burst-mode (BM) ICs with advanced features have been designed by INTEC_design, shown in Fig. 3.Beyond PIEMAN, INTEC_design continues the development of an advanced APD-based BM-TIA with high sensitivity and a wide dynamic range for a symmetric 10Gbps PON. This activity aims to comply as much as possible with the emerging 10Gbps ITU-T 10G-XGPON2 standardization.Furthermore, INTEC_design is involved in the EU-funded FP7 ICT EURO-FOS Network of Excellence[4]. This is a perfect platform to continue our R&D on the design of innovative 10Gbps BM ICs and their subsystem integration.

Acknowledgement - This work was supported in part by the Flemish Government under the IWT research contracts ELOCA, OptoCMOS and AutoFun, and in part by the European Commission under the research contracts POWERNET, PIEMANand EURO-FOS.

References[1] www.intec.UGent.be/design[2] www.ist-powernet.org[3] www.ist-pieman.org[4] www.euro-fos.eu

Fig. 3.: 10 Gbps burst-mode OLT prototypes developed in the IST PIEMAN project: BM-transimpedance amplifier (BM-TIA), BM-limiting amplifier (BM-LA), BM Clock and Data Recovery (BM-CDR)


- 70 -

1

All-Optical Mixer Based on an 850nm Emitting VCSEL Y. Ben-Ezra1, M. Ran1, B.I. Lembrikov1*, M. Haridim1, A. Leibowitch1

1 Holon Institute of Technology (HIT), Department of Engineering, 52 Golomb Str., Holon 58102, Israel * Corresponding author: Phone: +972 3 5026684, Fax: +972 3 5026685, E-Mail: [email protected]

Abstract - We present an all-optical mixer for up-conversion of MB OFDM UWB signals, based on VCSEL non-linearity. The measurements show that low conversion loss values of 7dB and low EVM values can be achieved.

Chapter 1 – Radio over fiber (RoF) systems are incorporated in many desirable communication applications, such as signal processing of microwave signals and photonic integrated circuits. Transmission of ultrawide band (UWB) signals over fiber can provide an effective solution for high rate data links in indoor networks. Among other types of UWB technologies, the multi-band orthogonal frequency division multiplexing (MB OFDM) is the most promising and the most available technology. The MB OFDM can be implemented at various parts of the spectral mask for UWB of 3.1-10.3GHz. An efficient and low cost optical microwave (MW) mixercan provide a flexible solution for transition between different MB OFDM signals in order to facilitate transition between existing systems. Recently, a low cost optical mixing based on vertical cavity surface emitting laser (VCSEL) non-linearity has been reported, presenting experimental results for up-conversion by 2.1GHz using a 1550nm VCSEL [1]. We extend the use of VCSEL based optical mixers for the 850nm band which is becomming attractive for use in UWB systems as mentioned above. We investigate, theoretically and experimentally, the up conversion of MB OFDM signals in a multimode (MM) VCSEL operating at 850nm. Experimental results for the conversion efficiency and error vector magnitude (EVM) are presented. They are inn agreement with the simulation results. The influence of the VCSEL's bias current and the CW power is also investigated.

Chapter 2 – A schematic diagram of the experimental set up used in our measurements is shown in Fig.1. e use a Finisar multimode VCSEL of 850nm, biased by a stabilized HP E3620A current source. A Wisair UWB development kit was used to provide the MB OFDM signals with a central frequency of 3.432GHz and bandwidth 5.28GHz. The UWB power was kept at -14dBm as required by the Federal Communication Commission (FCC). Using an RF combiner, the MB OFDM signal was combined with a signal of frequency fLO from an Agilent E8257C local oscillator (LO) and then it was fed as the modulating signal into the VCSEL. The

VCSEL output was detected using a Finisar PIN photodiode (PD) including a transimpedance amplifier, after a 3m MM fiber (MMF) link. Due to relatively high conversion efficiency of the optical mixer, no optical or RF amplifier was required in the link. The LO frequency fLO was varied from 2GHz up to 6GHz. An Agilent 89600 VSA vector signal analyzer was used to monitor the detected signals.

Fig. 1. The experimental set-up for the proposed mixing method

The measured values of the conversion loss as a function of the VCSEL bias current for several values of fLO presented in Fig. 2, show that the mixing efficiency is evaluated by the conversion loss of the mixer, which is determined mainly by the non-linearity level of the VCSEL's P I− characteristics

and its transfer function 21S . The conversion loss is

in the range of 8-30 dB, and depends on both the bias current and the detuning frequency (Fig.2). The quality of the up-converted signal passing through the whole link was evaluated by EVM measurements for two values of the LO power, -10dBm, and -3dBm (Figures 3,4) In these experiments, the lowest values of EVM to ensure signal detectability are -5dB, -3dB and -1dB for data rates of 480Mbps, 200Mbps and 53Mbps, respectively. The relatively low value of EVM for fIF =7.656GHz can be attributed to the relatively flat region of |S | in the vicinity of fIF

=7.656GHz. The EVM values for -3dbm are on


- 71 -

2

average 3dB better than for -10dbm, as expected due to the high power in the former case.

Fig. 2. Measured values of the conversion loss as a function of a VCSEL bias current for a UWB RF signal power of -10 dBm and different up-converted frequencies

Fig. 3. Measured values of EVM as function of a VCSEL bias current for a UWB RF signal power of -10 dBm and different up-converted frequencies

Fig. 4. Measured values of EVM as a function of a VCSEL bias current for a UWB RF signal power of -3 dBm and different up-converted frequencies

Simulations for the link shown in Fig.1, based on the commonly used models for VCSEL [2], PIN PD and MMF [3], were carried out in order to optimize the VCSEL design as a mixing device. The typical valuesof the following parameters were used: UWB central frequency 3.432GHz, UWB bandwidth 528MHz, average power -14dBm, sampling frequency 33.792GHz, symbol length 242.4nsec, guard time 70.08nsec, total bits per symbol 200, data bits persymbol 150, and total number of symbols 600. The simulation results indicate that the conversion loss exhibits qualitatively the same behavior as the measurements. These simulations can be used to study the impact of different design parameters of VCSEL on the mixing performance.

Chapter 3 – We have presented a new method for up-conversion of MB-OFDM based on the nonlinearity of VCSEL. Our experimental results show that using this approach conversion losses aslow as 7dB and low values of EVM can be achieved, at low bias currents. The mixing properties of VCSELs have been also studied by simulations that are in good agreement with the experimental results.

Acknowledgement - This work was supported in part by the European Project UROOF-Photonic Components for UWB over Optical Fiber (IST-5-033615).

References[1] S.B.Constant,, Y.Le.Guennec, G.Maury,

N.Corrao, and B.Cabon, "Low-cost all-optical up-conversion of digital radio signals using a directly modulated 1550-nm emitting VCSEL", IEEE Photonic Technology Letters, VOL 20, No. 2, pp 120-122 (2008).

[2] P. V. Mena, J. J. Morikuni, Member, Kang, A. V. Harton, K. W. Wyatt, "A Simple Rate-Equation-Based Thermal VCSEL Model", Journal of Lightwave Technology, vol. 17, No. 5, May 1999, 00. 865- 672.

[3] P. Pepeljugoski, S.E. Golovich, A. John Ritger, P. Kolesar, and A. Risteski, "Modeling and simulation of next-generation multimode fiber links", Journal of Lightwave Technology, vol. 21, No. 5, May 2003, pp. 1242-1255.


- 72 -

1

The Possibility of UWB Signal Detection in a Thin SiGe Layer

B.I. Lembrikov1,*, Y. Ben Ezra1

1 Holon Institute of Technology (HIT), Department of Engineering, 52 Golomb Str., Holon 58102, Israel * Corresponding author: Phone: +972 3 5026684, Fax: +972 3 5026685, E-Mail: [email protected]

Abstract - We developed a theoretical model of UWB modulated optical signal detection in a thin layer SiGe-on-Si.

Chapter 1 – The efficient detection of an ultra wideband (UWB) radio frequency (RF) envelope is required in UWB over optical fiber (UROOF) technology. The variations of the photocurrent at the optically controlled load of the microstrip (MS) line produce electromagnetic (EM) waves that propagate along the MS line towards the output port of an optically controlled microstrip convertor (OCMC) from which they are probed by a coaxial line of the

same characteristic impedance, 0Z . The temporal

response of this version of OCMC is determined mainly by the carrier lifetime in the substrate, which for silicon is about 1 sμ [1], [2]. The effective

bandwidth of this version of OCMC is not large enough being in the range of several tens MHz. Recently, it has been demonstrated experimentally that devices based on thin SiGe on Si layers of a thickness about one up to several micrometers can operate successfully as UWB RF signal detectors

providing a bandwidth of about ( )10 20 GHz÷ [3].

In this paper, we developed a novel physical model and its analytical description for the case of a thin SiGe layer with a high optical absorption and carried out the numerical estimations which clearly show that a 1μm SiGe layer on a Si substrate provides a bandwidth of about 20GHz at the 3dB level which cannot be achieved for Si.

Chapter 2 – For an infinite in the ,x y directions layer

of a thickness d in the z direction placed on a semi-infinite in the z direction substrate (Fig. 1), it can be shown by solving the boundary problem for the optical waves [4] that the time averaged total optical

intensity totoptI in the layer has the form.

( ) ( )( )

( )( )

30

2

23

2

2

2 cos[ cosh

1 sinh ]

to topt

ZI z I z d

Z

Zz d

Z

θ α

α

= −

⎛ ⎞⎜ ⎟− + −⎜ ⎟⎝ ⎠

Here 1,2,3Z are the wave impedances of the media,α is an optical absorption coefficient,

2

11 20 2

12

2 cos; ;

2effopt

opt eff b

eff

E AZ PI P A r

ZA D Z

θπ

+

= = =

( ) ( ) ( )2

2 1 31 32 22

2 2

sinh 1 coshZ ZZ Z

D d dZ Z

γ γ+⎡ ⎤

= + +⎢ ⎥⎣ ⎦

optP is the optical power of the incident wave in the

free space 0z < , 1E + is the incident wave

amplitude, 1 2 3 3, ,2 rk i k

c c

ω α ωγ β ε= = + = ,

β is the propagation constant, and br is the light

beam radius. The denominator 2

D can reach a

minimum value under a realistic quasi-resonance assumption sin 0; , 1, 2,...d d m mβ β π= = = for

1opt mλ μ∝ and ( )0.5 2d mμ∝ ÷ . Evaluation

of the photocarrier concentration in the framework of the drift-diffusion model is based on the continuity equations for the photoinduced electron and hole

concentration ( ),n z t and ( ),p z t [5]. Near the

illuminated surface of the semiconductor the stronginjection mode and ambipolar diffusion are realized

when 0 0,n n p>> and the ambipolar mobility

aμ vanishes [6]. In our case the thin layer is entirely

occupied by the strong injection mode region. Undersuch conditions continuity equations reduce to the ambipolar diffusion equation [5], [6]

( ) ( )2

0 12g ,a

n n nD z g z t

t z τ∂ ∂

= − + +∂ ∂

where ,aD τ are the ambipolar diffusion coefficient

and carrier lifetime, respectively, ( )1 ,g z t is the time

dependent part of the carrier generation rate given by

( ) ( ) ( ) ( ) ( )1 0

,,

totoptI z t

g z t f t g z f th z

ην∂

= =∂


- 73 -

2

η is a quantum efficiency, and ( )f t is the UWB RF

envelope of the optical carrier. We are interested in

the time-dependent part ( )1 ,n z t of the photocarrier

concentration which is responsible for the UWB RF signal detection. The averaged over the layer

thickness d Fourier transform ( )1 ,N z ω of

( )1 ,n z t can be used as the frequency response of

the illuminated layer when ( ) ( )f t tδ= . The

analytical expression of ( )1 ,N z ω is too

complicated, and we do not present it here. The results of the numerical evaluation of the response

function ( )1N ω for the typical values of material

parameters of SiGe on Si are presented in Fig. 2.

Chapter 3 - The numerical estimations based on the proposed analytical model of the thin layer SiGe/SiOCMC structure with an detecting layer thickness of

about ( )0.5 2d mμ= ÷ clearly show that a

bandwidth of at least 20GHz can be achieved. The resonant conditions are essential for the layer thickness because the reflection from the SiGe/Si interface in such a case reaches its maximum value.The proposed structure is simpler as compared to resonant-cavity-enhanced (RCE) photodetectors with distributed Bragg reflector (DBR) layers in the substrate.

Acknowledgement - This work was supported in part by the European Project UROOF-Photonic Components for UWB over Optical Fiber (IST-5-033615).

References [1] R. Gary, J.-D. Arnould, and A. Vilcot, Semi-

analytical computation and 3D modeling of the microwave photo-induced model in CPW technology, Microwave and Optical Technology Letters, Vol. 48, No. 9, September 2006, pp. 1718-1721.

[2] R. Gary, J.-D. Arnould, and A. Vilcot, Semi-analytical modeling and analysis in three dimensions of the optical carrier injection and diffusion in a semiconductor substrate, Journal of Lightwave Technology, Vol. 24, No. 5, May 2006, pp. 2163-2170.

[3] Z. Huang, Ning Kong, X. Cuo, M. Liu, N. Duan, A.L. Beck, S.K. Banerjee, and J. C. Campbell, 21-GHz-bandwidth Germanium-on-Silicon photodiode using thin SiGe buffer layers, IEEE Journal of Selected Topics in Quantum

Electronics, Vol. 12, No. 6, November/December 2006, pp. 1450-1454.

[4] N.N. Rao, Elements of Engineering Electromagnetics, 5th edition, Prentice Hall (2000).

[5] H. Mathieu, Physique des semiconducteurs et des composants electroniques. Masson, Paris (1998).

[6] J.-D. Arnould, R. Gary, and A. Vilcot, 3D Photo- induced Load Modeling for Optically Controlled Microstrip Line, Microwave and Optical Technology Letters, vol. 40, pp. 356-359 (2004).

Fig. 1. Illuminated SiGe layer on a Si substrate.

Fig. 2. Normalized transfer function ( )1N ω for

different SiGe layer thicknesses d=0.5; 1.0;1.5;2.0μm


- 74 -

Advanced phase detection for electro-optical phase-locked loop (EO-PLL)

L.Naglic1,*, L.Pavlovic1, M.Vidmar1

1 University of Ljubljana, Faculty for Electrical Engineering, Trzaska 25, 1000 Ljubljana, Slovenija * Corresponding author: Phone: +386 1 4768423, Fax: +386 1 4768424, E-Mail: [email protected]

Abstract - A phase detector with a reduced spurious output is presented in this article. The parallel operation of two conventional phase detectors (fast-logic EXOR gates) is at the beginning presented theoretically and then verified with a practical experiment.

Introduction: Circuits, including optical components, are usually many orders of magnitude larger than the optical wavelength. Correspondingly, the time delays in any feedback loop are many orders of magnitude larger than the optical time periods. Without optical-injection locking [1], feedback loops can only be built with narrow-linewidth optical sources, corresponding to a large coherence length. Even in this case the delay of the feedback loop that includes both optical and electrical components has to be kept very small. A typical feedback loop, to be used as a millimetre-wave frequency synthesizer, is an electro-optical PLL and is shown in Fig 1.

Fig. 1: Electro-optical PLL.

Even when using very narrow linewidth (1 MHz) solid-state lasers, the complete feedback-loop delay has to be kept less than a few nanoseconds [2] to stay within the linear range of the phase detector and avoid cycle slips. Since filtering introduces delay, little if any filtering can be applied to the output of the phase detector. The linear range of the phase detector can be increased by inserting a frequency divider in the loop. Frequency dividers bring two major drawbacks: (1) the loop gain is reduced and (2) the spurious output of the phase detector includes more spectral lines at lower frequencies. Unfortunately, frequency dividers are the only practical solution when the optical linewidth and/or the loop delay can not be further reduced.

Design of an advanced phase detector: Double-balanced mixers or fast-logic EXOR gates are usually used as phase detectors. Their linear phase response is limited to less than /2. The loop filter of a second-order PLL is usually a Proportional-Integrating (PI) regulator. The major spurious product is the second harmonic of the comparing frequency generated inside the phase detector. This spurious product produces an unwanted frequency modulation of the optical source. This effect is particularly harmful to the optical spectrum when optical mixing with an optical comb generator, electrical harmonic mixing or electrical frequency dividers are used inside the loop. The second harmonic can be cancelled if two identical phase detectors are being operated at 90-degree phase shift at their inputs. In this case their outputs contain the second harmonic at 180-degree phase shift. When the two outputs are summed, the desired phase information is summed while the unwanted second harmonic is cancelled. A practical implementation includes two master-slave toggle flip-flops to obtain 90-degree phase shifts. Block diagram of the improved phase detector is shown in Fig 2.

Fig. 2: Circuit diagram of the improved phase detector.

Since 90-degree phase-shifted signals are available, a third phase detector EXOR#3 can be built in for a lock detection and integration. The latter is required to automatically enable and/or disable the search logic required by the PI regulator.


- 75 -

Experiment: Electro-optical PLLs require integrated packaging [3] in order to minimize loop delays. Since such an integrated component was not available yet, standard (fibre) pigtailed optical components were used in our experiment: a commercial DFB (Distributed Feedback) laser (estimated linewidth around 10 MHz) as a slave VCO (Voltage Controlled Oscillator) laser, an external-cavity laser as a master laser, a 10 GHz-bandwidth TIA (Transimpedance Amplifier) photo-detector, a broadband 30 dB MMIC (Monolithic Microwave Integrated Circuit) amplifier for the beat signal, an electrical ECL (Emitter Coupled Logic) divider and fast (200 MHz) ECL logic for the phase detector (XOR gate). Due to the large loop delay and wide laser linewidth, the electrical beat signal at 2.7 GHz had to be divided by 1024 to increase the phase-detector linear range. The spurious output from a single EXOR-gate phase detector at twice the comparing frequency is clearly seen as a frequency modulation on the electrical 2.7 GHz beat signal on Fig. 3 (red line). The improved phase detector cancels the spurious output at twice the comparing frequency, therefore produces a much cleaner beat signal as seen on Fig 3 (black line). The acquisition range of a P-only loop is around 60 MHz at 2.7 GHz, but this range is enormously increased by using search logic and an integrator in a PI feedback.

Fig. 3: Single-EXOR phase-detector optical-beat spectrum (red line) and an advanced phase-detector optical-beat spectrum (black line).

High-speed low-delay phase detector: In the case where a low total-loop delay and sufficiently narrow laser linewidth is assured, there is no need to extend the linear range of the phase detector and consequently no need for the frequency divider in the loop. For the mentioned conditions a phase detector was developed with a bandwidth exceeding several GHz and a sub-ns delay. It comprises low-cost high-speed logic for the phase and lock detection (PI regulation loop) and also a search circuitry for an automatically locking upon power-up.

The estimated total loop delay of 0.9 ns in our experiment with a microwave CCO (current-controlled oscillator), in place of the integrated optical components (dual mode laser and photodiode), enabled a successful phase lock of a wide signal linewidth of 5 MHz (without any frequency division) as seen in figure 4.

Fig. 4: Free-running (red line) and phase-locked (blue line) spectrum of a noisy microwave CCO oscillator.

Conclusion: A phase detectors presented in this article are particularly useful in electro-optical phase locked loops, where group delay and therefore filtering inside the loop has to be kept minimal and at the same time optical comb mixing, electrical harmonic mixers or electrical dividers may be present in the loop. The idea of a parallel operation of two mixers can be further extended to any practical number of mixers, resulting in an increased linear phase response range without compromising the output with the spurious signals. This allows an arbitrary extension of the detector phase range with simple electrical circuitry, which is able to operate at much higher frequencies than the solution proposed in [4].

Acknowledgment: The described work was developed as a part of the EU project under contract number 035317 (IPHOBAC), financed by the European Community.

References[1] A.C. Bordonalli et. al., J.Lightwave Tech.,

1999, 17(2), pp 328-342. [2] Ramos, R.T. et. al., Electron. Lett., 1990, 26,

pp. 389-391 [3] L.N. Langley et.al., IEEE Trans. MTT, 1999,

47(7) pt.2 pp 1257-1264 [4] D. Weinfeld et. al., Eighteenth Convention of

Electrical and Electronics Engineers in Israel, 1995, pp. 4.1.3/4


- 76 -

An Antenna-Integrated Photonic Millimeter-Wave Transmitter V. Rymanov1*, M. Weiß1, A. Steffan2, S. Fedderwitz1, A. Stöhr1 and D. Jäger1

1 Universität Duisburg-Essen, ZHO-Optoelektronik, Lotharstr. 55, 47057 Duisburg, Germany 2 u2t Photonics AG, Reuchlinstr. 10-11, 10553 Berlin, Germany


Abstract – In this paper, a small-scale antenna-integrated photonic millimeter-wave transmitter based upon a traveling-wave p-i-n structure is presented, comprising a transit time optimized layer structure. In conjunction with the applied planar log-periodic toothed antenna structure, we demonstrate 30-325GHz operation.

Introduction – Photonic millimeter-wave (mm-wave) generation is of great importance for many emerging markets as it allows extremely compact size, wideband tunability as well as broadband modulation capabilities and further enables the utilization of optical fibers as a low-loss mm-wave transport medium. Applications comprise broadband mm-wave transmitters for communication (e.g. within the V-, E- and F-band), mm-wave radar, mm-wave synthesizers, imaging and radio astronomy [1]-[4]. A straight forward way to generate a high-frequency mm-wave signal within the optical domain is to apply light from a dual-wavelength laser source to a photodetector converting the optical signal into an mm-wave which is further radiated via a broadband antenna. For communication purposes the generated mm-wave signal may further be modulated by a broadband data signal. However, realizing such a photonic mm-wave transmitter by using discrete components may cause high costs due to several bulky mm-wave components like coaxial cables or rectangular waveguides to connect the mm-wave photodetector with the antenna, further on mm-wave transitions and connectors leading to power losses, increased module size and higher costs. Therefore an integrated solution would be beneficiary in terms of costs and size if detector and antenna are integrated to one single chip [5]. Fig. 1 shows the operation principle of such an antenna-integrated photodetector. Light is coupled by a single mode fiber (SMF) to a passive optical waveguide (POW) which further transports the optical signal to a high-frequency photodetector operating within the mm-wave range. After detection, the converted electrical signal is coupled to an electrical waveguide, further to the feeding point of a planar antenna and transmitted. Due to the large difference in the dielectric constant between air and semiconductor, the generated mm-wave is mainly radiated through the substrate. For efficient beam-forming and coupling to free-space, an additional

quasi-optics is necessary which was already reported in [6].

optical fiber input

millimeter-wave radiation

photodetector

passive optical waveguide

planar antenna

electrical waveguide

Fig. 1. Operation principle of the developed component.

In this paper, we report on an ultra-broadband InGaAs(P)/InP traveling-wave (TW) photodetector with integrated antenna operating within the whole mm-wave range (i.e. 30-325GHz). At first, we discuss the layer structure and the physical properties of the developed photodetector based upon a modified partially non-absorbing traveling-wave p-i-n structure comprising a partially p-doped layer. Further on, we discuss the developed device from a component’s sight of view, i.e. monolithically integrated optical waveguide, photodetector, electrical waveguide and broadband planar antenna on InP-substrate. We have packaged the component to a photonic mm-wave transmitter module with optimized mm-wave radiation efficiency and radiation pattern. Finally, we present experimental results on the fully packaged devices.

Photodetector Operation Principle – The developed layer structure is schematically shown in Fig. 2. Our approach comprises a traveling-wave p-i-n structure with a transit time optimized layer structure. Regarding the layer structure, the amount of slow photo-generated carriers (holes) is significantly reduced which would otherwise


- 77 -

contribute to the photocurrent and therefore degrade the RF-performance.

Ti/Pt/Au

Ge/Pt/AuGe/Pt/Au

lower cladding:compensated InP:Fe-substrate

contact layer:p+-InGaAs:Zn

drift region:n.i.d.-InP

drift region/spacer layers:n.i.d.-InGaAsP

intrinsic region/absorber:n.i.d.-InGaAs

p-doped absorber:p-InGaAs:Zn

diffusion blocker 2:p-InGaAsP:Zn

diffusion blocker 1:p-InGaAsP:Zn

passive waveguide core:n.i.d.-InGaAsP

n-doped layer/passivewaveguide core:n+-InGaAsP:Si

upper cladding:p+-InP:Zn

x/nm

y/μm

z

Fig. 2. Schematic cross section (middle) of the developed traveling-wave photodetector, enlarged active section (left hand) and 3-dimensional model (right hand).

In detail, the main improvements are a partially p-doped absorbing layer on the one side and a partially non-absorbing i-layer on the other side. A structure based on this principle, in conjunction with a thin absorber in the drift layer is expected to deliver higher photocurrent without compromising frequency response [7],[8]. Another key benefit is the applied traveling wave principle, which differs from a lumped element in a non-RC time limited response exhibiting superior high-frequency performances. This was already demonstrated e.g. in [9]-[12]. The layer structure has been successfully grown in metal organic vapor phase epitaxy. Images recorded by transmission electron microscopy show good agreement between thicknesses of designed and fabricated epitactic layers whereas analysis by x-ray diffraction shows a maximum lattice mismatching of only 0.5%. The schematic cross section describing the specific layer functions is further shown in Fig. 2. The dielectric ridge loaded optical waveguide consists of a 50nm p-doped and additional 50nm InGaAs absorber. The drift region consisting of three InGaAsP spacer layers and the InP layer is located below the intrinsic absorber. The thickness of the non-absorbing InP drift layer is 220nm. Beneath the drift region, two InGaAsP layers, n+-doped and non-doped, form the passive optical waveguide core. Photoluminescence wavelength for non-absorbent core was determined to be 1.21 m at room temperature which means that the photomixers can be operated not only at 1.55 m but also at 1.3 m. The whole structure is grown on compensated InP substrate.

Antenna-Integrated Photodetector – Fig. 3 shows the schematics of the developed mm-wave transmitter. An applied photonic mm-wave signal using a lensed SMF is coupled to the POW, transported to the broadband photodetector and o/e-converted. After o/e-conversion, the electrical signal is coupled to a microstrip circuitry and further fed to the center of the planar log-periodic toothed antenna (LPTA).

TW-photomixer

microstripfeed line

passive opticalwaveguide

log-periodic toothed antenna structure

Fig. 3. Schematic structure of the developed component comprising passive optical waveguide, photodetector, low-loss antenna feed and a broadband LPTA-structure.

A well-suited design was applied for the optical waveguide for achieving high coupling efficiencies. In that regard, BPM CAD simulations were carried out to calculate the overall optical coupling efficiency from a lensed SMF to the active photomixer section, i.e. lensed SMF to POW and POW to active photomixer section. It was found, that the maximum efficiency is as high as 56% if a proper geometrical design is applied. In Fig. 4, the realized antenna-integrated photomixer is illustrated, comprising of a 2x2mm2 log-periodic antenna and an approx. 1mm POW for optical feed, expending from the front surface to the active photomixer section. The photomixer, exhibiting a 70μm microstrip feed line between the active photomixer section and the antenna center, is positioned close to antenna center for low electrical losses between photomixer output and antenna feeding point. The antenna-integrated photomixers are fabricated using conventional photolithography, wet chemical etching and metal evaporation. Lift-off technique is employed to realize the metal contacts. Electrical passivation at the interface between the active photomixer section and the coplanar output and antenna feeding point via microstrip circuitry, respectively, is implemented by a pure baked-out polyimide bridge to prevent broken microstrip feed


- 78 -

lines lying above the passivation. The devices are thinned down to 125μm.

passive optical waveguide

Polyimide

TW-photomixer

n-contact

p-contact microstrip feed line

Fig. 4. Photographs of the developed component. Top view of the fabricated antenna-integrated photomixer (left hand) and close up view of the center of the planar log-periodic toothed antenna with microstrip circuitry, active photomixer section and passive optical waveguide (right hand).

Packaging & Characterization – For characterizing the developed antenna-integrated photomixer, we have constructed a photonic transmitter package comprising techniques for low mm-wave loss and an efficient antenna beam generation, which was already shown in [6]. We have packaged two devices to complete modules which are shown in Fig. 5. The package consists of two DC pins to allow biasing of the antenna-integrated photodetectors, a single mode fiber with FC/APC connector and a quasi-optics (i.e. a high-resistive silicon lens) for efficient mm-wave transmission.

Fig. 5. Top view of a packaged, antenna-integrated photomixer (left hand) and side view (right hand) showing the quasi-optics for efficient radiation pattern generation.

We have characterized the modules using a set of power detectors with rectangular waveguide input and attached horn antenna (i.e. WR22 to WR03) to cover a frequency range of 30-325GHz. The measured frequency responses are shown in Fig. 6. As can be seen, both detectors exhibit a very similar response behavior with a power difference of approx. 2dB indicating a good reproducibility of the processed devices. The response exhibits a very smooth behavior with a total roll-off of about 25dB except a fall-off within 60GHz. This is attributed to an insufficient antenna size for lower-frequency operation (i.e. below 75GHz). Improved antenna designs are currently under work.

-50

-40

-30

-20

-10

0

10

25 75 125 175 225 275 325

Frequency (GHz)

Rel

ativ

e D

etec

ted

Pow

er (d

B)

IPHOBAC Transmitter No. B9.2971(UDE 300G LP.20.7.3.A)IPHOBAC Transmitter No. B9.2970(UDE 300G LP.20.6.3.B)

Fig. 6. 30-325GHz characterization of the developed components using a set of power detectors.

Conclusion – In this paper, we have presented an antenna-integrated photonic mm-wave transmitter based upon a traveling-wave p-i-n diode with an advanced partially p-doped and partially non-absorbing layer structure. The advanced photodetector layer structure in conjunction with the applied broadband LPTA-structure as well as applied optical and electrical coupling techniques allow a broadband operation. After packaging to modules, operation within 30-325GHz was demonstrated for two antenna-integrated photomixers.

Acknowledgement - This work was carried out within in the framework of the European integrated FP6 project IPHOBAC under grant no. 35317.

References[1] www.ist-iphobac.org[2] A. Hirata et al, “120-GHz-band millimeter-

wave photonic wireless link for 10-Gb/s data transmission,” IEEE Trans. Microw. Theory Tech., vol. 54, no. 5, pp. 1937-1944, 2006.

[3] X. C. Zhang, “Terahertz Wave Imaging: Horizons and Hurdles,” Phys. Med. Biol., vol. 47, no. 21, pp. 3667-3677, 2002.

[4] A. Stöhr et al, “All-optical radio-independent millimeter-wave radio-on-fiber system with lean antenna base stations,” Proc. Int. Topical Meeting Microwave Photonics (MWP), pp. 213-216, 2002.

[5] A. Stöhr, C.C. Renaud, D. Moodie, A.G. Steffan, L. Pavlovic, D. Jäger, A.J. Seeds, M. Robertson, A. Umbach, M. Vidmar, M. Weiß, V. Rymanov, S. Fedderwitz, “Optical Millimeter-Wave Generation using 1.55μm Photodiodes with and without Integrated Antennas,” 2008 URSI General Assembly, Chicago, Illinois, USA, 2008.

[6] M. Weiß, A. G. Steffan, S. Fedderwitz, G. Tsianos, A. Stöhr, D. Jäger, “Highly-Compact Fibre-Optic Package for 30-300GHz Wireless Transmitter Modules,” 2nd Electronics System-Integration Technology Conference, London, UK, pp. 1111-1114, 2008.


- 79 -

[7] K. J. Williams, “Comparisons between dual-depletion-region and uni-travelling-carrier p-i-n photodetectors,” IEE Proc.-Optoelectron., vol. 149, no. 4, Aug. 2002.

[8] D. A. Tulchinsky, X. Li, N. Li, S. Demiguel, J. C. Campbell, K. J. Williams, “High-Saturation Current Wide-Bandwidth Photodetectors,” IEEE Journal of Selected Topics in Quantum Electronics, vol. 10, no. 4, July/Aug. 2004.

[9] A. Stöhr, A. Malcoci, A. Sauerwald, I. C. Mayorga, R. Güsten, D. Jäger, “Ultra-wideband traveling-wave photodetectors for photonic local oscillators,” IEEE Journal of Lightwave Technology, vol. 21, no. 12, pp. 3062-3070, 2003.

[10] A. Malcoci, A. Stöhr, K. Lill, F. Siebe, P. van der Wal, A. Sauerwald, R. Güsten, D. Jäger, “Optical submillimeter-wave generation

employing antenna integrated ultra-fast travelling-wave 1.55μm photodetectors,” IEEE MTT-S International Microwave Symposium - IMS 2003, vol. 1, pp. 143-146, 2003.

[11] D. Jäger, A. Stöhr, “Microwave Photonics,” Cobra-Basic Research and Applications Colloquium, Inter-University Research Institute on Communication Technology, Eindhoven, Juny 18, 2003.

[12] A. Stöhr, R. Heinzelmann, A. Malcoci, D. Jäger, “Optical heterodyne millimeter-wave generation using 1.55μm traveling-wave photodetectors,” IEEE Transactions on Microwave Theory and Techniques, vol. 49, no. 10, pp. 1926-1933, 2001.


- 80 -

Millimeter-wave frequency generation with dual-wavelengthDFB laser

S. Ginestar1*, F. van Dijk2, A. Accard2, O. Legouezigou2, F. Poingt2, F. Pommereau2, L.Legouezigou2, F. Lelarge2, B. Rousseau2, J. Landreau2, J-P. Vilcot1 , G.H. Duan2

1�� !��"��#�� $#�%��%&��&%��'��"� �� (��)�"�*

��"� ��+ ��+'!�,��-�.��&/0%�� )�� 1�� 2�#��2�344�4��/&5�&&/��)�(2�344�4��/&5�&%%�

+��2��6��7��8��9�7��+��/7:�

�� + ��

�� ;� ��+<��<��6�� 96�� : � �<� �=)$ �� <�� 9�� : ��<��<��7�":��6�� ,��:��<��<�� ,�� :�� .� ��6��::�� ,��<�� <� ��6�� :�� 7 � �� ::��:� ��7�"��.��<� �<�� ::�� :��9��6�� ::� �� 6�� :�,�� ,��9�� . � �� <��7 � �� : � �� :�� .��6�� ::�� : � �� <� � �9��<��7 �"��,��.��:�<�� .��,��<��6�� ( �� >-? �� : �� .�<��6��,��6��,��<��:��6�� @/A7� �� ,�� :�,�� .��9� � :�� 9��:�� :�� .��.�B�� : � 9� ��<��+6�� .��9 � �� :�(��,��:�� .��7� �� ::��.��:�� .��<�� 6��,�.��:��9��,.�9��:.�� C� �� ::�� 7��6�6��<��6��6��:��9�� :��9��=)$��:��:��6��,+ >- �:�� . ��7 �� ,�� -�� 6 � �� 9��9�� <�� :�� .��(6��9��+�6��9��7

�� + � �� <�� : � �<� �� D9+�� ,�� :��+,� B�E=)$F �� +�� 7 �� $�� <�� 6�� . ��::�� Δ/ �� Δ�7 � �� .�� <��<��:��<��6�,.�� 9��

,��9 � �6��(. � EG�+�$F7 � �� 96�� : � %��G�"�#��9�<��G�"�#�,��7�":��+,��9 � ��6�. ��: � �� 6�� 66��#��.��<�� .��<��+��<��,.��H'#�@�A7�� :��<��:� ��<�� "� � �� 9�� ,�� =)$�,��7 � �� <�� 9�� "��,9��:��7

�� + �� :�� : � �� =)$ �� <�� : ��,�� 9"7 �)�� /��<� � �� 6�� 6� ��9 � �,�� :��<�� :��:��=)$�� 9��9"�� : � �� <�� %�9"� �/��9"��9"7�

��

��

��

��

��

��

��

��

��

�

��

��

��

��

��

!"��$ !"��$ !"��$

�� H6�� 6� ��9

�� <�� ::�� ,�� :�� 6��E��9"� � ��9"F �<�� 7�0�9� ��6�� 6��+��:�� .��:��/G>-7� ��9��,�� :�� . ��6�� :�� :�� :�� :�� =)$�� E�� :��:��=)$�� B�6��9"F7


- 81 -

��<��<�� <�� G>-�6��7 � �� 6� ��9 ��: � �� ,�� <�� 9�� 6� ��9��.��7 �� (�� C� �� B�� <��7� �� <�� : � �� 6�� <�� 9�� : � ��.�� 9�� @4A � �� <��:�� <��9�� .��6��+��7�7��6��:��6�� 6� ��9 � ��.��7� ��6�� :�� :��9��,�� ,�� 6�� 6� �� :�)�� /7 � �� 6�� <�� ,��9�� :��<�� :�� <�� ,�� 9�� 6 � ��6�� 6�,��7�7��G>-7

�� >��.��,��:�� .��,��.��6�� <��

�� 6�� <�� 4�>- � �� /0�>-�:�� .��:��9��G>-��G>-7��9��

��,�� 6 � �� /�� G>- � �� ,�� :��<��7 �� <�� /� ��>- � :��9��G>-��6��G>-��.��,��:��6�� <�� <�� 9�� :�� 9�� 7

�� $�� 6�� 6� ��9 � �� 6��+�� 6� ��9 � �: � � � ��+<�� =)$ � �� ,�� 7 �"��,��.�:��9��4��G>-��,��,��<��<��:�� ,��<�/��>-�:��:�� G>-7

��!�"�� + � �� <��B �<�� <��:��9�<��B��:��6��6��C� � � �#>H$"* �� <��B ��: �( �� 7

#�$��@/A *7 �$�� 7 �$�� 7 �� >+� #7�

��$7��I"��+�6�� B�� .��G>-��,�.��I� �#�� 7��5��67�*��:7��H6�7 �*�997�E*H*F��/��'��7�0��667��4+��0�E��/F

@�A �7�G��"7�" ��)7�'��=�CB��J+#7�'�� G+>7 �=�� I�� :�� +9�� I��#�K��6�� //+/5��.��,��.�E��F

@4A 7 �HB�� L7 �L�B� �� "7 ��B�.�9�� I��9��:��9��9��:��6�� 6� ��9I� � �� !�� '��7�/%E/%F��667�%4�+%4/�E/&��F


- 82 -

Performance of 802.11g signals over a multimode fibre-fed distributed antenna system incorporating optical splitting

L. C. Vieira*, A. Nkansah, P. Assimakopoulos, and N. J. GomesUniversity of Kent, Department of Electronics, Canterbury, CT2 7NT, UK

* Phone: +44(0)1227823244, Fax: +44(0)1227456084, E-Mail: [email protected]

Abstract An experimental investigation on the distribution of IEEE 802.11g signals using low cost indoor RoF system incorporating an optical power splitter. The performance of the system was assessed in terms of EVM and RF power.

Introduction In this paper, we present an experimental study of the performance of a Radio-over-Fibre (RoF) system for transmission of IEEE 802.11g signals, considering the effects of the optical link and RF amplifiers on the Error Vector Magnitude (EVM) and link gain. Two optical links are considered, one with an optical splitter and the other without for comparison purposes.RoF based Distributed Antenna Systems (DAS) have been reported as an option to provide coverage improvements of wireless LAN (WLAN) signal transmission [1,2]. Also, low-cost radio-over-multimode-fibre links, using 850-nm vertical-cavity surface-emitting lasers (VCSELs), have been shownto achieve the IEEE 802.11g requirements [3].The use of Remote Antenna Units (RAU) in a WLAN, connected through a RoF-DAS network, means that the access points can be situated centrally improving maintenance and installation costs. In most cases one dedicated optical link is used per RAU [1-3]. In this paper, a different optical link isevaluated, with only one laser feeding two RAUs, via an optical splitter.

Experimental set-up The IEEE 802.11g signal (54 Mbps, 52 active subcarriers) was generated from an Agilent E4438C Vector Signal Generator (VSG),directly modulating an ULM-Photonics 850-nmVCSEL (maximum optical power 6mW) at 2.45 GHz. The laser was initially connected through a 50-μm-core MMF patch cord to an Appointech InGaAs PIN photodiode (PD), and the rms EVM and output power were measured after the optical link by an Agilent E4440A Vector Signal Analyser (VSA) connected to a laptop with Agilent VSA software. The input power at the laser was varied from -28.7 dBm to 3.7 dBm, with the maximum power chosen to be around 10 dB below the 1 dB compression point of the laser. After that, two other cases were investigated by (a) inserting a 300-m length of 50-μm-core MMF into the link and (b) emulating a two-RAU scenario using an optical splitter at the end of the MMF in a star formation. The experimental scheme for the latter case is depicted in Fig. 1.Finally, the effects of RF amplification were also investigated, adding two amplifiers (Mini CircuitsZX60-2522M-S) and one 10-dB attenuator in theRoF link as also illustrated in Fig. 1.

Fig. 1.: The Experimental set-up. The parts where there is optical transmission are depicted in blue colour. CU=Central Unit, RAU=Remote Antenna Unit, MMF=Multimode Fibre, ATT=Attenuator, Amp=Amplifier.


- 83 -

Results The signal power and rms EVM (%)averaged over 100 measurements were recorded at the output of the PD. The results of the first set of experiments are shown in Fig. 2.

Fig. 2.: Gain and EVM for three cases: (a) laser directly connected to PD; (b) 300-m length of MMF optical linkwithout optical splitter; (c) 300-m length of MMF with optical splitter.

The RF link loss increases by about 9 dB for theoptical splitter case. This is a result of the increase inoptical insertion loss of about 4.5 dB, due to the splitter and additional connectors. The loss can be compensated by increasing the RF amplifier gain at the RAU. The measured EVM was 2 % and 1.6 % for the 300-m length MMF optical link with and without splitter respectively, at around -5 dBm input power, confirming results previously reported for a similar link without optical splitter [3]. Moreover, the measured EVM with the optical splitter was less than 3.1 %, from -16 to +2.8 dBm input range. The increase in EVM is due to the higher RF loss and corresponding increase in Noise Figure (NF).

Fig. 3. Gain and EVM for 300-m length of MMF non-splitting optical link with 30 dB RF gain and splitting optical link with either 30 dB or 40 dB RF gain.

The results with and without an optical splitter,including the RF amplifier effects are shown in Fig.3. With RF amplifiers inserted at the output of the optical link, the EVM measured for the link with optical splitter, at the worst case, was about 1.2 % higher than the link without the optical splitter. The EVM is very similar for both 30 dB and 40 dB RF gain. The results show that the EVM is below the maximum 5.6% allowed by the IEEE 802.11g standard for a large range of input powers. However, it has to be noted that the inclusion of the wireless link will further degrade the EVM.Although the average input power at the laser is at least 10dB below the laser compression point, there will be clipping due to the peak-to-average power ratio (PAPR) of the 802.11g signal.The results presented here are for the downlinktransmission. New experiments are needed to investigate the uplink transmission performance.

Conclusion Link gain and EVM, due to the inclusion of an optical splitter, were experimentally investigated for IEEE 802.11g transmission on low-cost radio-over-multimode fibre link. These resultssuggest that an optical link incorporating an optical splitter, experiences degradation in performance at low input powers with respect to a link without an optical splitter. However, with compensating amplifiers at the RAU both types of optical link have similar performance. The advantage of the optical splitting case is the reduction in the number of optical transmitters and optical fibres required.

Acknowledgement - This work was partially carried out within the framework of the European Integrated project FUTON (FP7 ICT-2007-215533).

Luis C. Vieira is sponsored by the Brazilian Government through CNPq and UTFPR, whose support is gratefully acknowledged.

References[1] M. J. Crisp, S. Li, A. Watts, R. V. Penty and I.

improvements of 802.11g signals using a

Technology, pp. 3388-3395, Nov. 2007.[2] P. Assimakopoulos, A. Nkansah and N. J.

Use of commercial Access Point employing spatial diversity in a Distributed Antenna Network with different fiber lengthsIEEE Int. Topical Mtg. on Microwave Photonics, MWP 2008, pp 189-192, 2008.

[3] A. Nkansah, A. Das, I. J. Garcia, C. Lethien, J-P. Vilcot, N. J. Gomes, J. C. Batchelor, D. Wake,

-band radio signals over a multimode fibre fed indoor

Comp. Lett., pp. 627-629, Nov. 2006.


- 84 -

Static nonlinear distortion modelling of radio-over-fibre systems

L. C. Vieira1,*, N. J. Gomes1, A. Nkansah1, F. van Dijk2, A. Enard2, F. Blache2, M. Goix2

1University of Kent, Department of Electronics, Canterbury, CT2 7NT, UK2Alcatel-Thalès III-V Lab, joint lab of Bell Labs and Thales Research & Technology, Palaiseau, France

*Phone: +44(0)1227823244, Fax: +44(0)1227456084, E-Mail: [email protected]

Abstract The static nonlinear distortions of a 1550-nm direct modulation radio-over-fibre (RoF) link are modelled using AM/AM and AM/PM measurement techniques. However, this methodology needs to be better validated by new experiments and simulations.

Introduction Radio-over-fibre (RoF) has been used in the last few years as an alternative for connection between a central office (CO) and remote antenna units (AU) for cellular networks or for connection ofwireless local area network (WLAN) access points with distributed antenna units [1-2]. The main advantages of the RoF technology are the reduction of the cost of installation and maintenance of the AUs and the improvement in the coverage area of the CO or of the WLAN access point.The optical link portion of the RoF system usually has adequate bandwidth to support wide-band and high-capacity services. However, the nonlinear distortion of the optical link, due mainly to the laser diode, may impose serious limitations on the system performance, especially in a multi-user environment[3].Although the photodiode (PD) can cause some nonlinear effects on a directly-modulated RoF link,the light-current transfer function of the laser is considered the main source of static nonlinearity [4]. Generally, the optical fibre nonlinearity can be neglected for lower-power RoF links.In this work, a behavioural (black-box) modeling technique is applied to model the nonlinear distortions of a direct modulation RoF link. More specifically, the RoF link model was obtained using AM/AM and AM/PM measurement techniques. In these modeling approaches, the nonlinear effects of the link are observed when the input RF power is varied, that is, the nonlinear amplitude and phase characteristics are obtained from the input-output measured data.

Experimental set-up The model was extracted from an experimental directly-modulated RoF link,which is depicted in Figure 1. The experimental set-upconsisted of a 1550-nm distributed feedback (DFB)laser from Alcatel-Thales III-V Labs, a FC/APC-FC/PC fibre patch cord and an Appointech InGaAsPIN photodiode, with bandwidth of 2.5 GHz and

responsivity of 1 A/W. A Mini-Circuits bias-T was used to bias the PD at -5V. The laser was biasedaround the middle of its light-current characteristicthrough its internal bias circuit, with its temperaturemaintained at 25oC. A sinusoidal signal of 1 GHz, with 12-dB dynamic range, was applied by a HP 8722ES vector network analyzer (VNA), and the output power and phase shift were measured after the optical link, with the input power varied in order to find the saturation region of the laser. Each recorded output signal was averaged over 16 measurements.A highly linear amplifier (Mini-circuits ZHL-30W-252+) was used to drive the laser in order to achieve the dynamic range for this characterization, being also necessary to include a 16-dB RF attenuator to protect the VNA. The experimental set-up was calibrated against the response of the RF cables, attenuator and amplifier used.

Fig. 1.: Experimental set-up

Results The output voltage amplitude at the fundamental frequency can be represented by the odd-order distortion terms from a polynomial formulation.In this work, the AM/AM amplitude distortion is modelled by the following 5th-order polynomial

ApApApAg 13

35

5)( (1)

where A is the input amplitude.


- 85 -

It was found that better fitting resulted when using a rational function for the phase distortion. Thus, the AM/PM characteristic is modelled by the following

qAqApApApApAp

A1

21

22

33

44)(

The input amplitude in (2) is normalized by mean 4.309 and standard deviation 1.704.The coefficients for (1) and (2) are given in Table I. The AM/AM and AM/PM characteristics of the RoF link are shown in Figures 2 and 3. These models were fitted using Trust-Region algorithm, there being obtained a root mean square error (RMSE) of 0.0002for the AM/AM and an RMSE of 0.02177 for the AM/PM.

Table I.: Coefficients for AM/AM and AM/PM models of the RoF link.

Coefficients for AM/AM modelp1 p3 p5

0.01902 6.649e-005 -1.267e-006

Coefficients for AM/PM modelp p1 p2 p3 p4

138.3 -149.8 61.49 -0.0493 -0.2494q q1

2.255 -2.443

Fig. 2.: AM/AM characteristic of the RoF model.

Fig. 3.: AM/PM characteristic of the RoF model.

Conclusion The AM/AM and AM/PM model were extracted from an experimental directly-modulated RoF link, and the resulting model has been presented here. As this work reports preliminary results on RoF link nonlinear modelling, this methodology still needs to be better validated against other experimental and simulation results. For the future steps, new experiments will be carried out using modulated input signals and testing other types of lasers in order to improve this methodology.

Acknowledgement - This work was carried out within the framework of the European UnionIntegrated Project FUTON (FP7 ICT-2007-215533).

Luis C. Vieira is sponsored by the Brazilian Government through CNPq and UTFPR, whose support is gratefully acknowledged.

References

Microwave Photonics, pp. 21-24, Awaji, Japan, Nov. 2002.

[2] A. Das, A. Nkansah, N.J. Gomes, I. Garcia, J. -cost

multimode fiber-fedIEEE Microwave Theory Tech., vol. 54, no. 8, pp. 3426-3432, Aug. 2006.

[3 ve asymmetric linearization of radio over fiber links

IEEE Transactions on Vehicular Technology, vol. 51, pp. 1576-1586, Nov. 2002.

[42004.


- 86 -

Dynamic Binary Interleaving Codes for ECMA-368 standard and future Multi-Gigabit Wireless Systems (OMEGA project)

I. Siaud1*, A.M. Ulmer-Moll2

1 Orange Labs RESA/WIN Dept., 4, rue du Clos Courtel BP 91 226, 35 512 Cesson Sévigné Cedex, FRANCE 2 Orange-Labs RESA/WASA Dept., 4, rue du Clos Courtel BP 91 226, 35 512 Cesson Sévigné Cedex, FRANCE

Phone: +33 299 124 629, Fax: +33 299 124 098, E-Mail: [email protected]

Abstract – This paper proposes and evaluates new binary interleaving schemes intended for ECMA-368 proposal and future WPAN Multi-Gigabit Wireless Systems designed in the ICT OMEGA project. A dynamic binary interleaving set-up utilizing a Turbo-Like interleaver is proposed as an alternative to the third stage of ECMA-368 binary interleaving process.

Introduction –This paper presents a novel binary interleaving process intended to cope with narrow band interferers and introduce diversity upon UWB systems. The system utilizes time-variant binary interleaving process as an alternative to the cyclic shifter stage of the ECMA-368 standard binary interleaving [1]. Cyclic shifter has been proposed to introduce diversity upon 6 "on air" transmitted OFDM symbols by shifting small blocks of bits through 6 OFDM symbols corresponding to the block binary interleaving depth. It strengthens interleaving benefits in presence of slow time-variant channel and narrow band interferers. Here, we simply propose a time-variant block interleaver varying from OFDM symbol to OFDM symbol following a periodic pattern composed of n different interleaving rules. The concept is denoted Dynamic Interleaving Codes (DIC) and Dynamic Binary Interleaving Codes (DBIC) when it is performed uponblocks of encoded bits. This concept has been assessed upon data sub-carrier interleaving process with a high number of sub-carrier per OFDM symbol [4]. The system [4] refers to an optimized UWB-OFDM system characterized by a high spectrum efficiency involving a high number of data sub-carrier per OFDM symbol anda dedicated interleaving algorithm denoted Turbo-Like (TL) interleaver providing high interleaving spreading. TL interleaver is characterized by a turbo-based structure generating flexible interleaving patterns [3] in a low complexity manner. TL interleaver has been proposed in the IEEE802.15.3c proposal, in IPHOBAC for WPAN mm-wave radio over Fiber (RoF) applications and evaluated upon optimized multi-RF band Multi-GigaBit Wireless systems (MGWS) [3][5][6]. Here, BDIC is preferentially applied to encoded bits of the ECMA-368 system due to the weaknumber of data sub-carrier per OFDM symbol and we select the TL interleaver. The model allows shortened binary interleaving depth reducing latency on the system. BDIC combined with the TL interleaver is compared to ECMA-368 and TL binary interleavers

considering 1 and 6 OFDM interleaving depth respectively.

The ECMA-368 PHY layer system- The ECMA-PHY layer proposal is based on UWB-OFDM transmission combined with Multi-Band process to introduce time and frequency diversity on radio communications and cope with narrow band interferers. UWB transmission refers to transmission bandwidth size set to 528 MHz, according to FCC part 15 decisions. ECMA-368 system operates in the UWB spectrum mask defined in the {3.1-10.6} GHz. The UWB spectrum is split into 14 sub-channels (bands) spread over 6 band-groups. MB process is performed within every band group where Time Frequency Codes are utilized to carry out time frequency hopping between 6 adjacent OFDM symbols using sub-channels forming the band-group.

Fig. 1:UWB spectrum mask A total of 110 sub-carriers (100 data carriers and 10 guard carriers) are used per sub-channel. 12 pilot sub-carriers allow for coherent detection. Spreading techniques are implemented for low data rates to strengthen FEC coding redundancy. Frequency Domain Spreading (FDS) duplicates 50 symbols in the OFDM symbol following an Hermitian symmetry. Time Domain Spreading (TDS) consists in transmitting simultaneously the same OFDM signal using two separate RF sub-channels associated with a band-group and select the best one. Information bits are firstencoded with a convolutional code of 1/3 mother code rate, and a constraint length of 7. Encoded bits are punctured, interleaved using a three stage interleaver and mapped to data symbols. FDS is eventually implemented and OFDM modulation is performed with a zero-forcing suffix to cope with ISI and RF sub-


- 87 -

channel switching latency. A binary interleaving process completes the MB processing as detailed in the next section. Time related parameters are given in the Table I.

TABLE I : TIME RELATED PARAMETERS

Parameter Value Parameter Value

# of data subcarriersNSD

100 Cyclic prefix duration TCP

70 ns

# of pilot carriersNSP

12 IFFT/FFT period

TFFT ns 242.42

# of DC carriers NDC 1 Symbol interval TSYM

ns 312.42

# of guard carriersNGC

10 FFT size NFFT128

Channel bandwidthBFFT MHz

528 FEC generator polynomials

(133, 165, 171)

ECMA-368 Interleaving design –Binary interleaving is performed using a three stage interleaving applied upon block of encoded bits associated with six "on air" OFDM symbols included in a TFC. Interleaving size varies with the considered interleaving stage the spreading factors. The first stage performs binary spreading over 6 consecutive OFDM symbols based on a matrix interleaver composed of Nr=NCBPS rows (NCBPS

is the number of coded bits per OFDM symbol) and Nc

=6/NTDS columns, NTDS is the time spreading factor set to 2 when TDS occurs. NCBPS depends on FDS factor, the number of bits per OFDM symbol NBPSC and FEC code rate r ,NCBPS=NBPSC NSD r /NFDS . NSD is the number of data sub-carrier per OFDM symbol.The second stage is intra-symbol tone interleaving permuting bits upon different sub-carriers within one OFDM symbol using a matrix interleaver with a fixednumber of columns set to 10 and an interleaving depth K'=NCBPS (NCBPS=NTINT 10). The third stage cyclically shifts block of jNcycl bits within the span of 6 "on air" OFDM symbols. Within each block of NCBPS bits, cyclic shifts are proportional to the OFDM symbol index in the TFC pattern and Ncycl

(Fig. 2). The cyclic shift parameter is given by Ncycl = NTDS NCBPS/6. This allows exploiting frequency diversity, both for time-domain spreading and fixed-frequency-interleaved (FFI) codes. Turbo-Like Interleaving design – The TL interleaver [3][4][6] is built on a turbo-based structure to generate in a flexible way different interleaving patterns using a basic module I (k). TL interleaver provides high interleaving spreading thanks to its algebraic expression. It preserves some dedicated partitioning of data allowing interleaving combined with dedicated data mapping. The system is equivalent to p parallel interleavers where p is one parameter of the interleaver. To perform a comparison with ECMA, the interleavingblock size is set to 6 OFDM symbols.

��

��

� ��

��

��

��

��

��

�

��

��

��

��

��

��

� ��

��

��

��

��!�

Ncycl 2 Ncycl

6 Ncycl

Fig. 2: ECMA-368Binary interleaver (Ncycl=33)

Fig. 3: BDIC implementation for ECMA-368standard

Performance – Link level performance are given with the CEPD propagation model detailled in [3][6][7]. 60 GHz and UWB multipath signatures are equivalent in the case of short range distances [7]. Mode #1 correspond to the ECMA binary set-up with an interleaving depth K1=NBPSC6, mode #2 corresponds to the TL interleaver implementation with the same K1

interleaving depth, mode #3 corresponds to BDIC implementation with 3 different TL permutation rules considering a shortenened interleaving depth set to one OFDM symbol (K2=NBPSC). 16-QAM and QPSK modulation are considered to acheive 1 Gbps on the air interface. BER results are given for LOS and moderate NLOS channels with a delay spread set to 2-.3 ns and 7-8 ns. In the next table, we give the interleaving


- 88 -

spreading ΔL(s) associated with the 3 modes in the case of 16-QAM. ΔL(s) corresponds to the minimum distance between bits separated by s-1 bits for a given interleaving rule L(k).

TABLE II : INTERLEAVING SPREADING PARAMETERS

16-QAM ΔL(s=1) ΔL(s=2) ΔL(s=4) ΔL(s=400)

Mode#1 60 120 240 400

Mode #2 {p,q,j}={8,2,3}

1007 386 772 400

Mode#3 Rule#1 {16,2,1} 143 114 172 200 Rule#2 {2,2,3} 117 166 68 200 Rule#3 {4,2,3} 119 162 76 200

BER results exhibit benefits of DBIC combined with shortened interleaving depth when the propagation channel is selective or when we consider 16-QAM modulated sub-carriers with moderate code rate. For16-QAM modulated sub-carriers assorted with 1/2 code rate, DBIC presents similar performance as the TL interleaver in a static implementation and K1

interleaving depth. With a code rate set to r=3/4, BER performance presents an error floor for the mode#3 if the channel is LOS, due to small interleaving depth and limited interleaving efficiency under QPSK and LOS situations. DBIC does not take advantages of frequency selectivity of the channel and presents degradations facing to static interleaving with higher interleaving depth. Conclusions – BER performance prove that DBIC is an efficient technique to reduce interleaving latency when we consider efficient interleaver such as the TL interleaver. Gains appear under selective configurations

that are compliant with high data rates and modulation levels. These assumptions result from TL interleaving properties :high interleaving spreading are selected within every data symbols and between data symbols,ensuring gains in he case of high modulation levels. This work completes data sub-carrier DIC set-up and assesments evaluated on mm-wave and optimized UWB-OFDM systems[4][6]. The next step of these studies will be to harmonize binary and sub-carrierinterleaving process in a dynamic manner and evaluate the system in presence of narrow band interfers. Extensions are in progress for MIMO UWB-OFDM techniques.

References [1] Standard ECMA-368, "High Rate Ultra Wideband PHY and

MAC Standard", 3rd Edition - December 2008. [2] http://www.ist-iphobac.org [3] Siaud., I, Ulmer-Moll, A.M., "Turbo-like Processing for

Scalable Interleaving Pattern Generation: application to 60 GHz UWB-OFDM systems", IEEE Int. Conf on Ultra Wide Band, Singapore, Sept. 2007.

[4] Siaud., I Ulmer-Moll A.M, "A novel adaptive sub-carrier interleaving application to millimeter-wave WPAN OFDM Systems (IST MAGNET project)", IEEE Int. Conf. on Portable Information Devices, Orlando USA, March 2007.

[5] IST-IPHOBAC deliverable D2.2, June 2008. [6] "Short-Range Wireless Communications", ISBN-13: 978-0-

470-69995-9 - John Wiley & Sons Ed. chapter 18 February 2009

[7] E. Grass, I. Siaud, S. Glisic, M. Ehrig and all, "Asymmetric Dual-Band UWB / 60 GHz Demonstrator", Proc. of IEEE Personal Indoor Mobile Radio Communications, PIMRC'08,September 2008.

�"#��$��%�&'#(

�)��*�+

�)��*��

�)��*�

�)��*��

�)��*��

�)��*��

�)��*��

�)� )� �)� +)� �)� )� ��)� ��)� ��)��,-./��0�12

1�3

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%�5�.�%��,6�

�"#��$��%��*&��

�)��*��

�)��*�

�)��*��

�)��*��

�)��*��

�)��*��

)� �)� +)� �)� )� ��)� ��)� ��)� ��)� ��)��,-./��0�12

1�3

�4��!�%��5�.�%��,6� �4��!�%�5�.�7��,6�

�4��!�%��5�.�%��,6� �4��!�%�5�.�%��,6��4��!�%��5�.�%��,6� �4��!�%�5�.�%��,6�

�4��/�"#��$��&'#(

�)��*��

�)��*�

�)��*��

�)��*��

�)��*��

�)��*��

)� +)� )� ��)� ��)� �)� �+)� �)� ��)� ��)�

�,-./��0�12

1�3

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��/�"#��$��*&��

�)��*��

�)��*�

�)��*��

�)��*��

�)��*��

�)��*��

+)� )� ��)� ��)� �)� �+)� �)� ��)� ��)� �)��,-./��0�12

1�3

�4��!�%�5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%��5�.�%��,6�

�4��!�%��5�.�%��,6� �4��!�%�5�.�%��,6�

Fig. 4: BER performance.


- 89 -


- 90 -

Nonlinear optical Ti:PPLN wavelength conversion modules for free-space communication at 3.8 µm

K.-D. Büchter1,*, C. Langrock2, H. Herrmann1, M.M. Fejer2, W. Sohler1

1 Universität Paderborn, Angewandte Physik, Warburger Str. 100, 33098 Paderborn, Germany 2 Edward L. Ginzton Laboratory, Stanford University, Stanford, California 94305, USA

* Corresponding author: Phone: +49 5251 60-5870, Fax: +49 5251 60-5886, E-Mail: [email protected]

Abstract –. All-optical transmitter and receiver modules for free-space communication at 3.8 µm have been developed, essentially consisting of Ti-indiffused, periodically poled LiNbO3 waveguides. Conversion of C-band radiation to/from the mid infrared is demonstrated.

Middle infrared (MIR) radiation can be advantageousin free-space transmission links due to reduced scattering and scintillation effects compared to shorter wavelength radiation [1], and an attractivetransmission window exists in the atmosphere at around 3.8 µm [2]. This is why we have developed all-optical wavelength conversion modules from 1.55µm to 3.8 µm and back [3]. The converters are basedon χ(2)-nonlinear difference frequency generation (DFG) in periodically poled lithium niobate (PPLN) waveguides. The all-optical conversion enables data-format independent wavelength conversion. Therefore, transmission systems can be realized using readily available C-band components at both ends ofthe transmission line, allowing amplitude, frequency, and phase modulation schemes.

The concept of nonlinear frequency conversion is flexible, and easily transferrable to other wavelengths. The benefits of optical waveguides arethe high conversion efficiencies of the nonlinear processes and the simplified use of single mode optical fibers for easy device handling.

Module design is shown in Fig. 1. We use titanium indiffused waveguides of 18 µm width and 83 mm length, on a LiNbO3 substrate with a periodically poled area of 80 mm. A 3 mm long taper region is included to improve coupling to the fundamental waveguide modes. A pump at 1100 nm wavelength is used to convert a signal at 1550 nm to the MIR band.

For characterization purposes, a tunable C-band external cavity laser (ECL), followed by an erbium doped fiber amplifier (EDFA), is used as signal source. Pump and signal are joined by a WDM coupler and fiber butt-coupled to the waveguide on the 5.3° angle-polished input side of the sample, using an 8°-polished glass ferrule. Angled endfacesare used to suppress back-reflections, and thus Fabry-Perot interference effects, within the devices. A poling periodicity of 26.65 µm is used to provide quasi-phasematching for the DFG process. The devices are operated at about 200°C to suppress photorefractive effects. The phasematching characteristics of transmitter and receiver modules, along with a calculated curve, are given in Fig. 2, top. Phase-matching occurs at a signal wavelength of 1554.5 nm, at a temperature of197°C. The measured curves show an asymmetry which is due to a parabolic inhomogeneity gradient along the waveguide.

1551.5 1554.5 1557.5

0.0

0.5

1.0

Gen

erat

ed id

ler

pow

er [a

.u.]

Signal wavelength [nm]

Transmitter Receiver Theory

3770 3760 3750Idler wavelength [nm]

0 5000 10000 15000 200000

4

8

12

16 Ppump

= 150 mW P

signal= 50 mW

λp = 1100 nm

λs = 1554 nm

λi = 3763 nm

Gen

erat

ed id

ler

pow

er [m

W]

Pump power x signal power [mW2]

Fig. 2.: Phase-matching characteristics (top) and power characteristics (bottom) of the transmitter module.

WDMcoupler

Taper Region

MIR @3800nm

Pump @ 1100nm

Signal @ 1550nm

T-Stabilized OvenPM-Fibers Dielectric Mirror

(Receiver only)

PPLN SubstrateTi-Diffused Waveguide

Fig. 1.: Transmitter / receiver design scheme.


- 91 -

In Fig. 2, bottom, the power characteristics of thetransmitter module is given. Here, measured power levels of the generated 3.76 µm idler radiation (which are corrected for residual losses beyond the waveguide) are shown (symbols). Alongside the measured powers, theoretical curves are given. Two different cases are shown here: In the first case, pump power is fixed at 150 mW (coupled to the waveguide), which means that the gain of the EDFA signal amplifier changes along the abscissa. At high signal levels, a roll-off of the curve is evident due to pump depletion. In the second case, signal power iskept fixed at 50 mW, while pump power is varied. Here, a superlinear behavior becomes evident at high pump power levels, which is due to parametric gain in the nonlinear process. At low power levels, an internal conversion efficiency of 69 %/W is determined, normalized to the pump power level. Measurements and theoretical curves match surprisingly well, although the impact of the observed inhomogeneities was not considered in the calculation. So actually, measured power levels areslightly above the theoretical expectation. This is due to some uncertainty concerning the mode-distributions in the deep MIR-waveguide. Also, the loss figures at near-infrared wavelengths are not precisely known and difficult to measure due to themultimode characteristics of the waveguide. After characterization of the individual modules, atransmission experiment has been performed, using both, transmitter and receiver modules. The setup is shown in Fig. 3. Here, the 1100 nm pump radiation is split using a ratio of 50:50, to act as pump for both modules. Via DFG the MIR idler wave is generated in the transmitter module, which is coupled out of the waveguide using a CaF2 lens of 8.3 mm focal length. Two gold mirrors are used in the 1.5 m free-space path to steer the beam to the receiver. Residual pump radiation is removed from the transmission path using an anti-reflection coated Ge filter. Using a secondCaF2 lens, the 3.8 µm radiation is coupled to the receiver waveguide, where it is again converted to 1550 nm by DFG. At a MIR power level of 4.85 mW in the FSO link, generated by pump and signal powers of 150 and 80 mW, respectively, coupled to the transmitter waveguide, about 100 µW of signal power is regenerated in the receiver waveguide. The transmission therefore equals -29 dB from waveguideto waveguide. Of this, around -14 dB are caused by coupling losses and residual losses in the Ge filter. Another -6 dB arise on each end due to waveguide / fiber coupling. In conclusion, a C-band–MIR–C-band transmission line was set up with an overall transmission of -41 dB, with -15 dB due to the parametric process;

-26 dB are caused mainly by coupling losses in-between waveguides and fibers. Parametric losses can be reduced by pumping with higher power levels,potentially leading even to parametric gain. Free-space coupling losses from waveguide to waveguide can be reduced by the use of optimized bulk optics. Data transmission experiments via the link were performed successfully; transmitting an analogue QPSK modulated signal at 2.488 Gbit/s, low bit error rates could be demonstrated [4]. The noise impact of the system is negligible, and the signal to noise ratio of the parametric process exceeds 106 [5].

Acknowledgement - This work was supported by a subcontract of CeLight, Inc.

References [1] R. Martini et al., “High-speed modulation and

free-space optical audio/video transmission using quantum cascade lasers,” IEEE Elect. Lett., vol. 37, pp. 191-193 (2001).

[2] W. A. Traub, M. T. Stier, “Theoretical atmospheric transmission in the mid- and far-infrared at four altitudes,” Appl. Opt., vol. 15, No. 2, pp. 364-377, (1976)

[3] D. Büchter, C. Langrock, H. Herrmann, M. Fejer, W. Sohler, “All-optical Ti:PPLN wavelength conversion modules for free-space optical transmission links in the mid-infrared“, Opt. Lett., vol. 34, No. 4, pp. 470-472 (2009)

[4] E. Ip, D. Büchter, C. Langrock, J. M. Kahn, H. Herrmann, W. Sohler and M. M. Fejer, “QPSK Transmission over Free-Space Link at 3.8 μm using Coherent Detection with Wavelength Conversion,” ECOC2008

[5] S. Orlov, W. Grundkötter, D. Hofmann, V. Quiring, R. Ricken, H. Suche, and W. Sohler: "Mid infrared integrated optical parametric generators and oscillators with periodically poled Ti:LiNbO3 waveguides" in "Mid-Infrared Coherent Sources and Applications", M. Ebrahimzadeh and I.T. Sorokina (Eds.), NATO Science Series - B: Physics and Biophysics, Springer, Dordrecht (2008).

HWPs

PM Fibers

Transmitter Module

Receiver Module

GeFilter

WDM

WDM

MIR-FSO Link(3.8 µm)

1.55 µm

1.55 µm

Pump Unit

Input Signal

1.1-µmFiberLaser

Collimators

PBSIsolator

Output Signal

Ti:PPLN Sample

Ti:PPLN Sample

Fig. 3.: Transmission link experiment: both modules are combined to form a free-space transmission line with twofold wavelength conversion.


- 92 -


Access networks

Session Chair: B. CharbonnierFrance Telecom R&D, Lannion (France)

Optical Communication Networks in Support of User Mobility and Networks in Motion (invited)S. Azodolmolky, C. Tsekrekos, K. Kanonakis, T. Papachristos, I. Tomkos Athens Information Technology, Athens (Greece)

Optical Transmission Link Design for a Distributed Broadband Wireless System D. Wake, A. Nkansah, N.J. Gomes University of Kent, Canterbury (United Kingdom)

Simultaneous Direct Detection of Signals Carried on Baseband and SubcarrierM. Chacinski, U. Westergren, P-Y. Fonjallaz, R. Schatz Kista Photonic Research Centre (KPRC), Kista (Sweden)

Quality of Service Control in a multi-access integrated network based on Virtual Private LAN Service L. Rea1, A. Valenti1, S. Pompei1, L. Pulcini1, M. Celidonio1, D. Del Buono2

G.M. Tosi Beleffi21 Fondazione Ugo Bordoni, Roma (Italy) 2 Ministero dello Sviluppo, Economico Dip. Com, Roma (Italy)

Photonic Millimeter-Wave System for Broadband Wireless Access M. Weiß1, A. Stöhr1, S. Fedderwitz1, V. Rymanov1, B. Charbonnier2, D. Jäger1

1 University of Duisburg-Essen, Duisburg (Germany) 2 France Telecom R&D, Lannion (France)

60 GHz Wireless Signal Generation with 10 Gb/s 4QAM Modulation using Photonic Vector Modulator R. Sambaraju Universidad Politécnica de Valencia, Valencia (Spain)


- 93 -


- 94 -

European workshop on photonic solutions for wireless, access, and in-house networks

20 May 09

Optical Communication Networks in Support of User Mobility and Networks in Motion 1

20 May 09 Optical Communication Networks in Support of User Mobility and Networks in Motion 1

“Optical Communication Networks in Support of User Mobility and Networks in Motion”

“Optical Communication Networks in Support of User Mobility and Networks in Motion”

Siamak Azodolmolky, Christos Tsekrekos, Kostis Kanonakis,Thalis Papachristos, and Ioannis TomkosAthens Information Technology (AIT)

{sazo, tsek, kkan, itom}@ait.edu.gr, [email protected]


18-20 May 2009Duisburg, Germany

Session 7, Access Networks

ATHENS INFORMATION TECHNOLOGYATHENS INFORMATION TECHNOLOGY


OutlineOutline

• Introduction• Past, Current and Future Activities• Activities in the framework of “BONE WP23”

– Typical project on Optical communication networks in support of user mobility and networks in motion

– WP objectives– Planned activities

• Conclusions• Questions & Answers


- 95 -


20 May 09



Introduction (1/3)Introduction (1/3)

• With the growth of new wireless access technologies being developed and deployed, including WiMAX, optical wireless (or free-space optics) , and Wi-Fi, wireless traffics has already contributed to an increasing amount of share of the total traffic carried by wireline networks.

• At the same time, demands for higher bandwidth,longer reach and better interoperability are stronger than ever.

• The convergence among various wirelesstechnologies as well as wireline technologies(mainly fiber optic based) is a key to satisfy such demands at a low CAPEX and OPEX.


Introduction (2/3)Introduction (2/3)• Upcoming networking concept based on user mobility and

ubiquitous connectivity. • Individual subscribers carrying around their short-range

Personal Area Network (PAN).• Slow (people moving on foot) or fast (cars, trains, airplanes)

mobility patterns • Virtual Home/Virtual office

– Web surfing– Video on demand– Online gaming– Infotainment– E-mail– Video conferencing– Business applications

• Anytime, anywhere:– In car, on train, on plane,…


- 96 -


20 May 09



Introduction (3/3)Introduction (3/3)• Moving networks will need to communicate with each other or the

outside world.• Unique new form of network, namely the “network in motion” is

required.• Intelligent components and devices needed to provide guaranteed

content delivery efficiently and in a secure manner.• Use of optical network solutions in aggregation and core part is

essential and requires extensive research in both networking andtechnology areas

• Collaborative research towards three directions is required:– Intelligent technologies and design challenges for wireless access (RoF, FSO,

or conventional wireless solutions with optical fiber feed)– Networking properties and switching characteristics for the aggregation and

core networks (e.g. Switched Ethernet based solutions or advance schemes like OBS/OPS)

– Control plane and signaling algorithms and protocols for networks in motion (e.g. MAC layer design or network layer approaches with QoS quarantines, resource reservation approaches etc.)


Past, Current and Future ActivitiesPast, Current and Future Activities

• ORCLE Project (DARPA) (03~04)• FAst MOving USers (FAMOUS Project) – INTEC,

Ghent University• Broadband in Trains (BIT) – AIT• IEEE Comm. Soc. FiWi (Sub-TC)• Cisco® Motion• Integrated Service Networks• EU FP7 Call 4 proposals

– ACCORDANCE

• BONE - Typical Project (WP23)


- 97 -


20 May 09



ORCLE Project (1/2)ORCLE Project (1/2)

• Optical & RF Combined Link Experiment • High bandwidth access to/from battle space based

on “Mobile Free Space Optics”


ORCLE Project (2/2)ORCLE Project (2/2)

• Bronze portal on top of gimbals positioning stand• Providing FSO line of sight for moving objects (planes, cars,

trains,…)

• Source: ORCLE project, http://www.darpa.mil/STO/Solicitations/orcle/index.htm


- 98 -


20 May 09



Fast Moving Users (FAMOUS), INTEC, Ghent UniversityFast Moving Users (FAMOUS), INTEC, Ghent University

• Conceptual architecture:– Access network (handover issue)– Aggregation network (Dynamic traffic

condition)– Core network (Service Gateway)

• Development of the access network and the aggregation network were the main focus of the FAMOUS project.


Broadband in Trains (BIT) (1/2)Broadband in Trains (BIT) (1/2)

• Access and distribution layers of the proposed solution

• Free-Space-Optic (FSO)• Wireless backup link• Novel MAC protocol for

dynamic traffic consideration


- 99 -


20 May 09



Broadband in Trains (BIT) (2/2)Broadband in Trains (BIT) (2/2)

• Proposed architecture


IEEE Comm. Soc. TC FiWiIEEE Comm. Soc. TC FiWi

• IEEE Comm. Soc. – Emerging Technologies committee– Integrated Fiber & Wireless Technologies (TCFiWi) sub-committee– TCFiWi addresses architectures, techniques, and interfaces for the

integration of fiber and wireless network segments in a unifiedwired-wireless infrastructure (across core, access, and long-haul).

– Its objective is to enhance interoperability and resource sharing among wired and wireless segments so that mixed networks can provide better support for converged multimedia service independent of:

• Users' locations, • Terminal device capabilities,• Access media

– http://www.comsoc.org/socstr/org/operation/comm/subemerging.html


- 100 -


20 May 09



Cisco® Motion (1/2)Cisco® Motion (1/2)

• Cisco® Motion is a new practical approach to resolve challenges experienced when enabling business mobility. By making it easier to implement and manage, Cisco Motion helps IT meet and exceed business mobility demands.

• By delivering a holistic approach to business mobility, Cisco Motion helps customers:– Converge disparate networks– Manage the increasing quantity and diversity of mobile

devices– Deliver device-to-network-to-application integration and

security– Create an open platform for the development of mobility

applications


Cisco® Motion (2/2)Cisco® Motion (2/2)

• Cisco® Motion solution– Unified Wireless Network Migration Services– Wireless LAN Secure Mobility Services– Wireless LAN Voice Services– Context-Aware Services– Wireless LAN Guest Access Services– Application Services– Wireless LAN Optimization Service– Unified Communications Services


- 101 -


20 May 09



Integrated Service Networks (1/4)Integrated Service Networks (1/4)

Wavelength interleaving improves spectral efficiency, by using the optical spectrum between the carrier and the sideband of mm-wave RoF signals.

Lim et al., Tutorial, OFC 2009.

1, 2, 3, … : carriers of the RoF signalss1, s2, s3, … : single sidebands (SSBs) of the RoF signals



Integrated network for mm-wave (RF), microwave (IF) and baseband (BB) services, based on wavelength interleaving.

Bakaul et al., IEEE PTL, 18(21), p. 2311, 2006.

RFc: carrier of the RoF signalRFs: SSB of the RoF signal


- 102 -


20 May 09



Integrated network for mm-wave RoF and baseband (BB) services based on wavelength interleaving and a 4×4reconfigurable optical cross-connect which supports dynamic channel allocation and can serve user mobility.

Vegas Olmos et al., IEEE/OSA JLT, 26(15), p. 2506, 2008.




User mobility can be accommodated by assigning a RoF channel to a user and optically routing this channel to the AP at the range of the user.


- 103 -


20 May 09



ACCORDANCE (1/2)ACCORDANCE (1/2)• ACCORDANCE introduces a novel ultra high capacity (even reaching the

100Gbps regime) extended reach optical access network architecture based on OFDMA (Orthogonal Frequency Division Multiple Access) technology/protocols, implemented through the proper mix of state-of-the-art photonics and electronics. Such architecture is not only intended to offer improved performance compared to evolving TDMA-PON solutions but also inherently provide the opportunity for convergence between optical, radio and copper-based access.

• ACCORDANCE hence aims to realize the concept of introducing OFDMA-based technology and protocols (Physical and Medium Access Control layer) to provide a variety of desirable characteristics, such as increased aggregate bandwidth and scalability, enhanced resource allocation flexibility, longer reach, lower equipment cost/complexity and lower power consumption, while also supporting multi-wavelength operation. In addition, it enables the convergence of the optical infrastructure with standard wireless solutions, thus offering a way to integrate dominant wired and wireless technologies in a hybrid access network supporting seamless ubiquitous broadband services.


ACCORDANCE (2/2)ACCORDANCE (2/2)

• Proposed architecture


- 104 -


20 May 09



Building the Future Optical Network in Europe (BONE)Building the Future Optical Network in Europe (BONE)

• Typical project on Optical communication networks in support of user mobility and networks in motion (WP23)– Duration: 2 Years– Involved partners:19 – Proposed research topics (1st year): 27– Total PM allocation: 14.75PM– Current status:

• Dynamic repository of research topics• Repository of expertise• Joint activities for the first year are planned

– More information:• BONE Deliverable D23.1 (http://www.ict-bone.eu)


Description of WorkDescription of Work• Three main activities running in parallel for the duration of the project

– Technology-oriented one: Focus on investigation and development of novel approaches to support rapid handover and high bandwidth connectivity.

– Aggregation network-oriented: Switching solutions with rapid reconfiguration characteristics

– Network- and control layer-oriented: Study new MAC, routing and signallingprotocols to support networks in motion.

• The three activities cover research areas that can be initially developed independently

• However it is important to define a common knowledge platform about possible solutions and the properties of these novel network approaches that support seamless connectivity of various wireless users in a rapidly reconfigurable environment.

• Purpose of this knowledge platform:– Provide basic requirements and characteristics for the novel technology and

networking solutions – Identify limitations and challenges and consequently push technology and

networking towards these directions.• Join together efforts that could possibly evaluate or even demonstrate

complete solutions in support of the objectives.


- 105 -


20 May 09



Novel Wireless Access Technologies Novel Wireless Access Technologies

• Investigation of new technological solutions• Development of intelligent devices for the mobile

access part of the network. • Solutions should allow both a slow and fast mobility

patterns• Achieve connections in a dynamic and contentious

manner• Possible solutions that will be investigated are

based on:– Conventional Wireless technology– Radio over Fibre (RoF)– Free space optics (FSO)


Switching TechnologiesSwitching Technologies

• Possible network topology:– Fixed networks connecting high speed base stations with the

backbone core– Mobile users passing through base stations require immediate access

and high speed connectivity– Node reconfiguration becomes an important issue

• Most appropriate switching solutions based on data-oriented switching– Meet both high capacity and fast reconfiguration requirements– Solutions may range from simple Ethernet-based up to the more

advanced ones (OBS/OPS)

• Comparative studies to identify optimum solutions in terms of:– Efficiency– Number of users supported – Added complexity


- 106 -


20 May 09



Networking ProtocolsNetworking Protocols

• Effort concentrated mainly on development of control and signaling protocols for seamless end-to-end connectivity of mobile users

• Focus on the following areas:– MAC optimization and design issues for fast base station

identification and handover– Optimum signaling requirements and resource reservation solutions– Fast path identification algorithms and data switching– QoS guarantees in fast reconfigurable networks

• Previous aspects important for networks in motions due to the specific characteristics and requirements described earlier

• Studies closely related with the general switching approaches proposed in the relevant activity


Proposed Research Topics (1/2)Proposed Research Topics (1/2)• Wireless access technologies in support of networks in motion (RoF)• Wireless access technologies in support of networks in motion (FSO) • Switching technologies in support of networks in motion• Networking protocols in support of networks in motion• MAC optimization and design issues for fast base station identification

and hand over• Converged MAC algorithms for unified optical wireless functionality• Optimum signaling requirements and resource reservation solutions• Fast path identification algorithms and data switching • QoS quarantines in fast reconfigurable networks • Radio over fiber transmission and in the optical beam forming of the

antennas• Millimeter (mm) wave wireless communication systems and 70 GHz radio

front-end technology• UWB Radio-over-fiber transmission in indoor environments using

different media• Design and fabrication of coherent and envelope detection wireless

photonic receivers• FSO with studies and experiments for networks in motions solutions


- 107 -


20 May 09



Proposed Research Topics (2/2)Proposed Research Topics (2/2)• Hardware implications issues for networks in motion Ericsson,• Optical switching architectures capable of supporting user mobility.• Radio-over-fiber techniques for enabling Personal Network concepts• Robust radio-over-fiber techniques• Node architectures and the control/monument layer requirements • Radio over fiber technology with the focus on plastic optical and other

MM fiber• Mobile access networks based on free-space optical technology• Optimizing service delivery in a converged hybrid optical-wireless

network• Protocol routing over hybrid optical wireless mesh networks• End-to-End QoS and Service Delivery over Heterogeneous Network

Access• All-optical Routing Architecture of Radio Signals using Label

Processing Technique for In-building Optical Networks• Mitigating the Impact of Traffic Pattern Variations on Multi-Layer

Optical Networks• State of the art definition for components supporting FSO networks

in motion


Joint Activity 1Joint Activity 1

• Hardware implications issues for Networks in motion

• Partners:– Ericsson– Universität Duisburg-Essen (UDE)

• Objectives– In this JA the reliability of components and harware

implications issues for network in motion and sensor applications will be investigated.

• Target outcome:– Improved design and reliability of hardware for networks

in motion is the main outcome of this JA


- 108 -


20 May 09




• State of the art definition for components supporting FSO networks in motion

• Partners:– Instituto de Telecomunicacoes (IT)– Superior Institute of Communication and Information Technologies

(ISCOM)– Athens Information Technology (AIT)

• Objectives:– Free space optic is one of the options for high-bandwidth (and cost

effective) means of support for networks in motion. In this JA, existing techniques and component features will be collected andthe performance of different techniques will be benchmarked/compared

• Target outcomes:– A comprehensive survey on the state-of-the-art technologies for FSO-

based network in motion



• Converged MAC algorithms for unified optical wireless functionality

• Partners:– University of Hertfordshire (UH)– University College of London (UCL)– Athens Information Technology (AIT)

• Description:– A unique control layer (for SLA and service differentiation

provisioning) is required to implement dynamic resource allocation based on conventional NRZ over PONs.

– To enable efficient integration, an effective mapping mechanism is required between PON priority queue for equivalent QoS.

• Objectives and target outcomes:– Unified MAC protocol for legacy PON and wireless networks– Hardware implementation of the underlying network architecture


- 109 -


20 May 09




• UWB Radio-over-fiber transmission in indoor environments using different media

• Partners:– Universidad Politecnica de Valencia (UPVLC)– Athens Information Technology (AIT)

• Objectives:– RoF transmission of both OFDM and IR UWB signals for in-building

applications, e.g. offices or home– Testing different media such as standard single mode fiber (SSMF),

multimode fiber (MMF), plastic optical fiber (POF)

• Target outcomes:– Performance evaluation of UWB signals in RoF transmission for in-

building applications (QoS, bit rate, spectral efficiency, maximum reach, etc.)


Joint Activity 5Joint Activity 5• Optimizing service delivery in a converged hybrid optical-

wireless network• Partners:

– Universidad Autónoma de Madrid (UAM)– Athens Information Technology (AIT) – Technische Universiteit Eindhoven (TUE)

• Objectives:– Optimization of service delivery and resource usage under user

mobility.– Study of re-configurability in various networking layers and dynamic

mechanisms for quick network adaptation to accommodate broadband mobile networking.

• Target outcomes:– Key solutions in various networking layers (architecture, protocols,

tools)– Verification and benchmarking of the realized innovations with

respect to the existing solutions


- 110 -


20 May 09



Joint Activity 6 (1/2)Joint Activity 6 (1/2)

• All-optical Routing Architecture of Radio Signals using Label Processing Technique for In-building Optical Networks

• Partners:– Universidad Autónoma de Madrid (UAM)– Technische Universiteit Eindhoven (TUE)– Athens Information Technology (AIT)

• Objectives:– Optical routing performed in a home communication controller

(HCC), which arranges communications between rooms and routes signals to the proper rooms.

– Investigation of a new all-optical HCC architecture that forwards the radio signals to the specific rooms based on the label information.

• Target outcomes:– Investigation (experimentally) of a new all-optical HCC architecture

for routing of RoF signals with label processing techniques


Joint Activity 6 (2/2)Joint Activity 6 (2/2)

• HCC interfaces the access network and the in-building network, providing optical routing functionality.


- 111 -


20 May 09



BONE WP23 (JA Summary – Year 1)BONE WP23 (JA Summary – Year 1)


ConclusionsConclusions

• Networks in motion will play a central role in the people’s live in near future

• A brief overview of past, current and future activities introduced in this presentation, which shows the importance of the topic

• In the framework of BONE WP23 three main directions has been identified:– Intelligent technologies and design challenges for

wireless access in networks in motion– Investigation of networking properties and switching

characteristics for the aggregation and core networks in support of networks in motion

– Development of control plane and signaling protocols for networks in motion


- 112 -


20 May 09



Q & AQ & A

• Thank you!

• For more information visit:– http://www.ict-bone.eu

• You can also contact:– Ioannis Tomkos ([email protected]) (BONE WP23 Leader)– Siamak Azodolmolky ([email protected])– Christos Tsekrekos ([email protected])

• Electronic version of this presentation is available at:– http://personals.ac.upc.edu/siamak/publications.html– (Under “Technical Presentations” anchor)


- 113 -


- 114 -

Optical Transmission Link Design for a Distributed Broadband Wireless System

D. Wake*, A. Nkansah and N.J. GomesBroadband and Wireless Communications Group, University of Kent, Canterbury, UK.

* Corresponding author: Phone: +44 1229 823244, E-Mail: [email protected]

Abstract An analogue optical transmission link has been designed for connecting central units and remote access units in a distributed broadband wireless system. The link is low cost and its performance is such that it has minimal impact on the wireless range of the system.

Introduction The European integrated project FUTON is developing a network concept for next generation wireless systems involving the use of distributed antenna systems and centralized processing in order to achieve cell throughputs of the order of 1 Gbps [1]. This level of cell throughput will require wide radio channel bandwidths, high levelmodulation schemes and multiple MIMO channels. The FUTON architecture consists of a central unit(CU) which contains the centralized processing functions and a number of remote access units (RAUs) which contain the radio functions. This paper looks at the design and performance of the optical transmission links that connect the CUs to the RAUs.

Link Requirements The key radio parameters of the FUTON distributed broadband wireless system (DBWS) are given in Table 1. The DBWS base station (BS) has digital IQ input and output signals, similar to the open base station architectures developed for UMTS and WiMAX [2]. There are four Tx and four Rx signals; two for MIMO and two for sectorization. Control and management channelsare omitted here for simplicity. To minimize cost in the fibre distribution network, all channels should preferably be supported using a single optical wavelength. An optical power budget of 10 dB has been allowed to take into account fibre attenuation, splice and connector losses and the insertion loss of any passive optical components in the fibre network.

Link Design The obvious choice of transmission technology would appear to be digital since the DBWS BSs have digital inputs and outputs. However digital links would require a line rate of more than 20 Gbps to transport a total of 400 MHz of spectrum (4 x 100 MHz) per link direction for the radio channels.These links, and the associated multiplexers and demultiplexers, would be very costly. The most cost effective type of design uses analogue optical links [3]. Frequency translation must be used (transmission over the fibre is at IF) because the radio channels are all at the same carrier frequency and a single optical wavelength is preferred. This paper looks at low costanalogue optical links using uncooled DFB lasers and pin photodiodes. A simplified layout of the link design is shown in Fig 1. In the downlink direction, the four IQ signals from the BS are converted to analogue and then to IF using the IQ modulators. They are power combined and applied to the optical link. At the RAU the composite signal is split and filtered and then converted to RF. The opposite sequence applies to the uplink. A reference signal is also transmitted (not shown) in order to frequency lock the local oscillators in the RAU.

Optical Link and Amplifier ParametersParameters for the optical link design were obtained by constructing a link using low cost devices (anuncooled Teradian DFB laser and an Appointechphotodiode). At a frequency of 1.8GHz, a laser bias current of 45 mA and an optical loss of 10 dB the link had a gain of -36 dB, a P1dB of 17 dBm, an IIP3 of 30 dBm and an EIN of 130 dBm/Hz. EVM measurements were performed so that the maximum input power to the link could be found. Fig 2 shows the variation of EVM with input power for the IEEE802.11n standard at a channel bandwidth of 40MHz and IFFT size of 128 for 16-QAM and 64-QAM modulation. This standard was chosen because it was the closest available to the DBWS specification. The EVM requirement for 16-QAM and 64-QAM is 14.1 % and 5.6 % respectively, based

Table 1: Key radio parameters of the FUTON DBWS

Parameter ValueRadio channel bandwidth Up to 100 MHz

MIMO 2 x 2Sectors per RAU 2

Modulation scheme Up to 256-QAMMaximum EIRP 46 dBm

Minimum approach distance 2 mMaximum mobile Tx power 33 dBmMinimum mobile Tx power -10 dBm

RAU antenna gain 10 dBiPath loss exponent 2 (open)


- 115 -

on CNR requirements of 17 dB and 25 dB [4]. The EVM requirement for 256-QAM is 2.2 %, based on an expected CNR requirement of 33 dB. For a single channel, the maximum input power levels where these limits are reached are 11.2 dBm, 7.3 dBm and 4.7 dBm for 16, 64 and 256-QAM respectively. Note that these values must be reduced by 6 dB in order to accommodate the four radio channels in the design.The maximum gain of the uplink amplifier chain is limited by the minimum approach distance and the maximum input power to the uplink laser (fixed by the 256-QAM value). For the parameters given above, the maximum uplink amplifier chain gain is 48 dB. The maximum gain of the downlink amplifier chain is limited by the maximum EIRP, the optical link gain and the maximum input power to the link(fixed by the 256-QAM value). For the values given above, the maximum downlink amplifier chain gain is 73 dB. Although these values are high, note that the gain is split between RF and IF frequencies and therefore should not cause instability problems.

Wireless Range Calculations Using the parameters given in Table 1 we first calculate the wireless range for the case where there is no optical link as a benchmark. Assuming a receiver noise

figure of 5 dB and an implementation loss of 5 dB, the sensitivity of both the base station and mobile receiver is -67 dBm, -59 dBm and -51 dBm for 16, 64 and 256-QAM respectively. The wireless range is calculated from the maximum path loss and is shownin Table 2. Range is limited by the uplink because the mobile Tx power is relatively low. The optical links add noise to the system and the wireless range is therefore reduced. The wireless range with the optical links is also given in Table 2 using the optical link and amplifier parameters given in the previous section. It is clear that the optical links have minimal impact on wireless range.

Conclusions A link design has been produced for the distributed broadband wireless system currently being specified by the FUTON project. This design meets the requirement of low cost by using frequency translated analogue optical transmission on a single wavelength for the four broadband radio channels.The additional noise introduced by the optical links has minimal impact on the wireless range.

Acknowledgement - This work was partially supported by the European Commission under Grant Agreement No. FP7-ICT-2007-215533 (FUTON).

References[1] www.ist-futon.eu[2] www.cpri.info and www.obsai.com[3] A Comparison of Remote Radio

Head Optical Transmission Technologies for Next Generation Wireless Systemsto ECOC 2009.

[4] EN 300 744 v1.1.2, 2008.

Fig. 1.: Simplified link design. S1 and S2 are the two sectors; Tx1, Tx2, Rx1 and Rx2 are 2x2 MIMO channels.

0

2

4

6

8

10

12

14

16

-8 -6 -4 -2 0 2 4 6 8 10 12

Erro

r Vec

tor

Mag

nitu

de, %

Input Power, dBm

16-QAM limit

64-QAM limit

256-QAM limit

Fig. 2.: EVM as a function of input power for analogue optical link using IEEE 802.11n at 40MHz bandwidth for 16-QAM (red) and 64-QAM (blue) modulation.

Table 2: Wireless Range, m (uplink limited)

16QAM 64QAM 256QAMWithout optical link 2157 859 342

With optical link 1990 792 315


- 116 -

Simultaneous Direct Detection of Signals Carried on Baseband and Subcarrier

M. Chaciński 1,*, U. Westergren 1, P-Y. Fonjallaz 1, R. Schatz 1 1 Kista Photonic Research Centre (KPRC), Royal Institute of Technology (KTH),

Electrum 229, 164-40 Kista, Stockholm, Sweden * Corresponding author: Phone: +46 8 790 4054, Fax: +46 8 790 4090, E-Mail: [email protected]

Abstract - Simultaneous 10Gbps baseband and 1Gbps on 27GHz subcarrier modulation and transmission over 58km long standard single mode fiber (SSMF) was realized. The signals were added to modulate a single light-intensity modulator. An optical circulator and a narrowband fiber Bragg grating (FBG) filter were used to separate them in the optical domain and permit simple direct detection. Introduction – The optical fiber is more and more employed all the way into the home to enable fast internet connection. In some cases it is interesting to utilize the same fiber to carry different signals of different formats [1] that can be easily separated and detected. One example is cable TV. Although broadcast television can be sent via internet, it is still common to use a separate network for this service. A simple solution to extend the use of an existing fiber optic link to also distribute digital or analog cable TV would hence be attractive. The extra data channel could alternatively be used to carry header or service information of the main channel. In this work we propose a cost effective system to increase the use of a single 100GHz wide channel dedicated to high-speed digital data by inserting one or more additional signals on electrical subcarriers. The data on the subcarrier is recovered with passive optical filtering and direct detection without the need of a local oscillator in the receiver [2]. The advantage compared to wavelength division multiplex is that the additional signals can utilize the same transmitter and WDM channel. The inserted signals can be any kind of data stream (i.e. digital microwave radio) which could carry service such as optical header, another data stream, or additional signal. To study the concept, we have investigated a case with one high-speed baseband data signal combined with another high-speed signal modulated on a microwave frequency subcarrier, where both signals are used to modulate the intensity of the light in a fiber optical transmitter. System – In order to send and detect more than one signal at a time, some parts of the transmitter as well as on the receiver side need to be adapted. The modifications have to be made to avoid reconfiguration of existing networks such as electro optical (EO) converters and data receivers. Figure 1 (left) shows the building block of the transmitter.

Fig.1. Schematic of the transmitter (left) and receiver unit (right side).

A baseband data stream of 10Gbps Non-Return-to-Zero (NRZ) Amplitude-Shift-Keying (ASK) Pseudo-Random-Bit-Sequences of word length 27-1 (PRBS7) was supplied from a Pulse-Pattern Generator (PPG, Anritsu MF1758A). An additional signal, composed of 1Gbps ASK NRZ PRBS7 was created by another PPG (Anritsu MP1701A) which modulated a 27GHz electronic subcarrier generated with an Analog Signal Generator (ASG, Agilent E8252D PSG). Low-pass filters on each output of the PPGs were introduced to limit the frequency spectrum to the necessary minimum. The up-converted signal and the baseband modulation were added by a 3dB power splitter. In order to protect instruments and prevent signal distortion, sets of attenuators and amplifiers were introduced to form broadband isolators. At the input of an EO converter, the power level of the 27GHz carrier was chosen to be at least 10dB lower than the power of the 10Gbps baseband modulation. Thus the amplitude of the subcarrier signal was lower than the power at ‘0’ level of the 10Gbps signal, to minimize the perturbation to the baseband modulation. An optical carrier of 6dBm at 1.53µm wavelength was generated by a Tunable Laser Module (TLM, Agilent 8163A) and intensity modulated by a commercial 30GHz Mach-Zehnder modulator (Sumitomo Inc). The amplitude of the modulation signal and the bias of the modulator were adjusted to optimize the signal quality. The baseband (10Gbps) modulation of the optical carrier generates two sidebands next to the carrier

ASGLO

MixerPPG

1 GbpsPPG

10 Gbps

TLMOUTPUT

M-ZModulator

ASGLO

MixerPPG

1 GbpsPPG

10 Gbps

TLMOUTPUT

M-ZModulator

Fiber

FBG

10 GbpsReceiver

1 GbpsReceiver

EDFA

EDFA

BPF

OpticalCirculator

Fiber

FBG

10 GbpsReceiver

1 GbpsReceiver

EDFA

EDFA

BPF

OpticalCirculator


- 117 -

frequency. The subcarrier with 1Gbps signal will generate one additional sideband on each side of the optical carrier but separated from it by the subcarrier frequency, 27GHz. Since the subcarrier itself is double sideband modulated with the 1Gbps signal, the width of these subcarrier sidebands will be twice the spectral width of the 1Gbps signal. If the subcarrier itself is strong enough compared to the 1Gbps modulation, each subcarrier sideband can be regarded as an intensity modulated optical carrier. Hence, by optically filtering out one of the two subcarrier sidebands and directly detecting it with an intensity detector, the 1Gbps signal can be detected without any local oscillator. The theoretical minimum subcarrier frequency is set by the sum of the spectral width of 10Gbps signal and the 1Gbps signal. In practice it is determined by the selectivity of the optical filters and the amount of cross talk that can be tolerated. The schematic of the receiver is depicted on the right side of Fig 1. The design aim has been to incorporate the subcarrier signal in an existing baseband system with minimum extra complexity. The optical filtering and separation of the signals were done with a Fiber Bragg Grating (FBG) and an optical circulator. For the purpose of our experiment, the FBG [2] was created in Ge-doped fiber by UV luminescence. The gratings have about 20GHz bandwidth for the reflected light and >20dB signal suppression, where the flank is ca 20GHz wide. The FBG reflection peak was adjusted to reflect the lower (long wavelength) subcarrier sideband which was, via the circulator, detected by a PIN diode to recover the 1Gbps data. The optical carrier together with the baseband modulation and the upper subcarrier sideband was transmitted through the FBG and detected by another PIN diode. The remaining short wavelength subcarrier sideband that is not removed by the FBG will affect the 10Gbps signal after intensity detection in two ways. Firstly, it will give a spectral peak around the subcarrier frequency via mixing of the baseband and subcarrier spectral components. In a 10Gbps system, this peak will be filtered away by the receiver filter. Secondly, it will via self-mixing of the subcarrier signal give rise to crosstalk that contributes to the noise level of the 10Gbps signal. However, by keeping the power of the subcarrier signal more than 10dB lower than the power of the baseband signal, this degradation of the baseband signal can be kept small. In contrast the degradation of the subcarrier signal due to the base band modulation is more severe leading to a significant increase of the noise level. This degradation is due to both crosstalk at detection from the 10Gbps signal caused by

insufficient selectivity of the optical filter and due to nonlinear distortion in the modulator. The filter selectivity can be improved using cascaded FBGs, a microdisc [3] or ring resonator. Results – In Fig 2, the optical spectra of the modulated light are shown at the optical receiver input, the 10Gbps baseband signal transmitted through the FBG and the reflected 1Gbps subcarrier signal. Cases are shown where only one signal is present, and when the modulation is distorted due to the simultaneous transmission of both signals. As a reference the FBG filter characteristic is also shown.

-50

-30

-10

10

1529.8 1530 1530.2 1530.4 1530.6

complete10Gbps1GbpsFBG

Wavelength (nm)

Pow

er (d

B)

-50

-30

-10

10

1529.8 1530 1530.2 1530.4 1530.6



Wavelength (nm)

Pow

er (d

B)

Fig.2. Optical spectra of modulated signals, curves

correspond to the input light to the receiver (10Gbps together with 1Gbps on subcarrier), the part transmitted

through FBG and the reflected beam from the FBG. Finally, transmission over 58km standard single-mode fiber without dispersion compensation was tested and eye diagrams were recorded using a Digital Communication Analyzer (HP83480A). The results are depicted in Fig. 3 (10Gbps baseband signal) and Fig. 4 (1Gbps subcarrier signal). The eyes for 1Gbps were measured with a 1.25Gbps low pass filter. We found that the noise in the 10Gbps eyes was generated by the amplifier that was combined with attenuators and used here as a broadband isolator. The 10Gbps eyes for 0km and 58km, where a 50Ohm termination resistor instead of the amplifier was connected to the 3dB coupler, were clearly open. Due to single-sideband filtering in the receiver the chromatic dispersion will not significantly affect the subcarrier signal despite the high subcarrier frequency. The simultaneous detection of both sidebands would cause fading problems depending on the length of the fiber.


- 118 -

Fig.3. Eye diagrams of received 10Gbps signal with

electrical filter after transmission over 0km and 58km of SSMF, on the left and right, respectively. The eyes on the

top correspond to signals distorted by the subcarrier modulation, while in the bottom part of the figure the

subcarrier signal was off.

Fig.4. Eye diagrams of received 1Gbps signal with

electrical filter after transmission over 0km and 58km of SSMF, on the left and right, respectively. The eyes at the

top correspond to signal distorted by the baseband modulation, while in the bottom part of the figure the

10Gbps signal was off. The 1Gbps signal was also replaced by a Digital-Video-Broadcast-Terrestrial signal. This is a 64QAM signal covering the spectral range 0.4-0.9GHz. Successful transmission of 10Gbps data and TV signal was realized.

Conclusion – We have proposed and demonstrated a method to extend an existing 10Gbps fiber-optic link to carry an additional signal for e.g. distribution of cable TV on the same 100GHz WDM channel. The extra signal is transmitted at a high frequency subcarrier and optically filtered out in the receiver by utilizing a fiber Bragg grating, a circulator and an extra detector. No local oscillator is needed in the receiver and the output of the detector can be directly fed to e.g. a DVBT receiver. Successful transmission over 58km of standard single mode fiber of a 10Gbps signal extended with either a 1Gbps ASK signal or a DVBT signal on a 27GHz subcarrier show that the method is dispersion tolerant and does not severely affect the quality of the original 10Gbps baseband signal.

Acknowledgement - The work was carried out with partial financial support of the European Commission via: HECTO, ISIS, and IPHOBAC. The measurement equipment was funded in part by the Knut&Alice Wallenberg foundation, KAW.

References [1] H.Ohtsuka, O.Kagami, S.Komaki, K. Kohiyama,

M.Kavehrad, “256-QAM Subcarrier Transmission Using Coding and Optical Intensity Modulation in Distribution Networks” IEEE Photonic Techn. Lett. ’91 vol 3, No 4.

[2] Z. Yu, A. Djupsjobacka, M. Popov, P-Y. Fonjallaz, ”Direct Detection of Direct Optically Filtered Millimeter-Wave Signals”. IEEE Photonic Techn. Lett. ’07 vol.19, No4.

[3] P. Koonath,T. Indukuri, B. Jalalib “Add-drop filters utilizing vertically coupled microdisk resonators in silicon” Applied Physics Lett. Feb.’05 86, 091102.


- 119 -


- 120 -

Quality of Service Control in a multi-access integrated network based on Virtual Private LAN Service

L. Rea1,*, A.Valenti1, S. Pompei1, L. Pulcini1, M.Celidonio1, D. Del Buono2,G.M. Tosi Beleffi21 Fondazione Ugo Bordoni, Rome, Italy

2 Istituto Superiore delle Comunicazioni e Tecnologie dell’Informazione, Rome, Italy* Corresponding Author: Luca Rea: Phone +39 06 5480 2215, [email protected]

Abstract— In this work, we experimentally demonstrate how to achieve triple play services in Gigabit Ethernet (GbE) networks with access based both on Passive Optical Networks (PONs) and 60 GHz IP-Based radio-link. To guarantee QoS in these heterogeneous access conditions, the combination of the Virtual Private LAN Service (VPLS) forwardingprocess in the metro-core network and of the VLAN tagging one in the Edge (Radio-PON) segment has been implemented.

Introduction-The pervasive introduction of optical fibersin access networks is one of the fundamental requirements to deliver wide bandwidth services. Amongthe different technologies available on the market, PON permits to realize fiber access architectures that are simple to be implemented and at low costs [1].

However, it has to be pointed out that it is not possible to bring the fiber infrastructure in all orographic areas, or in places with a small number of potential end users. For this reason, a novel IP-Based radio access has been proposed and implemented with the goal to demonstrate interoperability of different access techniques (Radio-PON) guaranteeing a good end-to-end QoS.

Recently released spectrum at 60 and 80 GHz inducedICT industries to develop a new generation of high speed radio products. These equipments can reach data rates up to gigabit-per-second and allow to realize cost effective radio architectures, even if, the 60 GHz radio-link described in this paper, can operate at a bit rate of 100 Mbit/s.

60GHz and 80GHz wireless links [2] have emerged as considerable solutions to extend gigabit networks between two or more sites or in the access network. This solution offers a good trade-off between bandwidth availability and cost-efficiency opportunities.

In these heterogeneous scenarios including PON and wireless accesses, it is essential the introduction of techniques to control the QoS to assure Service Level Agreements also in conditions of traffic congestions as well as the bandwidth bottleneck introduced by adoptingradio solution.

Networks need procedures to guarantee End-to-End QoS properties, from access to the core and, in Ethernet environment, the Virtual Private LAN Service (VPLS) [3] is a suitable technique since it provides multipoint Ethernet connections employing MPLS Label Switched Paths (LSPs), allowing to achieve excellent network performances in terms of traffic management and QoS. VPLS is a Layer 2 Virtual Private Network (VPN) where the customers seem to belong to the same LAN, regardless of their real geographic position. VPLS works on routing elements called Provider Edge (PE), and VPLS procedures can be extended, by means of VLAN tagging technique, to routing elements (Customer Edge, CE) that do not support VPLS. Therefore, by means of VPLS-VLAN tagging we can define Class of Service (CoS) in End-to-End paths, crossing access and core networks [4].

In this paper, we experimentally show that, by means of VPLS-VLAN tagging technique operating in wide area GbE networks, including EPONs and 60 GHz IP-based radio-link, we are able to set-up upstream/downstream data streams that maintain bandwidth and QoS characteristics, also when traffic congestion occurs in some network segment.

Fig. 1. Experimental set-up representing a core (PEs) network with FTTx access network based on EPON and 60GHz IP-Based radio-link.

Test-Bed- The Test-Bed shown in Fig. 1 is made up of a core and access section [4]. The core part consists of four routers, Juniper M10 with ZX GbE interfaces (1550 nm), that operate as PEs and are fully meshed using the fibers deployed in the Roma-Pomezia-Roma cable (50 Km with round trip in Pomezia). The edge (metro) part is composed of three Cisco 3845 edge routers (that behave as CEs), connected with three Juniper routers by means of GbE fiber transmission. The access section is composed by an EPON and a 60 GHz IP-Based radio-link. The EPON (AN5116-03 ePON FiberHome) includesan Optical Line Termination (OLT) and eight Optical Network Units (ONUs). The OLT and the ONUs are connected by means of a single mode fiber with the downstream wavelength at 1490 nm and the upstream wavelength at 1310 nm.

The access IP-Based radio-link is implemented by two radio station terminals spaced 260 m apart, even if they can operate over longer distances.

The radio equipments used to perform the tests are the model Sencity Link provided by Huber Suhner. This


- 121 -

device is a receiver-transmitter terminal operating in the band 59.475-62.325 GHz, transmitting at 30 dBm EIRP, using the FDD duplexing technique and QPSK modulation.

The directive antenna is integrated into an external unit, it is characterized by a gain of 37 dBi and it is powered by a PoE (Power over Ethernet). The latter feature simplifies the manageability, both from an installation point of view, since generally the sites where the antennas are located (the roof of a building) are not often equipped with power supply, as well as regards the access to the web-based interface to configure the transmission parameters.

Network measurements were carried out by using a software network analyzer, NetIQ Chariot, that allows usto evaluate some network parameters like throughput, jitter and data loss. However, here we only report, for sake of brevity, throughput measurements.

To test the impact of the network congestion, a traffic generator (Smartbits 6000) is included in the Test Bed forintroducing a background traffic up to 1 Gb/s in the above mentioned test link.Results-The first step has been the characterization of 60 GHz IP-based radio-link in a stand-alone configuration; 60 GHz bandwidth availability is 100 Mbit/s, anyway we consider a reliable behavior for bit rate under 90 Mbit/s. In Fig 2 is shown the throughput measured in presence of rain.

Fig. 2. Agreement between Throughput (Mb/s) and Rain Rate (mm/h) during 4 hours observation

After the characterization, radio-link has been connected to the test bed to create a radio access area, as shown in Fig 1.

We set the VPLS-VLAN tagging, with a higher priority Class of Service that we called Gold Class (as reported in [5]), among CE1, and the ONUs to the CE3 connected to the radio-link, corresponding to a typical configuration for an operator that delivers services from a server to the users. In this configuration, we considered the downstream scenario sending data flows from Server to the user at the ONU1 output (Client 1) and to the user on CE3 (Client 2). We observe the same behavior of both access EPON and 60GHz IP-Based radio-link, and for the sake of brevity we report just Radio downstream results. The advantages of our method are illustrated by Fig. 3,where we report the throughput at the Client 2(downstream) for 40 Mbit/s flows, both in the absence (default Class of Service, i.e. Best Effort) and in the presence of VPLS-VLAN tagging, when the congestion occurs. When the VPLS-VLAN tagging is applied (with a Gold Class that allows to achieve the best QoS

performance), the throughput does not show any reduction. Conversely, without VPLS VLAN tagging the services supported by 60 GHz IP-based radio-link would be strongly degraded due to traffic congestion. Measurements carried out on jitter and data losses confirmed such aspects.

Fig. 3. Throughput of the 40 Mb/s flows at the Client 2 (downstream) with and without VPLS/VLAN tagging

Also measurements of perceived quality were carried out. In particular, we sent an MPEG2 High Definition (HD) video streaming both with and without VPLS-VLAN Tagging. As shown in Fig. 4, when the VPLS-VLAN Tagging is applied, there are no degradations and the video is received with a perfect quality. On the contrary, when tagging is not included, the video isstrongly degraded.

Fig. 4. Screenshot from MPEG HD video with VPLS-VLAN Tagging(left) and without VPLS-VLAN Tagging (right)

Conclusions-We believe that illustrated architecture, based on VPLS–VLAN Tagging in wide area GbE networks, allows different access networks to well satisfy requirements of both users and operators in terms of QoS. Such results suggest that, by means of such an approach,very reliable paths can be defined, guaranteeing QoS and bandwidth.Acknowledgment- The work described in this paper was carried out with the support of the BONE-project ("Building the Future Optical Network in Europe"), a Network of Excellence funded by the European Commission through the 7th ICT-Framework Programme.References[1] G. Kramer, B. Mukherjee, G. Pesavento, “Ethernet PON (ePON):

Design and Analysis of an Optical Access Network”, in Photonic Network Commun. vol. 3, pp. 307-319, July 2001

[2] Nan Guo, Robert C. Qiu, Shaomin S. Mo, Kazuaki Takahashi, “60-GHzMillimeter-Wave Radio: Principle, Technology, and New Results”, EURASIP Journal on Wireless Communications and Networking, September 2006

[3] G. Chiruvolu, A. Ge, D. Elle-Dit-Cosaque, M. Ali, J. Rouyer, “Issue and Approaches on Extending Ethernet beyond LANs” in IEEE Commun. Mag. Vol. 42, pp. 80-86, March 2004

[4] A. Valenti, S. Pompei, L. Rea, F. Matera, G.M. Tosi Beleffi, M. Settembre, “Network Performance Investigation in a Wide Area Gigabit Ethernet Test Bed adopting All Optical Wavelenght Conversion”, in Photonics Technol. Letters, Vol. 20, pp. 2144-2146, December 2008

[5] A. Valenti, S. Pompei, L. Rea, F. Matera, “Experimental Investigation on Optical Gigabit Ethernet network reliability for high-definition IPTV services”, in J. of Opt. Networking, Vol. 7, pp. 426-435, May 2008


- 122 -

Photonic Millimeter-Wave System for Broadband Wireless Access

M. Weiß1,*, A. Stöhr1, S. Fedderwitz1, V. Rymanov1, B. Charbonnier2 and D. Jäger1

1 Universität Duisburg-Essen, ZHO-Optoelektronik, Lotharstr. 55, 47057 Duisburg, Germany 2 France Telecom R&D, 2 Av. Pierre Marzin, 22300 Lannion, France


Abstract – In this paper, a 60GHz photonic millimeter-wave point-to-point link system for broadband wireless access is investigated supporting 10 Gb/s Ethernet data rates of up to 12.5 Gb/s. We have achieved error-free transmission of 10.3125 Gb/s over 40m.

Introduction – The past few years have witnessed the emergence of several new “bandwidth hungry” applications such as high-definition TV (HDTV), peer-to-peer communication or video-over-IP. In consequence, bandwidth requirements for access are expected to rise, targeting 1 Gb/s per user in the near future even up to 10 Gb/s in a mid-term period which requires corresponding fiber infrastructures like 10 Gb Ethernet passive optical networks (10G EPON, IEEE P802.3av). A similar development can be forecasted for wireless access allowing flexibility and portability. Even the introduction of 10 Gigabit Ethernet (10 GbE) wireless standards is expected, supporting the convergence of wired and wireless systems in the access, but available wireless systems are currently operating in the Mb/s-range [1]. A solution to this bottleneck is seen in the development of wireless systems operating at much higher carrier frequencies namely in the millimeter-wave (mm-wave) range where more bandwidth is available. Especially around 60 GHz a bandwidth of about 7 GHz is allocated for wireless communications depending on country and region. Consequently, broadband wireless systems operating at around 60 GHz are currently being studied worldwide, and potential applications may include broadband wireless access for rural areas, disaster recovery for metro access networks or wireless corporate access.

In this paper, we report on a very high speed point-to-point radio link operating in the 60 GHz frequency band for future mobile network backhauling or high speed wireless LAN bridging. At first, we present the setup of the constructed 60 GHz wireless system using optical on-off-keying (OOK). So far, 60 GHz radio-over-fiber (RoF) systems were mainly considered for short-range in-house applications due to the high atmospheric gaseous attenuation as well as the influence of chromatic dispersion to the fiber range when the millimeter-wave is transported within the fiber-optic domain. In this paper, we further study the maximum fiber length and the maximum wireless

distance that the constructed 12.5 Gb/s photonic wireless system can accommodate.

System Setup – The constructed broadband 60 GHz RoF test bed is shown in Fig. 1. It consists of 4 main building blocks for photonic 60 GHz generation with a subsequent broadband data modulation unit, a photonic wireless 60 GHz transmitter and a wireless receiver. Details on the constructed system are given in [2],[3].

f =30 GHzLO/2

Photonic Carrier Generator Data Modulator

PA-1

EDFA-1 PC-1 MZM-1 MZM-2PC-2

12.5 Gb/s Data2 -1 PRBS

31

PA-2

LNA-1

OBPF

RF to BasebandMixer

OAEDFA-2

ECL: external cavity laserEDFA: er-doped fiber amplifierLNA: low noise amplifierMZM: mach-zehnder modulator

PA: power amplifier

OA: optical attenuatorOBPF: optical band pass filter

PC: polarisation controllerPD: photodetector

LNAPhotonic 60GHz Transmitter

Wireless 60GHz Receiver

f =60 GHzLO-2

ECL

P

f0 f0

P

f0 f0

fLO

P

f0 f0

fLO

P

f0 fLOfLO/2

P

f0 fLOfLO/2

PDError

Detector

Fig. 1. Schematic of the developed 60 GHz RoF system

A key element of the system is the applied analogue modulation scheme, double sideband modulation with suppressed carrier (DSB-SC). The first Mach-Zehnder modulator (MZM-1) is biased to the minimum transmission point (MITB) to suppress the optical carrier and to solely generate the two optical sidebands by applying a 30 GHz LO signal to the modulators RF electrode. This scheme does not only reduce the requirements for the LO source (i.e. 30 GHz instead of 60 GHz), but also improves the tolerance to chromatic dispersion which is discussed


- 123 -

later in the paper. The second modulator (MZM-2) is conversely biased to the quadrature point (QP) to operate within the linear regime and thus to ensure an undistorted data modulation. After fiber-optic transmission, the radio-over-fiber signal is detected at the photodetector, amplified to about +11 dBm and transmitted using a 20 dBi medium-gain antenna. After wireless transmission, the signal is received by an identical antenna and detected within a coherent receiver. By using a low noise amplifier (LNA-1) and a low-loss custom design mixer, very high sensitivities are achieved thus reducing transmit power requirements.

Experimental results – We have performed medium-range out-door experiments over 20 m and 40 m wireless span for data rates of up to 10.3125 Gb/s and 12.5 Gb/s. This corresponds to the gross data rate for 64/66-coded and 8/10-coded 10 GbE. All measurements were carried out at the university campus, limiting the maximum wireless path length to 40 m due to surrounding buildings. Transmitter and receiver were placed at a height of 120 cm above ground. The transmit power was controlled by an optical attenuator (OA) as shown in the system setup in Fig. 1.

−60 −55 −50 −45 −40 −35 −30 −25−12−11

−10

−9

−8

−7

−6

−5

−4

Received Power (dBm)

log(

BE

R)

20m − 10.3125Gb/s20m − 12.5Gb/s40m − 5.0Gb/s40m − 7.5Gb/s40m − 10.3125Gb/s

Fig. 2. BER measurements after 20 m and 40 m wireless transmission applying data rates of up to 12.5 Gb/s.

Fig. 2 consists BER measurements after 20 m and 40 m wireless transmission applying data rates of up to 12.5 Gb/s. From the results, a sensitivity of -46 dBm for error-free (BER<10-9) transmission of 10.3125 Gb/s is observed. The sensitivity after 40 m wireless path length is slightly better than for 20 m which is attributed to reflections from buildings. The system even achieved 12.5 Gb/s wireless transmission over 20 m, however as can be seen from Fig. 2, error-rates were limited to about 5.10-7 in that case. This was because of a limit in the power link budget. As former performed indoor experiments over 2.5 m

wireless span have revealed no error-floor even for 12.5 Gb/s [2], we expect that error-free transmission of 12.5 Gb/s is possible by simply using either a slightly higher RF gain or antennas with a higher directivity.

Fiber-optic range – Due to the absence of an optical carrier (DSB-SC) for photonic mm-wave generation, the system is inherently tolerant to chromatic dispersion. Conversely in conventional DSB-systems, beating of the dispersion-induced and thus phase-shifted optical sidebands with the optical carrier during o/e-conversion may results in destructive interference significantly limiting the maximum transmission distance [4],[5]. Fig. 3 shows a simulation on fiber length inducing a 3 dB power penalty versus data rate, for conventional DSB-modulation requiring an LO frequency of 60 GHz and for DSB-CS modulation with an LO frequency of 30 GHz.

1.25 2.5 5 7.5 10.3125 12.510

−1

100

101

Data rate (Gb/s)

Fib

er le

ngth

indu

cing

3dB

DIP

P (

km)

fLO

= 30 GHz − DSB−CS

fLO

= 60 GHz − DSB

Fig. 3. Fiber length inducing a power penalty of 3 dB due to chromatic dispersion versus data rate, for conventional

DSB-modulation and for DSB-SC modulation.

As can be seen, a 3 dB dispersion induced power penalty (DIPP) is achieved after some 100 m even for low data rates in the conventional case (DSB), e.g. approx. 0.36 km at 12.5 Gb/s transmission speed. For the case of DSB-CS, fiber-optic transmission distance is extended to a range of 2.02 km (12.5 Gb/s) and 2.52 km (10.3125 Gb/s) which would be already enough for access applications. Even higher fiber-optic spans can be achieved by applying optical single-sideband transmission which would however require additional components and thus increased costs.

Wireless range – Based upon the experimental results with wireless path lengths of 40 m, we further studied the potential for wireless range extension if


- 124 -

50 dBi high-gain antennas (e.g. Cassegrain) are applied compared to 20 dBi within the experiments. Typically, the 60 GHz band is mainly considered for short-range in-house communication (i.e. for future personal and local area networks) due to the high atmospheric gaseous attenuation especially compared to alternatives within the E-band and the W-band. However it would be beneficial to use the 60 GHz band for broadband point-to-point applications in terms of interoperability (together with 60 GHz in-house communication systems) and costs, as 60 GHz components are cheaper and more mature than their counterparts operating within higher mm-wave bands. The received power can be calculated by considering the free space propagation loss L0, loss due to atmospheric gaseous attenuation L1 as well as rain attenuation L2 [2][5]:

dLdLLGGPP rttdBmr 210, (1) where Pt is the transmitted power, Gt the transmitter antenna gain and Gr the receiver antenna gain. Also, to study the link availability with respect to weather conditions, different rain attenuation figures based upon sample rain data from a middle European country were considered. In detail, the rain attenuation figures L2 used later for a link availability of 99%, 99.99% and 99.999% are 1.3 dB/km (2 mm/h), 10.1 dB/km (25 mm/h) and 32.5 dB/km (85 mm/h), respectively [7].

50 55 60 65 70 7590

100

110

120

130

140

150

160

170

180

Frequency (GHz)

Tot

al p

ath

loss

Lpa

th (

dB)

40m100m200m500m1000m

Fig. 4. Total path loss within the V-band comprising free space path loss, gaseous attenuation and rain attenuation

(25 mm/h).

Fig. 4 shows the total path loss versus frequency including free space path loss, gaseous attenuation and rain attenuation applying ITU models and ITU atmospheric standard conditions [8],[9]. As can be seen, the path loss up to 100 m is comparatively flat over the whole V-band whereas for higher distances, gaseous attenuation gets more severe inducing a

peaking of the loss figure around 60GHz. This strongly limits the application of higher level amplitude modulation (e.g. m-PAM) schemes if large bandwidths are consumed but is however acceptable for on-off-keying based communication systems. Fig. 5 shows the received power versus wireless path length if 50 dBi gain antennas are used. The corresponding receiver sensitivities for achieving a BER of 10-9 for data rates of 10.3125 Gb/s and 12.5 Gb/s are also indicated by the dashed lines. As can be seen from Fig. 5, the maximum wireless distances for a 64/66-coded 10 GbE signal (BER=10-9) for link availabilities of 99.999%, 99.99%, and 99% are 700 m, 1100 m, and 1500 m, respectively.

102

103

104

−55

−45

−35

−25

−15

−5

Wireless path length (m)

BER 10e−9 / 12.5 Gb/s

BER 10e−9 / 10.3125 Gb/s

Rec

eive

d P

ower

(dB

m)

Fair99% Availability (rain 2mm/h)99.99% Availability (rain 25mm/h)99.999% Availability (rain 85mm/h)

Fig. 5. Maximum wireless path lengths under different weather conditions.

Conclusion – In this paper, we have presented a 60 GHz RoF system operating up to 12.5 Gb/s which fully supports the gross data rate for 64/66-coded and 8/10-coded 10GbE. We further studied the maximum fiber length and the maximum wireless distance that the constructed photonic wireless system can accommodate.

Here, we have theoretically shown a fiber-optic range exceeding 2 km even operating at 12.5 Gb/s transmission speed and discussed the benefits of the applied analogue modulation format for photonic mm-wave generation. We further studied the potential of the system for medium-range transmission. While applying high-gain antennas, we predict the wireless range to be within the km-range.

Acknowledgement – This work was carried out within in the framework of the European integrated FP6 project IPHOBAC under grant no. 35317.

References[1] www.ist-iphobac.org [2] M. Weiß et al., “60-GHz Photonic Millimeter-

Wave Link for Short- to Medium-Range


- 125 -

Wireless Transmission Up to 12.5 Gb/s,” IEEE J. Lightw. Technol., vol. 26, no. 15, pp. 2424-2429, Aug. 2008, (invited)

[3] M. Weiß et al., “60GHz Radio-over-Fibre Wireless System for Bridging 10Gb/s Ethernet Links,” 34th European Conference on Optical Communication, ECOC 2008, 2008

[4] J. M. Fuster et al., “Analysis of hybrid modulation techniques in MZ-EOM-based photonic mixers to overcome dispersion-induced power penalty in up-converting millimeter-wave fiber-optic links,” Microwave and Optical Technology Letter, vol. 23, no. 2, pp. 127 – 129, Sep. 1999

[5] J. M. Fuster et al., “Chromatic dispersion effects in electro-optical upconverted millimeter-wave fibre optic links,” Electronics Letters, vol. 33, no.23, Nov. 1997

[6] C. Kopp, “TROPPO - A Tropospheric Propagation Simulator,” Technical Report, Monash University, Australia, 2004

[7] V. Kvicera, M. Grabner, “Rain attenuation at 58 GHz: prediction versus long-term trial results”, EURASIP J. Wireless Comm., vol. 2007, no. 1, p. 46, Jan. 2007

[8] Recommendation ITU-R P838-3, International Telecommunication Union, 2005

[9] Recommendation ITU-R P676-4, International Telecommunication Union, 1999


- 126 -

60 GHz Wireless Signal Generation with 10 Gb/s 4QAM Modulation using Photonic Vector Modulator

Rakesh Sambaraju Valencia Nanophotonics Technology Center, Universidad Politecnica de Valencia, 8F, Valencia, Spain

* Corresponding author: Phone: +34963879746, E-Mail: [email protected]

Abstract – 10 Gb/s QPSK modulation based 60 GHz wireless signal was successfully generated using a photonic vector modulator.

Introduction – The past two decades have seen an exponential growth in the demand of capacity in the telecommunication networks, which is mainly accounted for the huge success of internet protocol and the high end multimedia applications like Youtube, VoD, HDTV, SDTV; etc. As the present core and metro networks are moving towards 40 Gb/s or even 100 Gb/s to quench the user bandwidth thirst, the access networks are moving towards 10 Gb/s with the latest technologies like 10G-EPON FTTx. Though FTTx can successfully reduce the existing “access bottle neck”, there are several scenarios where laying the fiber in the last few kilometer is very expensive. Such a case is the rural areas, where the fiber to the remote user is a challenge. Also geographically remote areas like mountains, across the rivers; fiber deployment is a challenging task. To avoid these digital divides, wireless access of 10 Gb/s to the user is crucial. Various research efforts are underway [1]-[3] to develop gigabit wireless links. For implementing these broadband wireless links, photonics has proved to be a good solution [2]-[5], due to the various advantages of optical communications, like huge bandwidth or scalability. Most of these photonic techniques [2], [3], [6] for generating the signals required in the broadband wireless links were based on direct upconversion of the baseband data to the RF frequency, which is an inefficient modulation technique in terms of electrical bandwidth. Recently, a novel technique for the generation of bandwidth efficient electrical modulation schemes like m-ary quadrature amplitude modulation (MQAM) called photonic vector modulation technique has been proposed [7], [8]. Photonic vector modulation has proved to be a good technique for generating multi gigabit per second wireless links, and up to 10 Gb/s 16-QAM mm-wave signal generation [9]. In this a 60 GHz wireless signal with 10 Gb/s capacity in a QPSK/4QAM modulation is generated using photonic vector modulation, and upto 40m, of air transmission is simulated using attenuators. Working Principle – As shown in Fig. 1, the generation of a QPSK-modulated carrier is performed

in three steps. First, the two DFB lasers are modulated by the I and Q baseband data streams. In this case, the lasers are also standard components for digital baseband applications. Second, a millimeter-wave LO tone externally modulates the incoming optical carriers using a Mach-Zehnder modulator (MZM) biased at its minimum transmission point to generate optical carrier suppression and generating the second harmonic Finally, the quadrature condition (the 90º phase shift between the I and Qcomponents) is obtained by delaying the output of the MZM of the Q arm using a tunable optical delay line. The photo detected signal is QPSK modulated electrical carriers. Experimental Setup – In this section, the technique of photonic vector modulation for generating spectrally efficient modulation formats is described. Using photonic vector modulation, millimeter wave wireless links with various advanced modulation formats like QPSK, 16-QAM can be generated, and up to 10 Gb/s 16-QAM modulated millimetre wave carrier generation is demonstrated [9]. In the photonic vector modulator, two DFB lasers at 1554.14 nm and 1558.17 nm wavelengths, with modulation bandwidth of 4 GHz are directly modulated by two (I and Q) 5 Gb/s 27-1 PRBS data streams respectively. Fig. 1 shows the schematic of the experimental setup.

Fig. 1.: Schematic of the experimental setup for 10 Gb/s 60 GHz carrier generation.

The optical carriers with the I and Q data modulated in an NRZ-OOK format with are individually modulated by an electrical carrier of fLO/2=30 GHz using two 45 GHz Mach-Zehnder modulators biased


- 127 -

at minimum transmission point. The bias of MZM is chosen such that an optical carrier suppression (OCS) modulation is generated, and harmonics separated at 60 GHz are generated. The Q-arm optical signal is now delayed by 1/4fLO which corresponds to a 90 degrees phase shift between the I and Q electrical carriers. The two optical signals are combined using a 3-dB coupler, and photo detected using a 100 GHz photodetector with responsivity of 0.5 A/W. The input power to the photo detector was measured as -14 dBm. The photo detector output is a 10 Gb/s QPSK modulated 60 GHz carrier. Based on the photo detector input optical power and the responsivity, the output RF power was calculated as -48 dBm. The 10 Gb/s 4-QAM signal was amplified using a low noise amplifier (LNA) with a gain of 16 dBm. The RF signal was later filtered using a bandpass filter with a bandwidth of 10 GHz. Another high power amplifier (HPA) with a gain of 27 dBm was used to amplify the signal further. To emulate the effect of wireless transmission, 50 dB attenuation was added, which can be translated into a distance of 40 m if antennas with a gain of 20 dBi are used. The RF signal was later amplified with another LNA and HPA before down converting using a broadband mixer. For demodulation of the QPSK signals, the RF signals were mixed with a copy of the 60 GHz local oscillator. An electrical phase shifter was used in the LO to tune the phase, and thus the I and Qbaseband components were demodulated, one at a time. The eye diagrams of the demodulated I and Qdata are shown in Fig. 2.

Fig. 2.: The Inphase and Quadrature demodulated components.

The baseband data were directly fed into a bit error ratio tester (BERT) to measure the BER of the demodulated signals. BER of 4 x 10-8 and 8.7 x 10-8

were measured for I and Q data respectively. The BER shows a good quality of the 10 Gb/s QPSK modulated 60 GHz carriers. The quality can be further improved by increasing the optical power input to the photo detector, and by using high gain antennas, which will also improve the achievable transmission distance

Conclusions – wireless access of 10 Gb/s is crucial in various accss scenarios to avoid the digital divide, and photonics prove as a good candidate. With the photonic vector modulator, spectral efficient modulation formats can be generated, and a 10 Gb/s QPSK 60 GHz carrier generation is demonstrated. The quality of the signal after 40 m is above the limit put by wireless standards.

Acknowledgement - This work was carried out within in the framework of the European integrated project IST-35317 IPHOBAC

References[1] V. Dayadyuk, J. D. Bunton, J. Pathikulangara, R.

Kendall, O. Sevimli, L. Stokes, and D. A. Abbott, “A multigigabit millimeter-wave communication system with improved spectral efficiency,” IEEE Trans. on Microwave Theory and Tech., 55, 2813-2820 (2007).

[2] A. Hirata, M. Harada, and T. Nagatsuma, “120-GHz wireless link using photonic techniques for generation, modulation, and emission of millimeter-wave signals,” J. Lightwave Technol., 21, 2145-2153 (2003).

[3] A. Wiberg, P. Perez-Milan, M. V. Anders, P. A. Andrekson, and P. O. Hedekvist, “Fiber-optic 40 GHz mm-wave link with 2.5 Gb/s data transmission,” IEEE Photon. Technol. Lett., 17, 1938-1940 (2005).

[4] A. J. Seeds and J. K. Williams, “Microwave photonics,” J. Lightwave Technol., 24, 4628-4641 (2006).

[5] J. Capmany and D. Novak, “Microwave photonics combines two worlds,” Nat. Photonics, 1, 319-330 (2007).

[6] R. W. Ridgway and D. W. Nippa, “Generation and modulation of a 94-GHz signal using electrooptic modulators,” IEEE Photon. Technol. Lett., 20, 653-655 (2008).

[7] M.A. Piqueras, et al, “Direct Photonic Generation of Electrical Vector Modulations at Microwave/Millimeter-wave Frequencies,” IEEE Photon. Technol. Lett., 17, 1947-1949 (2005).

[8] R. Sambaraju et al, “Generation of Multi-Gbps MQAM/MPSK Modulated Mm-Wave Carriers Employing Photonic Vector Modulator Techniques,” J. Lightwave Technol., 25, 3350-3357 (2007).

[9] R. Sambaraju et al, “Ten gigabits per second 16-level quadrature amplitude modulated millimeter-wave carrier generation using dual-drive Mach-Zehnder modulators incorporated photonic-vector modulator,” Opt. Lett., 33, 1833-1835 (2008).


- 128 -

Documents

Workshop Proceedings