Space Time Processing forFixed Broadband Wireless

A. Paulraj

Gigabit Wireless&

Stanford University

ISART6 -8 September, 2000

Boulder, CO

Essential Attributes of Broad BandWireless (BWA) Network

• High Speed > 4-5 Mbps• High QoS / Availability

• DSL / Cable like

• Low Costs $$$• High capacity• High coverage• CPE and Infrastructure unit costs

• Friendly• Easy CPE Install• Environmental/Regulatory

• Scalability (Multi-cell)• Evolution to portability

BWA Network Architecture

• Multi-cell architecture• Low BTS antennas (50 - 150 ft)• Under-the-eave subscriber antennas

• Significant Non-LOS propagation(Rayleigh fading)

• Strong multi-path• Low Doppler channels

Mobile cellular-like network with all its challenges but withMobile cellular-like network with all its challenges but with5550 Speed and 50 Speed and 5550 50 QoSQoS

NAU = Network Access UnitBTS = Base Transceiver StationBSC = Base Station Controller

BWA Network

NAU

IP

NGU

BSC

BTS

BTS

BTS

Voicegateway Network

OperationsCenter

SMS

AccessMux

PSTN

IPGateway

NAU

NGU = Network Gateway UnitSMS = Subscription Management System

Data Rate & QoS Challenges

FF 555550 50 High Data RateHigh Data Rate

F 550 QoS(0.999 Availability on Throughput, Delay, Jitter)

Capacity (Bits/Hz/Cell)

Coverage(Cell Radius)

50.4

50.250.4

Typical Propagation Scenario

BTS Directional Antenna

BTS

Ht 30’-150’

Ht 8’-20’

0.1 - 5 miles

CPE Directional Antenna

Distanceto mobilescatterers

10 0 101 102-170

-160

-150

-140

-130

-120

-110

-100

Path Loss (Mean Level)

Sign

al P

ower

(dB

) Super CellBTS ht: 1500ft

Macro CellBTS ht 150 ft

Range in km

Frequency: 2.5 GHzAnt Beamwidth: 90 deg

Super Cell: Free Space modelMacro/Micro Cell: Erceg’s model

K-Factor vs. Distance(Suburban BWA)

10-1

100

101

-20

-10

0

10

20

30

40

Distance in km

K-F

acto

r in

dBht = 15m, 90 deg. Rx antenna

hr = 3m

hr = 10m

GW and AT&T measured data similar

K = 0 is necessary assumption for reliable deployment

--- hr = 10m --- hr = 3m

RMS Delay Spread vs Distance(Suburban BWA)

10-1

100

101

-25

-20

-15

-10

-5

0

5

10R

MS

Del

ay S

prea

d in

Mic

rose

cond

s (d

B)

Distance in km

GW experimental data similar

Doppler Spectrum - BWA

- 3dB

fmaxdf±

0.05 ≤ fd≤ 0.5 Hz

Signal MeasurementsLow Doppler, 0.05 Hz

High Doppler, 0.5 Hz

C vs I Measurements

2 4 6 8 10 12 14 16 18 20-180

-170

-160

-150

-140

-130

-120Rx Ant 1

s

Pat

h L

oss

[-dB

]

C/I: 10.0212 dB

BTSCCI

2 4 6 8 10 12 14 16 18 20-180

-170

-160

-150

-140

-130

-120Rx Ant 2

s

Pat

h L

oss

[-dB

]

C/I: 6.632 dB

BTSCCI

Reuse factor 1

Reuse factor 1

20 40 60 80 1 0 0 1 2 0-180

-170

-160

-150

-140

-130

-120Rx Ant 1

s

Pat

h L

oss

[-d

B]

C / I : 8 .3504 dB

BTS

CCI

20 40 60 80 1 0 0 1 2 0-180

-170

-160

-150

-140

-130

-120Rx Ant 2

s

Pat

h L

oss

[-d

B]

C/I: 0.21909 dB

BTSCCI

MIMO Measurements

BTS CPE

H22

H12

H21

H11

Spatial Multiplexing (SM) OffersHigher Speeds

Each Eigen mode of the matrix channel can support aradio link. 11.4 ⇒ 1.5 speed up possible in 2 x 3 systems.

Tx RxH

Increasing Capacity/Coverage

• Coverage: Lower the C/N needed to close link -improves link budget

• Capacity: Lower C/I needed to close link -improves reuse factor– Use diversity, array gain, coding, etc to minimize

required C/N or C/I to support a target PER

– Use RLP and fragmentation to allow higher target PER

– Use spatial multiplexing to run parallel channels

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8100

120

140

160

180

200

220

Mode 6, Doppler 1Hz, MaxUtil 635 users, small files

link utilization [%]

mea

n th

roug

hput

[K

bps]

Mean Per 1% Mean Per 5% Mean Per 10%

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80

500

1000

1500

2000

2500

3000

Mode 6, Doppler 1Hz, MaxUtil 635 users, big files

link utilization [%]m

ean

thro

ughp

ut [

Kbp

s]

Mean Per 1% Mean Per 5% Mean Per 10%

- TCP windowing algorithm yields the difference in thethroughput of ‘small’ and ‘big’ files

- Throughput = File size/ Mean transfer delay

Throughput vs Utiliz. (vs PER)

Increasing Coverage

• Diversity is critical (Tx antenna, Rx antenna,frequency, pol)

• Directional CPE antennas

• Forward Error Correction (FEC) & ARQ

• Increase BTS & CPE height

Coverage vs Required C/N

0 5 10 15 20 25 301

2

3

4

5

6

7

8

Ran

ge in

Mil

es

1 x 1, NO RLP2 x 2, RLP Required C/ N

CDF of Tx-Rx Diversity - Measured Data

-40 -30 -20 -10 0 10 2010

-4

10-3

10-2

10-1

100

Signal Power About Mean in dB

Prob

abili

ty T

hat S

igna

l Pow

er <

Abs

ciss

a

Before Combining After Combining

1 x 2

2 x 2

Probability of signal falling 10 dB below mean is 300 times lower with 2 x 2 compared 1 x 1

Increasing Spectrum Efficiency

• Reduce CCI to improve reuse - Increase signal diversity

- Interference averaging

- Antenna sectorization, beam forming, interference cancellation

• Reduce ACI to improve reuse - Careful frequency planning

- Stringent emission masks, Rx filtering

- Interference suppression

• Use link adaptation to exploit available C/N or C/I levels

• Use Radio Link Protocol to allow link to use high BER

C/ I After Combining - Measured Data

C /I = 10 dB

-40 -30 -20 -10 0 10 2010

-4

10-3

10-2

10-1

100

C/I About Mean in dB

Prob

abili

ty T

hat C

/I <

Abs

ciss

a

Before Combining

1 x 2

2 x 2

After Combining

Probability of C/I falling 15 dB below mean is 300 times lower with 2 x 2 compared to 1 x 1

Conclusion

• Achieving high capacity/coverage in a multi-cell BWA(550 speed & 550 QoS compared to mobile cellular)network is a challenging problem.

• Many technologies help improve performance. MIMOantennas are a key leverage.

• Gigabit Wireless is a pioneer of MIMO for BWAnetworks