Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Underwater Acoustic OFDM: Past, Present, andFuture
Shengli Zhou
Dept. of Electrical and Computer EngineeringUniversity of Connecticut
http://uwsn.engr.uconn.edu
WUWNET’11
Dec. 2, 2011
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 1 / 26
Underwater Communications
Cable
Acoustic communications (ACOMM)
Electromagnetic communications (Wireless Fibre Systems)
Optical communications (Blue-Green Laser)
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 2 / 26
ACOMM Techniques
Frequency shift keying (FSK)I e.g., Teledyne Benthos, WHOI Micro-modem
Direct sequence spread spectrum (DSSS)I e.g., LinkQuest, DSPCOMM, Tritech
Single carrier phase-shift-keying (PSK) transmissionsI e.g., WHOI Micro-modem, Benthos (additional processing card)
Multicarrier modulation (in the form of OFDM)
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 3 / 26
OFDM: A Prevalent Choice for Broadband WirelessSystems
DSL Modem
WiFi (IEEE 802.11)
WiMax (IEEE 802.16)
3GPP-LTE
4G and beyond
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 4 / 26
Number of Publications on OFDM
1994 1996 1998 2000 2002 2004 2006 2008 2010 20120
2
4
6
8
10
12
14
Year
Num
ber
of P
ublic
atio
ns
OCEANS Conference PapersIEEE/JASA Journal Papers
1994 - 2005: sporadic effort and little progress
2006 - 2011: sustained effort and great progress
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 5 / 26
Outline
OFDM basics: Pros and Cons
Algorithm development
Prototype development
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 6 / 26
Underwater Acoustic Channel CharacteristicsThe sound propagates too slow!
I Long multipathI Fast variation
The SPACE’08 experiment, Martha’s Vineyard, depth 15 m
2 4 6 8 10 120
20
40
60
80
100
delay [ms]
ampl
itude
Fast-varying multipath channel with a large delay spread
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 7 / 26
Block Transmission over LTI ChannelsConsider a linear time invariant (LTI) multipath channel
h(t) =
Np∑
p=1
Apδ(τ − τp)
Time domain waveform distortion; intersymbol interference (ISI)arises; complex channel equalizer needed
y(t) = s(t) ∗ h(t)
Frequency domainY (f) = H(f)S(f)
If s(t) is carefully constructed with no ISI in frequency domain
S(f)|f=fm= s[m]
Then no ISI at the receiver side
Y (f)|f=fm= H(fm)s[m]
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 8 / 26
Basics of Orthogonal Frequency Division MultiplexingFrequency domain; fm − fk = (m − k) 1
T
S(f) =∑
k
s[k]sinc(
(f − fk)T)
, S(f)|f=fm= s[m]
−5 −4 −3 −2 −1 0 1 2 3 4 5
−0.2
0
0.2
0.4
0.6
0.8
1
1.2
Frequency
Fre
qu
en
cy F
esp
on
se
s[k] s[k+1]s[k−1]
Time domain waveform; g(t): rectangular pulse shaper
s(t) =∑
k
s[k]ej2πfktg(t), fk = fc +k
T
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 9 / 26
Pros and Cons of OFDM
Pros:I Convert a dispersive channel to a set of parallel simple channels
{
zm = H(fm)s[m] + nm
}K/2−1
m=−K/2
I Receiver complexity does not depend on the channel delay spread!
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 10 / 26
Pros and Cons of OFDM
Pros:I Convert a dispersive channel to a set of parallel simple channels
{
zm = H(fm)s[m] + nm
}K/2−1
m=−K/2
I Receiver complexity does not depend on the channel delay spread!
Cons:I Poor performance on faded subchannels.I Sensitive to the Doppler effect
F Doppler shifts destroy the subcarrier orthogonality, and hence leadsto intercarrier interference (ICI)
I Large peak-to-average power ratio (PAPR)
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 10 / 26
How to Drastically Enhance Performance in FadingChannels?
0 5 10 15 20 2510
−6
10−5
10−4
10−3
10−2
10−1
100
SNR(dB)
aver
age
BE
R
fadingAWGN
(a) BER vs SNR
−5 0 5 10 15 20 250
1
2
3
4
5
6
7
8
9
SNR(dB)
Cap
acity
fadingAWGN
(b) Capacity vs SNR
Fading channel drastically affects the uncoded performance
Fading channel has the potential for reliable data transmission
Solution: coded OFDM with strong codes, e.g., Turbo, LDPCcodes that are capacity-achieving
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 11 / 26
How to Deal With the Doppler Effect?
ICI is inevitable!
Need signal processing algorithms to address ICI explicitly
Signal processing tailored to
underwater channels
OFDM demodulation
One example: Progressive receiver [JSTSP’2011]
There are other alternative approaches
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 12 / 26
Progressive Receiver
Adapt the receiver to channel conditions automatically without any apriori information
Achieves both low complexity and robust performance over time-varyingUWA channels
H0 H1
H2 H3
z
..
.
..
...
.
..
..
..
..
...
.
..
...
.
..
...
.
..
...
.
..
.
= + ns
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 13 / 26
Receiver Structure
Nosuccess or
D = Dmax
Yes
Output
decisions
Nonbinary
LDPC decoding
ICI equalization
Noise variance
estimation
Channel estimation
Increase D;
provide soft
information
Pre-processing;
set D = 0
z = HDs + n 1. The system model keepsbeing updated
Increase the span of ICIin equalization model
Increase the maximumpossible Doppler spreadin channel estimation
2. Soft information from thechannel decoder is utilized
3. No extra pilot-overheadneeded
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 14 / 26
Block Success Percentage: SPACE08
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8
D = 3
D = 2
D = 1
D = 0
Success percentage vs. number of phones
S1 (60 m)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8
D = 3
D = 2
D = 1
D = 0
Success percentage vs. number of phones
S5 (1000 m)
Averaged over Julian dates 295-302
With 4 phones: 90% (D = 0), 95% (D = 1), and up to 98.8% (D = 3)
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 15 / 26
How to Alleviate the PAPR Impact?
6 8 10 12 14 1610
−3
10−2
10−1
100
Thresh [dB]
Pr(
PA
PR
>T
hres
h)
QPSK8−QAM16−QAM
Multicarrier without
PAPR control
Multicarrier with PAPR
control
Single−Carrier
QPSK: The gap between OFDM and single-carrier is about 6dB
QAMs: The gap between OFDM and single-carrier is about 4dBDesign considerations on power amplifier and transducer:
I Peak or average power constrained?
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 16 / 26
Summary
OFDM is an elegant scheme
It has a clear advantage for short-range long dispersive multipathchannels.
It is an appealing technique for shallow-water high-data-rateacoustic applications
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 17 / 26
WUWNet’07 Demo, Sept. 2007
Link A to B
Link B to A
Single-input single-output (SISO) OFDM in-air demonstration
The data rate is 3.1 kb/s, with QPSK modulation, rate 1/2 LDPCcoding, and bandwidth of 5.5 kHz
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 18 / 26
WUWNet’08 Demo, Sept. 2008
Multi-input multi-output (2 × 2) OFDM in-air demonstration
The data rate is 6.2 kb/s, with QPSK modulation, rate 1/2 LDPCcoding, and bandwidth of 5.5 kHz
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 19 / 26
WUWNet’09 and ’10 Demo, Nov. 2009 & Nov. 2010
Aqua-fModem Prototype: With keyboard input and LCD display
Floating-point TMS320C6713 DSP board; running @ 225 MHz
Fixed-point TMS320C6416 DSP board; running @ 1 GHz
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 20 / 26
WUWNET’11 Demo: A Network of Modems
One-hop network, RTS/CTS based MAC protocol
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 21 / 26
Current / Future Issues (1)Multi-input multi-output (MIMO)
I Co-located: Increase the data rate via spatial modulation
x1
x2
x Nt
. . .
y1
y2
y Nr
. . .
h11
h12
hNtNr
I Distributed MIMO, asynchronous MIMO
User 1 User 2
Re
civ
ers
User 3
v1
v3 = 0
v2
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 22 / 26
Current / Future Issues (2)
Interference (Sonar, impulse interference, multiuser interference)
0 0.5 1 1.5 2 2.5
x 106
−600
−400
−200
0
200
400
600Timedomain_for_637_phone_10
I Interference mitigation / avoidance / alignment / management
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 23 / 26
Current / Future Issues (3)
Networking issuesI MAC
I Routing
I Reliable data transfer
I Applications
How to efficiently interact with higher layers?
Joint optimization (cross-layer design)?
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 24 / 26
Thank you!
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 25 / 26
Environmental Impact: S1
295 296 297 298 299 300 301 302 3030
1
2
3
4
Julian Date in 2008
Wav
e he
ight
(m
)
295 296 297 298 299 300 301 302 3030
5
10
15
20
Julian Date in 2008
Win
d sp
eed
(m/s
)
295 296 297 298 299 300 301 302 303
0
1
2
3
4
5
6
7
8
9
All success, D
max = 0
All success, Dmax
= 1
All success, Dmax
= 2
All success, Dmax
= 3
With errors, Dmax
= 3
S1 (60 m)
“All success”: all 20 blocks in a file correctly decoded
ICI-ignorant receiver (D = 0) works well during calm days
ICI-aware receiver (D > 0) is needed when the channel conditionsbecome worse
Shengli Zhou (University of Connecticut) Plenary Talk Dec. 2, 2011 26 / 26