7

Advanced coding for Underwater Communication

Hui Junying, L Liu , Feng Haihong and Liu Hong

(Underwater Acoustics Institute, Harbin Engineering University, Harbin 150001)

1. INTRODUCTION

IT is well known that the recent advances in satellite communication and

mobile communication technologies are not paralleled with same
developments for digital underwater acoustic communication systems.
In particular recent literature on underwater acoustic communications [1]-[21]
deal extensively with the challenges in high speed underwater acoustic
communications (see also chapter one of this book). However, although high
speed underwater acoustic communication techniques for vertical channels in
deep water environments are well investigated, there is no extensive studies
on the horizontal channel case and the limitation of the multi path effects in
this case. This is because in the horizontal channel case, the communication
rate is low and error bit rate is high which add further complexity and
challenge to apply the same vertical communication techniques to this case.
In this chapter, we present the design and development of a new Pattern
Time Delay Shift Coding (PDS) scheme applied to underwater acoustic
communication system in shallow water environments. Simulation results for
a 2kHz bandwidth communication system of a bit rate of 300 bit/s, and BER
less than 10-4 are achieved in simulated environments using this coding
scheme. Furthermore, the performance of the new method is presented and
the experimental underwater tests are discussed to verify the theoretical
concepts and validate the performance of the technique.
227
R.S.H. [stepanian and M. Stojanovic (eds.), Underwater Acoustic Digital Signal Processing and
Communication Systems, 227-246.
© 2002 Kluwer Academic Publishers.
228

2. CODING IN UNDERWATER ACOUSTIC

COMMUNICATION SYSTEMS

In this section we present some of the theoretical concepts of the Pattern Time
Delay Shift Coding (PDS) technique.
The major challenge of hydro acoustic communication in underwater
horizontal acoustic channels is the multipath interference. There are two
aspects of the multipath interference presented in the underwater acoustic
channel: code's multipath propagation time delay spread and received wave
pattern distortion by multipath interference.
The model of the impulse response function of the coherent multipath
channel is given by:

I
N·!
h( T) = Aob( T- To)+ Aib( T- T i ) (1)
i=!
Where Ai and Ti are sound ray parameters through the receiver
corresponding to the amplitude and arrival delay time respectively.
The sound ray determining the impulse response function is named acoustic
Eigenray.
If signal transmitted from the sound source is denoted by z (t), the received
signal in the multipath channel is given by:

N-!
S( t ) = AoZ( t - To ) + I AiZ( t - Ti ) + n( t ) (2)
i=!
Where the first term on the right is the direct wave, the second are refracted
and reflected with boundaries. When the signals of multipath arrivals are
overlapped with the direct arrival at the same time, it will derive interference,
so the waveform of composite signal is different from the transmitted signal.
The third term is noise; including environment noise and ship self noise.
When the delay time difference between the multipath ray and the direct
ray is larger than the code's duration of a symbol, the adjacent codes (symbols)
will be overlapped and interfered to cause "intersymbol multipath
interference". When the delay time difference is smaller than the code's delay
time resolution, "inner symbol multi path interference" will be introduced.
The simplest method to overcome the intersymbol multipath interference is to
introduce enough rest-periods between transmitting of adjacent symbols, that
is, the symbol interval To is larger than the multipath delay spread Td .
However, such method results in a low communication rate.
The other approach is to use Frequency Shift Keying (FSK) method. In this
approach the relationship between the numbers of used frequency points and
 229
the code's duration is typical given by:
M =11 Td 11+1 (3)
To
Where II. II is round-off operator, To is the code's duration, Td is maximum
delay spread.
At the receiver system, M filters are used to separate the symbols. In the
time interval of To, every filter only outputs the code's direct wave and
multipath arrivals of same corresponding frequency, so the intersymbol
interference is overcome. Although, the FSK is a well known reliable
communication system used for multipath channel conditions, but it still
requires broad bandwidth especially for the high bit rate digital
communication conditions.
The PDS separates symbol with some pattern codes. At the receiver end,
copy-correlators are used to separate the codes. To every pattern code, the
copy-correlator's reference signals are different. The copy-correlator outputs
a large correlation peak only when the received wave pattern is the same with
reference wave pattern. If the cross correlation coefficients between every
two codes are low enough, the copy-correlator can choose the corresponding
code. In the time interval of To, the copy-correlator outputs the peak of one
pattern code's direct wave and multipath waves, so the intersymbol
interference is overcome.
In PDS system design, the key to suppress intersymbol multi path
interference is to design a set of pattern codes, the cross correlation
coefficients of which are low enough. If there are M kind of pattern codes, the
delay time spread can be overcome is Td , when Td =M To. The PDS system
uses pattern to separate codes, so it uses narrower frequency bandwidth, and it
is more reliable than the FSK system. The PDS system codes with the delay
time difference between the information code and its correcting code.
The multipath interference problem can be mitigated by well-known
channel compensation technique such adaptive channel equalization
technique. Such techniques are effective to overcome the intersymbol
interference and inner symbol interference to a certain extent. Their recovery
of the transmitted signal from the received signal is via estimates of the
channel's impulse response. However, this will decrease the influence of
multipath interference and decode more correctly. Hence, neither of the above
methods is effective to cancel the effects of inner symbol interference with
extremely small delay time difference (multipath opposite-phase interference,
i.e., the direct wave is 1t phase shift to the reflected wave).
230

The general PDS signal frame structure is shown in Fig.I.

Frame Informa Informa Informa Informa Informa Informa Informa

-tion -tion -tion -tion -tion -tion -tion
synchron
-zstion code code code code code code code

code 2 n

~ -.0 ~ Block N

Fig.1 PDS signal frame Structure

The first code is named frame synchronization code, which provides time
reference for decoding. The first block codes "Block 0" are correcting codes
that provide time delay reference and detection threshold reference for
corresponding pattern information recoding. The correcting codes may
suppress the measuring error of time delay difference due to the inner symbol
multipath interference. It's noted that the correcting codes have the same
pattern with their corresponding information codes, so time delay deviation
due to inner symbol multipath interference in the slowly varying channel is
the same. If the correcting codes are used as time delay reference, the inner
symbol multipath interference has little influence to measuring the time delay
difference. The block codes after 'Block 0" are the information codes used for
transmitting data or messages. The code pulse width is Tc ' the time width used
for coding information is 1'" and the code interval is To = ~ + T" the time
delay 'difference between information code and its reference code represents
the messages. The number of pattern of frequency modulated (PM) codes is
M in every block. The cross correlation coefficients between every
two-pattern codes must be low enough. Their normalization cross correlation
coefficients must be less than 0.35 at least.
The general block diagram of the PDS digital communication system is
shown in Fig.2.
 231

"~~i~~"~····*"·
source + Channel I

shift

Fig.2 Block diagram of PDS underwater

Acoustic communication system

The source coder produces data sequence. The channel coder modulates
them into PDS time delay shift codes relatively to their correcting codes. The
received signal is distorted by multipath interference and noise. Before the
beginning of received message, copy-correlator searches the frame
synchronization signal. Once the PM signal is detected, the copy-correlator
opens a time window immediately and produces the corresponding pattern
reference signal in every time window. The copy-correlation detects the
code's correlation peak from the noise background. The detected code has the
same pattern with the reference signal, so the time delay difference between
the detected code's correlation peak and its correcting code's correlation peak
can be precisely measured. "Block 0" provides the time delay difference
reference for every pattern code. The information codes' time delay shift
sequence is translated into data sequence or messages. For every pattern code
of "block 0", the amplitudes of output peak of the copy-correlator are
different. They are time-varying and influenced by inner symbol multipath
interference and the channel's propagation loss. That the pattern codes'
detection threshold value can be determined by the corresponding correcting
codes' correlation peak multiplying a constant, which is smaller than one, it is
named adaptive threshold. The adaptive threshold can suppress intersymbol
interference and noise. Because every two pattern codes' cross correlation
coefficient is low enough, the adaptive threshold can suppress the
intersymbol multipath interference.
232

In brief, if the pattern codes' cross correlation coefficient is low enough,

the adaptive threshold can suppress intersymbol multipath interference.
Taking the correcting code's correlation peak as the reference of time delay
difference can decrease the measuring deviation caused by the inner symbol
multipath interference, so it can suppress the inner symbol multipath
interference. To transmit one block of correcting codes during the channel
coherent time interval may match the PDS codes with the slowly time-varying
channel, that is, the adaptive threshold and time delay reference may track the
channel's time-varying pattern.

3. TIME DELAY DIFFERENCE ESTIMATION

Since the PDS communication system uses time delay shift as information
coding, hence the time delay difference estimation represent one of the key
techniques in the decoding process of the system. The higher is the estimation
precision required, the higher would be the communication rate.
This process may be derived from the following formula:
The PDS system communication rate is given by [22]:

BS = 1 [Og2(T. + 1)1 bit! s (4)

~+T. 8
Where .1 is the quantization unit of time delay shift, T.
is the time width for
information coding, (~ + T.) = Yo is the code interval. In generally, we take
8 = 1/0 (5)
Where II is a constant, 0 is standard deviation of time delay estimator.
To transmit data with low error rate II must be big enough. We usually take
1l=4~5.
The correlator is the best time delay estimator in a band limited white noise
background. In order to make the correlator's time delay resolution be higher
than sampling period, interpolation technique must be adopted. In this case
the "cosine interpolation" method is used.
Assume the correlation function R( 't) around the peak is an approximate
cosine function, then:

R( T) = A cos mT+ B sinmT = -J A2 + B2 cos( mT- ffJ) = E cos( mT- ffJ)

(6)
ffJ=tg-1BI A (7)
There are three unknown quantItIes in above formulas: amplitude E
(correlation peak value), oscillation frequency of correlation function ro and
 233
initial phase <1>. If three sample values around the correlation peak R(k-l), R(k),
R(k+l) are given, three equations can be derived form the formula (6). From
them, ro and <1> can be solved. The cosine interpolation time delay formulas are
given by:

1 -1 R( k - 1) + R( k + 1)
OJ=-cos (8)
rs 2R( k)

-1 R(k -l)cosllkrs - R(k)cos(k -l)1"s

(jJ = tg (9)
R(k) sin OJ(k -l)rs - R(k -l)sinOJ(k + l)rs

1'= krs +!!!... (10)

OJ
Where 'Z's is the sampling period, k is the sampling number of the discrete
correlation function peak. r
is the estimated time delay.
Correlator estimation precision of the cosine interpolation is tested by
simulation experiments. The experiment condition is: pulse width of FM
signal is 10ms, center frequency is 5kHz, frequency bandwidth B=2kHz,
signal-noise ratio (SNR) is 5dB. The time delay estimation variances from
2000 times independent experiments list in table 1. A set of deviation samples
is shown in Fig.3. The time delay variance is about 8.9Ils. The maximum
deviation in Fig.3 is less than 341ls. When 11 in formula (5) is selected with
values of 4 and 5(~=35.6Ils and 44.5Ils), so that the recoding error rate might
be low enough.
Table- 1 below shows the results of time delay estimation variances with
different signal-noise ratio with the following parameters:
Ts=lOms, fo =5kHz, B=2kHz,

SNR(dB) Time delay estimation Time delay estimation

variances (S2) Standard deviation(s)
23 1.2842xl0- 12 1.133xlO-6
20 2.5979xl0- 12 1.612xlO-6
17 5.1506xlO· 12 2.269xlO·6
14 I.OO29xl0- 11 3.166xl0·6
II 1.9015xlO- 11 4.361xl0-6
8 3.5932xlO· 11 5.994xlO·6
5 7.9533xlO- 11 8.917xlO·6
2 1.742IxI0· IO 1.320xl0-5
234

U7 tillS}
0l)4 . - , - - -

Q1l:!

0lr1

.aD1

.(Ul2

.o..tXl
.0.,:)< t i n
D s:.o ~ '1IXl %lXl

fig.(3) Time delay estimation results for :WOO times in dependent experiments
(cosine interpolation time delay estimation simulation result)

In a typical multipath channel, either noise or inner symbol multipath

interference can introduce time delay estimation deviation. The codes' time
delay resolution is equal to the correlation peak width which is about lIB. If
there is inner symbol multi path interference, the correlation peak waveform
may be distorted by multi path interference, and the peak position and
amplitude may change (see Figs.7, 9 and 13). The correlation peak may
separate into two or many peaks. The constructive multipath interference
(same-phased overlap with the direct wave and reflected wave) may enlarge
the correlation peak, while the destructive multipath interference
(opposite-phase overlap) may decrease the correlation peak. In other words,
in a multipath channel, time delay measure is not definite, but time delay shift
measure may achieve high precision. On the account of above, a block of
correcting codes are set up in PDS system, which provide corresponding time
delay shift reference for corresponding pattern information codes. For the
correcting code and its corresponding information code, which have the same
pattern wave, the influence caused by multipath interference is the same, so
multipath interference couldn't influent time delay difference estimation
precision seriously. While the adaptive threshold could track the channel's
varying and provides the reasonable threshold reference for every pattern
information code. Furthermore, it is well known that the channel in ocean
environment is a slowly time-varying coherent multi path channel.
 235

Hence, one block of correcting codes is needed to transmit during the channel
coherent time length. In generally, it is only needed to transmit the correcting
codes one times in a few ten seconds if the relative speed between the
transmitter and receiver is not high. The other factor to influence the time
delay estimation is the relative motion between the transmitter and the
receiver. Motion can fasten the channel's varying, so correcting code
transmitting must increase correspondingly. Doppler effect due to relative
motion will introduce time delay measuring deviation. LFM signal's
ambiguity function is coupled to Doppler and time delay. For different value
of Doppler, the time delay corresponding to the correlation peak position is
different. Because the correcting code and the information code have the
same pattern, their time delay estimation deviations caused by Doppler are the
same and the time delay difference deviation can be eliminated.

4. UNDERWATER ACOUSTIC PDS COMMUNICATION

MODEL

A simulated model of the block diagram of the PDS communication

system shown Fig.2 is developed to emulate the performance of the PDS
method. The acoustic communication model consists of the coding
modular, the decoding modular and the channel model. The typical
response output of the acoustic simulated channel model is shown in
FigA.
236
[Link]"Y II"'IItOC'L£
1.70 l:;)UU

,,I
1\
1 I
I
'\J
\ -I
1 '\
\
I
1
1 \
\,t.,p"iifill1ii61ijWffi..d'filOW"[Link].dUtlnC6ou"[Link],il:,,".J 1 ,\1
.0
BOTTO""" REF1...E~

---r---.--':'!
.0

I, I I 11 _ _ 20.0""
_ ::0.0,....

:.1 0","_
I
",7,....
I' DI- DD.C
1I't_
1'\0-
::tOO..,...
2:JO.O~

fig. (4) Simulation result of a coherent channel

In the Fig. above the upper window on the left displays a ray trace diagram
which is a set of ray trace transmitted from the source with equalized angle
interval to understand the acoustic field's basic characteristic. The eigenray
cluster is shown in the middle window on the left. The eigenrays are a series
of special sound ray passing through the receiver which the parameters (Ai> 'ti»
determine the channel's impulse response function are shown in the lower
window on the left of the Fig. The calculation error of time delay of all
eigenray is less than 3/ls. The number of egienray is related to the following
effects: transducer's directivity, geometric position of the transmitter and
receiver, sea depth and profile of sound velocity etc. Once the channel
impulse response function is calculated, the simulation program can predict
the received signals form for the transmitted source. The sound velocity
profile is shown in the upper window on the right. The simulation program
enables to input actual measuring sound velocity profile, so communication
quality in any condition could be observed. The middle window on the right
displays bottom reflection coefficient curve. The simulation program enables
to input arbitrary bottom reflection characteristics. The simulation parameters
are listed in the lower window on the right. All parameters could be changed
to analyze the communication quality. Some of the simulation results of this
simulated model are shown in Fig.5 and Fig.6 respectively.
 .•
237

~ ~~
~L""""''''''''LE
. ,.eo 1~10

'-~~"~~=====~~:: !
eo "

1.00 ...--.::.r1····I---n'
:::: :::: =-->····I'
[Link] ···I-=wr-···'
::! ::::..I~··I
...... ·~I
::::::: ::..I·'·-----=I·
.... ·..·.. ·· =
:~: :::::!:::::::: ..::!::::
·· .. ·..::: :::!::::: ::::ru··! \0' ._ ........... ~ ............ _\
=..-·,·
-......
-·· 1 "OTTO'" REFLECf

1 _, .00 ; ~~~~ ~~~~ f:~~ J~~j=~~~: ~~~~~:::::;: ::~:~:~::::::::y::::::X:,: ~l\L:::~;;~~:~i;;;~~:~;;~~

I 1 , _ _ _ _ _ _ ____ _ , Cl DO

....- ~O.D",

.... - 2C.O",

,~& Illrl':~j:II:I:li~~;;·~~~wrn
...riL...... ,.............. .. I
......-- , '.0",

_1.De
........ 4 .. .:::::: . :::::: .:::::
j ..•
:::;:. :::::.
•
::::::::. :: :~: •. :::::;::::::::::;::::::::::::::::::::
R1:.--~oo.O!"'l"\

....... ~ ..- ::)0

[~I__llilN~W~~·~m~'.~
o 18~ ........ s~- 11.0dB

~,,­

Ct: 1.81

Cr-: ,.el
coo.: CI

fig.(5) PSD acoustic communication result of simulation

(inner symboI~e~tr:':I~t!ve multi path interference)

I~I~"--------"~---"-----~---~J~o~

~~~~~~ec l~ "

!
'-OO ~
.:::1::::1:::::::1::1::1::::::::::::::::::::::::;::;::;::::::I!' ~.9C!fT"OU
./'::::I~ITI:::
•••• _ - . . . '" •••• _ ......._ •• _ ••. _ .......................... _ ••••••••• _ .... [Link]
•• } • • • • • • • • • • • • • • u
....... .. ~
~~:~ ~:~: ~:~~ ::~~ .~:: :::~:~: :~::::y:~:~:::I::::::::L,~~~:~ :::~'~~1~~~::
.Jl
_ \ .00 ( ::

I ~ ,. -- -- ___ un_ __n

o _
.0

-1 0""" e;; t+- [Link]

.... - 20.0""

I '" 11!;I.I~I;lllli~~~~.1
r.- 20..otn
Rv-D-OO..om
-...- ,0
-2.13 C 'I:I~""" ~- [Link]

~""'"

11iIRmiwill~:j,~~i
.. ~ .. __ 1
Ct: ,eel
Cr: l.e1

1"::: o
I
fig. (6) PSD acoustic communication result of simulation
(inner symbol constructive multipath interference)

In these Figs. (from top to bottom), the left windows show eigenray
diagram, transmitted PDS code signal, received wave disturbed by noise and
238
multipath interference and output signal form the copy-correlator respectively.
while the right window represent the sound velocity profile, bottom reflection
coefficient and main parameters of the simulation system.
Fig.5 and Fig.6 show the simulation results of inner symbol multi path
interference. The former considers the inner symbol destructive multipath
interference, while the later considers the inner symbol constructive multipath
interference. On the account of above, the former received signal is not more
regular than the later, and its amplitude is smaller. The copy-correlator output
peak amplitude of the former is smaller, and its correlation peak waveform is
distorted as shown in Fig.7. The simulation results of above channel
conditions are listed in the Table 2 and Table 3 respectively. It can be
concluded from these results that in the inner symbol destructive multipath
interference case, the time delay estimation deviation is increased
considerably with a corresponding error rate increase. In order to reach high
reliable communication effect in such channel, higher SNR is needed.

J-_.
H(Y)

W
~f o
=c ,)'!-\cvy>o
w ~
A? =-'v\rvv. . ~~'""--
~ ~ ~
c /v,
~
--=~
~
Y(ms)

fig.(7) The detail correlation peak with inner symbol

destructive multi path interference

Table 2 Time delay estimation deviation and error rate in inner symbol
destructive mul --. ---- ------------- -------"
,-----~-------

292 bitls 383 bitls 453 bitls

~
SRN a,(ms)
l7dB .......... 0.01769
Pe
0
alms)
0.01775
Pe
0
a,(ms)
0.01747
Pe
0.00053
IldB 0.02999 0 0.03047 0.00022 0.03061 0.0074
---.l.4d~ _0.0~952_ 0.00022 0.04102 0.0023 0.04046 0.0194

Table 3 Time delay estimation deviation and error rate in inner symbol
Itioath interference (simul .
292 bitls 383 bitls 453 bitls
~
SR a,(ms)
17dB -.. . . 0.003248 I 0
I Pe o,(ms) I Pe
0.003113 I 0
o,(ms) I Pe
0.003182 I 0
 239

5. EXPERIMENTAL RESULTS OF THE SYSTEM

In this section the experimental set up and the underwater test results of the
system are presented. The aim of these underwater tests is to evaluate the
performance of the PDS communication system and to study the influence of
the tested underwater channel on the performance of the system. The block
diagram of the experimental test configuration is shown in Fig.8.

Wide
Compu Pattern Signal Compu
-band Hydro
-ter genera -ter
trans phone
(1 ) -tor (2)
-ducer

Oscillo ransmitting receiver

-graph system system

Fig.8 Block Diagram of the Experimental system

The underwater experimental test system is consisting of transmitter system

and receiver sub-systems. The transmitter sub-system consist of the
transmitter DSP-based computer system I, pattern generator, power amplifier
and wide-band transducer (with non-directivity). The pattern generator is
composed of (TMS320C25) fixed point DSP system. The computer system is
used for source coding, channel coding and controlling the pattern generator
to transmit PDS codes. The electric power of the power amplifier is about
30W. There are 5 pattern codes with center frequency (fo) of 5kHz, band
width of 2kHz, frame synchronization code duration of 20ms,information
coding time length T s=10ms, duration of information code pulse is 10 ms,
sampling frequency f s=6fo. The communication bit rate is changed from 70
bitls and 300 bitls respectively.
The receiver system includes hydrophone (non-directivity), normalized
amplifier and filter, signal processor (TMS320C25) and computer II. The
240
signal processor calculates copy-correlation estimates time delay difference
and decode in real time. At the same time, raw data and correlation output is
transmitted from the signal processor to the computer II. The raw data and
output data are collected and displayed in computer II in real time. The
copy-correlation output also may be observed through the oscilloscope after
DIA conversion.
The reservoir trial is performed in a wide water area with the depth of 40m
at the SongHua Lake in JiLin province in China. The transmitting ship is fixed
beside an island in the middle of the lake, while the receiver ship drifts with
the speed of about 3 knot. Surface wave height is about 30cm. The transducer
and the hydrophone are dipped with the depth of 15m and 5m respectively.
The system communicates data over a horizontal range of 1500m. With the
drifting of the receiver ship, the system marks statistic error rate, collects data,
displays correlation output and receives signal waveform and decodes
continuously during the drift. The correlation output waveforms at different
distances are shown in Figs. 9-11. The maximum error rate obtained with
three trials of the setup was less than 10-4.
 241

R(t)

D.B
~. ~
o·
o
-~-

$
)~~10 15 20 2$ 1I5 4D
t(ms)

fig.(9) Correlation output wavefonn (about 90m)

R(t)

, J\~, ,
10 15
--~
_rv-f
20
! -
2$
~([Link]
30 1I5 4D
t(ms)

fig.(10) Correlation output wavefonn (about 500m)

R(t)

DB J
Q.8

0-4

D.2
DE crJ"V\o=c ocyJVVV"V\!V'..,,,,,,, . ."c",r- it(ms)
o 10 15 20 2$ 30 35 4D

fig.(ll) Correlation output wavefonn (about 800m)

The Figs above show the correlation peak of the frame synchronization code
varying with the distances tested. The distances (270m, 460m) with high
correlation peak as shown in Fig.12 correspond to the case of inner symbol
constructive multipath interference. For the test distance (300m) with low
correlation peak as shown also in Fig.12 correspond to the case of the inner
symbol destructive multipath interference. In the condition of destructive
multipath interference the maximum error rate were at the level of 10"4.
242

Rmax(t)

,eo
I'Gr......."'~"
150 ,

120 . . . ~~~~
., I

90

so

ro ~-
t(s;
o ~I------~----~------~----~------~------~----~----~
o 100 200 XX) 400 500 600 700 BOO

fig. (12) Correlation output peak varying with time (distance)

(relative speed is about 3 knot, experiment results)

~~~~
2 ~t----------------------------------------------------~

'5~~vJ\ I

0.5 .-
\

o ~I----~--~------------------------~------------~
o 200 4O.J EM) 800 10Xl 1200 1400 1600 1800 range(m)

fig.(13) Simulation correlation output peak varying with distance

The results shown in Fig. 13 confirm that that the simulation results of the
system's model correlate well with the corresponding results of the actual
underwater trial conditions. Fig.I4 and Fig.I5 give the trial's correlation
output waveform every second during a testing period of lOs. This illustrates
that the channel is stable enough even if the receiver ship drifts at speed of 3
knot.
243

! ~
<J)

Correlator output wave group in every 1s

.......
~
c
4)
;>
4)
f
.-s::::

-.
6
§'
0
bh

'CXI
~ 4)
~
~ -1 a
}
~

110
0
I-<
-«iI::
0
4)
0
U
--

2D

fig (14)
"<:t
.......
." . ." ."
'-'
E
•. .. OIl

R(tJ
0

0
d 0":1 d•

:~
a:: ::I " tH

D
0

D..I

0.-
,

0._

0
"I

0.1
::I

0
D.B
D._
<>
::I

D
d

D-
O

D..I
o
::I

D-
D

D.B
D
0
d

D-
<>

D.B
D
d

D.-

D.-
D.B
~ d 0

D.I
~

D.B
0._
~

0
0
~ " 0 Q 0
244

R(t)

. I\
f,
o.~ l-
0.' I-
~ rJ \r\~

:
o

_A~
o.e
0.' I-
o

~J\~~
0.1 I-
0.' I- 0" •

o
0.1

0.'
o
-
_~J\~,_
-
~J\A.~
0.8

0.' l-
o

~J\~~
O.B I-
0.' -
,
o
o.e I-
0.4

a
- _J~~
-
_r\!\~r-
0.8
0.4 I-
o

_.J\,\~ _
0.8

-
0.4

o
-
_J\~~
0.8
04

o t(ms)
o 10 \2 14
l' II 31

[Link]) Detail correlator output wave offig.(14)

6. CONCLUSIONS

High-speed underwater acoustic digital communication is constrained by

the channel multi path interference whose influence is presented in the inner
symbol multipath interference and the intersymbol multipath interference.
The later limits the communication rate, while the former influents the
communication reliability and error rate. To overcome these effects, a typical
FSK communication that divides the frequency band to circumvent the
intersymbol multipath interference and uses a group of filters to separate
 245
codes with different frequency is used. However, this methodology requires
wide bandwidth for high data rate transmission requirements.
In this chapter a new PDS communication system technique that separates
codes by a group of copy-correlators with different reference signals is
presented. Hence, it can circumvent the intersymbol multipath interference to
achieve high data rate with relatively narrow bandwidth. The key features of
this method are that the cross correlation coefficients between every
two-pattern code must be low enough and their relevant normalization cross
correlation coefficients are less than 0.35.
A simulation system and real underwater test results are presented to
evaluate and correlate the performance of the PDS communication system.
This study illustrates that PDS communication system provide a more robust
approach especially in shallow water channel conditions. In the underwater
trials, the system was able to transmits data with bandwidth of 2kHz and bit
rate of 300 bit /s with a maximum error rate around 10-4 •
Further work on using relevant channel equalization techniques can
improve further the performance of the system in terms of better BER and
higher transmission rates and this is could be a further development to the
method presented in this work.

7. REFERENCES

1. Dr. Rodney Coates, "Underwater Acoustic Communication", Sea

Technology, pp41-47, July, 1994.
2. [Link] and AZielinski, "Multipath rejection using Narrow Beam Acoustic
Link", IEEE,pp287-290, 1988.
3. Manora [Link], "A Multi-frequency Digital Communication Technique
for Acoustic Channel with Multipaths" IEEE, pp140-145, 1987.
4. Michiya Suzuk, et aI, "Digital Acoustic Telemety of color Video
Information", Proc. Ocans'89, Settle, Washington, 1989, pp892-896.
5. P. Bragard, G. Jourdain, and J. Martin, "Optimal Adaptive Algorithms
Behavior used in underwater Communication Signal Equalization", Signal
processing IV :Theories and applications, pp363-366, 1988.
6. Milica Stojanovic, Josko A Catipovic, John [Link], "Phase Coherent
Digital Communication for Underwater Acoustic Channels", IEEE Journal of
Oceanic Engineering, Vol. 19, No.1,pp100-111, January, 1994.
7. Qian Wen and James A Ritcey,"Spatial Diversity Equalization Applied to
Underwater Communications", IEEE Journal of Oceanic Engineering, Vol. 19 ,
No.2, pp227-241, April 1994.
8. [Link], J. R. Galloway, and M. R. Balderson,"Auto Aligning System
for Narrow Band Acoustic Telemetry", Occans'85, San Diego.
246
9. [Link], etal, "Reliable Underwater Acoustic Data Link Employing an
Adaptive Receiving Array",Proc. ofthe inst. of Acoustics, Vo1.9, April, 1987.
10. [Link], et al,"An Acoustic Communication System for Subsea Robot",
[Link]'89 Settle, Washington,1989,pp765-770.
[Link] Zhoubin, Feng Haihong, Hui Junjing "A New Uderwater Acoustic
Communication System", Journal of Acoustics(Chinese) VoLl8, No.5, Sept.,
1993,pp892-896.
12. [Link], G.S. Howe, and Adams," An Adaptive High Bit Rate Subsea
Communication System", European Conference on Underwater Acoustics,
1992,9, pp75-79.
13. Arthur B. Baggeroer, "Acoustic Telemetry-An Overview", IEEE Journal
of Oceanic Engineering, Vol. ,OE-9, No.4, pp229-235, October, 1984.
14. M. Stojanovic, [Link] and [Link], "Adaptive Multichannel
Combining and Equalization for Underwater Acoustic Communication",
[Link]. [Link]., Vol.94, No.3, Sept.,ppI621-1631, 1993.
15. [Link] and GJourdain, "Adaptive Equalization for Underwater Data
Transmission", IEEE, pp 1171-1174, 1989.
16. [Link], GJourdain, and G. Loubct, " Codind for Communication
through Muitipath Channels and Application to Underwater Casc", Signal
Processing ill: Thcorics and Applications, pp 1087-1090, 1986.
17. B. Solaiman, A. Glavicux, and A. Hillion, "Performance of Slow
Frequency Hopping BPSK System Using Convolutional Coding in
Underwater Acoustic Media", Proc. Icassp'88.
18. Josko [Link], "Performance Limitations in Underwater Acoustic
Telemetry", IEEE Journal of Oceanic Engineering VoLl5 No.3, pp 205-216,
July, 1990.
19. [Link], G. Jourdain, and Martin."Optimal Adaptive Algorithm
Behavior Used in Underwater Communication Signal Equalization", Signal
Processing IV:Theories and Application, 1988, pp363-366.
20. Salvatore D. Morgera, Keith A. Reuben, and Cedric Cole, "A
Microprocessor-Based Acoustics Telemetry System for Tide Measurement",
IEEE journal of Oceanic Engineering, Vol. Oel1, No.1, January, 1986, pp
100-108.
21. G.H. Sandsmark, "High Speed Underwater Acoustic Data Transmission:
A Brief Review", Proc. of Int. Workshop on Marine Acoustics ,pp133-140,
Mar, 1990, Beijing.
22. Hui Junjing. "Underwater Acoustic Channel"(Chinese), Defence Industry
Press, 1991.