http://www.paper.edu.

cn
China Ocean Engineering , Vol. 20 , No. 3 , pp . 499 - 508

2006 China Ocean Press , ISSN 089025487

Under w
ater Acoustic Communication Based on Pattern
Time Delay Shift Coding Scheme

YIN Jing2wei ( ) a , HUI J un2ying ( ) a , HUI J uan ( ) a ,

Y
AO Zhi2xiang ( ) a , b and WANG Yi2lin ( ) a

College of Underwater Acoustic Engineering , Harbin Engineering University , Harbin 150001 , China
b

College of Electric , Navy Engineering University , Wuhan 430043 , China

( Received 15 August 2005 ; accepted 18 April 2006)

ABSTRACT

Underwater acoustic communication based on Pattern Time Delay Shift Coding ( PDS) communication scheme is
studied. The time delay shift values of the pattern are used to encode the digital information in the PDS scheme , which
belongs to the Pulse Position Modulation ( PPM) . The duty cycle of the PDS scheme is small , so it can economize the
power for communication. By use of different patterns for code division and different frequencies for channel division , the
communication system is capable of mitigating the inter symbol interference ( ISI) caused by the multipath channel. The
2
data rate of communication is 1000 bits/ s at 8 kHz bandwidth. The receiver separates the channels by means of band
2
pass filters , and performs decoding by 4 copy correlators to estimate the time delay shift value. Based on the theoretical
2
analysis and numerical simulations , the PDS scheme is shown to be a robust and effective approach for underwater acous
2
tic communication.
Key words : underwater acoustic ( UWA ) communication ; pattern time delay shift coding ( PDS ) ; estimation of time de2
lay shift ; multipath channel ; inter2symbol interference ( ISI)

1. Introduction

Underwater acoustic (UWA) communication is a fast developing field , and its application is not

limited to military affairs , but is also extending into commercial fields. Catipovic ( 1990) , Stojanovic

( 1996) and Kilfoyle and Baggeroer ( 2000) pointed out respectively that the underwater acoustic chan2

nel are far from ideal . The available underwater acoustic bandwidth for shallow2water acoustic commu2
nication is limited to a few kHz depending on both the range and frequency. In addition , the acoustic
signals are affected by spatial2temporal variation of the multipath channel , and that may result in severe
inter2symbol interference ( ISI) ( Robert and Stojanovic , 2002) . These characteristics restrict the range
and bandwidth for reliable communication and lead to a low data transfer rate.

The key to the realization of real2time underwater acoustic communication is to overcome the ISI

caused by the multipath channel . T overcome the difficulties brought about by time2varying multipath ,
o

the design of commercially available UWA communication systems has so far relied mostly on the non2

coherent modulation techniques which provide a relatively low data rate. Robustness and simplicity of

implementation are their advantages ( Proakis , 1991) . In order to mitigate the ISI , the existing nonco2
herent systems employ guard time making the interval between subsequent pulses longer than the multi2

1 Corresponding author. E mail : yinjingwei @hrbeu. ebu. cn


500

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

path time delay. The insertion of guard time results in a reduction of the available data rate. In addi2
tion , the bandwidth of frequency separation of the Multiple Frequency Shift Keying (MFSK) system is

desired to be wider than the coherence bandwidth so that several frequency channels can work at the
same time , and this further reduces the system efficiency. Chitre et al . ( 2005) used the Orthogonal
Frequency Division Multiplexing ( OFDM) for UWA communication which is becoming a modulation

technique chosen for wireless communication , which can provide a large data rate with sufficient ro2
bustness. In an OFDM scheme , a large number of orthogonal , overlapping , narrow band sub2channels

or sub2carriers , transmitted in parallel , divide the available transmission bandwidth. When OFDM is

applied to UWA communication , the efficiency is reduced due to the high frequency separation among

the channels and relatively long guard time. Recently , phase coherent modulation has developed rapid2

ly which provides a data rate an order of magnitude higher than that of the existing noncoherent sys2
tems. Phase coherent modulation real2time systems ( Suzuki and Sasaki , 1992) have been implemented
mostly for the application to vertical and very short range channels , such as to the deep ocean vertical

path channel in a Japanese image transmission system.

(1999) uses the time delay shift values of the pattern to code the information and the duty cycle is

small for economization of the power. The PDS scheme adopts the code division and correcting codes in
2

order to enable every information2code to be against the ISI caused by the multipath channel . The sys2
tem is also more robust against distortions , multipath fading and noise ( potentially non2Gaussian) in
the horizontal channel due to the larger bandwidth. The communication system occupies 5 13 kHz ,

divided averagely into 4 subbands for 4 communication channels ; the data rate reaches 1000 bits/ s in

the horizontal range of 10 km.

2. Pattern Time Delay Shift Coding

The information is not modulated on the carrier wave form in the PDS scheme. The time delay

shift value of the pattern is used to encode the digital information and different values represent differ2
ent information ( Fig. 1) . The duration of every pattern is Tp and the encoding time window for infor2
mation coding is Tc ; so the duration of one symbol is T0 = Tp + Tc and the duty cycle is Tp/ T0 small2

er than 1. If one symbol takes n bits digital information , the quantization unit of time delay shift is
= Tc / ( 2 n - 1) . The band of the system is divided averagely for N communication channels and every

channel has L patterns which are orthogonal to each other. Fig. 1 shows the sketch of pattern time de2

lay shift coding of one block.

Fig. 1. Sketch of pattern time delay shift coding.

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

501

In Fig. 1 , = k is the time delay shift value , where k = 0 , 1 , , 2 n - 1 , and is the

The Pattern Time Delay Shift Coding scheme belongs to the Pulse Position Modulation ( PPM) .
For communication channel I , the PDS coding signal is given by :
+ L- 1

s I ( t) =

p [ t j

i =0

j =0

( j + L i ) T0 - k ij ] , k ij = 0 , 1 , , ( 2 n - 1)

( 1)

)
where pj ( t ) is the j2th pattern whose duration is Tp ; ( k ij is the time delay shift value of the ( L
+ j + 1) 2th information2code.
i

The data rate for one communication channel is given as :

= log2

Tc

+1

/ T0 = n/ T0 .

( 2)

It is easy to find from Eq. ( 1) that the data rate is related to the parameters T0 and n . The data

rate becomes lower when the duration of one symbol is longer , and it becomes higher when is
shorter.

2. 1 The Rule for Pattern Selection

The PDS system makes use of the multiform pattern for code division. The pattern selection is so

important that it influences directly the ability of anti2multipath interference. The selected patterns
should have a keen2edged auto2correlation peak and a low cross2correlation coefficient for each other in

order to mitigate the ISI effectively.

There are L = 5 patterns in every communication channel which has an ability of anti2multipath

interference time extension of 5 = 100 ms. The pattern is made up of 7 chips , namely , 7 CW
20
chips , which are connected end to end and have different frequencies in the bandwidth of 2 kHz. The
assembly and combinations for the 7 chips are 7 !. The phase of each chip is either 0 180 so the
or
,

total combination kinds of the pattern are 7 ! 7 . Five patterns are selected for one communication
2
channel which has a keen2edged auto2correlation peak and a low cross2correlation coefficient ( < 0. 3) .

The lower the cross2correlation coefficient , the less the interference among different patterns and the

more reliable the communication system. The normalized correlation waveforms of Channel I are shown

in Fig. 2. The diagonal windows in Fig. 2 are auto2correlation waveforms of each pattern , and the oth2

ers are cross2correlations.

2. 2 Code Structure of PDS System

The code structure of the PDS system is shown in Fig. 3.

( 1) The wakening code is used to waken the system to work. It is just sent at the beginning of

communication.

( 2) The channel impulse response ( CIR) is estimated by sending an LFM signal . At the receiv2

er , the copy2correlator of LFM signal will have a series of correlation peaks and the multipath time de2


502

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

Fig. 2. The normalized correlation waveforms of Channel I.

Fig. 3. The code structure of PDS system.

lay T ISI could be estimated , as shown in Fig. 4.

Fig. 4. Channel impulse response of shallow water for the source at the depth of 50 m ,
and the receiver at 60 m , the length of transmission being 10 km.

( 3) The synchronous2code also uses an LFM signal . It is used to fix the time base for receiving

the correcting codes and information codes. Between the synchronous code and the correcting code ,
there should be an interval of T ISI in order to mitigate the interference with the correcting code caused
by the synchronous code.

( 4) The correcting codes are made up of all the patterns used in the communication system. At

the receiver , they will be received as the reference signals for the copy2correlators. They give the crite2

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

503

rion of zero time delay shift and mitigate the inner symbol multipath interference. In order to receive
every correcting code exactly not interfered by the others , an interval of T ISI should be reserved follow2

ing each correcting code. There are totally 20 patterns for the 4 communication channels. One correct2
ing code occupies ( 20 + T ISI) ms , so the whole time occupied by the correcting codes is 20 ( 20 +
T ISI) ms. The correcting codes have 5 patterns for one communication channel ; they all take the digi2

tal information0 0 0 0 0 namely , the zero time delay shift . After the time base is fixed by the syn2
,
chronous code , these correcting codes are received and stored.

( 5) The information codes are behind the correcting codes which encode the digital information by

the time delay shift values of the pattern. There are a series of information code blocks and the number

of blocks depends on the duration of the relative stability of the underwater acoustic channel . There are

5 patterns for one information code block corresponding to the correcting codes for every communication
)
channel . Because of the 4 communication channels ( , , , and working at the same time ,
every block of the information code is the summation of 4 blocks coming from the 4 communication
)
channels ( , , , and .

2. 3 Communication Flow

The communication system parameters are defined as follows : Tp = 8 ms , Tc = 12 ms , and n =

5 bits. The data rate of one communication channel is 5 bits/ 20 ms = 250 bits/ s , so the total data rate

of the communication system is 1000 bits/ s.

The PDS underwater acoustic communication system includes 3 parts : source coding , channel

coding and decoding , as shown in Fig. 5.

Fig. 5. The underwater acoustic communication system.

The signal transmitted at last s ( t ) is the summation for the 4 communication channels shown in

Fig. 5.

s ( t ) = s ( t ) + s ( t ) + s ( t ) + s ( t ) .

( 3)

the receiver , the first step is to search the synchronous code by the auto2correlator. There are
At
a set of multipath signals corresponding to the multiple reflection of the incident wave on the interfaces ,


504

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

a set of correlation peaks will be obtained by the correlator. The time corresponding to the maximal
peak is regarded as the time base. Then the correcting codes are received and stored into the RAM as
reference signals for the copy2correlators. The signal s ( t ) is the summation for the 4 channels ( s ,

s , s , and s ) because the 4 communication channels work at the same time , so there are 4 band2

pass filters ( : 57 kHz ; , : 79 kHz ; : 911 kHz ; : 1113 kHz) at the receiver to

separate the signals for the 4 communication channels. Every communication channel has a copy corre2
2

lator for decoding (whose reference signal should change every other T0 in order to keep the pattern
same with their corresponding information codes) .

3. A Theory for Mitigating ISI

In this section discussed is the ability of PDS system to mitigate the ISI.

The model of the channel impulse response function of the coherent multipath channel is given as

( Hui , 1992) :

h ( t) =

A
( t - )
i

( 4)

where A i and are sound ray parameters corresponding to the amplitude and time delay , respectively.
i

Two fundamental mechanisms of multipath formation are reflection at the boundaries and ray bend2

ing. Multipath propagation will occur in surface or bottom bounces and may be responsible for severe

degradation of the acoustic communication signal , since it generates ISI. The acoustic channels may
have extremely heavy multipath extension such as shown in Fig. 4a , whose value depends on the water
depth and communication range.

Multipath extension is an important figure of merit for the underwater acoustic communication sys2

tem. It is important and necessary to mitigate the ISI caused by the multipath channel for achieving
highly reliable communication.

The communication system adopts orthogonal patterns for code division. There are 5 patterns for

one communication channel and the duration of each pattern is 20 ms , so the duration of one block of

information code is 100 ms , namely , the next same pattern appears 100 ms later. implying that the
system has the maximum ability of anti2multipath interference within 100 ms that mitigates the ISI ef2

fectively.

In addition , the received correcting codes serve as the reference signal for copy correlators which
2

provide time delay shift reference for corresponding information codes. On account of the relative sta2

bility of acoustic channels during the coherent time , the influences of the multipath interference on the
information code and the correcting code which have the same pattern are almost the same. So the cor2

recting code is beneficial to the mitigation of the inner symbol multipath interference and precise esti2

mation of the time delay shift .

4. Simulated Experiment Results

The underwater acoustic channel is considered as a coherent multipath channel which varies slowly

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

505

and constant during these information code blocks. The ray acoustic theory provides a basis for such a
propagation model .

The communication system occupies the 513 kHz band , which is divided averagely for 4 com2

munication channels. In this experiment , the sampling frequency is 48 kHz and the input SNR is 10
dB at the receiver.

The simulated experiments were performed in a shallow water channel . The sound2speed profile

for the experimental environment is shown in Fig. 6. The depth of the water is 105 m. The channel is
characterized by extended multipath and the multipath time delay is nearly 100 ms. The source and the
receiver are both positioned at the depths from 50 m to 100 m , and the acoustic channel is varied with
the change of their positions. The transmission range is 10 km.

Fig. 6. The sound2speed profile of the shal2

The shallow water transmission results are shown in Fig. 7 , the data rate being 1000 bits per sec2

ond. The original signal before it is sent out is shown in Fig. 7a ; it passes through the shallow water
channel as shown in Fig. 4a ; the received signal consists of multipath and noise , as shown in Fig.
7b.

Fig. 7. Shallow water transmission.

At the receiver , first , auto2correlation of the synchronous code is done to fix the time base for re2

ceiving the correcting codes. Second , the correcting codes are received and stored in the RAM as the


506

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

reference signals for the copy filters for the decoding of the corresponding information code. Finally ,
the information codes are received , the signals are separated for the 4 communication channels by
band2pass filters , and the information is decoded by copy2correlators through estimation of the time de2

lay shift .

Now let us discuss the effect of correcting codes in the PDS communication system. The correla2

tion output waveforms of one information code symbol are shown in Fig. 8.

Fig. 8. The correlation output waveform.

Fig. 8a shows the correlation output waveform for the correcting code serving as the reference sig2

nal for copy2correlators , and its correlation peak is shown in Fig. 8b ; Fig. 8c shows the correlation

output waveform for the original pattern serving as the reference signal for copy2correlators , and its cor2

relation peak is shown in Fig. 8d. The correlation peak is separated into two or more peaks by inner

symbol multipath interference , as shown in Fig. 8d. As indicated in Section 2 , the correcting codes

provide time delay shift reference for corresponding information codes and mitigate the inner symbol

multipath interference effectively , as shown in Fig. 8b. So the inner symbol multipath interference
could not influence the estimation precision of time delay shift in the PDS scheme by virtue of the cor2

recting code.

Many simulations are done in order to test the communication system , and some results are given

in Table 1. The horizontal range of communication is 10 km. The digital information for transmission is
4000 bits at the data rate 1000 bits/ s and it is transmitted 100 times for testing the bit2error2rate per2

formance.

5. Conclusion

Underwater acoustic communication based on the PDS scheme encodes the digital information in

time delay shift values of the pattern. The PDS communication system uses a number of different pat2

terns for code division and different frequencies for channel division , so the system has a high capabili2
ty of anti2multipath interference. And it could mitigate the inner symbol multipath interference in virtue

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

507

of the correcting code. The data rate is 1000 bit/ s with a stable reliability. The small duty cycle is
beneficial to the economization of the power for communication , which is valuable for power2limited

and bandwidth2constrained underwater acoustic communication. The PDS system has a great potential

of application for underwater acoustic communication networks.

The experimental results

Source depth
( m)

Receiver depth
( m)

50

50

0. 0010

0. 0006

70

0. 0064

0. 0163

80

60

0. 0657

90

0. 0015

80

0. 0029

50

0. 0510

0. 0021

100

0. 0025

50

0. 0105

0. 0003

60

0. 0001

0. 0021

90

0. 0005

0. 0030

50
50
60
60
70
80
80
90

100
100

Bit2error2rate ( %)

Mean

0. 0004

0. 0123

0. 0023

0. 0093

0. 0012

0. 0783

0. 0199

0. 0044

0. 0175

0. 0007

0. 0022

0. 0011

0. 0007

0. 0410

0. 0521

0. 0365

0. 0006

0. 0415

0. 0012

0. 0134

0. 0015

0. 0015

0. 0013

0. 0010

0. 0001

0. 0012

Notes : , , ,
communication channels ; Mean bit2error2rate of the communication system and e2
the

qual to the mean for , , , and .

In order to make the PDS system more available to underwater acoustic communication , the data

rate should be changed for more reliable communication to meet the demands of different acoustic chan2

nels. In addition , the measurement precision of the time delay shift should be improved for stable reli2

ability and the quantization unit of time delay shift should be reduced for a high data rate. For mitigat2

ing the ISI , Edelmann et al . ( 2002) , Heinemann et al . ( 2003) and Rouseff ( 2005) had made use

of the time reversal mirror ( TRM) . We are also researching on the application of the time reversal mir2

ror ( TRM) to underwater acoustic communication based on the PDS scheme , because TRM could
match the acoustic channel automatically and lead to an adaptive focusing which could mitigate the ISI
effectively.

References

Catipovic , J . A. , 1990. Performance limitations in underwater acoustic telemetry , IEEE J . Oceanic Eng. , 15 ( 3 ) :
205216.
Chitre , M. , Ong , S. H. and Potter , J . , 2005. Performance of coded OFDM in very shallow water channels and snap
2
ping shrimp noise , IEEE/ MTS OCEANS
2005 ,Washington DC , 1823.
K
ilfoyle , D. B. and Baggeroer , A. B. , 2000. The state of the art in underwater acoustic telemetry , IEEE J . Oceanic
Eng. , 25 (1) : 427.
Rouseff , D. , 2005. Intersymbol interference in underwater acoustic communication using time reversal mirror signal pro2
2
cessing , J . Acoust . Soc. Am . , 117 (2) : 780788.
Edelmann. G. F. , Akal , T. , Hodgkiss , W. S. , K , S. , Kuperman. W. A. and Song , H. C. , 2002. An Initial
im


508

http://www.paper.edu.cn
YIN Jing2wei et al . / China Ocean Engineering , 20 ( 3) , 2006 , 499 - 508

Demonstration of Underwater Acoustic Communication Using Time Reversal , IEEE J . Oceanic Eng. , 27 (3) : 602
609.
Heinemann , M. , Larraza , A. and Smith , K. B. , 2003. Experimental studies of application of time reversal acoustics
2
to noncoherent underwater communications , J . Acoust . Soc. Am . , 113 (6) : 31113116.
HUI Jun2ying , 1992. Underwater acoustic channels. 1 , Beijing , National Defense Industry Publisher , 3033. (in Chi2
nese)
HUI Jun2ying , LIU Li and LIU Hong , 1999. A study on Pattern time delay coding underwater acoustic communication ,
Chinese Journal of Acoustics , 24 (6) : 561572. (in Chinese)
Proakis , J . G. , 1991. Coded modulation for digital communications over Rayleigh fading channels , IEEE J . Oceanic
Eng. , 16 (1) : 6673.
Robert , S. H. and Stojanovic , M. , 2002 , Underwater acoustic digital signal processing and communication systems , 1 ,
Boston , K
luwer Academic Publishers , 47.
Stojanovic , M. , 1996. Recent advances in high rate underwater acoustic communication , IEEE J . Oceanic Eng. , 21
(2) : 125136.
Suzuki , M. and Sasaki , T. , 1992. Digital acoustic image transmission system for deep sea research submersible , Proc.
OCEANS ,Newport , RI , Sept. ,1992 , 567570.
92