You are on page 1of 9

Ad Hoc Networks 89 (2019) 161–169

Contents lists available at ScienceDirect

Ad Hoc Networks
journal homepage: www.elsevier.com/locate/adhoc

Time reversal communication using vertical particle velocity and


pressure signals in shallow water
Sunhyo Kim a, Hyeonsu Kim b, Seom-kyu Jung a, Jee Woong Choi b,∗
a
Maritime Security Research Center, Korea Institute of Ocean Science and Technology, Busan, 49111, Republic of Korea
b
Department of Marine Science and Convergence Engineering, Hanyang University-ERICA, Ansan, Gyeonggido 15588, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: Acoustic communication in shallow water is characterized by multipath channels, which cause significant
Received 25 April 2018 delay spreading, leading to inter-symbol interference. This inter-symbol interference necessarily causes
Revised 2 January 2019
a significant degradation in communication performance. Although a time reversal technique has been
Accepted 19 March 2019
reported to produce satisfactory performance in multipath dominant environments, this technique re-
Available online 20 March 2019
quires a large receiver array covering the water column to achieve reliable communication performance.
Keywords: In this paper, a time reversal single-input multiple-output system using the pressure signal and the ver-
Underwater acoustic communication tical component of the particle velocity is presented. The vertical component of the particle velocity was
Particle velocity estimated using the finite difference in gradient between pressures measured by two vertically adjacent
Time reversal technique receivers. The experiment was conducted in shallow water off the south coast of Korea, where the water
depth is 59 m and the bottom consists of silty clay. The results showed that the time reversal communi-
cation system based on vector quantities performed better than systems where only the pressure signals
were used.
© 2019 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/)

1. Introduction of spatially separated receivers was applied as an alternative to


the equalization process in underwater acoustic communications
When sound propagates through shallow-water waveguides, [13–15]. The time reversal technique has a relatively simple re-
which are characterized by large temporal and spatial variations, ceiver structure and reduces the ISI effectively in a multipath en-
it undergoes multiple interactions with the sea surface and bot- vironment [16–18]. The time reversal was then combined with the
tom interfaces. This produces time-varying delays and frequency phase tracking process to compensate for the doppler effect due to
spreads. Such a time-varying multipath channel makes underwater the source and receiver movements [13] and equalization process
communication challenging [1]. The time-delay spread often cov- to remove the residual ISI remaining after time reversal processing
ers hundreds of information-carrying symbols and thereby causes [14,15]. Although the time reversal technique is robust to multipath
severe inter-symbol interference (ISI). The presence of ISI signif- fading environment, it still requires a relatively large-scale receiver
icantly degrades the communication performance [2–5]. In addi- array because the spatial diversity improves the bit error rate (BER)
tion, sea-surface movement produces time-varying channels ex- performance of the time reversal communication system.
hibiting relatively short coherence time and a large Doppler spread In recent years, the physics of sound propagation in the ocean
[6–8]. has been investigated in terms of acoustic vector quantities such
Single-channel equalizer was mainly used in coherent acoustic as the acoustic particle velocity, acceleration and displacement
communications to compensate for the ISI until the 1990s [9,10]. [19,20]. In general, the acoustic vector sensor performs better than
After that, a single-channel receiver was extended to a multi- hydrophone because it measures acoustic particle velocity as well
channel receiver system to improve the communication perfor- as pressure at a single point. Accordingly, the acoustic vector sen-
mance. However, there was a limitation that multi-channel equal- sor has been applied in various areas such as sound source track-
izers required the high computational complexity [11–12]. In the ing, acoustic noise reduction and target detection [21]. However,
early 20 0 0s, the passive time reversal technique using an array few studies have used vector quantities in underwater communi-
cation [22].
In this paper, a time reversal technique using particle ve-

Corresponding author. locity and pressure signal is applied to underwater acoustic
E-mail address: choijw@hanyang.ac.kr (J.W. Choi).

https://doi.org/10.1016/j.adhoc.2019.03.008
1570-8705/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/)
162 S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169

communication. The x, y, and z-components of the particle veloc- 2.2. Time reversal
ity and the acoustic pressure signal are used as multiple outputs
of the receiver. The efficiency of the proposed method is verified The time reversal technique improves the signal to noise ratio
using experimental communication data. However, only the verti- (SNR) and reduces the ISI by focusing the sound at a single posi-
cal component of the particle velocity was measured during the tion using the channel impulse response (CIR). The CIR is estimated
underwater communication experiment, owing to the limitation of using a probe signal without any information regarding the ocean
the equipment available in our system, and it was used to demod- environmental parameters. The probe signal and the subsequent
ulate the communication data with the acoustic pressure signal. communication stream are transmitted by a source and received
This paper is organized as follows. Section 2 provides a descrip- at multiple receivers after propagating through a multipath chan-
tion of the time reversal process using the vertical component of nel. The received probe signal at each receiver element is matched
particle velocity and pressure signal. Section 3 describes the un- filtered with a replica of the emitted probe signal to estimate
derwater acoustic communication experiment conducted in shal- the CIR, hˆ (t ) for each receiver position. Then, the time-reversed
low water. Section 4 presents results concerning the characteris- CIRs are convolved with the received communication streams and
tics of the communication channel and the communication perfor- summed to obtain the focused communication stream, y (t).
mance obtained using the method suggested in this paper. Finally, N  

a summary and discussion are given in Section 5. y (t ) = pi (t ) ∗ hˆ i (−t ) . (2)
i
2. Time reversal communication based on particle velocities
where pi (t) is the communication sequence received at the i-th re-
2.1. Particle velocity ceiver. N is the number of receivers. This procedure is shown in
Fig. 2. This process is also known as “passive phase conjugation”
The particle velocity v(t) can be calculated by integrating Eu- [13,24]. The performance of the time reversal process can be evalu-
ler’s equation with respect to time. If two pressure receivers are ated using a q (t)-function, which is calculated by matched filtering
assumed to be close together, v(t) is the time integral of the pres- hˆ i (t ) with hˆ i (−t ).
sure gradient, which can be approximated using finite difference N 
 
approximation [23]. The particle velocity is: q(t ) = hˆ i (t ) ∗ hˆ i (−t ) . (3)

1 t
p2 ( τ ) − p1 ( τ ) i
v(t ) = dτ . (1)
ρ0 0 d When the outputs of the matched filtering of hˆ i (t ) with hˆ i (−t )
where p1 and p2 are the pressure signals at two receivers placed are added, the main lobe of the q (t)-function is added coherently,
close together, d is the receiver separation distance, τ is the time whereas the side lobes destructively interfere to sum to zero. In
variable, and ρ 0 is the water density. The particle velocity in a par- the ideal case of an infinite bandwidth of CIR with no multipath
ticular direction is equivalent to the acoustic pressure divided by arrivals, the q (t)-function becomes a delta function with zero ISI
the acoustic impedance. The acoustic impedance is ρ 0 c, where c is [16–18].
a sound speed. As the particle velocity is a vector quantity, it has
three components in the x, y, and z directions. If the three velocity 3. Experiment and site
components and the acoustic pressure can be measured simulta-
neously at a given point using a vector sensor, these signals can be The experiments to evaluate the performance of underwater
used as a single-input multiple-output system (SIMO) for under- acoustic communication using the particle velocity and pressure
water acoustic communication. This procedure is shown in Fig. 1. signal were conducted on October 20, 2015, in a shallow water re-
gion located at 34◦ 48 N, 128◦ 47 E, south of Korea. The nomi-
nal water depth at the site is 59 m [Fig. 3(a)]. An omni-directional
transducer (NEPTUNE, D-11) was used as a source, which was de-
ployed at a depth of 31 m from the side of a small fishing vessel.
The communication signals were received by two receivers (RE-
SON, TC-4014), which were deployed at an average depth of 26 m.
The two receivers were placed 2.5 cm apart, equivalent to λ/5 for
a frequency of 11 kHz, which was a center frequency of the com-
munication signal used here. The horizontal ranges between the
source and the receiver were approximately 72 and 405 m at 0219
and 0522 UTC, respectively. The sound speed profiles were mea-
Fig. 1. Block diagram of the SIMO communication system using the three particle
sured using conductivity-temperature-depth (CTD) casts [Fig. 3(b)].
velocity components and the pressure signal. h(t) represents a channel impulse re- The sound speed structures were similar in both cases and the
sponse. sound speeds decreased slightly with depth. A surficial sediment

Fig. 2. Block diagram of the passive time reversal receiver. hˆ i (t ) is a time-reversed form of channel impulse response estimated at the ith receiver.
S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169 163

Fig. 3. (a) Experimental layout for the underwater acoustic communication measurements and (b) sound speed profiles measured by conductivity-temperature-depth (CTD)
casts.

Fig. 4. Signaling frame of the communication sequence consisting of a probe signal, a training sequence, and a communication data stream.

sample was taken using a grab sampler. The mean grain size was cused communication data y (t) and the q (t)-function were ob-
analyzed to be 7.3φ , where φ = − log2 (d/d0 ), d is the grain diam- tained via the time reversal process [26]. The y (t) was passed
eter in millimeters, and d0 is a reference length of 1 mm, corre- through a phase tracking process to compensate for the phase dis-
sponding to silty clay. The sound speed in the surficial sediment tortion in the communication data. Phase tracking was carried out
at the experimental site was estimated to be about 1483 m/s based using a decision-feedback phase-locked loop (DFPLL) based on the
on the empirical relationship between the mean grain size and the maximum likelihood (ML) estimate over an average of 20 sym-
sediment sound speed [25]. The wind speeds during the acoustic bols prior to the channel equalization [27]. The baseband wave-
measurements were relatively weak (<3 m/s) and therefore the sea form was recovered by multiplying by e−iωt , where ω is the an-
surface remained very calm. gular frequency. After that, the baseband data were low-pass fil-
The communication sequence consisted of a 20-ms-long, 9– tered and passed through an adaptive decision feedback equalizer
13 kHz linear frequency-modulated (LFM) pulse as a probe signal, (DFE) [27] to compensate for the channel distortion by the resid-
followed by a pause lasting 0.48 s, followed by binary phase shift ual ISI. Recursive Least-Squares (RLS) was used to adaptively up-
keying (BPSK) sequences lasting 3.5 s with a center frequency of date the filter weights of equalizer, in which a forgetting factor
11 kHz (Fig. 4). The BPSK signal was composed of 3500 symbols was 0.995. One hundred training symbols are also used for adjust-
including a sequence of 100 training symbols prior to the commu- ing the equalizer coefficient to their optimum value in RLS. The
nication stream. The pulse shaping filter was a root raised cosine number of feedforward and feedback filter taps in the equalizer is
(RRC) filter with a roll-off factor of β = 0.25. The communication related to the delay time spread due to the multipath channel [5].
data packets were transmitted every 10 s and repeated 10 times The tap numbers covering sufficiently the delay time spread were
for each case. used in the DFE process. The communication performance was fi-
nally evaluated with the BER and output SNR.
4. Communication results
4.1. Channel impulse responses of the particle velocity and pressure
As mentioned in Section 2, the particle velocity can be es-
fields
timated using the finite difference between the pressure signals
measured by two adjacent receivers. Since two receivers were ver-
Fig. 6(a) shows the CIRs measured by the two vertically adja-
tically placed in the experiment, the vertical component of the par-
cent receivers at a source-receiver range of 72 m. The CIRs were
ticle velocity was only estimated. The acoustic pressure signal at
estimated by matched filtering a replica of the transmitted LFM
the central point between the two receivers was estimated using
waveform with the measured probe signals. For the experimental
geometry of the site, the main arrival structure consisted of two
p1 (t ) + p2 (t )
pM (t ) = . (4) energetic arrivals−direct (D) and sea surface (S) paths. Although
2
the bottom path was received just after the S arrival, the amplitude
Fig. 5 shows a block diagram of the SIMO system used to de- of the bottom reflecting path was lower than that of the D and S
code the communication data using the vertical component of paths due to the large bottom loss. The silty clay on the seabed
the particle velocity vz (t) and the pressure signal pM (t). The fo- of the experimental site is characterized by an acoustically soft
164 S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169

Fig. 5. Block diagram of time reversal using the vertical particle velocity and the pressure signal. The receiver consists of various components to perform finite difference
approximation, time reversal, and adaptive equalization.

Fig. 7. Channel impulse responses estimated using the vertical component of the
particle velocity and the pressure at the center of two receivers at a source-receiver
range of 405 m.

arrival group in terms of the pressure signal. This is because the D


path propagates almost horizontally, and it is a main cause of the
lower correlation value between the vertical particle velocity and
the pressure signal.
Fig. 7 shows the CIRs for the source-receiver range of 405 m.
The correlation coefficient between the vertical component of the
particle velocity and the pressure was estimated to be 0.75. As the
Fig. 6. Channel impulse responses for a source-receiver range of 72 m (a) estimated
by two vertically adjacent receivers and (b) estimated using the vertical component range increases, the horizontal components of the S and B paths
of the particle velocity and the pressure at the center of two receivers. increase. This causes the amplitude of the vertical component of
the particle velocity to be considerably smaller than that of the
pressure signal. However, the amplitude of the D path is still signif-
bottom, which causes a large bottom loss. Therefore, the acoustic icantly larger than those of the second arrival paths, and the cor-
interaction at the bottom limited the number of dominant paths relation coefficient (0.75) is higher than that for the range of 72 m,
at this range. The CIRs at the two receivers were highly correlated, which implies that the diversity between the channels decreases.
with a correlation coefficient of 0.92. The low correlation between the vertical particle velocity and
Fig. 6(b) shows a comparison between the CIR of the vertical the pressure signal provides a large diversity gain, which reduces
component of the particle velocity and the CIR of the pressure at the inter-symbol interference by minimizing the sidelobe levels of
the central point of two receivers obtained using Eq. (4). It has the q (t)-function [16,26]. Therefore, the q (t)-function can be used
been reported that the time reversal performance degrades as the to evaluate the performance of the time reversal communication
signals become more correlated [16,22]. Although the particle ve- quantitatively. Fig. 8(a) and (b) show the normalized q (t)-functions
locity and the pressure were measured at the same position, the of two channels at source-receiver ranges of 72 and 405 m, respec-
correlation was lower than that shown in Fig. 6(a), with a correla- tively. These were obtained using the vertical component of the
tion coefficient of 0.56. On the D path, the amplitude of the verti- particle velocity and the pressure signal. The q (t)-function at a
cal component of the particle velocity was smaller than that of the range of 72 m is more similar to the delta-function form than the
second arrival group, which consisted of the S and B paths, while q (t)-function at a range of 405 m. This implies that the communi-
the amplitude of the D path was higher than that of the second cation performance will be better at shorter ranges.
S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169 165

Fig. 8. Normalized q (t)-functions for two channels obtained using the vertical particle velocity component and the pressure signal at source-receiver ranges of (a) 72 and
(b) 405 m.

4.2. Communication performance analysis mance was evaluated using the TR alone, and then it was repeated
using TR combined with the PLL (hereafter referred to as TR+PLL),
Figs. 9 and 10 are scatter plots of the communication perfor- and finally, the DFE was combined with the TR+PLL (referred to
mance obtained using the vertical particle velocity and pressure as TR+PLL+DEF). The results are shown in (a)–(c) respectively. The
signal at the ranges of 72 and 405 m, respectively. First, the perfor- values of the BER and output SNR are presented above the scatter

Fig. 9. Scatter plots of time reversal communication performance obtained via (a) TR, (b) TR + PLL, and (c) TR + PLL + DFE receivers using the vertical particle velocity and
pressure signal at a range of 72 m.

Fig. 10. Scatter plots of time reversal communication performance obtained via (a) TR, (b) TR + PLL, and (c) TR + PLL + DFE receivers using the vertical particle velocity and
pressure signal at a range of 405 m.
166 S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169

plots. These are communication performance parameters showing


the number of bit errors and the accuracy of the demodulated sig-
nal [28].
At a range of 72 m, the TR alone had the worst communication
performance, with a BER of 34.1% (11,940/35,0 0 0 symbols) and an
output SNR of −1.2 dB. The demodulated symbols exhibited phase
rotation due to the Doppler effect, induced by movements of the
source and receiver [Fig. 9(a)]. This phase distortion was well com-
pensated for by the PLL, resulting in a low BER (0.04%, 0.03%) and
high output SNR (8.7, 9.1 dB), as shown in Fig. 9(b) and (c). There
was little difference between the performance results obtained by
TR+PLL and TR+PLL+DFE. This implies that TR+PLL using the ver-
tical component of the particle velocity and the pressure signal
were sufficient to mitigate the effects of ISI.
Figs. 10 shows the communication performance results at a
Fig. 11. Scatter plots of the time reversal communication performance obtained via
range of 405 m. The communication using only TR failed, with a TR + PLL + DFE receivers using the pressure signals measured by the two receivers
BER of 47.5% and an output SNR of −3.2 dB. When the TR+PLL was were placed 2.5 cm apart at the ranges of (a) 72 m and (b) 405 m.
used, the performance improved but was still not satisfactory, with
a BER of 4.4%. This is associated with the high correlation between
the vertical component of the particle velocity and the pressure tern is the oscilloscope display to evaluate the effect of channel
signal and the relatively strong side lobe of the q (t)-function, as noise and ISI on the performance of baseband communication sys-
discussed in Section 4.1. When the DFE was used after TR+PLL, the tem [29]. For example, if the eye pattern is closed, it means the
performance was more improved, resulting in a BER of 0.3% and an communication signal is distorted by the ISI or noise. The eye pat-
output SNR of 7.6 dB. terns for the results obtained using the vertical particle velocity
To compare these results to those obtained using the pressure and pressure signal are more open compared to those obtained us-
signal only, the time reversal process was repeated using pressure ing two pressure signals for both ranges. The results imply that the
signals received at the two vertically adjacent receivers at source- communication method using the particle velocity and pressure is
receiver ranges of 72 m and 405 m, and their results are shown in more robust to the ISI.
Fig. 11(a) and (b), respectively. At a range of 72 m, the BER and out-
put SNR obtained using TR+PLL+DFE were 0.8% and 6.7 dB, respec- 5. Summary and discussion
tively. In other words, the communication receiver using the ver-
tical particle velocity and the pressure signal resulted in a 2.4 dB In this paper, the results of underwater acoustic communication
improvement in the output SNR, compared to that using the pres- using acoustic vector quantity were presented. The pressure signal
sure signals. However, there was only a 1.2 dB improvement in the and the vertical component of the particle velocity were used in
output SNR at a range of 405 m. This is because, at this range there the SIMO time-reversal communication system. Although time re-
was a high level of correlation between the vertical particle veloc- versal communication reduces the ISI and increases the SNR signif-
ity and the pressure signal. icantly, the communication performance is inversely proportional
Fig. 12(a) and (b) show the eye patterns for a range of 72 m for to the spatial efficiency. Thus, it is difficult to use the time-reversal
the receiver using the vertical particle velocity and pressure signal communications system in small underwater platforms. As vector
and the receiver using two pressure signals, respectively. The eye quantities such as the acoustic particle velocity and acceleration
pattern for a range of 405 m are shown in Fig. 13. The eye pat- have a direction as well as a magnitude, the x, y, and z components

Fig. 12. Eye patterns for (a) the receiver using the vertical particle velocity and pressure signal and (b) the receiver using two pressure signals at a range of 72 m.
S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169 167

Fig. 13. Eye patterns for (a) the receiver using the vertical particle velocity and pressure signal and (b) the receiver using two pressure signals at a range of 405 m.

can be measured at a single point using a vector sensor. Then, the composed of silty clay with a mean grain size of 7.3φ . The com-
three particle velocity components and the pressure can be con- munication measurements were made at source-receiver ranges
sidered as four communication channels. Unfortunately, a vector of 72 m and 405 m. At a range of 72 m, the TR+PLL had satis-
sensor was not operated in the communication experiment pre- factory performance with a BER of 0.04%. However, at a range
sented in this paper. Instead, two vertically adjacent receivers were of 405 m, the BER was 4.4%. After passing the received signal
used. Therefore, only the vertical component of the particle veloc- through the DFE combined with the TR+PLL, the BER was re-
ity could be estimated by the finite difference approximation of duced to 0.3%. The communication performances obtained using
the pressure gradient between the pressure signals measured by our system were compared to those obtained from a system that
the two receivers. The approximated vertical particle velocity was uses only the pressure signals. The result showed that there were
then used as multiple outputs with the pressure signal in the SIMO 2.4 dB and 1.2 dB improvements in the output SNR at ranges of 72
system. and 405 m, respectively. The lower performance gain in the case
The underwater communication experiment was conducted in of the larger source-receiver range seems to have been caused by
shallow water off the south coast of Korea in October of 2015. the higher correlation between the pressure signal and the particle
The seafloor at the experimental site was estimated to be flat and velocity.

Fig. 14. BER performance of time reversal communication as a function of input SNR at source-receiver ranges of (a) 72 m and (b) 405 m. Blue squares and red circles indicate
the mean BER for 50 repetitions for two pressure signals and for vertical particle velocity and pressure signal, respectively. Error bars indicate the standard deviations of the
mean.
168 S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169

In conclusion, time reversal receiver using the vertical particle [4] A. Song, M. Badiey, H.C Song, W.S. Hodgkiss, M.B. Porter, Impact of ocean vari-
velocity and the pressure signal provided a better communication ability on coherent underwater acoustic communications during the Kauai ex-
periment (KauaiEx 2003), J. Acoust. Soc. Am. 123 (2) (2008) 856–865.
performance than the time reversal receiver using only the pres- [5] S.U. Son, H. Kim, J. Joo, J.W. Choi, Multipath effects on high-frequency coher-
sure signals. The communication performance at the shorter range ent acoustic communications in shallow water, Jpn. J. Appl. Phys. (2013) 52
(72 m) was much better than that at the longer range (405 m), 07HG03.
[6] T.C. Yang, Properties of underwater acoustic communication channels in shal-
showing almost error-free communication. low water, J. Acoust. Soc. Am. 131 (1) (2012) 129–145.
To further investigate the performance of time reversal sys- [7] P.A. Walree, Propagation and scattering effects in underwater acoustic commu-
tem using particle velocity and pressure signal, the communi- nication channels, IEEE J. Ocean. Eng. 38 (4) (2013) 614–631.
[8] H. Yu, A. Song, M. Badiey, F. Chen, F. Ji, Iterative estimation of doubly selective
cation performance as a function of input SNR was simulated
underwater acoustic channel using basis expansion models, Ad Hoc Netw 34
for the cases considered in this paper. The simulation process is (2015) 52–61.
as follows; First, the channel impulse responses for the source- [9] D. Hirsch, B. Wolf, A simple adaptive equalizer for efficient data transmission,
IEEE J. Trans. Commun. 18 (1970) 5–12.
receiver ranges of 72 and 405 m, which are shown in Figs. 6 and
[10] J.G. Proakis, J.H. Miller, An adaptive receiver for digital signaling though chan-
7, were convolved with the communication sequence shown in nel with intersymbol interference, IEEE J. Tran. Inf. Theory 15 (1969) 484–497.
Fig. 4, and the additive white gaussian noise (AWGN) was added [11] M. Stojanovic, J.A. Catipovic, J.G. Proakis, Adaptive multichannel combining and
with SNR in the range of -2 − 12 dB at an interval of 1 dB. The equalization for underwater acoustic communication, J. Acoust. Soc. Am. 94
(1993) 1621–1631.
50 AWGNs were randomly generated for each SNR step and [12] P. Balaban, J. Salz, Optimum diversity combining and equalization in digital
the signals were passed through the passive time reversal pro- data transmission with applications to cellular mobile radio, IEEE J. Trans.
cess to decode the communication data. The performance simu- Commun. 40 (1992) 885–895.
[13] H.C. Song, W.S. Hodgkiss, W.A. Kuperman, M. Stevenson, T. Akal, Improve-
lation results are shown in Fig. 14. Overall the BER performance ment of time-reversal communications using adaptive channel equalizers, IEEE
of the time reversal receiver using the vertical particle veloc- J. Ocean. Eng. 31 (2) (2006) 487–496.
ity and the pressure signal is better than that of the time re- [14] H.C. Song, W.S. Hodgkiss, W.A. Kuperman, W.J. Higley, K. Raghukumar, T. Akal,
Spatial diversity in passive time reversal communications, J. Acoust. Soc. Am.
versal receiver using only the pressure signals. In addition, the 120 (2006) 2067–2076.
BER performance improves as input SNR increases. However, the [15] G. Zhang, H. Dong, Joint passive-phase conjugation with adaptive multichannel
BER for the range of 72 m improves more rapidly with increas- combining for coherent underwater acoustic communications, Appl. Acoust. 73
(2012) 433–439.
ing input SNR, compared to those for the range of 405 m. This
[16] T.C. Yang, Temporal resolutions of time-reversal and passive-phase conjuga-
is because there was less correlation between the vertical parti- tion for underwater acoustic communications, IEEE J. Ocean. Eng. 28 (2) (2003)
cle velocity and the pressure signals at shorter range, as discussed 229–245.
[17] G. Zhang, J.M. Hovem, H.E. Dong, Experimental assessment of different receiver
earlier.
structures for underwater acoustic communications over multipath channels,
Bottom loss depends on the sediment composition. Hence, the Sensors 12 (2012) 2118–2135.
vertical and horizontal components of multipath signals, includ- [18] L. Sun, B. Chen, H. Li, T. Zhou, R Li, Time reversal acoustic communication using
ing paths that have been reflected from the seabed, vary with the filted multitoned modulation, Sensors 15 (2015) 23554–23571.
[19] D.R. Dall’Osto, J.W. Choi, P.H. Dahl, Measurement of acoustic particle motion in
sediment composition and the grazing angle at the bottom inter- shallow water and its application to geoacoustic inversion, J. Acoust. Soc. Am.
face. This may lead to differences in the correlation coefficients be- 139 (1) (2016) 311–319.
tween the pressure and the particle velocity as the source-receiver [20] D.R. Dall’Osto, P.H. Dahl, J.W. Choi, Properties of the acoustic intensity vec-
tor field in a shallow water waveguide, J. Acoust. Soc. Am. 131 (3) (2012)
range varies. Although only the vertical component of the parti- 2023–2035.
cle velocity measured at a single point was used in this paper, the [21] G.L. D’Spain, J.C. Luby, G.R. Wilson, R.A Gramann, Vector sensors and vector
results imply that the communication based on vector quantities sensor line arrays: comments on optimal array gain and detection, J. Acoust.
Soc. Am. 120 (1) (2006) 171–185.
performs better in underwater acoustic communication environ- [22] A. Abdi, H. Guo, P. Sutthiwan, A new vector sensor receiver for underwater
ments. In order to apply this protocol in practical scenarios, further acoustic communication, J. Trans. Wireless. Commun. 8 (2009) 3326.
studies to quantify the efficiency of acoustic vector communication [23] F.J. Fahy, in: Sound Intensity, E & FN Spon, London, 1995, pp. 1053–1075.
[24] D. Rouseff, D.R. Jackson, W.L.J. Fox, C.D. Jones, J.A. Ritcey, D.R. Dowling, Un-
using vector sensors and/or vector sensor arrays will be carried
derwater acoustic communication by passive-phase conjugation: theory and
out. experimental results, IEEE J. Ocean. Eng. 26 (4) (2001) 821–831.
[25] D.R. Jackson, M.D. Richardson, in: High-Frequency Seafloor Acoustics, Springer,
New York, NY, USA, 2006, pp. 313–314.
Acknowledgments
[26] S. Kim, S.U. Son, H. Kim, K.H. Choi, J.W. Choi, Estimate of passive time reversal
communication performance in shallow water, Appl. Sci. 8 (1) (2018) 23.
This work was supported by the Agency for Defense Develop- [27] J.G. Proakis, in: Digital Communications, McGraw Hill, NewYork, NY, USA,
ment, Korea, under Grant UD170022DD, the Ministry of Oceans 2008, pp. 298–315.
[28] M. Stojanovic, J.A. Catipovic, J.G. Proakis, Phase-coherent digital communi-
and Fisheries (in a project titled the development of distributed cations for underwater acoustic channels, IEEE J. Ocean. Eng. 19 (1) (1994)
underwater monitoring & control networks) and a “Development 100–111.
of maritime defense and security technology” grant (PE99741), pro- [29] B. Sklar, in: Digital Communication, Prentice Hall, Los Angeles, USA, 2006,
pp. 151–152.
moted by the Korea Institute of Ocean Science and Technology
(KIOST). Sunhyo Kim received the Ph.D. degrees, from the
Hanyang University (Korea), Department of Environmen-
References tal Marine Sciences in 2016. He is currently a postdoctoral
scientist in the Department of Maritime Security Research
Center, Korea Institute of Ocean Science and Technology
[1] D.B. Kilfoyle, A.B. Baggeroer, The state of the art in underwater acoustic (KIOST), Korea, where he has been since 2019. His re-
telemetry, IEEE J. Ocean. Eng. 25 (20 0 0) 4–27. search interests include underwater acoustic communica-
[2] P.H. Dahl, J.W. Choi, The East China Sea as an Underwater Acoustic Communi- tion, underwater acoustic sensor network, Naval oceanog-
cation Channel: measurements of the Channel Impulse Response(U), U.S. Navy raphy, and underwater channel modeling.
J. Underw. Acoust. 56 (2006) 1–12.
[3] D. Rouseff, M. Badiey, A. Song, Effect of reflected and refracted signals on
coherent underwater acoustic communication: results from the Kauai exper-
iment (KauaiEx 2003), J. Acoust. Soc. Am. 126 (5) (2009) 2359–2366.
S. Kim, H. Kim and S.-k. Jung et al. / Ad Hoc Networks 89 (2019) 161–169 169

Hyeonsu Kim received the Ph.D. degrees, from the Inha Jee Woong Choi received the Ph.D. degrees, from the
University (Korea), Department of Electronic engineering Hanyang University (Korea), Department of Earth and Ma-
in 2017. He is currently a postdoctoral researcher in the rine Sciences in 20 02. From 20 02 to 2007, he had been
Department of Marine Science and Convergence Engi- a Research Associate at the Applied Physics Laboratory,
neering, Hanyang University-ERICA, Korea, where he has University of Washington, Seattle (USA). He is currently
been since 2017. His research interests include underwa- a Professor in the Department of Marine Science and
ter acoustic communication, underwater acoustic sensor Convergence Engineering, Hanyang University-ERICA, Ko-
network, and underwater channel modeling. rea, where he has been since 2007. In recent years, his
main areas of research have included acoustic propaga-
tion in shallow water, acoustic interaction with ocean
boundaries, geoacoustic inversion, and underwater acous-
tic communication. Dr. Choi is a member of the Acoustical
Society of America and the Acoustical Society of Korea. He
is currently Secretary General of the Western Pacific Commission for Acoustics.
Seom-kyu Jung received the Ph.D. degrees, from the Inha
University (Korea) in 2002. He is currently a Principal
research scientist in the Department of Maritime Secu-
rity Research Center, Korea Institute of Ocean Science and
Technology (KIOST), Korea. In recent years, his main areas
of research have included Shallow seismic data process-
ing and stratigraphy, Side scan sonar data processing and
interpretation, 3-D multi-beam bathymetric data process-
ing and analysis, Naval oceanography, and Geoacoustical
characteristics of ocean bottom sediments.

You might also like