Underwater Wireless Sensor Networks 2015

International Journal of Distributed Sensor Networks
Underwater Wireless Sensor

Networks 2015
Guest Editors: Dongkyun Kim, Juan C. Cano, Wei Wang, Floriano De Rango,
and Kun Hua
Underwater Wireless Sensor Networks 2015
Guest Editors: Dongkyun Kim, Juan C. Cano, Wei Wang,

Floriano De Rango, and Kun Hua
Copyright © 2015 Hindawi Publishing Corporation. All rights reserved.
This is a special issue published in “International Journal of Distributed Sensor Networks.” All articles are open access articles distributed
under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, pro-
vided the original work is properly cited.
Editorial Board
Jemal H. Abawajy, Australia Yao-Jen Chang, Taiwan Andrei Gurtov, Finland
Miguel Acevedo, USA Naveen Chilamkurti, Australia Mohamed A. Haleem, USA
Cristina Alcaraz, Spain Wook Choi, Republic of Korea Kijun Han, Republic of Korea
Ana Alejos, Spain H. Choo, Republic of Korea Qi Han, USA
Mohammod Ali, USA Kim-Kwang R. Choo, Australia Zdenek Hanzalek, Czech Republic
Giuseppe Amato, Italy Chengfu Chou, Taiwan Shinsuke Hara, Japan
Habib M. Ammari, USA Mashrur A. Chowdhury, USA Wenbo He, Canada
Michele Amoretti, Italy Tae-Sun Chung, Republic of Korea Paul Honeine, France
Christos Anagnostopoulos, UK Marcello Cinque, Italy Feng Hong, China
Li-Minn Ang, Australia Sesh Commuri, USA Chin-Tser Huang, USA
Nabil Aouf, UK Mauro Conti, Italy Haiping Huang, China
Francesco Archetti, Italy Alfredo Cuzzocrea, Italy Xinming Huang, USA
Masoud Ardakani, Canada Donatella Darsena, Italy Jose I. Moreno, Spain
Miguel Ardid, Spain Dinesh Datla, USA Mohamed Ibnkahla, Canada
Muhammad Asim, UK Amitava Datta, Australia Syed K. Islam, USA
Stefano Avallone, Italy Iyad Dayoub, France Lillykutty Jacob, India
Jose L. Ayala, Spain Danilo De Donno, Italy Won-Suk Jang, Republic of Korea
Javier Bajo, Spain Luca De Nardis, Italy Antonio J. Jara, Switzerland
N. Balakrishnan, India Floriano De Rango, Italy Shengming Jiang, China
Prabir Barooah, USA Paula de Toledo, Spain Yingtao Jiang, USA
Federico Barrero, Spain Marco Di Felice, Italy Ning Jin, China
Paolo Barsocchi, Italy Salvatore Distefano, Italy Raja Jurdak, Australia
Paolo Bellavista, Italy Longjun Dong, China Konstantinos Kalpakis, USA
Olivier Berder, France Nicola Dragoni, Denmark Ibrahim Kamel, UAE
Roc Berenguer, Spain George P. Efthymoglou, Greece Joarder Kamruzzaman, Australia
Juan A. Besada, Spain Frank Ehlers, Italy Rajgopal Kannan, USA
Gennaro Boggia, Italy Melike Erol-Kantarci, Canada Johannes M. Karlsson, Sweden
Alessandro Bogliolo, Italy Farid Farahmand, USA Gour C. Karmakar, Australia
Eleonora Borgia, Italy Michael Farmer, USA Marcos D. Katz, Finland
Janos Botzheim, Japan Florentino Fdez-Riverola, Spain Jamil Y. Khan, Australia
Farid Boussaid, Australia Gianluigi Ferrari, Italy Sherif Khattab, Egypt
Arnold K. Bregt, Netherlands Silvia Ferrari, USA Hyungshin Kim, Republic of Korea
Richard R. Brooks, USA Giancarlo Fortino, Italy Sungsuk Kim, Republic of Korea
Ted Brown, USA Luca Foschini, Italy Andreas König, Germany
Davide Brunelli, Italy Jean Y. Fourniols, France Gurhan Kucuk, Turkey
James Brusey, UK David Galindo, Spain Sandeep S. Kumar, Netherlands
Carlos T. Calafate, Spain Ennio Gambi, Italy Juan A. L. Riquelme, Spain
Tiziana Calamoneri, Italy Weihua Gao, USA Yee Wei Law, Australia
José Camacho, Spain A.-J. García-Sánchez, Spain Antonio Lazaro, Spain
Juan C. Cano, Spain Preetam Ghosh, USA Didier Le Ruyet, France
Xianghui Cao, USA Athanasios Gkelias, UK Joo-Ho Lee, Japan
João Paulo Carmo, Brazil Iqbal Gondal, Australia Seokcheon Lee, USA
Roberto Casas, Spain Francesco Grimaccia, Italy Yong Lee, USA
Luca Catarinucci, Italy Jayavardhana Gubbi, Australia Stefano Lenzi, Italy
Michelangelo Ceci, Italy Song Guo, Japan Pierre Leone, Switzerland
Shancang Li, UK Michael J. O’Grady, Ireland Olli Silven, Finland
Shuai Li, USA Gregory O’Hare, Ireland Hichem Snoussi, France
Qilian Liang, USA Giacomo Oliveri, Italy Guangming Song, China
Weifa Liang, Australia Saeed Olyaee, Iran Antonino Staiano, Italy
Yao Liang, USA Luis Orozco-Barbosa, Spain Muhammad A. Tahir, Pakistan
I-En Liao, Taiwan Suat Ozdemir, Turkey Jindong Tan, USA
Jiun-Jian Liaw, Taiwan Vincenzo Paciello, Italy Shaojie Tang, USA
Alvin S. Lim, USA Sangheon Pack, Republic of Korea Luciano Tarricone, Italy
Antonio Liotta, Netherlands Marimuthu Palaniswami, Australia Kerry Taylor, Australia
Donggang Liu, USA Meng-Shiuan Pan, Taiwan Sameer S. Tilak, USA
Hai Liu, Hong Kong Seung-Jong Park, USA Chuan-Kang Ting, Taiwan
Yonghe Liu, USA Miguel A. Patricio, Spain Sergio Toral, Spain
Leonardo Lizzi, France Luigi Patrono, Italy Vicente Traver, Spain
Jaime Lloret, Spain Rosa A. Perez-Herrera, Spain Ioan Tudosa, Italy
Kenneth J. Loh, USA Pedro Peris-Lopez, Spain Anthony Tzes, Greece
Juan Carlos López, Spain Janez Perš, Slovenia Bernard Uguen, France
Manel López, Spain Dirk Pesch, Ireland Francisco Vasques, Portugal
Pascal Lorenz, France Shashi Phoha, USA Khan A. Wahid, Canada
Jun Luo, Singapore Robert Plana, France Agustinus B. Waluyo, Australia
Michele Magno, Italy Carlos Pomalaza-Ráez, Finland Honggang Wang, USA
Sabato Manfredi, Italy Neeli R. Prasad, Denmark Jianxin Wang, China
Athanassios Manikas, UK Antonio Puliafito, Italy Ju Wang, USA
Pietro Manzoni, Spain Hairong Qi, USA Yu Wang, USA
Álvaro Marco, Spain Meikang Qiu, USA Thomas Wettergren, USA
Jose R. Martinez-de Dios, Spain Veselin Rakocevic, UK Ran Wolff, Israel
Ahmed Mehaoua, France Nageswara S.V. Rao, USA Chase Wu, USA
Nirvana Meratnia, Netherlands Luca Reggiani, Italy Na Xia, China
Christian Micheloni, Italy Eric Renault, France Qin Xin, Faroe Islands
Lyudmila Mihaylova, UK Joel Rodrigues, Portugal Chun J. Xue, Hong Kong
Paul Mitchell, UK Pedro P. Rodrigues, Portugal Yuan Xue, USA
Mihael Mohorcic, Slovenia Luis Ruiz-Garcia, Spain Geng Yang, China
José Molina, Spain Mohamed Saad, UAE Theodore Zahariadis, Greece
Antonella Molinaro, Italy Stefano Savazzi, Italy Miguel A. Zamora, Spain
Salvatore Morgera, USA Marco Scarpa, Italy Hongke Zhang, China
Kazuo Mori, Japan Arunabha Sen, USA Xing Zhang, China
Leonardo Mostarda, Italy Olivier Sentieys, France Jiliang Zhou, China
V. Muthukkumarasamy, Australia Salvatore Serrano, Italy Ting L. Zhu, USA
Kamesh Namuduri, USA Zhong Shen, China Xiaojun Zhu, China
Amiya Nayak, Canada Chin-Shiuh Shieh, Taiwan Yifeng Zhu, USA
George Nikolakopoulos, Sweden Minho Shin, Republic of Korea Daniele Zonta, Italy
Alessandro Nordio, Italy Pietro Siciliano, Italy
Underwater Wireless Sensor Networks 2015, Dongkyun Kim, Juan C. Cano, Wei Wang,
Floriano De Rango, and Kun Hua
Volume 2015, Article ID 623042, 2 pages
Design and Detection of Multilinear Chirp Signals for Underwater Acoustic Sensor Networks,
En Cheng, Shengli Chen, and Fei Yuan
Doppler Estimation Based on Frequency Average and Remodulation for Underwater Acoustic
Communication, Chan-Ho Hwang, Ki-Man Kim, Seung-Yong Chun, and Sang-Kook Lee
High Throughput Receiver Structure for Underwater Communication, Chang-Uk Baek and Ji-Won Jung
Underwater Sensor Network Applications: A Comprehensive Survey, Emad Felemban,

Faisal Karim Shaikh, Umair Mujtaba Qureshi, Adil A. Sheikh, and Saad Bin Qaisar
Proteus II: Design and Evaluation of an Integrated Power-Efficient Underwater Sensor Node,
Wouter A. P. van Kleunen, Niels A. Moseley, Paul J. M. Havinga, and Nirvana Meratnia
Hindawi Publishing Corporation
http://dx.doi.org/10.1155/2015/623042
Editorial
Dongkyun Kim,1 Juan C. Cano,2 Wei Wang,3 Floriano De Rango,4 and Kun Hua5
1
School of Computer Science & Engineering, Kyungpook National University, Daegu 702-701, Republic of Korea
2
Universidad Politécnica de València, Camino de Vera, S/N, 46022 València, Spain
3
Department of Computer Science, San Diego State University, Campanile Drive, San Diego, CA 92115, USA
4
Department of Informatics, Electronic and Sistemistic, University of Calabria, 87036 Cosenza, Italy
5
Department of Electrical and Computer Engineering, Lawrence Technological University, Southfield, MI 48075, USA
Correspondence should be addressed to Dongkyun Kim; dongkyun@knu.ac.kr
Received 23 November 2015; Accepted 23 November 2015
Copyright © 2015 Dongkyun Kim et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Recently, the research in Underwater Sensor Networks constant operations running on a small rechargeable battery
(UNWSNs) has gained a noticeable pace due to their vast over the time period of a month. The designed node uses a
variety of applications. Moreover, the 75% of earth’s surface host CPU for the network protocols and processing sensor
is covered with water in the form of rivers, canals, seas, and data and a separate CPU performs signal processing for
oceans. To explore and make use of this vast unexplored the ultrasonic acoustic software-defined Modulator Demod-
aquatic environment, new technologies provide us with a ulator (MODEM). Furthermore, a Frequency Shift Keying
number of applications such as pollution monitoring, disaster (FSK) modulation scheme supporting the configurable rates
prevention, and tactical surveillance for scientific, environ- of symbols, the Hamming error-correction, and estimation
mental, and military purposes. For this purpose, we came of Time-of-Arrival (ToA) enabling underwater positioning is
across enormous research and development efforts being presented in this paper. From the application perspective, the
made in UNWSNs that are a fusion of wireless technology on-board sensors, an accelerometer as well as a temperature
with extremely small micromechanical sensors having smart sensor, can be utilized to measure or sense the various envi-
sensing, intelligent computing, and communication capabil- ronmental conditions. For the evaluation purpose, several
ities. UNWSN is a network of autonomous sensor nodes, test results were obtained to validate the low-power opera-
which are spatially distributed deep underwater to sense the tions. Experimental results show that the node achieves the
water-related properties such as quality, temperature, and one-month lifetime and is able to perform communication in
pressure using acoustic channels. Therefore, applying terres- high-reflective environments and performs estimation with
trial WSN protocols into UNWSN is not straight forward and an accuracy of about 1-2 meters.
therefore researchers and industrials are still looking for more E. Felemban et al. presented a detailed survey on various
suitable and serious efforts to bring UNWSN into a complete applications supported by the underwater sensor communi-
reality. cations in the paper entitled “Underwater Sensor Network
This special issue presents recent schemes for underwater Applications: A Comprehensive Survey.” The authors have
networks, especially featuring current and future research classified underwater communication applications into five
waypoints in the field of underwater wireless sensor net- categories. These include applications for monitoring, dis-
works. aster prevention, military objectives, navigation, and sup-
In the paper titled “Proteus II: Design and Evaluation of porting sports events offshore and onshore. Furthermore,
an Integrated Power-Efficient Underwater Sensor Node,” the the authors subclassified these classes. Currently, the pace
authors have designed and evaluated an integrated low-cost of research in the area of Underwater Sensor Networks
underwater sensor node with the capacity of reconfigurabil- (UNWSNs) is slow due to the difficulties in applying most
ity. The proposed type of underwater sensor nodes allows of the terrestrial WSNs state-of-the-art to the underwater
2 International Journal of Distributed Sensor Networks
environment. Most of the underwater deployments depend Networks” got motivations by finding reduced power con-
on acoustic signals for enabling communication combined sumption and improved performance of multiple access in
with special sensors having the capacity to take on various UNWSNs and proposed a Multilinear Chirp-Code Division
environmental changes of the oceans. However, sensing and Multiple Access (MLC-CDMA) scheme. To be specific, the
subsequent transmission tends to vary as per different subsea differences between single slope chirp signal and multilinear
environments; for example, deep sea exploration requires chirp signal are analyzed in the paper. For example, a new
all together a different approach for communication as detection technique called mixing-change rate along with
compared to shallow water communication. In this survey, fractional Fourier transform (MCR-FrFT) is proposed at the
the authors provide most recent developments in UNWSN receiving end, to detect the multirate chirp (MRC) signal
applications and their deployments for monitoring and con- and reduce complexity of the node. Moreover, the authors
trol of underwater conditions. Finally, the challenges and have presented two steps to detect MCR-FrFT. Extensive
opportunities faced by recent deployments of UNWSN are simulation results are presented at the end of the paper. By
provided for researchers and other relevant personnel to using the MCR-FrFT, the computation of detection can be
follow the research roadmap. decreased to 50% compared with direct FrFT. Moreover, by
The authors in “High Throughput Receiver Structure for using the MCR-FrFT technique, the different users’ signal can
Underwater Communication” highlighted the effect of long be rapidly detected and separated.
multipath spreads in underwater acoustic channels. Those
extended paths may cause intersymbol interference as well Acknowledgments
as resulting in Doppler shift due to the relative source-
receiver motion. Therefore, in order to improve the system The guest editors would like to thank the authors for the
performance and throughput in the presence of these major great level of the contributions included in this special issue
hindrances, this paper proposes consecutive and iterative and the work developed by many experts who participated in
BCJR equalization for a long packet size. For this purpose, the review process providing constructive comments to the
the long packet has been divided into small consecutive authors to improve the quality of the papers.
packets, and previous packets are used to estimate channel
information required for compensation of the subsequent Dongkyun Kim
packets. Simulation results show that, in comparison with Juan C. Cano
various decoding schemes, the proposed work gets less error Wei Wang
rates. Moreover, the simulation scenario is based on real Floriano De Rango
experimental data measured on a lake in Mungyeong city. Kun Hua
C.-H. Hwang et al. in their paper titled “Doppler Esti-
mation Based on Frequency Average and Remodulation for
Underwater Acoustic Communication” proposed the more
reliable Doppler estimation method based on the existing
technique. Furthermore, this paper also proposed a remod-
ulation method that is able to estimate the time-varying
channel parameter during the payload. The authors further
compared the performance of the proposed remodulation
method with the conventional method in the simulation
environment with white Gaussian noise and the Doppler
effect. It has been reported that the uncoded BER of the pro-
posed method was lower than that of the conventional packet
method. The communication performance got improved
when the recursive frame length was shorter. The results of
the sea experiment illustrated the prospective for an improve-
ment in the performance, when the communication system
used the proposed remodulation method. When recursive
frame length is 250 symbols, communication performance
with the previous frequency estimation method depreciated,
whereas the proposed frequency average method presented
stable performance. In particular, it is evident from the results
that the communication performance is improved by up to
significant levels using the two proposed methods.
As we have seen so far, in Underwater Sensor Networks
(UNWSNs), few key problems have attracted more and more
attention, including, but not limited to, the power consump-
tion, performance of multiple access, and complexity of the
node. The authors in a paper titled “Design and Detection
of Multilinear Chirp Signals for Underwater Acoustic Sensor
http://dx.doi.org/10.1155/2015/371579
Research Article
Design and Detection of Multilinear Chirp Signals for
Underwater Acoustic Sensor Networks
En Cheng, Shengli Chen, and Fei Yuan

Key Laboratory of Underwater Acoustic Communication and Marine Information Technology, Ministry of Education,
Communication Engineering Department, Xiamen University, Xiamen 361005, China
Correspondence should be addressed to Fei Yuan; yuanfei@xmu.edu.cn
Received 2 February 2015; Revised 30 April 2015; Accepted 4 May 2015
Academic Editor: Juan Carlos Cano
Copyright © 2015 En Cheng et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In Underwater Acoustic Sensor Networks (UW-ASNs), some key problems have attracted more and more attention, including
power consumption, performance of multiple access, and complexity of node. Motivated by finding reduced power consumption
and improved performance of multiple access in UW-ASNs, a multilinear chirp-Code Division Multiple Access (MLC-CDMA)
scheme is proposed. The differences between single slope chirp signal and multilinear chirp signal are analyzed in the paper. At
the receiving end, a new detection technique called mixing-change rate along with fractional Fourier transform (MCR-FrFT) is
proposed to detect the multirate chirp (MRC) signal and reduce complexity of node. There are two steps to realize the detection
technique MCR-FrFT. By using the MCR-FrFT, the computation of detection can be decreased to 50% compared with direct FrFT.
The simulation results indicate that, using the MCR-FrFT technique, the different users’ signal can be separated and detected rapidly.
1. Introduction been used in underwater wireless communication (UWC)

[5]. With the help of multilinear chirp (MLC) signals used in
Underwater Acoustic Sensor Networks (UW-ASNs) call for terrestrial sensor networks and because chirp signal modula-
deployment of multiple autonomous underwater units or tion technique has been used widely in underwater acoustic
sensors can be used for a wide range of marine applications, communication [8], the MLC-CDMA scheme is proposed for
including oceanography, environment monitoring, under- UW-ASNs in the paper.
sea exploration, mine reconnaissance, disaster prevention, The technology of multiple access using chirp signals
tsunami warning, equipment monitoring, military oversight, were first referred by Cook [9]. MAI mitigation methods
and navigation [1–6]. These nodes are manually or randomly used in common CDMA system can be directly used in
scattered in different depths in underwater environments this chirp CDMA, while using frequency-shift keying (FSK)
may cause undesirable MAI [10]. However, there are some
to collect specific data from deep or shallow water [7]. To
disadvantages of this method. The data rate is limited by
enable their collaborative operation in a shared physical
the total length of spreading codes, the bandwidth efficiency
channel, multiple access communications must be established
is low especially for single slope chirp- (LFM-) CDMA
[6]. Code Division Multiple Access (CDMA) is the most [6], and the performance of multiple access is not efficient.
promising physical layer and multiple access technique for In [11–13], multilinear chirp (MLC) signals are used in
UW-ASNs [1]. As with other spread spectrum methods, multiple access technique CDMA. Since these signals can be
chirp spread spectrum uses its entire allocated bandwidth to occupying the same bandwidth, this method is bandwidth
broadcast a signal, making it robust to channel noise. Further, efficient [10]. And the differences between traditional chirp
because the chirps utilize a broad band of the spectrum, signals (LFM) and MLC are analyzed in this paper, including
chirp spread spectrum is also resistant to multipath fading the performance of multiple access. On the other hand,
even when operating at very low power [6]. Additionally, neither simple nor feasible detection method is mentioned in
chirp signal and a low Doppler sensitive signalhave also previous papers. In order to make the system more feasible
Th /2
f
f
1 1
2 2
B B
M
M
Th t Th t
(a) (b)
Figure 1: The difference in time-frequency characteristic between LFM signals and MLC signals. (a) Time-frequency graph for LFM signals.
(b) Time-frequency graph for MLC signals.
and effective, a novel detection method called MRC-FrFT system in UW-ASNs, comparing the difference of multiple
(multirate change-fractional Fourier transform) is proposed access performance between single slope chirp and MLC
in the paper too. signal, proposing MCR-FrFT to reduce the complexity and
There are some methods to estimate the parameter of computation of the system.
chirp signal. In [14], the method of time-delay frequency The remainder of this paper is organized as follows. Sec-
mixing (rate reduction) to convert the chirp signal with tion 2 introduces the related work and the signals designed
different chirp rate into MFSK signal is mentioned. In for different nodes in UW-ASNs. Section 3 describes the new
[15], authors present dechirp pulse compression processing detection technique of MCR-FrFT. Section 4 analyses the
technique. The two methods are effective to single slop chirp performance of MCR-FrFT with different kinds of received
signal but unsuitable for MLC signals due to numbers of signals, and in Section 5 the simulation results are shown.
multipliers and no fixed chirp rate. In [12, 16], matched filter Finally, in Section 6, we draw the conclusions.
receivers are used, and at least 2𝑀-filters are needed (𝑀 is the
user number). It adds the complexity and power consumption
of node. What is more, it is difficult to achieve. In very 2. Chirp Signal Model
simple terms, the fractional Fourier transform (FrFT) is a Single slope chirp (LFM) signals are used in UW-ASNs for
generalization of the ordinary Fourier transform. Specially, node localization in our research group. And the multiple
the FrFT implements the so-called order parameter 𝑝 which
access technology using single slope chirp signals is being
acts on the ordinary Fourier transform operator [17]. In
studied in our group too. Our work indicates that the
other words, the 𝑝th order FrFT represents the 𝑝th power of
demodulation is quite complicated when there is not only one
the ordinary Fourier transform operator. The FrFT presents
the best localization performance in a certain FrFT domain, node signal received at the same time. One of the reasons
which is fit for the detection and estimation of multicompo- is the unfairness between users caused by different time-
nent linear frequency modulation (LFM) signals [18–20], and bandwidth products, and the multiple access using single
some fundamental properties of fractional Fourier transform slope chirp signal is not effective. Based on our work and
can be found in [21]. Some improved algorithms based on paper [10, 11], to overcome the deficiency above, we choose
FrFT are also proposed, such as EEMD-FRFT [22] and the MLC signals instead, where time-frequency characteristic
STFT [23], but these methods need 2 different 𝑝 values to is as shown in Figure 1(b).
complete the parameter estimation. So for MLC signal, the Figure 1(a) is the time-frequency characteristic of LFM
main drawback of FrFT is heavy computation. To solve these signals, while Figure 1(b) is the time-frequency characteristic
drawbacks, MCR-FrFT is proposed in the paper. The system of MLC signals.
used in underwater acoustic channel [24] is discussed too, For LFM signals, the 𝑚th user’s signal could be expressed
and it shows that the system can be used in underwater as
acoustic channel normally. The contribution of our work
can be divided into three parts: proposing MLC-CDMA 𝑆𝑚 (𝑡) = cos (2𝜋𝑓0 𝑡 ± 𝜋𝜇𝑚 𝑡2 ) , (1)
International Journal of Distributed Sensor Networks 3
where 𝜇𝑚 = (𝑀 + 1 − 𝑚)𝐵/(𝑀 ⋅ 𝑇ℎ ), and the corresponding the signal duration. The slopes of general MLC signal can be
time-bandwidth product is 𝐷𝑚 = ((𝑀 + 1 − 𝑚)/𝑀)𝐵𝑇ℎ . expressed as
For MLC signals, each signal is characterized by two different
slopes and the general expression for these MLC signals can (𝑀 + 1) − 𝑚
be expressed as 𝜇𝑚𝑓 = ⋅ Δ𝑓,
𝑇ℎ /2
(5)
𝑆𝑚1 (𝑡) = 𝑆𝑚𝑓1 (𝑡) + 𝑆𝑚𝑏1 (𝑡) , 𝑚Δ𝑓
(2) 𝜇𝑚𝑏 = .
𝑇ℎ /2
𝑆𝑚0 (𝑡) = 𝑆𝑚𝑓0 (𝑡) + 𝑆𝑚𝑏0 (𝑡) .
𝑇ℎ is the duration time of each MLC signal per bit, 𝑓𝑐
The subscripts “1” and “0” mean bit 1 and bit 0. The
is the carrier frequency, and Δ𝑓 is the frequency separation
subscripts “𝑓” and “𝑏” indicate the first half and second half
between successive nodes at 𝑡 = 𝑇ℎ /2. The MLC signal for
of the duration of one bit signal. And
each node has the same bandwidth 𝐵 = (𝑀 + 1)Δ𝑓 and its
time-bandwidth product is given by
𝑆𝑚𝑓1 (𝑡) = 𝐴 cos (2𝜋𝑓𝑐 𝑡 + 𝜋𝜇𝑚𝑓 𝑡2 ) ,
(3)
𝑆𝑚𝑓0 (𝑡) = 𝐴 cos (2𝜋𝑓𝑐 𝑡 − 𝜋𝜇𝑚𝑓 𝑡2 ) 𝐷 = 𝐵𝑇ℎ = (𝑀 + 1) 𝑇ℎ ⋅ Δ𝑓. (6)
when 0 ≤ 𝑡 < 𝑇ℎ /2, and The time-bandwidth product 𝐷𝑚 of single slope chirp
signal is a variable that leads to different node signals having
𝑇ℎ different process gain, while 𝐷 of MLC signal is a constant to
𝑆𝑚𝑏1 (𝑡) = 𝐴 cos [2𝜋 (𝑓𝑐 + (𝑀 + 1 − 𝑚) Δ𝑓) (𝑡 − ) ensure each node signal has the same process gain.
2
The performance of multilinear chirp is mainly deter-
𝑇ℎ 2 mined by the cross-coherence between the different node
+ 𝜋𝜇𝑚𝑏 (𝑡 − ) ], signals. Ideally the signals should be orthogonal with zero
2
(4) cross-coherence to cancel the multiple access interference
𝑇 (MAI). The biggest cross-coherence coefficient existed in
𝑆𝑚𝑏0 (𝑡) = 𝐴 cos [2𝜋 (𝑓𝑐 − (𝑀 + 1 − 𝑚) Δ𝑓) (𝑡 − ℎ ) adjacent node and adjacent signal.
2
Set 𝐴 = √2𝐸/𝑇ℎ and the biggest cross-coherence 𝜌max
𝑇ℎ 2 takes the form
− 𝜋𝜇𝑚𝑏 (𝑡 − )]
2
𝜌max
when 𝑇ℎ /2 ≤ 𝑡 < 𝑇ℎ , where 𝑚 is the user number, 𝑚 = 󵄨󵄨 (7)
1 𝑇 ∗ 󵄨
1, 2, . . . , 𝑀, and 𝑀 is the total number of nodes in UW-ASNs. = ∫ 𝑆𝑚 (𝑡) ∗ 𝑆𝑚+1 (𝑡) 𝑑𝑡󵄨󵄨󵄨󵄨 .
𝜇𝑚𝑓 is the 𝑚th node’s slope within the first half of signal 𝐸 0 󵄨󵄨bit 1 and bit 1 or bit 0 and bit 0
duration and 𝜇𝑚𝑏 is the complement slope of the 𝑚th node
within the second half of signal duration, together for bit Substituting the corresponding values of 𝑆𝑚 (𝑡) from (1)–
“1.” −𝜇𝑚𝑓 and −𝜇𝑚𝑏 are two negative slopes given to the 𝑚th (5) and neglecting the integration over the higher frequencies,
node for bit “0,” respectively, for the first and second half of we get
𝜌max
1 𝑇ℎ 𝐵 2
{
{ 𝑇ℎ ∫0 cos (𝜋 (𝑀 + 1) 𝑇ℎ 𝑡 ) 𝑑𝑡,
{ Signal slope chirp signal (8)
={ 𝑇ℎ /2 𝑇ℎ 2
{
{ 1 [∫ cos (2𝜋 𝐵 𝐵 𝑇 2𝐵 𝑇
𝑡2 ) 𝑑𝑡 + ∫ cos (2𝜋 (𝑡2 − ℎ ) − 𝜋 (𝑡2 − ℎ )) 𝑑𝑡] combined chirp signal.
{ 𝑇ℎ 0 (𝑀 + 1) 𝑇ℎ 𝑇ℎ /2 𝑀+1 2 (𝑀 + 1) 𝑇ℎ 2
Taking V = √𝐵/(𝑀 + 1)𝑇ℎ 𝑡, 𝑢 = √𝐵/(𝑀 + 1)𝑇ℎ (𝑡−𝑇ℎ /2),

then (8) can be changed to
√𝐵𝑇ℎ /(𝑀+1)
{
{∫ cos (𝜋V2 ) 𝑑V
𝑀+1{{ 0
𝜌max =√ (9)
𝐵𝑇ℎ {
{
{
√𝐵𝑇ℎ /4(𝑀+1) √𝐵𝑇ℎ /4(𝑀+1)
𝐵
{∫ cos (2𝜋V2 ) 𝑑V + ∫ cos (2𝜋√ 𝑢 − 2𝜋𝑢2 ) 𝑑𝑢.
{ 0 0 (𝑀 + 1) 𝑇ℎ
1
0.9
0.25
0.8
0.7 0.2
Cross-coherence
Cross-coherence
0.6
0.5 0.15
0.4
0.1
0.3
0.2
0.05
0.1
0 0
5 10 15 20 25 30
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0.02
Node number, M
Time (s)
Single chirp signal Single chirp signal
Combined chirp signal Combined chirp signal
(a) (b)
Figure 2: The cross-coherence function 𝜌. (a) As a function of time with 𝑀 = 15. (b) As a function of nodes number 𝑀 from 2 to 32 with
the time-bandwidth product 𝐷 = 100.
User m
MCR FrFT Decision Bit 1
Received
signal
BPF Mixer FrFT
Frequency User m
synthesizer MCR FrFT Decision Bit 0
PN code
generator
Figure 3: Receiver block diagram for MLC-CDMA.
From (9), 𝜌 has an oscillatory nature as a function of time. is the same. So the MLC signals are selected in the paper for
Set 𝐵 = 5 kHz; when 𝑀 = 15, 𝑇ℎ is from 1/𝑓𝑠 to 0.02, we chirp-CDMA. And in the next section, the detection method
get Figure 2(a), and when 𝑇ℎ = 0.02, 𝑀 = [2 : 32], we get for this kind of MLC signal is described.
Figure 2(b).
The difference between single slope chirp signal and MLC
signal in cross-coherence is shown in Figure 2(a). The empty 3. The Detection Block Diagram
circle (o) is the cross-coherence coefficient of MLC signal,
The detection technique MCR-FrFT and its block diagram
and the solid dot (⋅) is the cross-coherence coefficient of
are shown in Figure 3. The received signal is divided into
single slope chirp signal. From Figure 2(a), when there is
three branches after mixing with PN Code. The first branch
the same number of nodes in the UW-ASNs, such as 15, the
and third branch use fixed local signal, 𝑥 local 𝑢(𝑡) and
performance of multiple access using MLC signals is better
𝑥 local 𝑑(𝑡), respectively, to implement MCR (the detail about
than the one using single slop chirp signal. For example, when
MCR is introduced in Section 3.1). And the FrFT (the detail
the duration time is 0.01 and that time product is 100, the
about FrFT is introduced in Section 3.2) is used to complete
cross-coherence coefficient of single slop chirp signal is about
parameter estimation of chirp signal that represents bit “0”
0.1 bigger than MLC signal. The time-bandwidth product is
or bit “1” for different users. That is, there are two steps to
100 in Figure 2(b); here the cross-coherence as a function
complete MCR-FrFT. Here, set 𝐴 = 1.
of node number that is the performance of multiple access
changed depend on the number of nodes. At the same time,
Figure 2(b) tells us that the performance of multiple access 3.1. Mixing-Change-Rate (MCR). The purpose of MCR is to
using MLC signal is much better when the number of nodes change the set of chirp rates of MLC signal into a new set. It
rf (t) r0 (t) Consider 𝑟𝑓 (𝑡) = 𝑟(𝑡) ⋅ 𝑥 local(𝑡), 𝑟0 (𝑡) = 𝑟𝑓 (𝑡) ∗ ℎLPF (𝑡),
r(t)
LPF where ℎLPF (𝑡) is the impulse response function of low pass
filter (LPF).
xlocal(t) 3.2. Fractional Fourier Transform (FrFT). FrFT, which can

Figure 4: The block diagram of MCR. be expressed as rotating the signal an angle of 𝛼 to 𝑢-
axis anticlockwise on the time axis, is a generalized Fourier
transformation form and is the linear projection of the signal
includes two parts, multiplier and low pass filter. The block to the rotary frequency space. The FrFT definition of signal
diagram of MCR is shown in Figure 4. 𝑥(𝑡) is
That is,
𝑥 local (𝑡) +∞
𝑋𝑝 (𝑢) = ∫ 𝐾𝑝 (𝑡, 𝑢) 𝑥 (𝑡) 𝑑𝑡. (11)
{ 𝐵 2 −∞
{𝑥 local 𝑢 (𝑡) = cos (2𝜋𝑓0 𝑡 + 𝜋 𝑇 𝑡 )
{ (10)
={ ℎ
{𝑥 local 𝑑 (𝑡) = cos (2𝜋𝑓0 𝑡 − 𝜋 𝐵 𝑡2 ) .
{ And the transformation kernel is
{ 𝑇ℎ
{ 1 − 𝑗 cot 𝛼 1 1
{
{
{
√ exp [𝑗 ( 𝑡2 cot 𝛼 + 𝑢2 cot 𝛼 − 𝑡𝑢 csc 𝛼)] , 𝛼 ≠ 𝑛𝜋
{ 2𝜋 2 2
𝐾𝑝 (𝑡, 𝑢) = {𝛿 (𝑡 − 𝑢) , 𝛼 = 2𝑛𝜋 (12)
{
{
{
{
{𝛿 (𝑡 + 𝑢) , 𝛼 = (2𝑛 ± 1) 𝜋.
Simple component LFM signal with noise is where 𝑛(𝑡) is white noise and its FrFT is
𝑟 (𝑡) = 𝐴 0 cos (2𝜋𝑓𝑐 𝑡 + 𝜋𝑢𝑡2 ) + 𝑛 (𝑡) , 0 ≤ 𝑡 ≤ 𝑇ℎ , (13)
+∞
𝐹𝛼 𝑟 (𝑢) = 𝑋𝑝 (𝑢) = ∫ 𝑟 (𝑡) 𝐾𝑝 (𝑡, 𝑢) 𝑑𝑡
−∞
(14)
1 − 𝑗 cot 𝛼 (𝑗(1/2)𝑢2 cot 𝛼) 𝑇ℎ (2𝑗𝜋((1/2)𝑡2 (𝑢+cot 𝛼/2𝜋)+𝑡(𝑓𝑐 −𝑢 csc 𝛼/2𝜋)))
= 𝐴 0√ 𝑒 ∫ 𝑒 𝑑𝑡 + 𝑛󸀠 (𝑡) ,
2𝜋 0
𝑇
where 𝑛󸀠 (𝑡) = ∫0 ℎ 𝑛(𝑡)𝐾𝑝 (𝑡, 𝑢)𝑑𝑡, 𝐹𝛼 𝑟(𝑢) is FrFT, 𝑝 is the uses the local signal to convert the chirp rates of MLC signal
transformational order, 𝛼 is the angle of rotation, and 𝛼 = into a new set of chirp rates. The second level adopts the FrFT
𝑝𝜋/2. When 𝑝 = 1, FrFT will degenerate into Fourier to estimate the parameter.
transformation, and when 𝑝 = 0, FrFT is just the original
signal. And the inverse of an FrFT with an angle 𝛼 is the FrFT 4.1. Single Node Signal. Assuming that only the 𝑚th node’s
with an angle −𝛼. signal is received, substituting the corresponding value of
From (14), one LFM signal is an impulse function only 𝑆𝑚1 (𝑡) from (2)–(5), we get
in the appropriate fractional Fourier domain. Select the right
angle of rotation (or 𝑝 value) to do FrFT for LFM signal, then 𝑟 (𝑡) = 𝑆𝑚1 (𝑡) + 𝑛 (𝑡) = cos (2𝜋𝑓0 𝑡 + 𝜋𝜇𝑚𝑓 𝑡2 )
the amplitude of energy aggregation of LFM signal will show
obvious peak. The energy of noise, which cannot show energy 4𝑡 − 𝑇ℎ
⋅ rect ( )
aggregation at any fractional Fourier domain and cannot 2𝑇ℎ
show the peak, distributes on the whole time-frequency plane (15)
𝑇ℎ
evenly. + cos [2𝜋 (𝑓0 + (𝑀 + 1 − 𝑚) Δ𝑓) (𝑡 − )
2
4. The MCR-FrFT for Different 𝑇ℎ 2 4𝑡 − 3𝑇ℎ
+ 𝜋𝜇𝑚𝑏 (𝑡 − ) ] ⋅ rect ( ) + 𝑛 (𝑡) .
Received Signals 2 2𝑇ℎ
In this section, different received signals have been discussed, 𝑛(𝑡) is the underwater acoustic channel noise. After the
such as single user’s signal and three users’ signal. There are MCR, the high frequency terms have been filtered and the
two levels of MCR-FrFT at the receiving end. The first level signal becomes as follows:
1
𝑟0 (𝑡) = 𝑤 (𝑡) +
2
{ 4𝑡 − 𝑇ℎ 𝑀+1−𝑚 𝐵 𝑇 𝑇 2 4𝑡 − 3𝑇ℎ
{
{ cos (𝜋 (𝜇𝑚𝑓 − 𝜇) 𝑡2 ) ⋅ rect ( ) + cos [2𝜋 ( 𝐵− ) (𝑡 − ℎ ) + 𝜋 (𝜇𝑚𝑏 − 𝜇) (𝑡 − ℎ ) ] ⋅ rect ( ), the first branch (16)
{ 2𝑇ℎ 𝑀+1 2 2 2 2𝑇ℎ
⋅{
{
{ 4𝑡 − 𝑇ℎ 𝑀+1−𝑚 1 𝑇 𝑇 2 4𝑡 − 3𝑇ℎ
{cos (𝜋 (𝜇𝑚𝑓 + 𝜇) 𝑡2 ) ⋅ rect ( ) + cos [2𝜋 ( 𝐵+ 𝐵) (𝑡 − ℎ ) + 𝜋 (𝜇𝑚𝑏 + 𝜇) (𝑡 − ℎ ) ] ⋅ rect ( ) , the third branch.
{ 2𝑇ℎ 𝑀+1 2 2 2 2𝑇ℎ
Comparing the third branch and the first branch in (16), the MLC signal composed of two different negative slopes for
after MCR, the MLC signal composed of two positive slopes bit “0” in the first branch can be ignored too.
in the third branch is just only a short time duration (less Substituting the corresponding values 𝜇 of (5) and (10)
than 1/6 duration time), as shown in Figure 5(b). So the MLC into (16), neglecting the third branch, we get
signals for bit “1” can be ignored in the third branch; similarly,
𝑀 + 1 − 2𝑚 2 𝑇ℎ
{cos [𝜋 𝐵𝑡 ] , 0≤𝑡≤
1{
{ (𝑀 + 1) 𝑇ℎ 2
𝑟0 (𝑡) = 𝑤 (𝑡) + { 2 (17)
{cos [2𝜋 (𝑀 + 1 − 2𝑚) 𝐵 ⋅ (𝑡 − 𝑇ℎ ) − 𝜋 𝑀 + 1 − 2𝑚 ⋅ 𝐵 (𝑡 − 𝑇ℎ ) ] ,
2{ 𝑇ℎ
≤ 𝑡 ≤ 𝑇ℎ .
{ 2 (𝑀 + 1) 2 𝑀+1 𝑇ℎ 2 2
The 𝑤(𝑡) = (𝑛(𝑡) ⋅ 𝑥local(𝑡)) ∗ ℎLPF (𝑡) is the noise of the channel and its reflection properties, which determine
component after MCR. From (17), by MCR, a new set of chirp the number of significant propagation paths, their relative
rates is got at the corresponding branch and the 𝑤(𝑡) does not strengths, and delays. Strictly speaking, there are infinitely
change the chirp rate at the condition. The major character of many signal echoes, but those that have undergone multiple
the new set of chirp rates is that the two slopes are opposite, reflections and lost much of the energy can be discarded,
as shown in Figure 5(b) with dashed line. leaving only a finite number of significant paths. At this point,
Figure 5(a) is the original MLC with two positive slopes, the channel impulse response can be expressed as
and after MCR, the slopes are changed as shown in Fig-
ure 5(b). At the corresponding branch, we get a new MLC 𝑁−1
signal with the set of chirp rates, where the two values are ℎ (𝑡) = 𝐴 0 𝛿 (𝑡 − 𝜏0 ) + ∑ 𝐴 𝑖 𝛿 (𝑡 − 𝜏𝑖 ) . (18)
𝑖=1
opposite, as dashed line shows in Figure 5(b). The solid line in
Figure 5(b) is the MCR result at the other branch (no middle
Assuming that the emission signal is 𝑠(𝑡), after underwa-
branch).
ter acoustic channel, the received signal can be expressed as
According to Section 3.2, the parameter estimation based
on FrFT is not sensitive to noise, and the result mainly 𝑁−1
depends on the signal. Figure 5 shows that, using the MCR, 𝑟 (𝑡) = 𝐴 0 𝑠 (𝑡 − 𝜏0 ) + ∑ 𝐴 𝑖 𝑠 (𝑡 − 𝜏𝑖 ) + 𝑛 (𝑡)
the set of chirp rates is changed from two different slopes 𝑖=1
(Figure 5(a)) to a new set of chirp rates, where two slopes (19)
𝑁−1
are opposite. With the fundamental property of FrFT and
= ∑ 𝐴 𝑖 ⋅ 𝑠 (𝑡 − 𝜏𝑖 ) + 𝑛 (𝑡) ,
DFrFT mentioned in [17] that only one 𝑝 value is needed 𝑖=0
to complete the new MLC signal with two slopes opposite
while the original MLC signal needs two different 𝑝 values where 𝑁 is the total number of propagation paths and 𝑖 = 0
to complete its parameter estimation, the computation using corresponds to the direct path. 𝐴 𝑖 is the amplitude of the 𝑖th
MCR-FrFT can almost be reduced to 50% compared with propagation path at receiving side. Sometimes it represents
direct FrFT without MCR. the gain of this path. 𝜏𝑖 is the propagation delay of the 𝑖th
propagation path.
4.2. Multipath Underwater Channel. The impulse response of Take the 𝑚th node as an example, and set 𝑠(𝑡) = 𝑆𝑚1 (𝑡);
an underwater acoustic channel is influenced by the geometry then
𝑁−1 𝑁−1
2 4 (𝑡 − 𝜏𝑖 ) − 𝑇
𝑟 (𝑡) = ∑ 𝐴 𝑖 ⋅ 𝑠𝑚 (𝑡 − 𝜏𝑖 ) + 𝑛 (𝑡) = ∑ 𝐴 𝑖 [cos (2𝜋𝑓0 (𝑡 − 𝜏𝑖 ) + 𝜋𝜇𝑚𝑓 (𝑡 − 𝜏𝑖 ) ) ⋅ rect ( )
𝑖=0 𝑖=0 2𝑇
(20)
(𝑀 + 1 − 𝑚) 𝑇 𝑇 2 4 (𝑡 − 𝜏𝑖 ) − 3𝑇
+ cos (2𝜋 (𝑓0 + 𝐵) (𝑡 − 𝜏𝑖 − ) + 𝜋𝜇𝑚𝑏 (𝑡 − 𝜏𝑖 − ) ) ⋅ rect ( )] + 𝑛 (𝑡) .
𝑀+1 2 2 2𝑇
×104
The new sets of chirp rate by MCR
2.9
1800
2.85 1600
2.8 1400
1200
Frequency (Hz)
Frequency (Hz)
2.75
1000
2.7
800
2.65
600
2.6
400
2.55 200
2.5 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s) Time (s)
Bit “1,” first branch

Bit “1,” third branch
(a) (b)
Figure 5: (a) The new sets of chirp rate by MCR. (b) Show the difference of chirp rates after MCR between the first branch and the third
branch with the frequency within about [0, 1900] Hz.
After MCR, we get
1𝑁 2 (𝑀 + 1 − 𝑚) 𝐵𝜏𝑖 (𝑀 + 1 − 2𝑚) 𝐵 2 (𝑀 + 1 − 𝑚) 𝐵 2
𝑟0 (𝑡) = ∑𝐴 𝑖 [cos (2𝜋 𝑡−𝜋 𝑡 + 2𝜋𝑓0 𝜏 − 2𝜋 𝜏 )
2 𝑖=1 (𝑀 + 1) 𝑇 (𝑀 + 1) 𝑇 (𝑀 + 1) 𝑇 𝑖
(𝑀 + 1 − 2𝑚) 𝐵𝑇 − 4𝑚𝐵𝜏𝑖 𝑇 (𝑀 + 1 − 2𝑚) 𝐵 𝑇 2 (𝑀 + 1 − 𝑚) 𝑚𝐵 (21)

+ cos (2𝜋 (𝑡 − ) − 𝜋 (𝑡 − ) − 2𝜋 𝐵𝜏𝑖 − 2𝜋𝑓0 𝜏𝑖 + 2𝜋 𝜏 2 )]
2 (𝑀 + 1) 𝑇 2 (𝑀 + 1) 𝑇 2 𝑀+1 (𝑀 + 1) 𝑇 𝑖
+ 𝑤 (𝑡) .
The signals of different nodes are detected according to to (5), we get

the different sets of chirp rates, and (21) shows that the 𝑟 (𝑡) = 𝑆11 (𝑡) + 𝑆21 (𝑡) + 𝑆30 (𝑡) (22)
multipath and noise of underwater channel do not change the
chirp rate or generate any new chirp rate compared with (17). and, similarly to (15), (16), and (17), after MCR, the signal can
For fixed 𝑀, 𝑇, and 𝐵, the chirp rates are only determined by be expressed as
𝑚. So the detection result will not be affected by multipath
underwater acoustic channel. 1 𝑀−1
the first branch 𝑟0 (𝑡) = ⋅ [cos (𝜋 𝐵𝑡2 )
2 (𝑀 + 1) 𝑇ℎ
4.3. Multiple Users’ Signal. From Sections 4.1 and 4.2, we can 𝑀−3 4𝑡 − 𝑇ℎ
see that the noise of underwater channel does not change the + cos (𝜋 𝐵𝑡2 )] ⋅ rect ( )
(𝑀 + 1) 𝑇ℎ 2𝑇ℎ
chirp rate, so it can be ignored temporarily in the process of
analysis. When there is not only one MLC signal received, (𝑀 − 1) 𝐵 𝑇
+ [cos (2𝜋 (𝑡 − ℎ )
assume that there are three MLC signals received: user 1 bit 2 (𝑀 + 1) 2
“1,” user 2 bit “1,” and user 3 bit “0.” Here, two cases are
considered: the first one is that all these signals are received at 𝑀−1 𝑇 2
−𝜋 𝐵 (𝑡 − ℎ ) )
the same time and the other one is that these signals are not (𝑀 + 1) 𝑇ℎ 2
received at the same time.
If these signals are received at the same time, substituting (𝑀 − 3) 𝐵 𝑇
+ cos (2𝜋 (𝑡 − ℎ )
the corresponding values of 𝑆11 (𝑡), 𝑆21 (𝑡), and 𝑆30 (𝑡) from (2) 2 (𝑀 + 1) 2
𝑀−3 𝑇 2 4𝑡 − 3𝑇ℎ Assuming 0 < 𝑡1 < 𝑇ℎ /2 < 𝑡2 < 𝑇ℎ , and 0 ≤ 𝑡 ≤ 𝑇ℎ , after

−𝜋 𝐵 (𝑡 − ℎ ) )] ⋅ rect ( ),
(𝑀 + 1) 𝑇ℎ 2 2𝑇ℎ MCR, we get
1 𝑀−5 1 𝑀−1 4𝑡 − 𝑇ℎ
the third branch 𝑟0 (𝑡) = ⋅ cos (𝜋 𝐵𝑡2 ) 𝑟0 (𝑡) = [cos [𝜋 𝐵𝑡2 ] ⋅ rect ( )
2 (𝑀 + 1) 𝑇ℎ 2 (𝑀 + 1) 𝑇ℎ 2𝑇ℎ
4𝑡 − 𝑇ℎ (𝑀 − 5) 𝐵 𝑇 (𝑀 − 1) 𝐵 𝑇 𝑀−1
⋅ rect ( ) + cos (2𝜋 (𝑡 − ℎ ) + cos [2𝜋 (𝑡 − ℎ ) − 𝜋
2𝑇ℎ 2 (𝑀 + 1) 2 2 (𝑀 + 1) 2 (𝑀 + 1) 𝑇ℎ
𝑀−5 𝑇 2 4𝑡 − 3𝑇ℎ 𝑇ℎ 2 4𝑡 − 3𝑇ℎ

−𝜋 𝐵 (𝑡 − ℎ ) ) ⋅ rect ( ). ⋅ 𝐵 (𝑡 − ) ] ⋅ rect ( )
(𝑀 + 1) 𝑇ℎ 2 2𝑇ℎ 2 2𝑇ℎ
(23)
𝑀−3 2
+ cos [𝜋 𝐵 (𝑡 − 𝑡1 ) ]
According to (23), the new sets of chirp rates can be (𝑀 + 1) 𝑇ℎ
shown in Figure 6(b).
The solid line in Figure 6(b) is the MCR result of the 4 (𝑡 − 𝑡1 ) − 𝑇ℎ
⋅ rect ( )
solid line in Figure 6(a) in time-frequency characteristic at 2𝑇ℎ
the third branch while the dot line and circle line are at the (27)
first branch. (𝑀 − 3) 𝐵 𝑇
+ cos [2𝜋 (𝑡 − 𝑡1 − ℎ )
Above all, taking bit “1” as an example, the MLC signals’ 2 (𝑀 + 1) 2
rates changed by MCR can be expressed as
𝑀−3 𝑇 2
𝜇1𝑓 𝜇1𝑏 −𝜋 𝐵 (𝑡 − 𝑡1 − ℎ ) ]
(𝑀 + 1) 𝑇ℎ 2
[ 𝜇 𝜇2𝑏 ]
[ 2𝑓 ]
[ ] 4 (𝑡 − 𝑡1 ) − 3𝑇ℎ
[ . .. ]
[ . ] 𝐵 ⋅ rect ( )
[ . . ]− 𝑇 2𝑇ℎ
[ ] ℎ
[𝜇 ]
[ (𝑀−1)𝑓 𝜇(𝑀−1)𝑏 ] 𝑀−5 2
+ cos [𝜋 𝐵 (𝑡 − 𝑡2 ) ]
[ 𝜇𝑀𝑓 𝜇𝑀𝑏 ] (𝑀 + 1) 𝑇ℎ
𝐵 𝐵 2𝑡 − 𝑇ℎ − 𝑡2
[ 𝜇1𝑓 − 𝑇ℎ 𝜇1𝑏 −
𝑇 ] (24) ⋅ rect (
2𝑇ℎ − 2𝑡2
)] .
[ ℎ ]
[ 𝐵 𝐵 ]
[ 𝜇 − 𝜇2𝑏 − ] Assuming that 𝑡1 = 𝑇ℎ /4, 𝑡2 = 3 ∗ 𝑇ℎ /4, then the sets
[ 2𝑓 𝑇 𝑇ℎ ]
[ ℎ ]
[ .. .. ] of chirp rates of original and its MCR result are shown in
=[
[ . . ].
] Figure 7.
[ ]
[ 𝐵 𝐵] In Figure 7(b), the dots are the received signal without
[𝜇(𝑀−1)𝑓 − 𝜇(𝑀−1)𝑏 − ]
[ 𝑇ℎ 𝑇ℎ ] time-delay, the empty circle is the received signal where the
[ ]
[ 𝐵 𝐵 ] time-delay is less than half of the per bit signal duration, and
𝜇𝑀𝑓 − 𝜇𝑀𝑏 − the solid line is the one where the time-delay is bigger than
[ 𝑇ℎ 𝑇ℎ ]
half of per bit signal duration but less than per bit signal
Substituting the corresponding values of 𝜇 in (3) and 𝑚 = duration.
1, 2, . . . , 𝑀, the new sets of chirp rates after MCR are got as From the above, the result of MCR of MLC signal is also a
MLC signal but the new set of chirp rates is composed of two
𝑀 𝑀 opposite slope values.
−
[𝑀 + 1 𝑀 + 1] Since the inverse of an FrFT with an angle 𝛼 is the FrFT
[ ]
[𝑀 − 1 𝑀 − 1] with an angle −𝛼, each new MLC signal from MCR result
[ − ]
[𝑀 + 1 𝑀 + 1] needs only one 𝑝 value to realize its parameter estimation,
󸀠 𝐵 [[ .. ..
]
] while the MLC signal before MCR needs two different 𝑝 val-
𝜇 = [ ]. (25)
𝑇ℎ [ . . ] ues to complete its parameter estimation. So the computation
[ 2 2 ]
[ − ] can be reduced to half of the original one.
[𝑀 + 1 𝑀 + 1]
[ ]
[ 1 1 ]
− 5. Simulation
[𝑀 + 1 𝑀 + 1]
If these signals are not received at the same time, there The simulation tool MATLAB is used to evaluate the perfor-
is relative time-delay between one another; substituting the mance of MCR-FrFT. Set the total number of nodes 𝑀 = 15,
corresponding values of 𝑆11 (𝑡), 𝑆21 (𝑡), and 𝑆31 (𝑡), we get the bandwidth 𝐵 = 5 kHz, the signal duration time 𝑇ℎ =
0.04 s, and the sample frequency 𝑓𝑠 = 90 kHz. The noise
𝑟 (𝑡) = 𝑆11 (𝑡) + 𝑆21 (𝑡 − 𝑡1 ) + 𝑆31 (𝑡 − 𝑡2 ) . (26) is white Gaussian noise and the SNR is 0 dB. FrFT used in
×104 The sets of combined chirp rate The new sets of chirp rate by MCR
3
2.95 2000
2.9 1800
2.85 1600
Frequency (Hz)
1400
Frequency (Hz)
2.8
1200
2.75
1000
2.7
800
2.65
600
2.6 400
2.55 200
2.5 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s) Time (s)
Bit “1,” user 1 Bit “1,” user 1, the first branch

Bit “1,” user 2 Bit “1,” user 2, the first branch
Bit “0,” user 3 Bit “0,” user 3, the third branch
(a) (b)
Figure 6: The sets of chirp rates of MLC signal received at the same time. (a) The original slopes of three MLC signals. (b) The new sets of
chirp rates after MCR.
×104 The sets of combined chirp rate The new sets of chirp rate by MCR
3
1800
2.95
1600
2.9
1400
2.85
Frequency (Hz)
1200
Frequency (Hz)
2.8
1000
2.75
800
2.7
2.65 600
2.6 400
2.55 200
2.5 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s) Time (s)
Bit “1,” user 1 Bit “1,” user 1

(a) (b)
Figure 7: The signals of user 1, user 2, and user 3 for bit “1” are not received at the same time. (a) The slopes of MLC signals. (b) The slopes
after MCR corresponding to (a).
MATLAB is discrete fractional Fourier transform (DFrFT) Figure 8(a) is the waveform of received signal including
and each 𝑝 at the receiving end is given by 𝑝 = 2/𝜋⋅𝑎cot[(𝑀+ only one MLC signal at the SNR that is 0-dB. Figure 8(b)
1)𝑓𝑠 /(𝑀 + 1 − 2𝑚)𝐵] + 1. In the condition of only one node indicates that, using the MCR, the parameter of MLC signal
signal, let 𝑚 = 1; then 𝑝 = 1.0309, and the received signal’s can be detected correctly with the SNR that is 0-dB for AWGN
waveform and its 𝑝 order FrFT are shown in Figure 8. The channel. Of course, 𝑀 times FrFT with different 𝑝 values have
ordinate is amplitude and abscissa is time or 𝑢 value. been taken at the first and the third branches, respectively,
The received signal The received signal

5 6
4
4
3
2 2
1
0
0
−1 −2
−2
−4
−3
−4 −6
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
(a)
(a)
250
400
350
200
Amplitude (V) 300
150 250
Amplitude (V)
200
100
150
100
50
50
0 0
0 500 1000 1500 2000 2500 3000 3500 4000 0 500 1000 1500 2000 2500 3000 3500 4000
N/1 N/1
User 1, bit “1” User 1, bit “1”
(b) User 2, bit “1”
(b)
Figure 8: (a) The waveform of received signal. (b) The result of The third user included in received signal
detection. 250
200
but only when 𝑝 = 𝑝1 = 1.0309 at the first branch we get

Amplitude (V)
150
Figure 8(b); that is, only the 1st node bit “1” exists.
In the condition of Section 4.3 the waveform of 𝑠(𝑡) with
noise is shown in Figure 9(a), and when 𝑝 = 𝑝1 = 1.0309 and 100
𝑝 = 𝑝2 = 1.0265 at the first branch, we get Figure 9(b); that is,
user 1 bit “1” and user 2 bit “1” exist; when 𝑝 = 𝑝3 = 1.0221 at
the third branch, Figure 9(c) can be got which indicates that 50
user 3 bit “0” exist.
From (27), set the time-delay between the first node and 0
the second node that is 3𝑇ℎ /8, and the time-delay between the 0 500 1000 1500 2000 2500 3000 3500 4000
first node and the third node is 5𝑇ℎ /8. The waveform of the N/1
received signal 𝑠(𝑡) is as shown in Figure 10(a) and its MCR-
FrFT result is as shown in Figure 10(b). But only node 1 and User 3, bit “0”
node 2 can be detected, and node 3 cannot be detected. The (c)
result indicates that the received signals whose duration time Figure 9: (a) The waveform of received signal. (b) and (c) are the
is bigger than 𝑇ℎ /2 can be detected. result of MCR-FrFT.
The received signal

6 250
4 200
Amplitude (V)
150
0
100
−2
50
−4
−6 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 500 1000 1500 2000 2500 3000 3500 4000
N/1
User 1, bit “1”

User 2, bit “1”
(a) (b)
Figure 10: (a) The waveform of received signal. (b) The result of its FrFT.
The received signal

6 250
4
200
2
Amplitude (V)
150
0
100
−2
50
−4
−6 0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0 500 1000 1500 2000 2500 3000 3500 4000
N/1
User 1, bit “1”

(a) (b)
Figure 11: (a) The waveform of received signal. (b) The result of its MCR-FrFT result.
For Section 4.2, set 𝜏 = [0 0.005 0.008 0.013 0.017], signal or MLC signal can be expressed as
𝐴 = [1 0.8 0.7 0.5 0.3], 𝑁 = 5, 𝑚 = 1, and then the
waveform of 𝑠(𝑡) is shown in Figure 11(a), and only when the
𝑝1 = 1.0309 we can get the detection result as is shown in 1 𝐸
𝑃𝑒 = erfc (√ (1 − 𝜌)) . (28)
Figure 11(b). 2 𝑁0
According to Figure 11(b) and the value of 𝑇 and 𝜏, those
signals where the time-delay is smaller than 𝑇/2 can be √𝐸/𝑁0 is the SNR and 𝜌 is the cross-coherence coefficient
detected. that is expressed by (8) or (9); when 𝑀 = 10, we get
According to [13], the relationship between bit error rate 𝜌 = { 0.1589
0.0554 , and then the BER curve can be drawn as in
(BER) and cross-coherence coefficient of single slope chirp Figure 12(a).
100 101
100
−1
10
10−1
BER
10−2
BER
10−2
10−3
−3
10
10−4
10−4 10−5
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 9 10
SNR (dB) SNR (dB)
Signal chirp signal 4 users, MLC 8 users, MLC

MCR 4 users, FSK 8 users, FSK
(a) (b)
Figure 12: BER for (a) LFM and MLC signals. (b) FSK and MLC FH-CDMA techniques for an AWGN channel.
It can be seen from Figure 12(a) that the BER performance Conflict of Interests
of the MLC signals is better than the single slope chirp signals.
The reason is that the cross-coherence coefficient of single The authors declare no conflict of interests.
slope chirp signals is bigger than the one of MLC signals.
And the BER is a monotone decreasing function as cross- Acknowledgments
coherence coefficient. Figure 12(b) shows the performance
comparison of the MLC and FSK FH-CDMA, and the MLC This work was supported by the Nation Natural Science
scheme always outperforms the FSK scheme. For example, for Foundation of China (61471308, 61001142, and 61301097) and
8 users and BER = 0.01, the MLC scheme is about 1-dB better the Fundamental Research Funds for the Central Universities
than FSK scheme. (2013121023).
References
6. Conclusions
[1] D. Pompili, T. Melodia, and I. F. Akyildiz, “A CDMA-based
Code Division Multiple Access (CDMA) is the most promis- medium access control for UnderWater acoustic sensor net-
ing physical layer and multiple access technique for UW- works,” IEEE Transactions on Wireless Communications, vol. 8,
ASNs. Since the chirp spread spectrum is resistant to multi- no. 4, pp. 1899–1909, 2009.
path fading and is of low Doppler sensitivity, it is used in UW- [2] F. Favaro, P. Casari, F. Guerra, and M. Zorzi, “Data upload from
ASNs widely. And the technology of multiple access using a static underwater network to an AUV: polling or random
chirp signals was referred to. To improve the performance of access?” in Proceedings of the Yeosu OCEANS, pp. 1–6, IEEE,
multipath access, a new scheme with MLC signal has been Yeosu, Republic of Korea, May 2012.
proposed in terrestrial sensor networks but not in UW-ASNs. [3] J. G. Proakis, E. M. Sozer, J. A. Rice, and M. Stojanovic, “Shallow
In this paper, the scheme of multilinear chirp-CDMA (MLC- water acoustic networks,” IEEE Communications Magazine, vol.
CDMA) used in UW-ASNs is proposed; the comparison of 39, no. 11, pp. 114–119, 2001.
LFM (linear chirp signal) and MLC signals shows that MLC [4] I. F. Akyildiz, D. Pompili, and T. Melodia, “Underwater acoustic
is a better choice. And MCR-FrFT, a detection technique for sensor networks: research challenges,” Ad Hoc Networks, vol. 3,
MLC-CDMA, is proposed too. The MCR-FrFT is a receiver- no. 3, pp. 257–279, 2005.
based scheme, which reduces both the complexity and the [5] C. He, J. Huang, Q. Zhang, and K. Lei, “Reliable mobile
computations of the system. The simulation results show that underwater wireless communication using wideband chirp
our MLC-CDMA system with MCR-FrFT signal detection signal,” in Proceedings of the WRI International Conference on
technique achieves higher performance of multiple access in Communications and Mobile Computing (CMC ’09), pp. 146–
UW-ASNs than FSK-CDMA. Further, maybe we can use the 150, Yunnan, China, January 2009.
detection technique to estimate the time-delay and use it in [6] H. Shen, S. Machineni, C. Gupta, and A. Papandreou-Suppap-
the location system. pola, “Time-varying multichirp rate modulation for multiple
access systems,” IEEE Signal Processing Letters, vol. 11, no. 5, pp. [23] L. Shen, Q. Yin, M. Lu et al., “Linear FM signal parameter esti-
497–500, 2004. mation using STFT and FRFT,” Chinese Journal of Electronics,
[7] M. T. Kheirabadi and M. M. Mohamad, “Greedy routing in vol. 22, no. 2, pp. 301–307, 2013.
underwater acoustic sensor networks: a survey,” International [24] M. Stojanovic, “Underwater acoustic communications: design
Journal of Distributed Sensor Networks, vol. 2013, Article ID considerations on the physical layer,” in Proceedings of the 5th
701834, 21 pages, 2013. Annual Conference on Wireless on Demand Network Systems
[8] B. B. Chen, Research and implementation of medium-range and Services (WONS ’08), pp. 1–10, Garmisch-Partenkirchen,
underwater communication system based on chirp signal [MS Germany, January 2008.
dissertation], Xiamen University, Xiamen, China, 2012.
[9] F. Y. Meng and X. M. Gu, “A combined chirp signal modulation
technique for multiple access system,” Information Technology
Journal, vol. 10, no. 2, pp. 416–421, 2011.
[10] P. Zhang, H. Liu, and J. H. Hu, “A novel multiple-access
scheme for chirp UWB,” in Proceedings of the IEEE Wireless
Communications and Networking Conference (WCNC ’07), pp.
1631–1635, Kowloon, Hong Kong, March 2007.
[11] P. Moungnoul, T. T. Hung, and T. Paungma, “Investigation of
multi-linear chirp FH-CDMA over fading channel model,” in
Proceedings of the 5th International Conference on Information,
Communications and Signal Processing (ICICS ’05), pp. 1480–
1484, Bangkok, Thailand, December 2005.
[12] S. E. El-Khamy, S. E. Shaaban, and E. A. Thabet, “Efficient
multiple-access communications using multi-user chirp modu-
lation signals,” in Proceedings of the 4th International Symposium
on Spread Spectrum Techniques and Applications, (ISSSTA’96),
pp. 1209–1213, Mainz, Germany, September 1996.
[13] X. W. Wang, X. Li, and M. R. Fei, “Anti-noise performance of
chirp spread spectrum system,” Communications Technology,
vol. 42, pp. 36–37, 50, 2009.
[14] K. Y. Wang, “A novel demodulation method of chirp-rate
modulation,” Microcomputer & Its Applications, vol. 30, pp. 71–
76, 2011.
[15] Y.-L. Wan, Q. Si, Y.-X. Lu, H. Wang, and X.-G. Wang, “Study
of dechirp pulse compression processing method for wideband
signals,” Acta Aeronautica et Astronautica Sinica, vol. 27, no. 5,
pp. 917–921, 2006.
[16] T. Leszczynski, “Matched filtration of LFM signals with non-
zero initial frequency,” Acta Physica Polonica A, vol. 124, no. 3,
pp. 494–497, 2013.
[17] E. Sejdić, I. Djurović, and L. Stanković, “Fractional Fourier
transform as a signal processing tool: an overview of recent
developments,” Signal Processing, vol. 91, no. 6, pp. 1351–1369,
2011.
[18] J. Song, Y. X. Wang, and Y. F. Liu, “Iterative interpolation for
parameter estimation of LFM signal based on fractional Fourier
transform,” Circuits, Systems and Signal Processing, vol. 32, no. 3,
pp. 1489–1499, 2013.
[19] R. Chen and Y. M. Wang, “Universal FRFT-based algorithm
for parameter estimation of chirp signals,” Journal of Systems
Engineering and Electronics, vol. 23, no. 4, pp. 495–501, 2012.
[20] F. Liu, Y. Huang, R. Tao, and Y. Wang, “Chirp-rate resolution of
fractional fourier transform in multi-component LFM signal,”
Journal of Beijing Institute of Technology, vol. 18, no. 1, pp. 74–
78, 2009.
[21] L. B. Almeida, “Product and convolution theorems for the
fractional Fourier transform,” IEEE Signal Processing Letters,
vol. 4, no. 1, pp. 15–17, 1997.
[22] H. Y. Hao, “Multi component LFM signal detection and param-
eter estimationbased on EEMD-FRFT,” Optik, vol. 124, no. 23,
pp. 6093–6096, 2013.
http://dx.doi.org/10.1155/2015/746919
Research Article
Doppler Estimation Based on Frequency Average and
Remodulation for Underwater Acoustic Communication
Chan-Ho Hwang,1 Ki-Man Kim,1 Seung-Yong Chun,2 and Sang-Kook Lee2

1
Department of Radio Communication Engineering, Korea Maritime and Ocean University, Busan 606791, Republic of Korea
2
Agency for Defense Development, Changwon, Republic of Korea
Correspondence should be addressed to Ki-Man Kim; kimkim@hhu.ac.kr
Received 2 March 2015; Accepted 15 April 2015
Academic Editor: Juan Carlos Cano
Copyright © 2015 Chan-Ho Hwang et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
Underwater acoustic communication channel is time varying and has Doppler effect due to wind, sea surface roughness, and
platform motion. We have to recognize the changing channel state and apply it to communication technique for increasing
transmission efficiency in the underwater acoustic channel. In this paper, we propose more reliable Doppler estimation method
based on frequency average and remodulation. We conducted the simulation and sea trial to evaluate the performance of the
proposed method. When the channel coding technique was not applied, the uncoded bit error rate performance of the proposed
method was improved about maximum 50.3% compared with conventional method.
1. Introduction the Doppler shift frequency is obtained according to the

matched filter branch that has the highest energy. In this
Underwater sensor networks (USNs) can be used for environ- method, the accuracy of the estimated value is dependent on
ment monitoring, disaster prevention, and military surveil- the number of matched filters. But increasing the number of
lance [1, 2]. Unlike terrestrial communication, underwater matched filter also increases the complexity of this method.
acoustic communication utilizes sound waves. The sound Another approach to measuring the Doppler-induced fre-
speed, which is slower than radio waves, is dependent on the quency offset is performed in the frequency domain. The
water temperature and the salinity. The sound transmission received signal is converted to the frequency domain and
path is changed by refraction according to the gradient of the received carrier frequency is estimated. However, this
vertical sound velocity [3]. When sound waves are passed method is difficult to use as the carrier frequency estimation
through the underwater, they are affected by attenuation, requires a high signal-to-noise ratio, and the computational
reflection of bottom and surface, scattering, ambient noise, load increases when obtaining a high resolution in frequency
and the Doppler effect caused by movement of the transmitter domain [7]. Another uses the time difference of arrival
and the receiver. Of these factors, the Doppler effect affects between the preamble and postamble signals [9, 10]. The time
the correlation time of the channel related to the transmission difference of arrival is compared to the period of the trans-
efficiency [4, 5]. These influences thus cannot be neglected in mitted packet to estimate the Doppler shift ratio. However,
the design of underwater acoustic communication systems. this method should be used in an environment with a fixed
To increase transmission efficiency, we need to estimate the Doppler shift between preamble and postamble signals. The
changing Doppler frequency and compensate for distortion. other uses a sample correlation to generate the frequency
In practice, several Doppler frequency estimation tech- offset estimate. The frequency offset can be estimated by
niques have been used for underwater acoustic communi- complex multiplications. As such, this method has a low
cation [6, 7]. One uses a bank of the matched filter which computational complexity but does not perform well for low
corresponds to the different Doppler shift frequency [8]. The signal-to-noise ratios. Moreover, the above methods are not
received signal is passed through the matched filters, and then suitable for use in the time-varying channel environment.
In this paper, we propose Doppler shift frequency esti- Doppler shift frequency [6]. For a continuous time signal with
mation methods to use in the time-varying channel. Two wide band, the ambiguity function is given as
kinds of methods are proposed. One is a Doppler estimation ∞
algorithm that has stable performance as a part of the 𝐴 𝑠 (𝜏, Δ) = (1 + Δ) ∫ 𝑠 (𝑡 − 𝜏) 𝑠 ((1 + Δ) 𝑡) 𝑑𝑡. (2)
communication system. From an overall communication −∞
system perspective, the other is a demodulation scheme that Here, 𝜏 is the delay and Δ is the Doppler shift frequency. If
is different from the conventional communication system. the received signal 𝑟(𝑡) is considered, we can define the cross-
There are differences between the conventional communica- ambiguity function:
tion system and the proposed system. First, the conventional
∞
communication system uses a Doppler estimation method
𝐴 𝑟𝑠 (𝜏, Δ) = (1 + Δ) ∫ 𝑟 (𝑡 − 𝜏) 𝑠 ((1 + Δ) 𝑡) 𝑑𝑡. (3)
based on samples correlation [11]. This estimation algorithm −∞
exploits only one correlation lag. As such, the estimated
value can be sensitive to variation in correlation. On the To estimate the Doppler shift, we have to search in Δ with
other hand, the proposed system uses the frequency average time alignment and find the maximum value of the cross-
estimation, which is based on multiple sample correlations. ambiguity function:
Because the frequency average estimation uses all correlation ∞
lag, the estimated Doppler frequency can maintain the 𝐴 𝑟𝑠 (0, Δ) = (1 + Δ) ∫ 𝑟 (𝑡) 𝑠 ((1 + Δ) 𝑡) 𝑑𝑡. (4)
−∞
consistency even if there is a variation in correlation. Second,
the conventional communication system simply demodulates When the value of (4) is maximized, Δ is estimated as the
the payload based on the estimated Doppler frequency in Doppler shift in the received signal. We conduct the corre-
the training sequence section and for this reason is sensitive lation between incoming signal and Doppler-shifted replicas
to time-varying channel. However, the proposed system can of the transmitted waveform called correlator-bank. There
estimate the channel parameters such as Doppler frequency is a constant frequency interval among each replica. When
during the data frame and improve the communication the frequency interval between replicas is narrow, Doppler
performance. frequency is estimated more accurately, but the amount
We present the ambiguity function and decision-directed of calculation is increased. Figure 1 shows the structure of
estimation. We also propose a more accurate Doppler esti- correlator-bank.
mation method that is based on decision-directed estimation
and propose a remodulation method that can update the
3. Proposed Methods for Doppler Estimation
channel parameters for the equalizer operation and estimate
the Doppler frequency during the data frame using two In this section, we propose two techniques for improving
Doppler estimation techniques. A simulation based on the communication performance in the Doppler environment.
Bellhop modeling and sea experiment was carried out to One is the Doppler estimation method which is less sensitive
compare the performance of the proposed method with a to variation than previous methods. The other is the remod-
conventional nonrecursive Doppler estimation. ulation method, which can estimate the varying channel
The remainder of this paper is organized as follows. parameter during the data frame.
Section 2 presents the Doppler estimation technique based
on ambiguity function which is using Doppler estimator at 3.1. Decision-Directed Estimation Based on Frequency Average.
first, and Section 3 presents proposed Doppler estimation Assume the modulation is PSK (phase shift keying) with
based on the previous method for second Doppler estima- perfect timing synchronization [13]. The received signal after
tor and proposes the remodulation method. In addition, matched filtering is given by
we present the frame formation and receiver structure for
the remodulation method. Section 4 applies the proposed 𝑟 (𝑘) = 𝑎𝑘 𝑒𝑗(2𝜋Δ𝑓𝑘𝑇+𝜃) + 𝑛 (𝑘) , (5)
method to the simulation and experimental data. Finally,
Section 5 gives a summary and conclusion. where 𝑎𝑘 is the 𝑘th symbol amplitude, Δ𝑓 is the Doppler shift
frequency, 𝑇 is the symbol period, 𝜃 is the carrier phase offset,
and 𝑛(𝑘) is the white Gaussian noise with zero mean. In the
2. Doppler Estimation Technique Based on training mode, modulation can be removed by multiplying
Ambiguity Function 𝑎𝑘∗ and ∗ means complex conjugate. The received signal with
The ratio between signal propagation speed and the source frequency and phase component is given by
relative velocity is defined as the Doppler shift frequency
𝑧 (𝑘) = 𝑎𝑘∗ 𝑎𝑘 𝑒𝑗(2𝜋Δ𝑓𝑘𝑇+𝜃) + 𝑎𝑘∗ 𝑛 (𝑘) . (6)
[12]. When the Doppler shift frequency is Δ, the signal with
Doppler shift frequency can be expressed as follows: 𝑧(𝑘) can be viewed as a complex sinusoid superimposed on
white Gaussian noise. Using (6), multiplying 𝑧(𝑘) by 𝑧∗ (𝑘 − 𝑖)
𝑟 (𝑡) = 𝑠 ((1 + Δ) 𝑡) , (1) can be expressed as follows:
where 𝑟(𝑡) is the received signal and 𝑠(𝑡) is the transmitted 1 𝐿−1
signal. The ambiguity function is represented by the matched 𝜒 (𝑖) = ∑ 𝑧 (𝑘) × 𝑧∗ (𝑘 − 𝑖) 𝑖 = 1, 2, 3, . . . , 𝑁. (7)
filter response against delay of the incoming signal and 𝐿 − 𝑖 𝑘=𝑖
Correlator 1 The first recursive data frame is demodulated using the

estimated channel parameters at the training section. Then,
the demodulated data frame is remodulated. The remodu-
Correlator 2 lated first recursive frame is used like a training signal for
demodulation of the second recursive frame. This process is
repeated. Figure 2 shows the proposed method.
Incoming Select
Correlator 3
signal maximum
3.3. Receiver Structure. We present the receiver structure
based on PSK (phase shift keying) in Figure 3. There is a
..
. buffer at the entrance of the receiver. An amount of the
received signal equal to the recursive frame size is saved into
the buffer. The lever behind the buffer is raised toward the
Correlator M
buffer after the training signal and initial first recursive data
frame are entered into receiver. Therefore, we assume that the
Figure 1: Structure of correlator-bank.
receiver knows the training length and recursive frame size
for operating the lever. There are other buffers in the receiver.
𝜒(𝑖) is the sample correlation. Here, 𝐿 is the number One saves the training sequence, and the other is used for
of train symbols, and 𝑁 is a correlation range not greater saving the demodulated recursive data frame. The receiver
than 𝐿/2. The frequency offset is estimated by using sample consists of frequency estimators, matched filter, PLL (phase
correlations [11]. Consider locked loop), DFE (decision feedback equalizer) based on
RLS (recursive least square) algorithm [17], and remodulator.
𝑁−1
1 The receiver operating order is as follows.
𝑓̂𝑑 = arg { ∑ [𝜒 (𝑖 + 1) 𝜒 (𝑖)∗ ]} . (8)
2𝜋𝑇 𝑖=1 A The training sequence is modulated and then consists
of correlator-bank with different Doppler shift repli-
The arg{ } means phase. Equation (8) is the previous
cas.
Doppler estimation algorithm. Since (8) is calculated by
using one index interval correlation, frequency offset can be B The coarse Doppler frequency of the received training
sensitive to variation of correlation values. The sensitivity can signal is estimated at the first frequency estimator.
be reduced by averaging the frequency from correlations of all C The received signal is downconverted with the esti-
indexes. The frequency offset using average is given by mated frequency.
1 𝑁−1 ̃ D The downconverted signal is passed through the
𝑓̂𝑑 = ∑ 𝑓 (𝑗) , matched filter and sampled to symbol interval.
𝑁 − 1 𝑗=1 𝑑
(9) E Second Doppler frequency estimation and recovery
𝑁−1 are conducted on the symbol spaced signal.
1
𝑓̃𝑑 (𝑗) = arg { ∑ [𝜒 (𝑖 + 𝑗) 𝜒 (𝑖)∗ ]} . F The distorted phase of signal is recovered by PLL.
2𝜋𝑇𝑗 𝑖=1
G The signal distortion caused by multipath propaga-
3.2. Frame Formation for Remodulation. In a conventional tions is removed by DFE.
communication system, the communication packet consists H The equalizer outputs are passed through the hard
of training section and data frame (or payload) section decision.
[14–16]. When a training section in which transmitter and 0 Data symbols other than demodulated training sym-
receiver are already known is received, the training signal bols are saved into the recursive frame buffer.
is used to update the equalizer tap coefficients for recovery
of the distorted signal through multipath propagation. This 1 The levers are raised toward recursive frame buffer.
is also used to estimate the Doppler frequency. However, 2 Recursive data symbols at recursive frame buffer are
after training mode is finished, the difference between the used like a train sequence.
estimated value and the actual value is increased, and this 3 A ∼ 0 procedure is performed.
is caused by time variance of the channel. These channel
variances reduce the reliability of communication. For this The above procedure is an example of a payload which
reason, we should continuously estimate and compensate is divided into two recursive data frames. If the number of
the Doppler frequency during data frame. In a conventional recursive data frames is increased, the number of buffers at
communication system, the payload is only demodulated the entrance of the receiver is also increased.
using the estimated channel parameters such as equalizer tap
coefficients and Doppler frequency in the training section. 4. Simulation and Experimental Results
By comparison, the proposed remodulation method is able to
estimate the varying channel parameter during the payload. 4.1. Simulation. Before communication performance was
The basic idea is as follows. The received payload is evaluated, we examined the difference between the sec-
divided into several data frames called recursive data frames. ond Doppler estimation techniques. At environment with
Channel parameter
Recursive frame Recursive frame Recursive frame

Training
#1 #2 #3
Demodulation
Remodulation
Channel parameter

Training
#1 #2 #3
Demodulation
Remodulation
Channel parameter

Training
#1 #2 #3
Demodulation
Figure 2: The proposed remodulation method.
Symbol spaced sampling

After training
Buffer ⑩ ③ ④
Received Matched
signal filter ⑤
Frequency Frequency
recovery recovery
based on cross- based on frequency
ambiguity ② average
function
Phase
recovery
⑥
① Decision
feedback
⑪ equalizer
⑦
⑨ After training
Recursive frame buffer
Hard
Remodulation decision
⑩
⑧
Symbol output
Training sequence
Figure 3: Receiver structure.
the Doppler effect, we estimated the Doppler frequency is result of a simulation for 100 times. Overall, the proposed
using both first estimator and second estimator. When 8 Hz method had an averagely small frequency offset error and
Doppler frequency existed, assumed frequency offset was narrow deviation. At 5 dB, the standard deviation of the
already calculated at the first Doppler estimator. Therefore, previous method and the proposed method was 0.0245 and
the estimated value of the second estimator is zero in the ideal 0.0108, while the mean value of the previous method and the
case. However, simulation results show that this estimation proposed method was 0.0757 and 0.0291, respectively. Both
has an error. Because we use training symbols at (6), the deviation and mean error of the proposed method are smaller
product of the received symbol with noise and complex than in the previous method.
conjugate training symbol makes an error as a symbol We executed the simulation using VirTEX (Virtual Time
amplitude difference between those symbols. This being so, series Experiment) [18] based on the Bellhop to evaluate
variance of estimation is large in a low SNR environment. We the performance of the proposed remodulation method.
can show the difference between the two cases in Figure 4 that To create the simulation environment, we used an actual
50 50
40 40
Number of occurrences
Number of occurrences
30 30
20 20
10 10
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Frequency offset (Hz) Frequency offset (Hz)
5 dB 15 dB 5 dB 15 dB
10 dB 20 dB 10 dB 20 dB
(a) (b)
Figure 4: Estimated frequency offset at second Doppler estimation: (a) previous method, (b) proposed method.
sound speed profile in Korea’s South Sea. We assumed 4.2. Sea Trial. We executed the experiment during June 2014
that surface and bottom were flat. The bottom information near Geoje island, Korea, to evaluate the proposed method.
for the simulation was as follows. Sound speed at bottom The setup for sea trial is shown in Figure 7. The water depth
was 1600 m/s, bottom density was 1.23 g/cm3 , and bottom of the experimental area was about 60 m, and the transmitter
attenuation was 0.95 dB/𝜆. The water depth was 85 m, and and the receiver were located at 20 m and 26 m below the
both the transmitter and the receiver were located at 60 m. ocean’s surface, respectively. The wind speed was 9 m/s, and
We also assume that the 5 Hz Doppler frequency occurred. wave height was 2.5 m. The ship towing a transmitter drifted
Figure 5 shows the sound speed profile, eigenrays, and chan- because the ship’s engine was turned off to avoid engine noise.
nel impulse response. Distance between transmitter and receiver was getting closer
Source signal was modulated to QPSK. Bit rate was 1 kbps to 1 km.
and carrier frequency and sampling frequency were 16 kHz The source signal has the same packet as described in
and 192 kHz, respectively. We disposed of the 0.2 sec silence Table 1. The carrier frequency is 5 kHz and sampling fre-
between training section and data section to avoid delay quency is 25 kHz. Channel coding is not applied to the signal.
distortion. Several parameters for communication are given Before transmitting the communication signal, we got the
in Table 1. scattering function shown in Figure 8 using the modulated
The performance regarding recursive frame length and M-sequence. We confirmed that Doppler frequency was from
SNR (signal-to-noise ratio) is shown in Figure 6. When 4 to 7 Hz and delay spread was about 50 ms.
recursive frame length is 1,000 symbols, the uncoded BER We got the uncoded BER in order to compare the
(bit error rate) result means the conventional communication performance of the proposed method with the conventional
packet method. The communication performance became system. Figure 9 shows the uncoded BER per 100 bits.
improved when the recursive frame length was decreased Since uncoded BER is cumulated, the last value means total
and SNR was increased. In particular, when SNR was 5 dB, uncoded BER. The data packet consists of 1,000 symbols
the BER of the proposed method was 3.7% lower than in (2,000 bits). Therefore, the result of 1,000 symbols belongs
the conventional method. In all SNR cases, the performance to the conventional data packet system. We can show that
of the proposed method was better than the conventional the performance of the proposed method was better than
method. conventional method. The shorter the recursive frame length,
We compared the performance using frequency average the better the performance. In particular, when the number
method at second Doppler estimator with previous esti- of recursive frames was 200 symbols, the uncoded BER was
mation. When SNR was 15 and 20 dB, the two cases had reduced to a minimum of 4.1%.
similar performance. However, the communication system We also compared the performance using the frequency
using proposed estimation had better performance, up to a average with the previous method at the second Doppler fre-
maximum 0.4% in 5 and 10 dB cases. We can confirm that quency estimation. Figure 9(a) shows the uncoded BER with
the proposed estimation method is advantageous in terms of second frequency estimation based on the previous method.
communication reliability. The outcomes were similar to those obtained through the
×10−4
0
0
14
10
10
20 12
20
30 10
30
Depth (m)
Depth (m)
Amplitude
40
40 8
50
50 6
60
60
4
70
70
80 2
80
90
1500 1510 1520 0 200 400 600 800 1000 0.65 0.7 0.75
Sound speed (m/s) Range (m) Arrival time (sec)
(a) (b) (c)
Figure 5: Simulation channel characteristic, (a) sound speed profile, (b) eigenrays, and (c) channel impulse response.
Table 1: Parameters for simulation.

Element Contents
Modulation QPSK (quadrature phase shift keying)
Data length 1000 symbols
Packet structure
Training sequence length 256 symbols
Recursive frame length 200, 250, 500 symbols
The number of replicas 200
1st Doppler estimator
Frequency interval 0.1
Phase lock loop Gamma 0.1
Feed-forward/feed-backward length 10/20
Decision feedback equalizer
Delta/lambda 999/0.999
frequency average method, except that recursive frame white Gaussian noise and the Doppler effect. We showed
length was 250 symbols. Compared to the frequency average that the uncoded BER of proposed method was lower than
method, estimation frequency per frame was different at third that of the conventional packet method. The communication
recursive frame, which had large error. This result indicates performance became improved when the recursive frame
that the performance of the communication system that used length was shorter. We also could confirm the performance
frequency average estimation is more reliable. difference based on whether or not the proposed frequency
average method at second frequency estimator was used. The
5. Conclusions results of the sea experiment showed the potential for an
improvement in performance when the communication sys-
In underwater acoustic communication, the Doppler effect tem used the proposed remodulation method. When recur-
decreases the transmission efficiency and distorts the source sive frame length is 250 symbols, communication perfor-
signal. The conventional communication packet system in mance with the previous frequency estimation method dete-
which payload is only demodulated using the estimated chan- riorated, whereas the proposed frequency average method
nel information in training mode is sensitive to time variance showed stable performance. In particular, we confirmed that
of the channel. In this paper, we proposed the more reliable the communication performance was improved by up to
Doppler estimation method based on the existing technique. 50.3% using the two proposed methods.
We confirmed that the proposed estimation method resulted
in a smaller estimation error. Furthermore, we also proposed
a remodulation method that is able to estimate the time- Conflict of Interests
varying channel parameter during the payload. We compared
the performance of the proposed remodulation method with The authors declare that there is no conflict of interests
the conventional method in the simulation environment with regarding the publication of this paper.
5 −10 0
Doppler frequency (Hz)

−8 −1
−6 −2
−4 −3
4 −2 −4
−5
(dB)
0
2 −6
Uncoded BER (%)
3 4 −7
6 −8
8 −9
10 −10
2 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Delay time (s)
1 Figure 8: Scattering function.
0
5 10 15 20 20
SNR (dB)
(a)
5 15
4 Cumulative uncoded BER (%)

10
Uncoded BER (%)
5
2
0
1 0 5 10 15 20
Segment
0 (a)
5 10 15 20
20
SNR (dB)
1000 symbols 250 symbols

15
Cumulative uncoded BER (%)
(b)
Figure 6: Uncoded bit error rate second Doppler estimation (a)

using the previous method, (b) using the frequency average method. 10
1∼1.5 km
Wind speed: 9 m/s 5
Buoy Tx
Rx Wave height: 2.5 m
0
0 5 10 15 20
20 m Segment
26 m Transmitter
Receiver 60 m
1000 250
500 200
(b)
Weight
20 kg
Bottom Figure 9: Comparison of performance by recursive frame length
with second Doppler estimation (a) using the previous method, (b)
Figure 7: Setup for sea trial. using the frequency average method.
Acknowledgments [15] J. K. Cavers, “An analysis of pilot symbol assisted modulation

for Rayleigh fading channels,” IEEE Transactions on Vehicular
This work was supported in part by Defense Acquisition Technology, vol. 40, no. 4, pp. 686–693, 1991.
Program Administration and Agency for Defense Develop- [16] S. Sampei and T. Sunaga, “Rayleigh fading compensation for
ment under the Contract UD130007DD and by Ministry QAM in land mobile radio communications,” IEEE Transac-
of Education and National Research Foundation of Korea tions on Vehicular Technology, vol. 42, no. 2, pp. 137–147, 1993.
(NRF) through the Human Resource Training Project for [17] S. Haykin, Adaptive Filter Theory, Prentice Hall, 2009.
Regional Innovation. [18] M. Siderius and M. B. Porter, “Modeling broadband ocean
acoustic transmissions with time-varying sea surfaces,” Journal
References of the Acoustical Society of America, vol. 124, no. 1, pp. 137–150,
2008.
[1] I. F. Akyildiz, D. Pompili, and T. Melodia, “Challenges for effi-
cient communication in underwater acoustic sensor networks,”
ACM SIGBED Review, vol. 1, no. 2, pp. 3–8, 2004.
[2] M. Chitre, S. Shahabudeen, and M. Stojanovic, “Underwater
acoustic communication and networks: recent advances and
future challenges,” Marine Technology Society Journal, vol. 42,
no. 1, pp. 103–116, 2008.
[3] X. Lurton, An Introduction to Underwater Acoustics: Principles
and Applications, Springer, 2002.
[4] T. C. Yang, “Measurements of temporal coherence of sound
transmissions through shallow water,” Journal of the Acoustical
Society of America, vol. 120, no. 5, pp. 2595–2614, 2006.
[5] G. D. Durgin, Space-Time Wireless Channels, Prentice Hall, New
York, NY, USA, 2003.
[6] B. S. Sharif, J. Neasham, O. R. Hinton, and A. E. Adams,
“Computationally efficient Doppler compensation system for
underwater acoustic communications,” IEEE Journal of Oceanic
Engineering, vol. 25, no. 1, pp. 52–61, 2000.
[7] R. Diamant, A. Feuer, and L. Lampe, “Choosing the right
signal: doppler shift estimation for underwater acoustic signals,”
in Proceedings of the 7th ACM International Conference on
Underwater Networks and Systems (WUWNet ’12), Los angles,
Calif, USA, November 2012.
[8] O. Rabaste and T. Chonavel, “Estimation of multipath channels
with long impulse response at low SNR via an MCMC method,”
IEEE Transactions on Signal Processing, vol. 55, no. 4, pp. 1312–
1325, 2007.
[9] L. Wan, Z. Wang, S. Zhou, T. C. Yang, and Z. Shi, “Performance
comparison of Doppler scale estimation methods for under-
water acoustic OFDM,” Journal of Electrical and Computer
Engineering, vol. 2012, Article ID 703243, 11 pages, 2012.
[10] L. Wu, J. Trezzo, D. Mirza et al., “Designing an adaptive acoustic
modem for underwater sensor networks,” IEEE Embedded
Systems Letters, vol. 4, no. 1, pp. 1–4, 2012.
[11] S. Tao, L. Xu-Wen, and C. Xiao-Ting, “Decision-directed esti-
mation of carrier frequency and phase for burst PSK trans-
mission,” in Proceedings of the 5th International Conference on
Wireless Communications, Networking and Mobile Computing
(WiCOM ’09), pp. 1–4, Beijing, China, September 2009.
[12] J. Rosen and L. Q. Gothard, Encyclopedia of Physical Science,
Facts on File, 2009.
[13] S. Tao, H.-J. Liu, X.-T. Chen, X.-W. Liang, and G.-D. Zhao,
“An improved frequency estimator for synchronization of PSK
signals with large frequency offset,” in Proceedings of the 4th
International Conference on Wireless Communications, Net-
working and Mobile Computing (WiCOM ’08), pp. 1–3, Dalian,
China, October 2008.
[14] H.-B. Li, Y. Iwanami, and T. Ikeda, “Symbol error rate analysis
for MPSK under Rician fading channels with fading compensa-
tion based on time correlation,” IEEE Transactions on Vehicular
Technology, vol. 44, no. 3, pp. 535–542, 1995.
http://dx.doi.org/10.1155/2015/481576
Research Article
High Throughput Receiver Structure for
Underwater Communication
Chang-Uk Baek and Ji-Won Jung

Department of Radio Communication Engineering, Korea Maritime and Ocean University, Busan 606-791, Republic of Korea
Correspondence should be addressed to Ji-Won Jung; jwjung@kmou.ac.kr
Received 24 April 2015; Accepted 28 June 2015
Academic Editor: Dongkyun Kim
Copyright © 2015 C.-U. Baek and J.-W. Jung. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
Acoustic channels are characterized by long multipath spreads that cause intersymbol interference. The way in which this fact
influences the design of the receiver structure is considered in this paper. To satisfy performance and throughput requirements,
we propose a consecutive iterative BCJR equalization scheme. To achieve a low error performance, we resort to the powerful BCJR
equalization algorithms to iteratively update probabilistic information between inner decoder and outer decoder. Also, to achieve
a high throughput, we divide a long packet into a group of small consecutive packets, estimate channel information of the current
packets, and use it to decode the next packets. Based on an experimental channel response, we confirm that the performance is
indeed improved for long packet size.
1. Introduction only transmit small packets, which reduces the throughput.

To achieve a high throughput, we divide a long packet into
It is well known that underwater channels are often hostile a group of small consecutive packets and use the estimated
for underwater (UW) sensor communications, which impose channel information of previous packets to compensate for
three major obstacles for coherent transceivers. One is the the current and next packets. In this paper, we employ an
excessive multipath delay spread in a UW channel, which iterative receiver structure with fine-tuned parameters to
usually causes the intersymbol interference (ISI). Another process experimental data from a fixed source to a fixed
obstacle is the Doppler shift due to the relative motion receiver at the data transmission rate of 1 𝑘-symbol/s. The
between the source and the receiver, which causes compres-
results indicate that the proposed algorithm works effectively
sion or dilation on the received signals. The last one is the fast
well and indicate how much coding gains can be obtained as
time varying phase drift due to random nature of the UW
acoustic channels. the iteration number increases.
Various methods to cope with the multipath effect have
been developed. A well-known method to counteract ISI 2. Generic Transmission Model of
is the decision feedback equalizer (DFE), which has been Underwater Communication
used in many UW sensor communication applications [1, 2].
However the use of DFE has difficulties when the multipath The baseband model of the generic transmission system is
with a number of arrivals has equal strength or low SNR [3]. shown in Figure 1. It shows that the iterative linear equalizer
The other way to cope with ISI is to use an iterative equalizer is a decision feedback equalizer, which deals with the outer
which consists of an outer loop in addition to the inner loop code of the receiver. An inner code consists of convolutional
BCJR decoder in the receiver. The assembly utilizes the error codes. The information data to be transmitted are encoded by
correcting capability of the convolutional codes to get an one-half rate convolutional codes with the constraint length
efficient equalizer [4, 5]. Alternatively, to cope with multipath of 7.
effect, we adjust the packet length according to the channel Convolutional codes are used extensively in numerous
coherence time. Due to the very short coherence time, we can applications in order to achieve reliable data transmission,
Convolutional Symbol Dk Packet

Interleaver
encoder (2,1,7) mapper generator
Underwater
channel
I
LD
e + Le DFE Packet
Deinterleaver
equalizer synchronization
−
BCJR P󳰀
decoder
− P
+ LIc Doppler and
Interleaver
LD ̂ ,𝜃
f phase estimation
c d
Turbo equalizer
Figure 1: Baseband model of transmission system with turbo equalization.
including both wireless and underwater communication. A where, 𝛾 is a constant value and sgn{⋅} is a sign function
popular convolutional code with a constraint length of 7 and defined by
a rate of 1/2 is widely used. For example, the (2, 1, 7) convolu-
tional code with the generator matrix of (171, 133)octal is used {
{ −1 when 𝑥 < 0
{
{
in [6]. The equalizer and the BCJR decoder are connected
sgn {𝑥} = {0 when 𝑥 = 0 (3)
through the interleaving and deinterleaving functions that {
{
{
{+1 when 𝑥 > 0.
update each other’s information recursively. The inner coded
bits are then subtracted from the input and interleaved. The
interleaved output is canceled a posteriori from the pro- The value of 𝐿𝐷𝑒 after the interleaver is computed as 𝐿𝐼𝑒 −𝐿𝐼𝑐
ceeding received signal. The interleaving function helps the and then input to the turbo decoder. The estimated extrinsic
receiver convergence. Generally, the decoding method of the value of 𝐿𝐷𝑐 at the decoder output is given by
convolutional code is a Viterbi decoding using a hard decision
value. However, to improve the receiver performance through 𝑃 (𝑥 = +1)
𝐿𝐷𝑐 = log . (4)
the iterations with DFE, a soft decision value is required. 𝑃 (𝑥 = −1)
Therefore, we use the BCJR algorithm with the soft decision
value. The BCJR algorithm is a well-known maximum a The extrinsic value 𝐿𝐷𝑐 is an error correction term, from which
posteriori probability decoding algorithm, which has been the post probability is calculated. The reinterleaving of the
proposed earlier for point-to-point communication applica- computed value, 𝐿𝐷𝑐 − 𝐿𝐷𝑒, is input to the DFE; then 𝐿𝐼𝑐 is
tions [7]. updated in order to compensate for the errors.
Let 𝐿𝐼𝑒 be the output value of DFE as the estimated
extrinsic value from the received signal. Let 𝑦[𝑘] be the 2.1. General Packet Structure. The packet structure is shown
equalizer input at time 𝑘; then the output of the DFE at time in Figure 2. A packet is preceded by a linear frequency
𝑘, denoted by 𝐿𝐼𝑒 [𝑘], is given by modulation beginning (LFMB) signal to indicate that the
packet is about to start. After silence period, the packet
𝑁𝑏 −1 𝑁𝑎
starts. The packet consists of preamble (𝑃) and payload (𝐷)
̂ 𝐼 [𝑘 − 𝑗] ,
𝐿𝐼𝑒 [𝑘] = ∑ 𝑐𝑖 [𝑘] 𝑦 [𝑘 − 𝑖] − ∑ 𝑏𝑗 [𝑘] 𝐿 (1)
𝑒 fields. The preamble of 256 symbols is used for the initial
𝑖=0 𝑗=0
channel and Doppler estimation. The payload field contains
500 coded QPSK symbols. The packet ends with another LFM
where 𝑐𝑖 [𝑘] (𝑖 = 0, 1, . . . , 𝑁𝑐−1 ) are the forward equalizer taps
signal identified as LFME (LFM End). During transmission,
at time 𝑘, 𝑏𝑗 [𝑘] (𝑗 = 0, 1, . . . , 𝑁𝑏 ) are the feedback taps at time
to avoid interpacket interference, the packet size should
𝑘, and 𝐿̂ 𝐼 [𝑘] is the slicer output, which is the constellation be small and the gap sufficiently long. These requirements
𝑒
point closest to 𝐿𝐼𝑒 [𝑘], as in [8, 9]. 𝑒𝑠 [𝑘] is the Sato error given apparently reduce the throughput.
by Figure 3 shows the procedure of using the preamble
to assist the demodulation of the data symbols. Using the
𝑒𝑠 [𝑘] = 𝐿𝐼𝑒 [𝑘] 𝑦 [𝑘] − 𝛾 sgn {𝐿𝐼𝑒 [𝑘]} , (2) preamble data, the frequency and phase offset are estimated
0.5 s 0.5 s 0.5 s 0.5 s
LFMB ... P D ... LFME ... LFMB ... P D ... LFME ...
0.2 s 1000 0.2 s 0.2 s 1000 0.2 s

symbols symbols
Figure 2: General packet structure (change the data symbol length from 1000 to 500).
Preamble
Data
data
(1000 symbols)
(256 symbols)
Doppler and
phase estimation
Figure 3: Frequency and phase offset compensation.
with a Doppler frequency and phase estimation algorithm, where 𝑅(𝑚) is the sample correlation function, given by
which will be described in Section 2.2. The estimated fre-
quency and phase offset are fed to the payload field to 1 𝐿−1
compensate for the Doppler frequency and phase offsets. 𝑅 (𝑚) = ∑ 𝑧 (𝑘) × 𝑧∗ (𝑘 − 𝑚)
𝐿 − 𝑚 𝑘=𝑚 (8)
After the whole data packets are compensated using the
estimated frequency and phase offsets, the compensated data 𝑚 = 1, 2, . . . , 𝑁.
are fed to the equalizer.
2.2. Doppler and Phase Estimation Algorithm. A preamble of 3. Proposed High Throughput
known symbols is inserted to the beginning of each packet Receiver Structure
for the carrier frequency and clock timing recovery. Among
To improve the system throughput, a new packet structure
various methods, data-aided (DA) estimations are normally
shown in Figure 4 is proposed. The lengths of 𝑃 and 𝐷𝑘 are
employed to attain a good performance with a short preamble
the same as those of the conventional packet structure shown
[10].
in Figure 2. The only difference is that the length of the data
Filtering the received waveform in a matched filter and
field is doubled, which allows us to significantly improve the
sampling it at proper times yield
throughput for the same kind of overhead.
𝑥 (𝑘) = 𝑑𝑘 𝑒𝑗(2𝜋𝑓𝑑 𝑘𝑇+𝜃) + 𝑛 (𝑘) , (5) This packet structure may suffer from performance
degradation in the fast time varying channel. However, the
where 𝑑𝑘 is a unit-amplitude symbol, 𝑓𝑑 is the carrier performance degradation can be alleviated by a recursive
frequency, and 𝜃 is the carrier phase. 𝑇 is the symbol frequency offset compensation procedure shown in Figure 5.
duration. 𝑛(𝑘) is a complex-valued independent Gaussian In Figure 5, the 2-step demodulation procedure is illustrated:
random variable. In the data-aided mode, the modulation In the first step, the first data packet is demodulated using the
can be easily stripped by multiplying 𝑥(𝑘) with 𝑑𝑘∗ (complex Doppler and phase information estimated from the preamble,
conjugate of 𝑑𝑘 ), which yields and in the second step, the second data packet is demodulated
𝑧 (𝑘) = 𝑑𝑘∗ 𝑑𝑘 𝑒𝑗(2𝜋𝑓𝑑 𝑘𝑇+𝜃) + 𝑛 (𝑘) 𝑑𝑘∗ using the Doppler and phase information estimated from
(6) the first data packet after being demodulated. In fact, we
= 𝑒𝑗(2𝜋𝑓𝑑 𝑘𝑇+𝜃) (1 + 𝑛̃ (𝑘)) . can use this approach to divide a long packet into small
consecutive packets and to demodulate the subsequent small
As seen in (6), 𝑧(𝑘) can be viewed as a complex sinusoid packets using the information obtained from the previous
embedded in the white Gaussian noise. A data-aided carrier small packets, thereby improving the system throughput in
frequency and phase estimator is derived based on the the fast time varying channel environments.
observation of a few consecutive samples {𝑧(𝑘), 0 ≤ 𝑘 ≤ The frequency and phase offset information resulting
𝐿 − 1}. For the purpose of large frequency offset estimation, from the preamble data is used to compensate for any
the following improved algorithm can be employed: frequency and phase offsets in the 𝐷1 packet. After the
1 𝑁−1 Doppler and phase offset are compensated, the 𝐷1 packet
𝑓̂𝑑 = arg { ∑ [𝑅 (𝑚 + 1) 𝑅 (𝑚)∗ ]} , (7) can be decoded through turbo equalization. Based on the
2𝜋𝑇 𝑚=1 demodulated 𝐷1 and reencoded and remapped 𝐷1 packet,
0.5 s 0.5 s 0.5 s 0.5 s
LFMB ... P D1 D2 ... LFME ... LFMB ... P D3 D4 ... LFME ...
0.2 s 500 0.2 s 0.2 s 500 0.2 s

256 symbols 500 256 symbols 500
symbols symbols symbols symbols
Figure 4: Packet structure.
Doppler and Table 1: Parameters of UWA channel experiment.

phase estimation
Source 500-bit text
Channel coding (2, 1, 7) convolutional code
Modulation QPSK
Preamble Data (1) Data (2)
data Packet size (𝐷𝑘 ) 500 symbols
(500 symbols) (500 symbols)
(256 symbols) Bit rate 1 kbps
Center frequency 16 kHz
Sampling frequency 192 kHz
Doppler and
Distance 400 m
phase estimation Water depth 43 m
Depth TX: 2 m, RX: 20 m
Figure 5: A recursive procedure of Doppler and phase compensa-
tion.
The transmitter emits a source signal with center fre-

we can estimate the frequency and phase offset that occurred quency of 16 kHz and data rate of 1 kbps. The hydrophone
in the 𝐷1 packet period and apply it to the 𝐷2 packet. The was at 20 m below the water surface, and the horizontal range
reencoded and remapped 𝐷1 packet acts as the “preamble” from transmitter was 400 m. The received signal was sampled
data of the 𝐷2 packet. at 192 kHz. The experimentation with two types of packet
The receiver structure for the proposed high throughput structure is conducted.
packet model is shown in Figure 6. As explained above, to In Figure 8, the underwater channel response is shown
achieve high throughput, we divide a long packet into small during a 0.1-second interval. This response was measured
consecutive packets and use the estimated channel informa- by using an LFM signal with a bandwidth of 4 kHz. The
tion from the previous packets to assist the demodulation of channel gains for the secondary and third arrivals fluctuate
the next packets. The proposed receiver consists of buffer, more rapidly. This sparse channel is affected by multipath
reencoder, and remapper. Buffer stores received packets in propagation by reflection from the surface and the bottom.
order to estimate the Doppler and phase offsets for the Experiment parameters are listed in Table 1.
current data packet. For example, to estimate 𝑓̂𝑑 , 𝜃 from Figure 9 illustrates the error performance of the conven-
the 𝐷1 packet in order to demodulate the 𝐷2 packet, the tional system and the proposed system. The received symbols
received 𝐷1 packet has to be buffered. After estimating 𝑓̂𝑑 , have a total of 1047 bit errors in the first and second data
𝜃, the previous packets are removed from the buffer. For the fields (𝐷1 and 𝐷2 ) before any error correction. We investigate
previous packets to be used as the preamble data for the four different types of decoding methods: The first one is
current packet, reencoder and remapper are necessary. Let us the hard decision Viterbi decoder, the second one is the soft
assume that 𝐷̂𝑘 is output of the reencoder and remapper for decision Viterbi decoder, the third one is the conventional
the received 𝐷𝑘 packet; Doppler and phase estimation block turbo equalization described in Section 2 (see Figure 3), and
the last one is the proposed receiver structure of Figure 5.
determines 𝑓̂𝑑 , 𝜃 from 𝐷𝑘 and 𝐷 ̂𝑘 . The estimates 𝑓̂𝑑 and 𝜃
The first and second decoders have 619 bit errors and 492 bit
from 𝐷𝑘 are then used to demodulate 𝐷𝑘+1 .
errors, respectively. For the third decoding method, 451 bits
errors are recorded after the first iteration, 408 bit errors are
4. Experimental Results recorded after the second iteration, and finally the number
of bit errors is reduced to 252 after the third iteration. In
We evaluate the performance of the proposed packet trans- comparison, the proposed method can correct all the errors
mission model in real underwater environments. The exper- after three iterations. From the experimental results, we have
iment was conducted on a lake of Munkyeong city, Korea, confirmed that the proposed algorithm is very effective in
in May 2014. The water depth was approximately 43 meters. improving the communication system performance for the
Figure 7 depicts the lake trial. fast time varying underwater communication channel.
Convolutional Symbol Dk Packet

Interleaver
encoder (2,1,7) mapper generator
Underwater
channel
LD
e
I
+ Le DFE Packet
Deinterleaver
equalizer synchronization
−
BCJR ̂ ,𝜃
f d P󳰀
decoder
− P
+ LIc Doppler and
Interleaver Buffer of Dk
LD phase estimation
c Dk
Turbo equalizer
Reencoder Remapper
̂ k−1
D
Figure 6: Proposed high throughput receiver model.
1200
1100 Uncoded error
1000
Number of errors
900
2m 800 Hard decision
700
Soft decision Iteration 1
20 m 600 Iteration 2
Transmitter 500
ITC 1001 Iteration 3
400
43 m 400 m 300 Iteration 1
200 Iteration 3
Receiver 100
Iteration 2
B&K 8106 0
Iteration
Bottom
Uncoded error Conventional
Figure 7: Illustration of the lake trial. Hard decision Proposed
Soft decision
Figure 9: BER comparison between the conventional and the

proposed methods.
0
−1
−2 5. Conclusions
−3
Magnitude (dB)
Underwater acoustic channels are characterized by long

−4
multipath spreads that cause intersymbol interference in
−5
addition to Doppler shift due to the relative source-receiver
−6
motion. To improve the system performance and throughput
−7
in the presence of these major obstacles, we have proposed
−8
a consecutive iterative BCJR equalization for a long packet
−9
size. To achieve the high throughput in the fast time varying
−10 channels, we divide the long packet into small consecutive
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Arrival time (s) packets and use previous packets to estimate channel infor-
mation needed to compensate for the subsequent packets.
Figure 8: Underwater channel response. The error performances of four different decoding algorithms
are compared based on real experimental data measured

on a lake in Munkyeong city. We demonstrate that the
proposed algorithm can correct all errors of received symbols,
thereby confirming that the proposed algorithm is very
effective for communications in the time varying underwater
communication channel.
Conflict of Interests
The authors declare that there is no conflict of interests
regarding the publication of this paper.
Acknowledgments
This work was supported by Defense Acquisition Program
Administration and Agency for Defense Development under
Contract UD130007DD. This research was financially sup-
ported by the Ministry of Education, Science Technology
(MEST), and National research Foundation of Korea (NRF)
through the Human Resource Training Project for Regional
Innovation.
References
[1] M. Stojanovic, J. A. Catipovic, and J. G. Proakis, “Phase-coher-
ent digital communications for underwater acoustic channels,”
IEEE Journal of Oceanic Engineering, vol. 19, no. 1, pp. 100–111,
1994.
[2] D. B. Kilfoyle and A. B. Baggeroer, “The state of the art in
underwater acoustic telemetry,” IEEE Journal of Oceanic Engi-
neering, vol. 25, no. 1, pp. 4–27, 2000.
[3] M. Tüchler, R. Koetter, and A. C. Singer, “Turbo equalization:
principles and new results,” IEEE Transactions on Communica-
tions, vol. 50, no. 5, pp. 754–767, 2002.
[4] R. Koetter, A. C. Singer, and M. Tüchler, “Turbo equalization,”
IEEE Signal Processing Magazine, vol. 21, no. 1, pp. 67–80, 2004.
[5] J. Ling, T. Yardibi, X. Su, H. He, and J. Li, “Enhanced channel
estimation and symbol detection for high speed multi-input
multi-output underwater acoustic communications,” Journal of
the Acoustical Society of America, vol. 125, no. 5, pp. 3067–3078,
2009.
[6] B. D. Trumpis and P. L. McAdom, “Performance of convo-
lutional codes on burst noise channels,” in Proceedings of the
National Telecommunications Conference (NTC ’77), pp. 36:3-1–
36:3-14, Los Angeles, Calif, USA, December 1977.
[7] L. R. Bahl, J. Cocke, F. Jelinek, and J. Raviv, “Optimal decoding
of linear codes for minimizing symbol error rate,” IEEE Trans-
actions on Information Theory, vol. 20, no. 2, pp. 284–287, 1974.
[8] J. Salz, “Optimum mean-square decision feedback equaliza-
tion,” The Bell System Technical Journal, vol. 52, no. 8, pp. 1341–
1373, 1973.
[9] K. Berberidis, T. Rontogiannis, and S. Theodoridis, “Efficient
block implementation of the LMS-based DFE,” in Proceedings of
the 13th Digital Signal Processing International Conference, vol.
1, pp. 143–146, July 1997.
[10] S. Tao, X.-W. Liang, and X.-T. Chen, “Decision-directed estima-
tion of carrier frequency and phase for burst PSK transmission,”
in Proceedings of the 5th International Conference on Wireless
Communications, Networking and Mobile Computing (WiCom
’09), pp. 1–4, Beijing, China, September 2009.
http://dx.doi.org/10.1155/2015/896832
Review Article
Underwater Sensor Network Applications:
A Comprehensive Survey
Emad Felemban,1 Faisal Karim Shaikh,1,2 Umair Mujtaba Qureshi,2

Adil A. Sheikh,1 and Saad Bin Qaisar3
1
Simplicity Labs, TCMCORE, STU, Umm Al-Qura University, Mecca 21955, Saudi Arabia
2
IICT, Mehran University of Engineering & Technology, Jamshoro 76062, Pakistan
3
SEECS, NUST, Islamabad 44000, Pakistan
Correspondence should be addressed to Faisal Karim Shaikh; faisal.shaikh@faculty.muet.edu.pk
Received 5 June 2015; Accepted 10 August 2015
Academic Editor: Dongkyun Kim
Copyright © 2015 Emad Felemban et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
There is no escaping fact that a huge amount of unexploited resources lies underwater which covers almost 70% of the Earth. Yet, the
aquatic world has mainly been unaffected by the recent advances in the area of wireless sensor networks (WSNs) and their pervasive
penetration in modern day research and industrial development. The current pace of research in the area of underwater sensor
networks (UWSNs) is slow due to the difficulties arising in transferring the state-of-the-art WSNs to their underwater equivalent.
Maximum underwater deployments rely on acoustics for enabling communication combined with special sensors having the
capacity to take on harsh environment of the oceans. However, sensing and subsequent transmission tend to vary as per different
subsea environments; for example, deep sea exploration requires altogether a different approach for communication as compared to
shallow water communication. This paper particularly focuses on comprehensively gathering most recent developments in UWSN
applications and their deployments. We have classified the underwater applications into five main classes, namely, monitoring,
disaster, military, navigation, and sports, to cover the large spectrum of UWSN. The applications are further divided into relevant
subclasses. We have also shown the challenges and opportunities faced by recent deployments of UWSN.
1. Introduction for the benefit of humans. The sensor nodes, stationary or

mobile, are connected wirelessly via communication modules
The 75% of earth’s surface is covered with water in the form of to transfer various events of interest [2]. Underwater com-
rivers, canals, seas, and oceans. Plenty of precious resources munication is mainly done with a set of nodes transmitting
lie underwater which are required to be explored. The their data to buoyant gateway nodes that relay the data to
key to successful explorations has always been technology nearest coastal monitoring and control station also called
dependent. Recent advances in technologies have led the remote station [3]. Generally, in UWSN acoustic transceivers
possibilities to do the underwater explorations using sensors are used for communication. The acoustic waves are low
at all levels which were not possible previously. Accordingly, frequency waves which offer small bandwidth but have long
underwater sensor network (UWSN) is emerging as an wavelengths. Thus, acoustic waves can travel long distances
enabling technology for underwater explorations. UWSN is a and are used for relaying information over kilometers [4].
fusion of wireless technology with extremely small microme- UWSNs are utilized for a wide range of applications such
chanical sensor technology having smart sensing, intelligent as monitoring the marine environment for scientific explo-
computing, and communication capabilities. UWSN is a ration to commercial exploitation and coastline protection to
network of autonomous sensor nodes [1] which are spatially underwater pollution monitoring, from water-based disaster
distributed underwater to sense the water-related properties preventions to water-based sports facilitation. UWSN offers a
such as quality, temperature, and pressure. The sensed data promising solution to ever demanding applications. However,
can be utilized by variety of applications that can be used UWSN applications are exciting but challenging at the same
Terrestrial communication
infrastructure
Onshore
infrastructure
AUV
2D underwater
communication
4D underwater
communication
de
ab
Se
Cluster 2
1D underwater
Cluster 1 3D underwater communication
1D underwater communication
communication
Offshore
AUV
Buoy
Underwater device/node Surface stations
Wireless link
Boat
Figure 1: Underwater sensor network architecture.
time. The reason lies in unpredictable conditions of water sense underwater properties or it can be deployed underwater
environment which creates serious constraints in the design for a particular period of time to sense information and then
and deployment of such networks. The focus of this paper is float towards the surface to transmit the sensed information
to survey the available UWSN application. The paper further to the remote station. It can be an autonomous underwater
focuses on classifying these applications and presenting a vehicle (AUV) which dives inside the water, sense or collect
summary for each class. the underwater properties, and relay the information to the
The rest of the paper is organized as follows: Section 2 dis- remote station. In 1D-UWSN the nodes can communicate
cusses the general architecture of UWSNs. Section 3 discusses using acoustic, Radio Frequency (RF), or optical commu-
the recent research contributions in the field of UWSN and nication. Moreover, the topological nature of 1D-UWSN is
provides a detailed application classification. The challenges star where the transmission across the sensor node and the
and opportunities for UWSN are detailed in Section 4. In the remote station is carried over a single hop.
end, Section 5 concludes the paper.
2.2. 2D-UWSN Architecture. Two-dimensional- (2D-)
2. Underwater Wireless Sensor UWSN architecture refers to a network where a group of
Network Architecture sensor nodes (cluster) are deployed underwater. Each cluster
has a cluster head (also called anchor node). The clusters
In this section, we discuss the common UWSN architectures are fixed as they are anchored at the underwater surface.
(Figure 1) which are basis for designing the UWSN applica- Each member of the cluster gathers the underwater data
tions. and communicates it to the anchor node. The anchor node
gathers the information/data from all its member nodes and
2.1. 1D-UWSN Architecture. One-dimensional- (1D-) UWSN relays it to the surface buoyant nodes. In 2D-UWSN, the
architecture refers to a network where the sensor nodes are communication is carried in two dimensions; that is, (i) each
deployed autonomously. Each sensor node is a stand-alone member of the cluster communicates with its anchor node
network itself, responsible for sensing, processing, and trans- with horizontal communication link while (ii) the anchor
mitting the information to the remote station [5]. A node in node communicates with the surface buoyant node with
this type of architecture can be a floating buoy which can vertical communication link. In 2D-UWSN, acoustic, optical,
UWSN applications
Monitoring Disaster Military Assisted navigation Sports
Water quality Habitat Exploration
Mines Submarines Surveillance
Marine life Fish farms Reef
Volcano,
Flood earthquake, and Oil spill
tsunami
Natural Pipeline and

resources cables
Figure 2: Classification of underwater wireless sensor network applications.
and RF communication can be used depending on the type on the distance and data between them and either acoustic
of application and nature of underwater environment. or radio can be used. As the transmission is to be directly
In 2D-UWSN, acoustics communication is preferred for relayed to ROV, the sensors which have large data and are
underwater anchor node and the surface buoyant node due close to ROVs can use radio links while the sensors which
to typically high distance between them. For the cluster have small data to transmit or are far from ROV can use
of nodes, the network arrangement can be star, mesh, or acoustics links [7, 8].
ring depending on the application requirement. The 2D-
UWSN can be used for both time-critical and delay tolerant 3. Underwater Wireless Sensor Network
applications [6].
Application Classification
2.3. 3D-UWSN Architecture. In this type of network, the UWSN applications are rapidly gaining popularity for
sensors are deployed underwater in the form of clusters enabling advances in the area of ocean monitoring and
and are anchored at different depths. Due to the deploy- observatory systems, deep sea surveillance, tracking of var-
ment of the sensors at variable heights, the communication ious entities of the aquatic environment, and unearthing
between the sensors goes beyond the two dimensions. There resources. UWSNs find their application in fields like off-
are three communication scenarios in this architecture: (i) shore oil and gas extraction, oil spills, military surveillance
intercluster communication of nodes at different depths, (ii) and reconnaissance, mine detection, pollution monitoring,
intracluster (sensor-anchor node) communication, and (iii) natural calamities like tsunami and hurricane forecast, coral
anchor-buoyant node communication. In all three types of reef and habitat monitoring of marine life, and fish farming. A
communication scenarios, acoustic, optical, and RF links can comprehensive classification of potential UWSN applications
be used. is shown in Figure 2. This section presents a survey of recent
developments in the domain of UWSN applications. At the
2.4. 4D-UWSN Architecture. Four-dimensional- (4D-) top level we classify the UWSN applications as monitoring,
UWSN is designed by the combination of fixed UWSN, disaster forecasting and management, military, navigation,
that is, 3D-UWSN and mobile UWSNs. The mobile and sports which are going to have a significant impact on
UWSN consists of remotely operative underwater vehicles the underwater world and provide benefits to the humans.
(ROVs) to collect data from the anchor nodes and relay We then further classified monitoring, disaster, and military
the data to the remote station. ROVs can be autonomous applications into relevant subclasses.
submersible robots, vehicles, ships, and even submarines.
Each underwater sensor node can be autonomous in relaying 3.1. Monitoring Applications. Underwater monitoring refers
the data directly to ROVs depending on how close that to a network of sensors which is deployed underwater to
particular sensor node is to the ROV. The communication monitor the underwater environment, its characteristics,
scenario between ROV and underwater sensor node depends properties, or any object of interest. These applications are
particularly related to monitoring the physical environment. relayed to a buoyant node having GPRS connectivity that will
Underwater monitoring applications can further be classified relay the information gathered to the remote station.
into (i) water quality monitoring, (ii) habitat monitoring, and The authors in [14] have presented a decentralized ocean
(iii) monitoring underwater explorations. pollution and wreckage detection mechanism based on
underwater sensor nodes equipped with short range acoustic
3.1.1. Water Quality. Water is precious resource and is pre- modems for communication.
mier factor for survival of living things under or above the
water surface. Therefore, it is important to monitor the quality 3.1.2. Habitat. Underwater habitat monitoring deals with the
of water. The underwater quality monitoring applications study of environment of any living organism living over or
vary from monitoring water quality of canals to oceans. in the water. It is one of the most interesting yet challenging
The authors in [9] have developed an application to fields of natural sciences. Habitat monitoring, when con-
monitor the quality of pool water for trout farms. For the sidered underwater, becomes even more challenging due to
growth of trout in a farm/pool, various parameters were vulnerable underwater conditions. UWSN can also be used
monitored such as chemical oxygen demand, ammonium for ecosystem monitoring applications. Underwater habitat
nitrogen (NH3 -N), pH, and electrical conductivity (EC). The monitoring applications can be further classified into (i)
parameters were monitored for 270 days between August 2011 marine life monitoring, (ii) fish farm monitoring, and (iii)
and April 2012. An algorithm was proposed by the authors underwater plants/coral reef monitoring applications.
which can display the information of the input and output of
all the four pools. The comparison was made using fuzzy logic (1) Marine Life. Marine life monitoring application deals
for evaluating the sensed data and notifying in the case of any with the monitoring environment of various species living
critical state whenever the parameters surpass the threshold above the water or underwater. ACMENet is European
values. funded project designed for underwater marine environment
monitoring [15]. ACMENet is an underwater acoustic sensor
In [10], the authors have developed two UWSN based network which is capable of monitoring marine environment
approaches to monitor the water quality at drinking water such as fishes and any human or nonhuman activity within
reservoirs. First approach is based on using sensors integrated the coverage area and used for geological surveys. The authors
with mini-AUV responsible for collecting the water samples have developed a protocol based on TDMA master-slave
and transmission of the information to the remote station. mechanism for data transfer across the sensor nodes.
Second approach consists of probes mounted with solar pan- The authors in [16] have developed an UWSN framework
els using IEEE 802.15.4-based wireless interface for collecting for aquatic monitoring. The framework has capability to
and transmitting the information. Both approaches were monitor marine environment at different levels whose data
compared in terms of total time needed for deployment, total can be used to develop and verify models for ecosystems, pre-
time needed for analysis, total cost of ownership including diction of possibility of any change in ecological phenomena,
deployment, total number of measurements per analysis for and also adapting to any environmental/underwater climatic
a site, the need for qualified operators, real-time analysis, and changes. The system has features of sensing, wireless trans-
probe diversity. It was shown that the second approach proved mission to remote station, visualizing, and alarming about
to be more applicable, efficient, and of low cost for continuous the events. The system is currently deployed in Queensland,
water quality monitoring. Australia. The authors have designed and developed an ad
In [11], the authors have developed WSN based system hoc system, based on clusters (star topology) where cluster
to monitor the water quality of rivers. The authors have head directly transmits data to the buoyant gateway. A remote
designed and proposed a sensor node capable of monitoring data transmission from the gateway to the control center is
pH of water. The sensor module consists of signal processing proposed. The proposed system was used for monitoring the
module for processing the sensed data, a signal conditioning underwater luminosity (Lum) and temperature.
module for comparing the data with the threshold values, The authors in [17] developed a Smart Environmental
and wireless Zigbee module for transmission of data to the Monitoring and Analysis Technology (SEMAT) for marine
remote station. The authors have successfully designed and research and monitoring water fields. The authors claim that
deployed the system for river quality monitoring in India. A SEMAT is cheaper marine environment observatory system
similar Zigbee based UWSN system is developed in [63] and compared to its counterparts and traditional systems. SEMAT
proposed for water quality monitoring of rivers. comprises plug-and-play smart sensors, short range wireless
In [12], the authors have developed a WSN based water transmission, and near-real-time analysis tools. SEMAT can
quality monitoring system. The proposed prototype monitors be used for quick and short term deployments especially in
the water temperature (𝑇), dissolved oxygen, pH, and EC in shallow coastal aquatic environments.
real time. The information is disseminated in graphical and The authors in [18] have developed an UWSN based
tabular formats to relevant stake holders through web-based coastal shallow water marine environment monitoring sys-
portal and mobile phone platforms. tem. The system is a prototype which has been successfully
An underwater sensor network to monitor the water tested and deployed at Mar Menor coastal lagoon, Spain. The
quality and pollution is proposed in [13]. Underwater sensor developed system is capable of collecting oceanographic data
nodes in their proposed test bench measure entities such as through various water specific sensors. The data is transmit-
pH, turbidity (Tur), and temperature. The collected data is ted to the sink node (buoy) which relays the information to
the remote station for meaningful results. The authors have are used to monitor the quality of the water. The authors have
also developed application using LabView and mapped it formulated Clustering Protocol of Water Sensor (CPWS)
with Google Maps. The authors conclude that the proposed network, an energy efficient routing protocol which increases
prototype is efficient in terms of communication, better the overall network lifetime making the system more reliable,
network lifetime, and data reliability. robust, and everlasting.
An intelligent marine vehicle sensor network architecture
has been developed in [19]. The authors have also designed (3) Coral Reef. Coral reefs are diversified underwater ecosys-
protocols for ocean monitoring applications. The authors tems. Coral reefs are built by microorganism living under-
in [20] have presented a sensor network application for water. Coral reefs are important part of marine ecology
habitat monitoring of sea shell. They have presented an entire and marine life. UWSN has potential of coral reef habitat
architecture. The proposed network is also capable of coastal monitoring. Following are UWSN based applications.
monitoring and is currently deployed in Zhejiang province, The authors of [15] have presented and developed UWSN
China. based coral reef observatory system and application. The
A good survey on applications of UWSN for marine envi- system is prototype design based on 4 core network elements
ronment monitoring is presented in [64]. The survey includes which are (i) stationary nodes, (ii) moving nodes, (iii) surface
different techniques, recent projects, and applications along buoy nodes, and (iv) remote station. The stationary nodes
with the latest advancement in the field. are deployed underwater at specific coral targeted places,
equipped with underwater condition monitoring sensors
(2) Fish Farms. Fish farming has become one of the most such as temperature, cameras, and pH sensors. The moving
demanding occupations and is considered as a good econom- nodes are autonomous underwater vehicles (AUVs) which
ical resource. Fish farming requires strict and most relevant are capable of deploying the nodes, collecting the senses
habitat conditions, which requires a strict monitoring of the information from the stationary nodes, and also relaying the
farm. Therefore, sensors are deployed to monitor the habitat information to the surface buoy nodes. The surface buoy
conditions of a particular fish to maintain and create a closed nodes transmit the information to the remote station where
environment for fish farming. scientist could investigate and monitor the sensed data. The
The authors in [21] have developed UWSN application prototype is highly adaptable to the harsh water conditions
which is able to assess the amount of uneaten feed and fiscal and is deployed in real time in Okinawa coast in Japan;
waste in fish farm. The system developed is particularly tested however there are some shortcomings such as battery and
in marine fish farm for fauna and flora. The authors devised network lifetime.
and proposed a group based wireless sensor network for accu- Alippi et al. have designed intelligent autonomous under-
rate quantitative analyses for monitoring water pollution. The water vehicle based Starbug project for monitoring and
authors in [9] developed, monitored, and maintained the locating coral reefs in shallow water [16]. The AUV is very
ecosystem for trout in trout farms by sensing the quality of carefully designed having a video transceiver, water quality
water. monitoring sensors, acoustic communication module, mag-
In [22] the authors have formulated a Zigbee based netic compass, pressure sensors, and GPS sensors. Starbug
UWSN for monitoring the fish farm properties. The authors project has proven to be platform with reliable monitoring
designed fish farm property specific sensor nodes which and great commercial potential. However, the authors have
sense pH, NH4 , and temperature and transmit the informa- emphasized on lowering the cost of the system.
tion to central unit through Zigbee based routing protocol. Pirisi et al. have designed and developed an intelligent
The authors have proposed the system for small fish ponds buoy for coral reef monitoring [25]. The buoy is intelligent
and pools. and is made by introducing a novel approach, optimizing the
The authors of [23] designed and developed WSN based energy by converting the sea wave energy, thus increasing the
system application which has the ability to monitor the large network lifetime. The design is brief in the research presented.
farms. The system has features of automatic monitoring, his- The prototype developed is tested in real time and is proposed
tory chart of the sensors data, and remote login for interested by the authors for potential applications for coral reef and
users. Authors have presented comparison of wireless tech- marine life and marine ecological systems and applications.
nologies and have proposed Zigbee technology as optimal in Bainbridge et al. have presented in their research [26]
this case. The developed system is able to monitor parameters one of the most practical, real-time, and mature UWSNs
like dissolved oxygen, pH values, temperature, the water level, for monitoring coral reefs. This UWSN is deployed at seven
and humidity and the information is wirelessly transmitted different coral reef locations in northeastern Australia and
to managers station. Furthermore, wireless cameras are used currently in operation since past two years. The system
and integrated with the system and the Internet for wireless designed is acoustic based, reliable in terms of ocean data
monitoring anytime and anywhere in the world. monitoring, and robust in terms of sustaining the harsh
In [24] the authors have developed an energy efficient environment.
sensor based network for monitoring fishes in ponds and
lakes. The framework has features of detecting depth of the 3.1.3. Underwater Exploration. There are large number of
pond and monitoring and maintaining the quality of water minerals present underwater which is required to be explored
by maintaining the acidic levels in the water. The depth is such as oil and gas. Also, one has to meet the water end
detected by using RGB sensors while pH and oxygen sensors while joining one end of the surface with another. While
water is covering the large part of the Earth’s surface, the control function in case of any leakage in the pipeline. The
remaining dry surface of earth is connected by laying cables prototype is tested in lab environment. The authors conclude
underwater. Underwater cables are deployed which bring with testing various WSN based technologies and protocols
some of the most basic social necessities such as oil, gas for more reliable network.
pipelines, and fiber optics cables. Therefore UWSN can be Saeed et al. have developed a prototype for monitoring oil
used to explore the precious resources and also to monitor and gas in pipeline underwater [30]. The prototype was tested
the underwater pipeline and cables. Underwater explorations in lab environment. The authors have concluded the initial
and applications are further categorized into (i) underwater work successfully; however the real-time implementation still
natural resource explorations and (ii) cable and pipeline mon- holds serious concerns in terms of its network lifetime and
itoring applications for underwater oil and gas explorations. reliability. Also Yu and Guo have designed and proposed an
efficient oil and gas pipeline monitoring system using UWSN
(1) Natural Resources. The authors in [65] have developed an [31].
UWSN which deployed to locate and investigate manganese
crust. The system developed uses remotely operative under- The UWSN monitoring applications are summarized in
water vehicle (ROV) integrated with acoustics communica- Table 1. It can be observed that when it comes to water
tion module and vision based mapping system and monitors quality monitoring, most of the recent applications have
in situ manganese crust depth up to 3000 m. The ROV is used RF to achieve their objectives. Considering any sort
operated in low altitudes. Thus the authors have successfully of water contaminating impurities such as oil spills, they
carried out experiments using the ROV based UWSN and are in restricted area of water; therefore, due to shorter
have concluded and proposed the system as useful tool for distances, RF signals fulfill the requirements. However, there
manganese crust detection and measurement underwater. is an equal emphasis on using RF and acoustic waves as the
The authors in [27] have proposed an UWSN which communication type. Since RF has limitations with respect
is a combination of underwater acoustic sensor networks to distance, acoustic has been considered for longer distances
and underwater mobile (ROVs, AUVs) sensor networks to ranging to a few kilometers. Acoustics are employed mostly
discover and excavate the mineral resources underwater. The for marine life, reef, and deep sea exploration applications,
proposed system can be used on large scale to scan large and since more coverage is required. A trend of employing
deep spaces in oceans for exploration. various architectures is also seen when dealing with these
An acoustic based UWSN proposed in [15] is also used applications, whereas, for marine life, reef, and exploration
for deep sea exploration and monitoring of coastal areas. applications, 3D and 4D architectures are more preferred
The system is specifically tested for coastal application by due to different depths and distances of the objects under
deploying cameras underwater. consideration. Since fish farms require a low coverage sensor
network, 1D and static 2D architectures will perform an
Detailed literature is present motivating the researchers
efficient job to cater the application.
to develop UWSN application for deep sea explorations and
application [1, 4]. The literature presents the communication
architecture and possible design of the system and highlights 3.2. Disaster. Generally, the natural disasters are inevitable.
suitable conditions and challenges. Among others, water based natural disasters are more dan-
gerous and produced huge destruction to the earth. Accord-
(2) Marine Cable and Pipeline. There are large number of ingly, disaster monitoring and preventive mechanisms are
oil and gas pipelines laid, interconnecting entire continents. very necessary. UWSN offers a wide range of applications
UWSN has potential to monitor underwater oil and gas for management and recovery of such disasters. More par-
pipeline monitoring applications. Applications are discussed ticularly, it relates to the monitoring of events that aggravate
below. a disasters aftermath. Along with inadequate resources for
Mohamed et al. have presented a very detailed survey over comprehensive monitoring of the vast area of water (e.g.,
underwater oil and gas pipeline monitoring using UWSN ocean), the task becomes even more challenging with occa-
[28]. The authors discussed the sensor network architecture sionally ruthless weather. Therefore, efficient monitoring of
which includes four probable networks which are underwater marine and aquatic dynamics is a significant research chal-
wireless acoustics sensor networks, underwater wired acous- lenge. UWSN monitoring strategies for disaster management
tics sensor networks, underwater wireless RF sensor net- and prevention can be formulated into a wide variety of
works and underwater wired networks. Furthermore, inte- applications such as floods, underwater volcanic eruptions,
gration of wired/acoustic and wired/RF can also be possible. underwater earthquakes and their resulting tsunamis, and
The authors have compared all the networks and proposed oil spills which lead to above-the-water and underwater
underwater integrated wired/acoustic sensor networks and ecological instabilities.
underwater integrated wired/RF sensor networks for oil and
gas pipeline monitoring applications. 3.2.1. Floods. The repercussions of a flood and its increased
Jawhar et al. present a framework for pipeline monitoring frequency have pushed the researchers to find ways of
using WSNs [29]. The paper presents reference model which timely flood alerts. The alerts need not only be placed in
could be used as applications. The prototype designed is based urban shores and hence require remote deployment. UWSN
on proposed architectural model which is capable of mon- helps develop solutions of underwater sensor deployments
itoring the pipelines underwater and also giving additional with over-the-water relay agents to calculate aquatic vitals.
Table 1: Comparison of UWSN monitoring applications.
Deployment Sensors Communication

App. Implementation
Architecture Salinity level Operable depth Type Number Type
Water quality
[9] 1D Pool Few meters O2 , NH3 -N, pH, EC Few RF Real-time
[10] 3D, 4D Reservoir Few meters 𝑇, EC, 𝐷, NO3 − , Tur Many RF, acoustic Real-time
[11] 1D River Few meters pH Many RF Real-time
[12] 4D Lake Few kilometers 𝑇, O2 , pH, EC Many Acoustic Real-time
[13] Static 2D Shallow water Few meters 𝑇, pH, Tur Many RF Nontime
[14] 3D Ocean Kilometers n/a Many Acoustic n/a
Habitat
Marine life
[15] 3D Sea 75 Km n/a Few Acoustic Real-time
[16] 3D, 4D Sea Meters 𝑇, 𝑃 Many Acoustic Real-time
[17] 3D Shallow water 1–4 kilometers 𝑇, 𝑃, 𝑆, Photo Few RF Real-time
[18] 3D Shallow water Kilometers 𝑇, 𝑃, 𝑆 Many Acoustic Real-time
[19] 3D Sea n/a 𝑇, 𝑃, 𝑆, 𝑉 n/a Acoustics Real-time
[20] 3D Sea n/a 𝑉 Few Acoustic Real-time
Fish farms
[21] Static 2D River 2–15 meters 𝑇, 𝑆, Tur Few RF Real-time
[9] 1D Pool Meters O2 , NH3 -N, EC, pH Many RF Real-time
[22] Static 2D n/a n/a 𝑇, pH, NH4 n/a RF Real-time
[23] Static 2D Pool, pond 10 Km 𝑇, 𝐿, 𝐻, O2 , pH Many RF, acoustic Real-time
[24] 3D River n/a 𝐷, O2 , pH Few Acoustic
Reef
[15] 4D River Few meters 𝑇, C, pH Few Acoustic Real-time
[16] 4D River 200 meters C, 𝑃, Comp, GPS n/a RF, acoustic Real-time
[25] 1D River n/a 𝑇 Few Acoustic Real-time
[26] 4D River n/a 𝑇, 𝑃, pH Few Acoustic Real-time
Exploration
Natural resources
[26] 4D River 3 Km 𝑇, 𝑆, 𝑃, 𝐷 Few Acoustics Real-time
[27] 4D River Meters n/a Few Acoustics Not deployed
[15] 3D Oceans Kilometers n/a Few Acoustic Real-time
Pipelines and cables
[28] 3D, 4D Pool Meters n/a Few RF, acoustic Not deployed
[29] 3D Pool Few meters Vib, Cur Few RF, acoustic Real-time
[30] Static 2D Pool Few meters n/a Few n/a Lab environment
[31] n/a n/a n/a n/a n/a n/a n/a
C = camera, 𝐷 = depth, 𝐻 = humidity, 𝐿 = water level, O2 = oxygen, 𝑃 = pressure, 𝑆 = salinity, 𝑇 = temperature, 𝑉 = velocity, Cur = current, Comp = magnetic
compass, EC = electrical conductivity, NH3 -N = ammonium nitrogen, NO3 − = nitrate, NH4 = ammonium, Photo = light, Tur = turbidity, Vib = vibration, and
GPS = global positioning system.
These vitals are gathered at remote station and inspected for is to observe the information and predict the flood. The
flood indications. transponder module is used for relaying the information in
Pasi and Bhave have addressed the issue of flood monitor- case of the flood. The designed system is simulated and tested
ing and alarming with the help of UWSNs [32]. The designed in perspective of its efficiency by varying the number of nodes
system consists of sensor module, observatory module, and deployed versus the time delay. The system is also tested for
transponder module. The sensor module is responsible for its localization error and the area of coverage.
monitoring water and gathering water-related parameters Tyan and Oh [33] have proposed an acoustic based
such as level, thrust, and intensity of water as flood indicators. UWSN for flood monitoring in rivers. The designed system is
The information is transmitted to the remote station for based on 4D-UWSN architecture which consists of underwa-
further observations. The purpose of observatory module ter deployed sensors, AUV, and a remote station. The sensor
relays the information acoustically to the AUV which collects thickness of oil spills in water which can expedite the cleaning
the information and transmits the information to the remote procedure.
station. The designed system is tested in 5000 m by 200 m In [38], the authors have devised an ad hoc UWSN that
wide river testbed. The authors have successfully tested and detects ocean pollution. They have specified different sensors,
deployed the network in real time. their synchronization algorithms, routing protocols, and the
Marin-Perez et al. [34] have designed a robust flood complete protocol stack. The work focuses on maximizing the
warning system based on UWSN. The designed system is network lifetime and QoS.
Zigbee based, with integrated sensors which collect the infor- Iwendi and Allen [39] have demonstrated the security
mation of different water levels. The system is also integrated issues in Niger-Delta Oil and Gas Sector, where they have
with SCADA system which is responsible for monitoring worked specifically on developing a network which has a very
the hydrological events such as rainfall and other weather secure key distribution scheme. Apart from key management
monitoring conditions. The system is tested and implemented system, the work also focuses on robust and fast clustering
in real time in flash-flood prone 650 km2 semiarid watershed algorithms with dynamic key allocation system.
in Southern Spain. In [40], the work focuses on designing a sensor that is
Udo and Isong [35] have designed a sensor based flood able to sense, process, and transmit information relating the
monitoring and detection system. The system takes into thickness and location of the oil spill. The work suggests two
account information such as humidity, temperature, water algorithms Light Sensor Array and Conductivity Array in
level, and amount of rainfall as flood indicators. The sensor order to find out the thickness of the oil spill. The work also
deployed in the sensor field senses the information and focuses on designing a simulator that shows real-time data on
transmits it to the remote station where, on crossing the a map, relating the thickness and the location of the oil spill.
threshold, the vicinities are notified through Short Message In [41], the authors focused on providing the feasibility
Service (SMS). The system currently covers 15 flood prone of using UWSNs for the use of monitoring oil spillage,
regions in Uyo metropolis in Akwa Ibom state, Nigeria. detection, and pipeline vandalization. The work also focuses
on describing the applications, features, network type, and
challenges, which makes UWSNs an intelligent choice to
3.2.2. Volcano, Earthquake, and Tsunami. Underwater earth-
make in detection of oil spillage and pipeline vandalization
quakes and volcanoes are natural disasters and causes for
in Nigeria.
endangering living things. These natural calamities can occur
The comparison of disaster applications is given in
anytime and anywhere over the surface of earth and are
Table 2. When disaster applications are surveyed, flood appli-
even more alarming when they occur underwater, depending
cations tend to use RF in conjunction with 2D architectures
upon the seismic and the geological changes that take place
more as compared to acoustics in conjunction with 4D due to
under earth. Thus, it is important to monitor such conditions.
the fact that flood related applications can be greatly covered
Kumar et al. [36] have discussed UWSN architecture and
by deploying sensors in the form of clusters. This ultimately
proposed 4D-UWSN for early warning generation in case
leads to low coverage UWSNs employing RF communication
of any hazardous event such as earthquakes and tsunamis.
paradigms. For applications related to volcanoes, simulation
The authors have proposed multicarrier communication and
and analytical models have been proposed to cater these.
OFDM for efficient underwater communication in such
For oil spill applications as well, analytical, simulation, and
scenarios.
testbed models employing RFs and acoustics have been
Casey et al. [37] have proposed an efficient architecture
considered with 2D, 3D, and 4D architectures.
for detecting tsunamis. The proposed sensor based architec-
ture utilizes seismic pressure sensors to predict the tsunami 3.3. Military. UWSNs are employed to assist military appli-
underwater and relay sensed informed by directed diffusion cations as well. These systems take the aid of different sensors
routing protocol. The architecture works on the sense and deployed for detection of different aspects of military appli-
response mechanism. The authors have detailed a number of cations. Different sensors such as cameras, imaging sonars,
approaches and have tested the proposed network in a testbed and metal detectors integrated with AUVs are used to assist
environment. in finding underwater mines, securing ports, and submarines
In [66], a detailed survey highlights underwater natural and are also used for monitoring and surveillance. These
disaster monitoring applications including underwater earth- applications can lead to economic solution to protect naval
quakes and volcanoes. To the best of our knowledge, only few forces.
real deployed applications are present to date for seismic and
tsunami detection. There is a lot of literature addressing the
3.3.1. Mines. Since a sensor can sense physical parameters, it
natural calamities through UWSN and formulating efficient,
is an intellectual practice to detect hidden mines underwater.
reliable, and robust application but hardly are there full
This can assist the military ships in a trouble-free voyage. The
applications designed.
mines are usually made up of ferrous materials and they can
be differentiated from underwater clutter based on the fact
3.2.3. Oil Spill. Manmade pollution is an important factor that the clutter is nonmetal. Therefore using metal detecting
to consider when talking about marine life health. Marine sensors can help find clutters underwater.
life is highly affected by oil spills pollution and therefore In [42], the authors have illustrated a developing sys-
UWSNs have contributed a way to find out the location and tem from Naval Surface Warfare Center, Florida, that uses
Table 2: Comparison of UWSN disaster applications.

App. Implementation
Architecture Salinity Operable Type Number Type
level depth
Floods
[32] 2D Canals Few meters Water level, thrust, and intensity Many RF Simulation, testbed
[33] 4D River Kilometers Depth Few Acoustic Testbed
[34] 2D River Kilometers Water level, rainfall Many RF Real-time
[35] 2D Rivers Kilometers Humidity, temperature, water level, and rainfall Many RF Real-time
Volcano, earthquake, and tsunami
[36] 4D Sea Kilometers Pressure Many Acoustic Analytical
[37] 3D n/a n/a Pressure Many n/a Simulation
Oil spill
[38] 3D Sea 50–500 m n/a Few Acoustic Analytical
[39] 4D n/a n/a n/a n/a n/a Simulation
[40] 2D n/a n/a Light, conductivity, and GPS n/a RF Testbed
[41] 2D n/a Few meters n/a n/a RF Analytical
a magnetic real-time tracking vector gradiometer (RTG) and 3.3.2. Submarines. Target localization is yet another appli-
integrates it with an autonomous vehicle to use underwater. cation of UWSNs that take the aid of AUVs to localize
Another system called the Buried Object Scanning Sonar submarines. Conventional ways of submarine localization
(BOSS) would be placed with the RTG in order to optimize require heavy computation. These UWSNs based systems
the performance of clutter avoidance. The sensor has been tend to provide economical solutions to the localization
placed in the nose of the vehicle to avoid electromagnetic problems.
interferences from the rest of the vehicle. In [47], the authors have investigated the localization of
Another work [43] focused on designing a system for submarines and focused on two problems that are limited
mine detections in which an AUV is employed with a sensor communication range due to acoustic transmission and
that is located on its side. The sensor is capable of producing incorporating sonar signal processing on the AUVs. The trial,
very high resolution images that can result in localizing called the GLINT09, was experimented close to Elba, an
the underwater mines. Here the possibility of error is also island in Italy at a depth of 110 meters. It was concluded in
addressed by capturing multiple view of the same area. The the experimentation that AUVs provide an optimal solution
system is efficient in a way that the route of the AUV can to localization of submarines with less manpower, less risk,
be predicted even before the deployment procedure by using and more space in order to realize the 3D imaging.
Synthetic Aperture Sonar (SAS).
The authors in [48] propose a paradigm based on dis-
Authors in [44] use a pattern analyzer by taking images
tributed sensor network for submarine detection in places
underwater. The pattern analyzer can detect mines and
where the submarines are not easily tracked by the sonar
similar structures. These images that are taken underwater
system. This paradigm estimates the Cramer-Rao lower
are compared with an atlas of images of mines that have been
bound for the purpose of accuracy of estimation. The results
provided by the Naval Surface Warfare Center.
are in the form of either presence of the submarine or
In [45], a simulation based model has been created that
no presence of the submarine (binary results). The results
considers five different types of AUVs and two sonar types
have also been extended from single submarine detection to
(Raytheon and Klein). The software model is implemented
multiple submarines detection.
in different areas such as the Chesapeake Bay and is tested
for different types of mines and area and the way the mine is In another work [49], Particle Swarm Optimization
cleared. The authors showed that the Raytheon type of sonar, (PSO) for estimating the sensor position considers the atten-
if used with any of the vehicle types, would consume more uation, water depth, and transmission range.
energy as compared to Klein sonar.
In another work [46], comparing the classifiers that 3.3.3. Surveillance. UWSNs are also used for surveillance
use machine-learning approach and computer vision. The purposes where the intrusion of any undesired entity is
work corresponds to taking the aid of semisynthetic data detected. This can be used for near- and offshore surveillance
for preprocessing of data before it is given as an input to purposes. This has a huge set of applications offered for near-
machine-learning and computer vision approaches. This way, shore surveillance such as detection of battle ships and arrival
the amount of data set given to the approaches is reduced. of logistics.
Table 3: Comparison of UWSN military applications.

App. Implementation
Mines
[42] 1D Sea n/a UUV-RTG Few Acoustic Real-time
[43] 1D Sea 20 m Camera n/a Acoustic Real-time
[44] 1D Sea n/a Camera Few Acoustic Software based
[45] 1D Bay n/a n/a Few Acoustic Simulation
[46] n/a Seabed n/a n/a n/a n/a Software
Submarines
[47] 2D Sea 110 m n/a n/a Acoustic Real-time
[48] 4D Sea n/a Localization n/a n/a Real-time
[49] 4D Seabed 0–100 m Localization n/a Acoustic Simulation
Surveillance
[50] 2D Sea 100–150 m n/a Many Acoustic Real-time
[51] 4D Sea 0–100 m n/a Many Acoustic Real-time
[52] 3D Seabed n/a Magnetic, radiation Many Wired Simulation
The GLINT10 [50] field test trials were performed in preferred. For surveillance, we have experimented with 2D,
order to test the warfare surveillance. The work focuses on 3D, and 4D architectures.
signal processing capabilities of the said project. The results
showed that AUVs are in probation to become a complete 3.4. Assisted Navigation. Underwater environment is ex-
autonomous system in dynamic environment but so far they tremely uneven, unexplored, random, and dark with increas-
do not possess enough capability of operating autonomously. ing depth. In such environment there is need of assistance for
Underwater acoustic network (UAN) [51] is an EU fund- navigating the vessels, boats, ships, and even swimmer and
ed project that was developed to protect offshore and near- explorers. Navigating assistive technologies which are most
shore equipment and infrastructures. The UAN11 project held common over the surface of water are not used underwater
in 2011 in Norway experiments the AUVs for the surveillance due to the change in medium of transmission. Therefore, in
of nodes deployed on the site. The experiment consisted of order to locate, guide, and navigate, there is a need of assistive
four nodes, two AUVs, and a node that is mobile on a vessel. navigation technologies, system, and application which could
The network resulted in a very robust performance by being be used underwater. In this prospect, UWSN can be used
able to deal with variations such as when a node was added to provide assistive navigation systems and applications.
or subtracted from the network. They also presented their Following are the UWSN applications discussed.
results based on average delivery ration, packet loss, and A detailed literature regarding underwater localization
round-trip time. could be found in [68]. The authors in [53] have used AUV
In [52], the authors have suggested a new architecture to assist in localization underwater. The designed system is
layout of underwater surveillance system which consists of acoustics based and uses acoustic localization techniques. The
sensors deployed on the surface first; then the depth and authors have tested the system in multiple stages (denser
the location where the maximum coverage is achieved are nodes) and compared the performance on the basis of local-
selected. The nodes are considered to bear different types ization coverage, accuracy, and communication costs. Thus
of sensors such as magnetic, radiation, mechanical, and they concluded that there lays a tradeoff between localization
acoustic. Data mining techniques are further used in order at multistage and higher communication cost.
to differentiate between different types of objects such as Guo and Liu [54] have proposed anchor-free localization
submarines, divers, mines, and vehicles. algorithm for underwater sensor location assistance. The
Full activated systems have also been deployed; for algorithm designed relies on the information of adjacent
example, in Japan [67], a tsunami monitoring system has been nodes. The proposed technique is tested for both static and
deployed that consists of 150 underwater ocean observatories. mobile UWSNs.
UWSN military applications are summarized in Table 3. For Carroll et al. [55] have proposed an on-demand under-
military applications, acoustics have been more preferred due water localization algorithm for locating the underwater
to longer distance coverage through the UWSN. Also com- sensors. The authors have developed a mathematical model of
bination of RF and acoustics is preferred mainly due to the the designed algorithm which has been embedded and tested
fact that RF travels faster compared to acoustics underwater; in real time in small pond and pools.
hence they facilitate quick transmission and responses in The summary of UWSN assisted navigation applica-
case of the emergencies. For mine detection, 1D architectures tions is given in Table 4. Navigation applications take the
have been more preferred due to stationary position of the aid of acoustic communication to cover longer distances.
mines. For submarines, 2D and 4D architectures have been Localization, tracking, and trilateration applications require
Table 4: Comparison of UWSN assisted navigation applications.

App. Implementation
[53] 4D n/a n/a Localization Many Acoustic Prototype
[53] 4D n/a n/a Tracking, sonar Many Acoustic Real-time
[54] 4D, 3D n/a n/a Localization n/a Acoustic Real-time
[55] 4D Pool, pond Few meters Triliterization Many Acoustic Real-time
Table 5: Comparison of UWSN sport applications.

App. Implementation
[56] 2D Pool n/a Accelerometer, gyroscope Few Acoustic Real-time
[57] n/a Pool n/a Magnetic field Few RF Real-time
[58] n/a Pool n/a Various Many n/a n/a
[59] n/a Pool 0–10 cm Camera, pressure, and force Few Acoustic Real-time
[60] n/a Pool n/a Various Many n/a n/a
[61] 2D Pool n/a Accelerometer Few Optical Real-time
[62] 2D Pool 10 cm Accelerometer Few Optical Real-time
an UWSN that communications with the ROV to acquire data has also been used as most of the applications that fall in
from the anchor nodes and relay the data to the remote station this category use a low coverage area (such as in pools).
(4D architecture). 2D architectures have been employed because numbers of
sensors in form of clusters are suspended in the water to
3.5. Sports. A wide set of UWSN applications exist for collect the required information that would communicate
underwater sports category as well. These applications are with the cluster head (acting as a remote station when
different from other UWSN applications in terms of velocity distances are short).
of the mobility of the nodes, sensing parameters, and so forth.
In [56] the authors have presented a wireless sensor 4. Challenges and Opportunities
network that is used to monitor the performance of a
While UWSN is a promising new field and may help in
swimmer or multiple swimmers. The networks bandwidth is
exploring the unfathomed world that lies underwater, there
put into consideration in this work, where the performance is
are many challenges and opportunities as well.
communicated to the coach and other swimmers simultane-
ously. Also in [69], a similar motive is achieved that requires Unpredictable Underwater Environment. Underwater condi-
real-time feedback to coaches. tions are extremely unpredictable. The anonymous high water
Another work, [57], focuses on finding the location of pressure, unpredictable underwater activities, and uneven
swimmers in a pool. This is done by utilizing magnetic depths of the underwater surface make it difficult to design
variations that are caused by the pools. This method is said and deploy UWSNs.
to be more accurate as compared to inertial positioning in
which postprocessing is required. Intricate Network Design and Deployment. Due to the unpre-
The authors in [58] have also investigated the localization dictable underwater environment, it is extremely difficult
of swimmers by analyzing the use of magnetic, inertial, to deploy the network underwater which works reliably
accelerometers, magnetometers, and gyroscope sensors. and wirelessly. The current tethered technology allows con-
In [59], the signal processing of the data achieved from a strained communication but it incurs significant cost of
swimmers performance is performed. This shortlists the data deployment, maintenance, and device recovery to cope with
that is sent to the coaches. volatile undersea conditions.
In [60], an optical system is used for transmission of
data in real time for communicating real-time feedback of Unscalability. Traditional underwater exploration relies on
swimmers to coaches. Also in [61], an optical wireless system either a single high-cost underwater device or a small-
is used but it focuses on communicating the stroke rate scale underwater network. Neither existing technology is
during swimming. Again in [62], a wireless optical modem suitable for applications covering a large area. Enabling a
based on FSK is used for communicating the swimmers scalable underwater sensor network technology is essential
feedback. for exploring a huge underwater space.
Applications in UWSN sport are summarized in Table 5.
As far as sports applications are considered, RF transceivers Unreliable Information. Underwater nodes are in continuous
are widely used; however, a trend of using optical signals motion due to the water currents; thus locating nodes
underwater becomes much more crucial. Traditional posi- is still a room for ample contributions particularly in the
tioning and localization systems do not work underwater. physical deployments of the systems on large scale.
Therefore, underwater conditions dismantle the location of
the nodes and the network topology which ultimately makes Conflict of Interests
the information transmission unreliable.
The authors declare that there is no conflict of interests
Requirement of Novel Protocols for UWSNs. In underwater regarding the publication of this paper.
communication, the medium of communication is water,
unlike air as in terrestrial sensor networks. Therefore, ter-
restrial sensor network communication protocols get void Acknowledgments
underwater. Mostly, acoustic signals are used for underwater
This work is supported by the NSTIP Strategic Technologies
communication over large distances, while radios are con-
Program no. 11-INF1688-10, King Abdul Aziz City of Science
sidered for short-distance, water surface communication. But
and Technology of the Kingdom of Saudi Arabia. The authors
radio signals transmit for long distances at extra low frequen-
thank the Science and Technology Unit for their logistic
cies, which requires large antennas and high transmission
support.
power [1], which can decrease the overall network lifetime of
UWSNs. Moreover, the propagation delay of acoustic com-
munication is very high compared to RF communication; References
hence many algorithms and protocols for terrestrial WSN
[1] I. F. Akyildiz, D. Pompili, and T. Melodia, “Underwater acoustic
cannot be adapted directly to UWSN.
sensor networks: research challenges,” Ad Hoc Networks, vol. 3,
no. 3, pp. 257–279, 2005.
Low Data Rates. Radio frequency (RF) communications are
not effective in underwater communication due to medium [2] S. Iyer and D. V. Rao, “Genetic algorithm based optimization
technique for underwater sensor network positioning and
effect on communication. Water absorbs much of the RF
deployment,” in Proceedings of the IEEE Underwater Technology
energy and hence only very short range communication is (UT ’15), pp. 1–6, IEEE, Chennai, India, Feburary 2015.
allowable using RF [4]. Instead, acoustic communication is
[3] E. Felemban, “Advanced border intrusion detection and surveil-
being used to transmit pulse signals and low fidelity informa- lance using wireless sensor network technology,” International
tion underwater due to its low bandwidth. Potential UWSN Journal of Communications, Network and System Sciences, vol.
applications such as measuring the amount of pollution from 06, no. 05, pp. 251–259, 2013.
a fishing farm at the seabed [21] require transmitting lots of [4] J. Heidemann, W. Ye, J. Wills, A. Syed, and Y. Li, “Research
data. However, with such low frequencies, it requires a lot of challenges and applications for underwater sensor networking,”
time to send such dynamic data. in Proceedings of the IEEE Wireless Communications and Net-
working Conference (WCNC ’06), pp. 228–235, April 2006.
Physical Damage to Equipment. The sensors used in underwa- [5] G. A. Hollinger, S. Choudhary, P. Qarabaqi et al., “Underwater
ter devices are susceptible to routine underwater challenges, data collection using robotic sensor networks,” IEEE Journal on
for example, algae collection on camera lens [70] and salt Selected Areas in Communications, vol. 30, no. 5, pp. 899–911,
accumulation, decreasing the effectiveness of sensors and so 2012.
forth. [6] J.-H. Cui, J. Kong, M. Gerla, and S. Zhou, “The challenges
of building mobile underwater wireless networks for aquatic
Cost. Finally, the energy requirements and cost of UASNs applications,” IEEE Network, vol. 20, no. 3, pp. 12–18, 2006.
are high compared to higher power and regular battery [7] M. C. Domingo and R. Prior, “Energy analysis of routing
replenishing techniques are quite costly. protocols for underwater wireless sensor networks,” Computer
The amount of challenges in designing of UWSNs makes Communications, vol. 31, no. 6, pp. 1227–1238, 2008.
it an interesting area for researchers to work on. With the [8] S. Yoon, A. K. Azad, H. Oh, and S. Kim, “AURP: an AUV-aided
advancement in sensor and wireless technologies, UWSNs underwater routing protocol for underwater acoustic sensor
have attracted a lot of researchers and have contributed networks,” Sensors, vol. 12, no. 2, pp. 1827–1845, 2012.
significantly to this field. However, the window is still wide [9] A. Yalcuk and S. Postalcioglu, “Evaluation of pool water quality
open for upcoming research and opportunities. of trout farms by fuzzy logic: monitoring of pool water quality
for trout farms,” International Journal of Environmental Science
and Technology, vol. 12, no. 5, pp. 1503–1514, 2015.
5. Conclusion [10] G. Tuna, O. Arkoc, and K. Gulez, “Continuous monitoring
of water quality using portable and low-cost approaches,”
In this paper, we have presented a comprehensive literature International Journal of Distributed Sensor Networks, vol. 2013,
review of UWSN applications and their classification. It was Article ID 249598, 11 pages, 2013.
observed that a good number of applications are assisted by [11] K. Menon, P. Divya, and M. Ramesh, “Wireless sensor network
UWSNs for the harsh underwater environment. UWSN has for river water quality monitoring in India,” in Proceedings of
become one of the prime focuses for researchers. If these the 3rd International Conference on Computing Communication
applications are properly exploited and catered, a lot of lives, Networking Technologies (ICCCNT ’12), pp. 1–7, July 2012.
time, and money can be saved. Although UWSNs have seen [12] A. Faustine, A. N. Mvuma, H. J. Mongi, M. C. Gabriel, A. J.
a tremendous amount of growth in the past few years, there Tenge, and S. B. Kucel, “Wireless sensor networks for water
quality monitoring and control within lake victoria basin: Communications, Networking and Mobile Computing (WiCOM
prototype development,” Wireless Sensor Network, vol. 06, no. ’12), pp. 1–5, September 2012.
12, pp. 281–290, 2014. [28] N. Mohamed, I. Jawhar, J. Al-Jaroodi, and L. Zhang, “Sensor
[13] M. Shakir, M. A. Khan, S. A. Malik, and Izhar-ul-Haq, “Design network architectures for monitoring underwater pipelines,”
of underwater sensor networks for water quality monitoring,” Sensors, vol. 11, no. 11, pp. 10738–10764, 2011.
World Applied Sciences Journal, vol. 17, no. 11, pp. 1441–1444, [29] I. Jawhar, N. Mohamed, and K. Shuaib, “A framework for
2012. pipeline infrastructure monitoring using wireless sensor net-
[14] A. Khan and L. Jenkins, “Undersea wireless sensor network works,” in Proceedings of the Wireless Telecommunications Sym-
for ocean pollution prevention,” in Proceedings of the 3rd posium (WTS ’07), April 2007.
International Conference on Communication Systems Software [30] H. Saeed, S. Ali, S. Rashid, S. Qaisar, and E. Felemban, “Reliable
and Middleware and Workshops, pp. 2–8, Bangalore, India, monitoring of oil and gas pipelines using wireless sensor
January 2008. network (wsn): remong,” in Proceedings of the 9th International
[15] G. Açar and A. E. Adams, “ACMENet: an underwater acoustic Conference on System of Systems Engineering (SoSE ’14), pp. 230–
sensor network protocol for real-time environmental moni- 235, June 2014.
toring in coastal areas,” IEE Proceedings: Radar, Sonar and [31] H. Yu and M. Guo, “An efficient oil and gas pipeline monitoring
Navigation, vol. 153, no. 4, pp. 365–380, 2006. systems based on wireless sensor networks,” in Proceedings of
[16] C. Alippi, R. Camplani, C. Galperti, and M. Roveri, “A robust, the 3rd International Conference on Information Security and
adaptive, solar-powered WSN framework for aquatic environ- Intelligent Control (ISIC ’12), pp. 178–181, IEEE, August 2012.
mental monitoring,” IEEE Sensors Journal, vol. 11, no. 1, pp. 45– [32] A. A. Pasi and U. Bhave, “Flood detection system using wireless
55, 2011. sensor network,” International Journal of Advanced Research in
[17] J. Trevathan, R. Johnstone, T. Chiffings et al., “SEMAT—the next Computer Science and Software Engineering, vol. 5, no. 2, pp.
generation of inexpensive marine environmental monitoring 386–389, 2015.
and measurement systems,” Sensors, vol. 12, no. 7, pp. 9711–9748, [33] S. Tyan and S.-H. Oh, “AUV-RM: underwater sensor network
2012. scheme for AUV based river monitoring,” Research Trend in
[18] C. A. Perez, M. Jimenez, F. Soto, R. Torres, J. A. Lopez, and Computer and Applications, SERCE, vol. 24, pp. 53–55, 2013.
A. Iborra, “A system for monitoring marine environments [34] R. Marin-Perez, J. Garcı́a-Pintado, and A. S. Gómez, “A real-
based on wireless sensor networks,” in Proceedings of the IEEE time measurement system for long-life flood monitoring and
OCEANS, pp. 1–6, IEEE, Santander, Spain, June 2011. warning applications,” Sensors, vol. 12, no. 4, pp. 4213–4236,
2012.
[19] S. Zhang, J. Yu, A. Zhang, L. Yang, and Y. Shu, “Marine vehicle
sensor network architecture and protocol designs for ocean [35] E. N. Udo and E. B. Isong, “Flood monitoring and detection sys-
observation,” Sensors, vol. 12, no. 1, pp. 373–390, 2012. tem using wireless sensor network,” Asian Journal of Computer
and Information Systems, vol. 1, no. 4, pp. 108–113, 2013.
[20] H. Yang, H. Wu, and Y. He, “Architecture of wireless sensor net-
work for monitoring aquatic environment of marine shellfish,” [36] P. Kumar, P. Kumar, P. Priyadarshini, and Srija, “Underwater
in Proceedings of the 7th Asian Control Conference (ASCC ’09), acoustic sensor network for early warning generation,” in
pp. 1147–1151, IEEE, August 2009. Proceedings of the Oceans, pp. 1–6, IEEE, Hampton Roads, Va,
USA, October 2012.
[21] J. Lloret, S. Sendra, M. Garcia, and G. Lloret, “Group-based
[37] K. Casey, A. Lim, and G. Dozier, “A sensor network architecture
underwater wireless sensor network for marine fish farms,” in
for tsunami detection and response,” International Journal of
Proceedings of the IEEE GLOBECOM Workshops (GC Wkshps
Distributed Sensor Networks, vol. 4, no. 1, pp. 28–43, 2008.
’11), pp. 115–119, IEEE, Houston, Tex, USA, December 2011.
[38] A. Khan and L. Jenkins, “Undersea wireless sensor network
[22] M. Lpez, S. Martnez, J. Gmez et al., “Wireless monitoring of the
for ocean pollution prevention,” in Proceedings of the 3rd
pH, NH4+ and temperature in a fish farm,” Procedia Chemistry,
International Conference on Communication Systems Software
vol. 1, no. 1, pp. 445–448, 2009.
and Middleware and Workshops (COMSWARE ’08), pp. 2–8,
[23] L. A. Abdul-Rahaim and A. M. A. Ali, “Remote wireless January 2008.
automation and monitoring of large farm using wireless sensors [39] C. O. Iwendi and A. R. Allen, “Wireless sensor network nodes:
networks and internet,” International Journal of Computer security and deployment in the niger-delta oil and gas sector,”
Science & Engineering Technology, vol. 6, no. 3, pp. 118–137, 2015. International Journal of Network Security & Its Applications, vol.
[24] N. Nasser, A. N. K. Zaman, L. Karim, and N. Khan, “CPWS: 3, no. 1, pp. 68–79, 2011.
an efficient routing protocol for RGB sensor-based fish pond [40] A. Koulakezian, R. Ohannessian, H. Denkilkian et al., “Wireless
monitoring system,” in Proceedings of the IEEE 8th International sensor node for real-time thickness measurement and localiza-
Conference on Wireless and Mobile Computing, Networking and tion of oil spills,” in Proceedings of the IEEE/ASME International
Communications (WiMob ’12), pp. 7–11, October 2012. Conference on Advanced Intelligent Mechatronics (AIM ’08), pp.
[25] A. Pirisi, F. Grimaccia, M. Mussetta et al., “Optimization 631–636, IEEE, Xi’an, China, August 2008.
of an energy harvesting buoy for coral reef monitoring,” in [41] O. N. H. Nweke and F. Henry, “Wireless sensor networks
Proceedings of the IEEE Congress on Evolutionary Computation basedpipeline vandalisation and oil spillagemonitoring and
(CEC ’13), pp. 629–634, IEEE, Cancun, Mexico, June 2013. detection: mainbenefits for nigeria oil and gas sectors,” The SIJ
[26] S. Bainbridge, D. Eggeling, and G. Page, “Lessons from the field- Transactions on Computer Science Engineering & Its Applica-
two years of deploying operational wireless sensor networks on tions, vol. 3, no. 1, pp. 1–6, 2015.
the great barrier reef,” Sensors, vol. 11, no. 7, pp. 6842–6855, 2011. [42] S. Kumar, A. Perry, C. Moeller et al., “Real-time tracking
[27] S. Srinivas, P. Ranjitha, R. Ramya, and G. Narendra, “Investiga- magnetic gradiometer for underwater mine detection,” in
tion of oceanic environment using large-scale uwsn and uanets,” Proceedings of the OCEANS ’04 and MTTS/IEEE TECHNO-
in Proceedings of the 8th International Conference on Wireless OCEAN ’04, vol. 2, pp. 874–878, November 2004.
[43] D. P. Williams, “On optimal AUV track-spacing for underwater [59] T. Le Sage, P. Conway, L. Justham, S. Slawson, A. Bindel, and
mine detection,” in Proceedings of the IEEE International Con- A. West, “A component based integrated system for signal
ference on Robotics and Automation (ICRA ’10), pp. 4755–4762, processing of swimming performance,” in Proceedings of the
IEEE, Anchorage, Alaska, USA, May 2010. International Conference on Signal Processing and Multimedia
[44] C. Rao, K. Mukherjee, S. Gupta, A. Ray, and S. Phoha, Applications (SIGMAP ’10), pp. 73–79, IEEE, Athens, Greece,
“Underwater mine detection using symbolic pattern analysis of July 2010.
sidescan sonar images,” in Proceedings of the American Control [60] R. M. Hagem, D. V. Thiel, S. G. O’Keefe, and T. Fickenscher,
Conference (ACC ’09), pp. 5416–5421, June 2009. “Optical wireless communication for real time swimmers feed-
[45] S. Khaledi, H. Mann, J. Perkovich, and S. Zayed, “Design of an back: a review,” in Proceedings of the International Symposium
underwater mine detection system,” in Proceedings of the IEEE on Communications and Information Technologies (ISCIT ’12),
Systems and Information Engineering Design Symposium (SIEDS pp. 1080–1085, October 2012.
’14), pp. 78–83, April 2014. [61] R. M. Hagem, S. G. O’Keefe, T. Fickenscher, and D. V. Thiel, “Self
[46] C. Barngrover, R. Kastner, and S. Belongie, “Semisynthetic contained adaptable optical wireless communications system
versus real-world sonar training data for the classification of for stroke rate during swimming,” IEEE Sensors Journal, vol. 13,
mine-like objects,” IEEE Journal of Oceanic Engineering, vol. 40, no. 8, pp. 3144–3151, 2013.
no. 1, pp. 48–56, 2015. [62] R. M. Hagem, D. V. Thiel, S. G. O’Keefe, A. Wixted, and
[47] M. J. Hamilton, S. Kemna, and D. Hughes, “Antisubmarine T. Fickenscher, “Low-cost short -range wireless optical FSK
warfare applications for autonomous underwater vehicles: the modem for swimmers feedback,” in Proceedings of the IEEE
GLINT09 sea trial results,” Journal of Field Robotics, vol. 27, no. SENSORS Conference, pp. 258–261, IEEE, Limerick, Republic of
6, pp. 890–902, 2010. Ireland, October 2011.
[48] S. Zhou and P. Willett, “Submarine location estimation via a [63] D. He and L.-X. Zhang, “The water quality monitoring system
network of detection-only sensors,” IEEE Transactions on Signal based on WSN,” in Proceedings of the 2nd International Confer-
Processing, vol. 55, no. 6, pp. 3104–3115, 2007. ence on Consumer Electronics, Communications and Networks
[49] R. B. Manjula and S. S. Manvi, “Coverage optimization based (CECNet ’12), pp. 3661–3664, Yichang, China, April 2012.
sensor deployment by using PSO for anti-submarine detection [64] G. Xu, W. Shen, and X. Wang, “Applications of wireless sensor
in UWASNs,” in Proceedings of the 12th Symposium on Ocean networks in marine environment monitoring: a survey,” Sensors,
Electronics (SYMPOL ’13), pp. 15–22, October 2013. vol. 14, no. 9, pp. 16932–16954, 2014.
[50] S. Kemna, M. J. Hamilton, D. T. Hughes, and K. D. LePage, [65] B. Thornton, A. Bodenmann, A. Asada, T. Sato, and T. Ura,
“Adaptive autonomous underwater vehicles for littoral surveil- “Acoustic and visual instrumentation for survey of manganese
lance,” Intelligent Service Robotics, vol. 4, no. 4, pp. 245–258, crusts using an underwater vehicle,” in Proceedings of the
2011. OCEANS, pp. 1–10, October 2012.
[51] A. Caiti, V. Calabrò, A. Munafò, G. Dini, and A. Lo Duca, [66] J. Lloret, “Underwater sensor nodes and networks,” Sensors, vol.
“Mobile underwater sensor networks for protection and secu- 13, no. 9, pp. 11782–11796, 2013.
rity: field experience at the UAN11 experiment,” Journal of Field
[67] T. Kanazawa, “Japan Trench earthquake and tsunami monitor-
Robotics, vol. 30, no. 2, pp. 237–253, 2013.
ing network of cable-linked 150 ocean bottom observatories and
[52] E. Cayirci, H. Tezcan, Y. Dogan, and V. Coskun, “Wireless its impact to earth disaster science,” in Proceedings of the IEEE
sensor networks for underwater survelliance systems,” Ad Hoc International Underwater Technology Symposium (UT ’13), 5, p.
Networks, vol. 4, no. 4, pp. 431–446, 2006. 1, March 2013.
[53] M. Waldmeyer, H.-P. Tan, and W. K. G. Seah, “Multi-stage AUV- [68] G. Han, A. Qian, C. Zhang, Y. Wang, and J. J. P. C. Rodrigues,
aided localization for underwater wireless sensor networks,” in “Localization algorithms in large-scale underwater acoustic
Proceedings of the IEEE International Conference on Advanced sensor networks: a quantitative comparison,” International Jour-
Information Networking and Applications Workshops (WAINA nal of Distributed Sensor Networks, vol. 2014, Article ID 379382,
’11), pp. 908–913, March 2011. 11 pages, 2014.
[54] Y. Guo and Y. Liu, “Localization for anchor-free underwater [69] N. Chakravorti, T. Le Sage, S. E. Slawson, P. P. Conway,
sensor networks,” Computers & Electrical Engineering, vol. 39, and A. A. West, “Design and implementation of an inte-
no. 6, pp. 1812–1821, 2013. grated performance monitoring tool for swimming to extract
[55] P. Carroll, S. Zhou, K. Mahmood, H. Zhou, X. Xu, and J.-H. Cui, stroke information at real time,” IEEE Transactions on Human-
“On-demand asynchronous localization for underwater sensor Machine Systems, vol. 43, no. 2, pp. 199–213, 2013.
networks,” in Proceedings of the IEEE Oceans, pp. 1–4, IEEE,
[70] M. Jaroensutasinee, K. Jaroensutasinee, T. F. S. Bainbridge et
Hampton Roads, Va, USA, October 2012.
al., “Sensor networks applications for reefs at Racha Island,
[56] T. Le Sage, A. Bindel, P. Conway, S. Slawson, and A. West, Thailand,” in Proceedings of the 12th International Coral Reef
“Development of a wireless sensor network for embedded Symposium, Cairns, Australia, July 2012.
monitoring of human motion in a harsh environment,” in
Proceedings of the IEEE 3rd International Conference on Com-
munication Software and Networks (ICCSN ’11), pp. 112–115, May
2011.
[57] J. Marshall, “Magnetic field swimmer positioning,” IEEE Sensors
Journal, vol. 15, no. 1, pp. 172–179, 2015.
[58] F. A. D. Magalhaes, G. Vannozzi, G. Gatta, and S. Fantozzi,
“Wearable inertial sensors in swimming motion analysis: a
systematic review,” Journal of Sports Sciences, vol. 33, no. 7, pp.
732–745, 2015.
http://dx.doi.org/10.1155/2015/791046
Research Article
Proteus II: Design and Evaluation of an Integrated
Power-Efficient Underwater Sensor Node
Wouter A. P. van Kleunen, Niels A. Moseley, Paul J. M. Havinga, and Nirvana Meratnia
University of Twente, 7500 AE Enschede, Netherlands
Correspondence should be addressed to Wouter A. P. van Kleunen; w.a.p.vankleunen@utwente.nl
Received 5 June 2015; Revised 24 August 2015; Accepted 25 August 2015
Academic Editor: Wei Wang
Copyright © 2015 Wouter A. P. van Kleunen et al. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
We describe the design and evaluation of an integrated low-cost underwater sensor node designed for reconfigurability, allowing
continuous operation on a relatively small rechargeable battery for one month. The node uses a host CPU for the network
protocols and processing sensor data and a separate CPU performs signal processing for the ultrasonic acoustic software-defined
Modulator/Demodulator (MODEM). A Frequency Shift Keying- (FSK-) based modulation scheme with configurable symbol
rates, Hamming error correction, and Time-of-Arrival (ToA) estimation for underwater positioning is implemented. The onboard
sensors, an accelerometer and a temperature sensor, can be used to measure basic environmental parameters; additional internal
and external sensors are supported through industry-standard interfaces (I2C, SPI, and RS232) and an Analog to Digital Converter
(ADC) for analog peripherals. A 433 MHz radio can be used when the node is deployed at the surface. Tests were performed
to validate the low-power operation. Moreover the acoustic communication range and performance and ToA capabilities were
evaluated. Results show that the node achieves the one-month lifetime, is able to perform communication in highly reflective
environments, and performs ToA estimation with an accuracy of about 1-2 meters.
1. Introduction or a satellite link may be used to communicate to the shore.

Next to static sensors, Autonomous Underwater Vehicles
Underwater Acoustic Sensor Networks (UASNs) can be (AUVs) may be used in the network to provide dynamic links.
used in many applications, such as monitoring underwater One of the first realizations of UASNs is the Seaweb
pipelines, monitoring underwater drilling, and performing program [2]. This program has performed several experiment
environmental monitoring [1]. Traditional approaches of deployments with networks of underwater sensors and has
underwater measuring involve deploying underwater sensors performed experiments with modulation, communication,
with data logging capabilities and retrieving the sensors time synchronization, and positioning.
after recording several months of data. Such approach of For our research purposes we require an underwater
monitoring provides no real-time monitoring, no reconfig- sensor node we can reconfigure to perform experiments
uration, and no failure detection capabilities. By providing with physical, Medium Access Control (MAC), networking,
communication capabilities to the sensors, it is possible to and underwater monitoring applications. In addition, we
perform online monitoring and reconfiguration and realize require the node to be flexible in terms of the internal and
underwater networks of nodes. external sensors that can be connected. To minimize network
Figure 1 gives an example of how an underwater network maintenance, maximizing battery lifetime is crucial. Our aim
can look like. Different communication links are used; for is to have a node lifetime of more than one month in low
example, in the underwater clusters, acoustic communication transmit duty cycle scenarios. Existing available commercial
can be used to communicate data from different sensors to a underwater sensor nodes do not fit these requirements.
cluster-head. The cluster-head may then use a long-range link In this paper we describe our design of an underwater
to forward the (possibly aggregated) data towards a surface acoustic sensor node which combines sensing, processing,
station. A wireless RF link will forward the data to the shore, and communication powered by an integrated rechargeable
Satellite Surface sink
Onshore sink
Surface station
Cluster Cluster
Cluster
Surface sink UW sensor
Surface station Vertical link
UW sink Horizontal multihop link
Figure 1: Example of an underwater network [1]. Different terrestrial communication links are used in the network, acoustic links between
sensors underwater, and RF links between buoys and onshore stations.
battery. It is organized as follows. In Section 2 we discuss and the MODEMs of Evologics [4]. The WHOI MicroModem
related work and existing MODEMs used for underwater [5] is a MODEM designed for navigation and communication
sensing. In Section 3 we discuss the design of our integrated with AUVs. These MODEMs implement functionality for
underwater sensor node. We discuss the functionality pro- underwater communication and in most cases an external
vided on the board, the software design of the MODEM, more CPU is required to perform networking and driving sensors.
specifically the modulation/demodulation of the FSK signal, The Kongsberg Mini node [3] uses Direct-Sequence
and the ToA estimation approach. Moreover we discuss the Spread Spectrum (DSSS) and Quadrature Phase Shift Keying
software functionality of the host processor and the net- (QPSK) modulation schemes. The Evologics MODEMs use
working protocols used in the experiment setup. In Section 4 their proprietary sweep-spread carrier system [6]. The WHOI
the results of the performance evaluation of the node are MicroModem is based on Frequency-Hopping Frequency
presented. We have measured the power consumption of Shift Keying (FH-FSK) and uses an optional floating point
the node in different operation mode, performed battery coprocessor to implement Phase Shift Keying (PSK) modu-
lifetime tests, and performed initial communication and ToA lation for data rates from 300 to 5000 bps. These modulation
estimation tests. schemes allow for higher symbol rates compared to FSK
but require more computational power and therefore require
2. Related Work more expensive and power consuming processors.
The underwater node described in [7] has features similar
There are several commercial underwater acoustic MODEMs to our proposed design; it uses a two-processor architecture.
solutions available, for example, the Kongsberg Mini node [3] The node uses a low-power controller CPU for system
management and a more powerful Digital Signal Processor

(DSP) for the MODEM. The whole system can be placed in
a low-power sleep mode when no acoustic communication
is performed. In receive mode, the node requires more than
600 mW to operate, which is too high for our purposes.
Many of the existing platforms were not designed for low-
power operation. Moreover, commercial platforms are closed
designs and the MODEMs software cannot be adapted to Figure 2: A picture of the Proteus board and enclosure. The board
perform experiments. and battery (not shown) are placed in the enclosure to make an
The MODEMs outlined above do not provide the network integrated underwater sensor node. Connectors on the lid of the
layer, the sensor integration, or the required data processing enclosure allow external sensors to be connected. The lid is also fitted
capabilities to deploy complete UASNs. For the network layer, with an RF connector for the 433 MHz radio antenna. The node can
a networking library could be used. be completely submerged; however no 433 MHz radio connection
An example of an external networking library which will be available when doing so. At the bottom end of the enclosure,
can drive various MODEMs using a separate CPU is the the (black) acoustic transducer is visible and the transducer operates
Sunset framework [8]. This framework allows using the same in the frequency range of 20–40 kHz.
code base for both simulation and deployment, easing the
development of and the experimentation with the network
layer. This flexibility comes at a cost however; the framework
Host CPU Modem
requires the Linux operating system, which runs on high-end
power consuming CPU. For example, their recommended
Gumstix [9] platform draws between 0.5 W and up to 1 W
even when completely powered down. Simply running such Data bus
a system in standby mode for a month would require a
minimum battery capacity of 216 Ah at 3.3 V, which is far from
practical. External Sd card/ 433 MHz
Underwater sensor nodes that fit our (research) require- Sensors sensor memory radio
ments are simply not available. We therefore designed our interface
own low-power extensible nodes from the ground up. Figure 3: A block diagram showing the architecture of the Proteus II
In [10] we briefly described the design of our first node. The host CPU can transmit information to other nodes via the
generation low-cost underwater node, the SeaSTAR node, MODEM. Various subsystems, such as the sensors, SD card, and the
based on a single ARM Cortex-M3 processor. The MODEM 433 MHz radio, are interfaced to the host processor via the internal
implemented FSK modulation and ToA estimation. The data bus.
node’s power consumption was still quite high because a
development board was used for the Cortex-M3. A 433 MHz
surface radio interface was used for collecting measurements
when the node was deployed at the surface. This proved to be the software-defined radio paradigm [11]. The MODEM
very useful for performing experiments and our new design also provides ToA estimation for time synchronization and
also incorporates a surface radio. positioning.
The MODEM needs to perform CPU-intensive decoding
of the acoustic signal, requiring the processor to run in
maximum performance mode. The host CPU will mostly be
3. Design of the Underwater Sensor Node dormant and run software timers to control the sensing and
transmission of data. This processor can therefore be run in
The following section describes the hardware and software
the very low-power run mode available on the CPU.
design of the Proteus II sensor node. First we will discuss the
The onboard sensors, which consist of an accelerometer
hardware components on the board and their functionality.
and a temperature sensor, can be used to measure basic envi-
The Proteus II sensor uses two processors, one for the
ronmental parameters such as surface wave characteristics,
MODEM functionality and one for the MAC and networking
node orientation, and water temperature. The node supports
protocols. After discussing the hardware design, we discuss
additional internal or external sensors through its industry-
the software design of the HOST and the MODEM processor.
standard digital serial interfaces (I2C, SPI, and RS232) and
We also discuss the physical modulation and packet format
can interface analog peripherals through a 16-bit ADC. A
of the acoustic communication and the ToA estimation
433 MHz radio provides an additional way of communication
approach used for time synchronization and positioning.
when the nodes are deployed on the water surface.
The Proteus II sensor node (see Figure 2) is built around
Figure 3 shows the architectural block diagram of the
two low-cost low-power 32-bit ARM Cortex-M4 processors.
Proteus II node.
The host CPU is responsible for the network protocols
and processing the data of onboard and optional exter-
nal sensors. The second CPU is dedicated to the acoustic 3.1. Hardware Architecture. Figure 4 shows the components
MODEM functionality, which is implemented following available on the Proteus II board. The board can be powered
Class D
3.3 V regulator 5 V regulator amplifier
433 MHz radio
RS232 level-converter Analog downmixer

and external
sensor interface Host CPU Modem CPU
SD card
Figure 4: A photograph of the Proteus II PCB, showing the location of various subsystems, such as the CPUs, the receive downmixer, the
amplifier, voltage regulators, SD card, and the 433 MHz radio.
from a battery or an external power in the range of 12 V– 3.2. Software Architecture: Host Processor. The host processor
24 V. The input voltage is converted to 3.3 V and 5 V by has many duties. It processes the data from the sensors,
two switching regulators. The regulators were selected for runs the network and routing layer software, drives the
high energy efficiency even during the ultra low-power MODEM, and controls the 433 MHz radio. The 433 MHz
(sleep) operation. The 3.3 V regulator powers all the digital radio is used for network debugging, configuring the sensors,
circuitry while the 5 V regulator powers the analog part of and performing time synchronization when the sensor has
the MODEM. When the MODEM is not operating, the 5 V been deployed at the water surface. In addition, the host
regulator is turned off to save power. To further save power, processor is responsible for the power management of the
the host processor is placed in low-power mode when the node.
node is sleeping. Because the host processor is an embedded processor,
In receive mode, the signal from the acoustic transducer is with limited memory (64 kb) available and no Memory
amplified, filtered, and downconverted to baseband, resulting Protection Unit (MPU), an operating system such as Linux
in a quadrature signal. A 24-bit stereo sigma-delta is used will not run on this processor. It is our believe that small,
to convert the received signal into the digital domain where energy optimized operating systems such as FreeRTOS [12]
it is further processed by the MODEM processor. The main or TinyOS [13] are more suited for energy-efficient sensors.
benefit of using a sigma-delta converter is the much relaxed Inspired by these small (real-time) operating systems, we
requirements on the antialiasing filters. This reduces the have implemented the Proteus II operating system. This runs
design complexity and saves board space. Both the downcon- on both the host and the MODEM processor. The operating
verter center frequency and the sample rate of the ADC are systems consist of a low-power time scheduling framework.
under software control. Different periodic tasks can be scheduled and the operating
In transmit mode, an FSK signal is generated by toggling a systems control that when idling, the host processor goes
general-purpose output pin through a timer on the MODEM into a low-power sleeping state with wakeup timers running.
processor. The output pin drives the class D amplifier and Moreover the operating system is able to handle events, such
acoustic transducer. The timer, which is under software as incoming SPI or UART communication or a radio packet
control, directly determines the frequency of the FSK signal. arriving from the 433 MHz radio, waking up the processor
The Proteus II node has several sensors and sensor inter- from low-power mode when such event occurs.
faces on board. The onboard sensors are an accelerometer Using this operating system, the different processors
(ADXL362) and a temperature sensor. Additional analog and can be placed in sleep mode. The MODEM processor and
digital sensors are supported through the onboard 16-bit associated MODEM electronics are put into sleep mode when
ADC and the serial interfaces (I2C, SPI, and RS232-level no communication is required. This is controlled by the host
UART), respectively. For data logging, the node can be fitted processor through the SPI interface. The host processor is able
with an SD card. Finally, the node features a 433 MHz radio to run on a wakeup timer and shut down the MODEM. In this
for RF communication. This, however, only works when the mode, the overall power consumption of the node is reduced
node is deployed at the surface. to less than 8 mW.
Figure 2 illustrates the Proteus II board and the designed In our test network we have implemented a time-division
enclosure. Rechargeable Lithium-Polymer (LiPo) battery multiple access (TDMA) network protocol designed for
packs of up to 5000 mAh@12 V fit into the enclosure to power performing experiments. Because we wanted to evaluate the
the board. Packet Delivery Rate (PDR) achievable by the node within
With the integrated sensors, data processing, and com- our test environment, we used TDMA protocol to avoid
munication capabilities the node is an integrated flexible collisions. The networking protocol uses both the RF radio
sensor platform. and the underwater MODEM to perform communication.
Mark Timing
detector recovery
Downconverter ADC +
−
Acoustic Space MLSE
transducer Bits out
detector detector
ToA
ToA out
correlator
Class D Modulator Bits in

amp.
Hardware Software
Figure 5: A block diagram of the hardware and software partitioning of the MODEM physical layer. In the receive path, the acoustic
transducer signal is sampled and downconverted to baseband. Two tone detectors detect the energy at the Mark (1) and Space (0) frequencies.
A maximum-likelihood sequence estimator (MSLE) performs the symbol detection and timing-recovery and then it outputs the received
bits. In the transmit path, a modulator converts the incoming bits into symbols. The symbols are amplified and transmitted by the acoustic
transducer.
Time is divided in 4-second timeslots and nodes in the The described networking protocol is geared towards a test
network are assigned a static timeslot before deploying the network, for experimenting with physical, MAC networking,
network. The gateway uses the RF radio to poll the different routing, and positioning algorithms, and the intended usage
sensors in their assigned timeslots. In the poll message to the of the network.
sensor the following data is included. Because we use a TDMA approach, the sensor is able
to sleep between the polling cycles, thereby reducing the
(i) Command. It indicates what type of message should be power consumption of the sensor. In other projects the
transmitted by the sensor: a Time-of-Flight (ToF) beacon, a host processor can be reconfigured for supporting different
sensor measurement or nothing should be transmitted at all. sensors, performing networking algorithms, and positioning
During our tests we transmit 3 ToF measurements and every or implementing network security using data encryption and
fourth packet is a measurement packet which includes mea- validation.
surements from the accelerometer and temperature sensor.
(ii) Baud Rate. It indicates what baud rate should be used for
the acoustic transmission when sending the acoustic packet. 3.3. Software Architecture: MODEM Processor. Following
the software-radio paradigm, the majority of the acoustic
(iii) Timestamp. It indicates the current time of the gateway. A transmit and receive functions are implemented in software
ToF beacon message includes this timestamp, such that when and run on the MODEM processor. Figure 5 shows the
the packet arrives at the gateway, the round-trip time can be hardware/software partitioning of the physical layer.
calculate. The Proteus II uses FSK modulation with fixed frequen-
cies around 21 kHz. FSK is chosen for its low computa-
The poll message indicates what type of packet should be tional complexity, its relatively good multipath resilience,
transmitted by the sensor and what settings should be used. and its constant envelope property, which greatly simplifies
Upon a poll message, the sensor will reply with a statistics the power amplifier design. Communication speed of the
message using the RF radio. MODEM is slow (10 to 200 baud), but the speed can be
improved in the future by using multitone FSK.
(iv) Battery Power. It indicates the battery power voltage In the receive mode, the quadrature baseband signal from
reading. the ADC is processed by two tone detectors, which operate
This setup allows us to configure the behaviour of the at 8 ksps sampling rate. Their outputs are resampled at 16
sensors and provides debugging information. It allows us times the baud rate to reduce the computational complexity
to centrally control the network, experiment with baud of the following processing blocks. A maximum-likelihood
rates, and perform ToF and PDR measurements. In future sequence estimator is used to determine the most likely 4-
work we would like to extend this with configuration of symbol sequence of the most recently received data. Then bits
the packet type (BCH or Reed-Solomon with configured are extracted and if necessary, the symbol timing is adjusted.
error correction overhead) and with routing information The resulting bit-stream is fed into the packet decoder. See
(indicating which sensors should send to which sensor). Figure 5 for an overview of the physical layer processing.
Sync Preamble Len Payload CRC Len Payload CRC
Figure 6: Proteus II packet format: the packet consists of sync 32-bits 32-bits 32-bits 32-bits
bits for time synchronization, a preamble with low autocorrelation
properties to allow for ToA estimation, and packet start detection. 32-bits 6-bits 32-bits 6-bits 32-bits 6-bits 32-bits 6-bits
ECC ECC ECC ECC
The payload of the packet is modulated using the Hamming error
correcting approach. The CRC is used for validation of the packet Figure 7: Hamming based error correction is performed on
data upon reception. subblocks of the payload + CRC. The number of error correction
bits for the Hamming code is fixed and error correcting capabilities
for the Hamming code cannot be configured.
The packet format used by the Proteus II node is defined
as follows. The packet consists of several sync bits (10101010),
techniques for estimating ToA of ultrawideband (UWB)
which allows the bit synchronization to be performed before
signals have been proposed. These algorithms can also be
the actual data of the packet data is received; see Figure 6. It is
used for estimating the arrival of an acoustic signal. We use
followed by a Barker sequence preamble [14]. The Barker code
the “search and subtract” algorithm [16], which works as
has low autocorrelation properties allowing accurate packet
follows:
arrival detection and ToA estimation. Bit synchronization
and ToA estimation are further improved by using data- (1) Find the strongest arrival in the cross-correlation.
whitening on the packet payload. Error protection is provided (2) Set the first arrival threshold to 1/4 of the peak of the
through a low-complexity Hamming code based Forward strongest arrival.
Error Correction (FEC) scheme and a Cyclic Redundancy
Check (CRC). (3) Subtract the strongest arrival from the cross-cor-
After the sync bits and Barker code preamble, the length relation result.
of the payload (including its CRC) is transmitted. The 16-bit (4) Search again for the next strongest arrival that is
CRC is added at the end of the packet to validate the packet received before all other strongest arrivals and is
upon reception. A CRC polynomial was selected which has stronger than 1/4 of the strongest arrival.
good bit-error detecting properties [15]. Every 32-bit block of (5) Subtract this arrival and go back to step (4).
the payload and CRC is extended with a 6-bit parity block
for storing the parity bits produced by the Hamming error The process of finding an earlier arrival in the cross-
correction code. This is shown in Figure 7. correlation and attempting to determine the first arrival of
Data-whitening is applied to the len + payload + CRC the signal rather than the strongest arrival is repeated up
bits of the packet. This is done to reduce the autocorrelation to 4 times or is halted earlier if no strong arrival peak can
properties of the packet payload, thereby improving the ToA be found. Repeating the algorithm (4 times) represents the
estimation performance, and to increase the number of Mark maximum number of earlier multipath arrivals we expect to
to Space transitions required by the demodulator to perform be able to detect.
bit synchronization.
4. Performance Evaluation
3.4. ToA Estimation. The MODEM performs ToA for sup-
porting ToF or Time-Difference-of-Arrival (TDoA) localiza- The following section describes the performance evaluation
tion. The ToA can be determined by cross-correlating the we have performed on the Proteus II node. We have evaluated
transmitted signal and the received signal which is buffered the power consumption of the node in different state and we
by the MODEM. The sampling rate of the signal is dependent make an estimation whether one-month lifetime is achiev-
on the configured symbol rate of the MODEM. The energy- able. We have performed a lifetime test with a high duty cycle
detector of the MODEM samples the signal at 16x the and a low duty cycle; in the low duty cycle test we achieve a
symbol rate. Therefore when the symbol rate of the MODEM lifetime of one month. Next to the lifetime tests, we evaluate
is increased, the sampling rate is increased and the ToA the PDR and ToA estimation of the node. Moreover we have
estimation is expected to be more accurate. evaluated the maximum communication range of the node.
Once a packet is correctly received, this can be deter-
mined by verifying the CRC, the fixed preamble, and a part of 4.1. Power Consumption. We have measured the power con-
the decoded packet is used to perform the cross-correlation. sumption of the node during acoustic transmission, acoustic
The Barker sequence in the header and the whitened data of reception, and idle and with the 433 MHz radio enabled.
the packet improve the cross-correlation capabilities of the Results are shown in Table 1. Transmission means that an
packet. acoustic packet is being transmitted; reception indicates that
After cross-correlation is performed, the correlation the MODEM is listening for the reception of an acoustic
shows the signal response of the acoustic channel. This packet. In the idle state the acoustic MODEM is turned off
response may contain several multipath arrivals of the signal. and the host processor is running a sleep timer to wake up the
The strongest arrival may not necessarily be the first or line- host periodically and radio that indicates the 433 MHz radio
of-sight arrival. Therefore the cross-correlation needs to be is turned on. As can be seen the acoustic MODEM transmits
searched to estimate the first path arrival. In [16] several at about 8 W; during reception the MODEM uses 300 mW.
Table 1: Power consumption of the node in different states.

Position 4
State Power (W) Description
Transmission 8W Active acoustic transmission
Reception 300 mW Active packet reception Gateway
Idle 60 mW Node sleeping on wakeup timer Position 3
Node sleeping on wakeup timer
Idle (fix) 8 mW after disabling the RS232 Position 1
level-converter Position 2
Radio 200 mW 433 MHz radio active
Table 2: Estimation of the battery capacity required for a battery

Figure 8: Illustration of the PDR setup deployment. The nodes
lifetime of one month.
were placed at different distances from the gateway and the water
State Duty cycle Power (mW) Total (mWh) continues under the shown building.
Idle 1 8 5952
Transmission 12/3600 8000 19840
Battery capacity required (mWh) 25792
Battery capacity 5000 mAh at 12 V (mWh) 60000
Initially with the MODEM turned completely off and the

host running on a wakeup timer and being in sleep mode
the complete node consumed 60 mW. This was higher than
what should have been achievable with the design. Later, we
discovered that the RS232 level-converter was still in standby
mode and after disabling the RS232 level-converter, the idle
power consumption of the node dropped to 8 mW. Using this
idle power consumption, one month of monitoring can be
easily achieved using different communication patterns. Figure 9: A Proteus II node deployed in the water at the test setup.
Table 2 gives an indication of the battery capacity The acoustic part of the node is underwater and the radio antenna is
required to run the node for one month (31×24 = 744 hours). above the water surface to allow a radio connection with the gateway.
The idle power consumption of the node is always present
(duty cycle is 1). For the transmission we assume a duty cycle
of 12 seconds for every hour (3600 seconds). The total battery All the nodes have a single-hop connecting to the gateway.
capacity required is about 25792 mWh, which indicates that Packets were transmitted for a period of 22 hours at different
a battery of 5000 mAh@12 V (60000 Wh) is more than symbol rates (50, 100, and 150 baud). The PDR rate was
sufficient, and leaves plenty of room for self-discharge of the calculated for this setup. Results shown in Figure 11 indicate
battery. The calculation indicates that a battery of 5000 mAh that about 30–40% PDR is achieved in this environment.
is more than sufficient to run the node for one month. Such When the symbol rate is increased to 150 bps, the delivery
a battery fits in the enclosure; therefore we can say that rate of two nodes decreases to only a couple of percent. In
we can achieve the required one-month operation; other this case, the multipath causes these nodes to be unable to
configurations are also possible. In Section 4.3 we perform communicate.
lifetime test with different transmission duty cycles. Figure 12 gives an indication of the pattern of packet
loss. Shown in the graph are the received packet numbers of
4.2. Packet Delivery Rate. Figure 8 shows our Packet Delivery the gateway, with their respective receive time. The packet
Rate (PDR) setup. Four nodes and one gateway were deployed numbers cycle from 0 to 64 before wrapping back to 0 and
in an artificial lake at the University of Twente campus. ideally all should be received. What can be seen is that certain
The acoustic communication is performed underwater while periods of time experience little packet loss, while other
a radio connection to the gateway is used to collect the periods of time experience a significant packet loss. What
measurement data. Acoustic communication was performed causes this effect is still unclear to us; possibly this is the result
in the 20–22 kHz band; the same frequencies were used for of the node slightly moving in the water and experiencing
the different baud rates. The deployment of the node with the different multipath effects for an extensive period.
acoustic transducer underwater and the radio antenna above
the water surface is shown in Figure 9. The test environment 4.3. Lifetime. In the same setup we have performed two life-
is shown in Figure 10; a lot of concrete is present in the form time tests. With a transmission interval of every minute, per-
of walls, pillars, and the lake floor. The communication is forming a transmission of 1 second every minute, the nodes
therefore highly influenced by many multipaths. were able to communicate for 5 days using a 2200 mAh@12 V
Table 3: Estimation of the battery capacity required for the lifetime

test over a duration of 5 days.
State Duty cycle Power (mW) Total (mWh)

Idle 1 60 7200
Transmission 60/3600 8000 16000
Battery capacity required (mWh) 23200
Battery capacity 2200 mAh at 12 V (mWh) 26400
Periods of extensive
Node 1 packet loss
70
60
Packet number
Figure 10: Picture of the deployment in the artificial lake at the 50
University of Twente campus. The environment is a small lake 40
between office buildings and consists of concrete walls and floor. 30
The gateway node was placed near an office to allow a network 20
connection to be made to campus network, allowing access to 10
0
the testbed through internet. The nodes are kept floating with the
14:24 15:36 16:48 18:00 19:12 20:24 21:36 22:48 0:00
radio antenna above the water surface, to allow a radio connection
Arrival time
to the gateway for collecting measurement data, and are fixed
with a sandbag on the floor. The acoustic transducer is deployed Figure 12: Arrival of packets at the gateway over the period of 10
underwater and acoustic communication is performed through the hours. The received packet numbers at the time they are received are
water. shown; packet numbers cycle from 0 to 64. What can be seen is that
certain periods of time show little packet loss, while other periods
of time experience significant packet loss.
60.00
Packet delivery rate (%)
50.00
4.4. Time of Arrival. In a different setup, ToA estimation was
40.00 evaluated. Using two nodes we estimate the range between
30.00 two nodes by using the radio and an acoustic transmission.
20.00 The master node transmits a radio packet to the slave node to
start a transmission. The slave node replies with an acoustic
10.00
packet and the master node determines the round-trip time
0.00 of the radio and acoustic transmission.
50 baud 100 baud 150 baud
Position 1 36.72% 26.55% 3.61% The distance between the two nodes was changed from
Position 2 36.72% 51.98% 33.81% the two nodes deployed completely next to each other (0-
Position 3 31.41% 19.44% 1.03% meter distance) to 10, 20, and 30 meters (setup shown in
Position 4 46.33% 49.04% 27.42% Figure 13). The distance of 0 meters is used to determine the
Avg. 37.80% 36.75% 16.47% processing delay of the measurement. The results can be seen
in Figure 14.
Figure 11: Results of the PDR test shown are the percentage of
packets that were correctly delivered to the gateway over a period Our results show that ToA of the MODEM is able to esti-
of 22 hours for different symbol rates. mate the distance between the nodes in the experiment. As
expected, when the symbol rate is increased the accuracy of
ToA measurement and consequently distance measurements
also increases. At the distance of 30 meters, the MODEM
rechargeable battery. In Table 3 we show the calculation of appears to lock on to a multipath arrival rather than the
the required energy for these 5 days. For the transmission direct line-of-sight. A relative distance of about 40 meters is
the duty cycle was 60 seconds for every hour (3600 seconds). measured, rather than the expected 30 meters. This result is
The energy required was 23200 mWh; this is in line with the same for all symbol rates.
the capacity of the battery we used (2200 mAh@12 V = ToA estimation based on modulated signals remains dif-
26400 mWh). The calculated consumed energy is in line with ficult and is affected by multipath. Even with the “search and
the battery capacity that we calculate. substract” algorithm described in Section 3.4, the MODEM
In another test we have equipped the node with a still locks on the multipath arrivals in some cases. ToA
5000 mAh@12 V battery. With a low duty cycle, a transmis- estimation based on wideband chirps is definitely more
sion of 2 seconds every 10 minutes, we have achieved a one- accurate. Experiments with a chirp signal in highly reflective
month lifetime of the underwater node. This shows that our environments were performed in [10, 17]. The performance
set goal, a lifetime of one month using a low duty cycle, is of the ToA estimation can be increased using multitone FSK
achievable. and therefore using a wider band for the signal.
battery for one month and, ideally, the node should also be
m
Position 4 of low cost.
30.04
Because existing commercial nodes do not allow recon-
figuration of the physical layer protocols in the MODEM and
existing research platforms were not available to us or did not
meet our energy efficiency criteria, we have designed our own
Position 3
underwater sensor node called Proteus II.
20 m
This node has an energy-efficient power circuit and is
designed for low-power operation; a lithium battery inside
the enclosure powers the sensor node. The node has inte-
grated sensors such as an accelerometer and temperature
Position 2 sensor and ADC allows connecting analog sensors. External
sensors can be connected through connectors available on the
enclosure.
A 433 MHz radio is available on the node to allow a
radio connection for debugging and collecting measurement
Position 1 data when the node is deployed at the water surface. In our
0
test a TDMA MAC protocol was implemented to prevent

Figure 13: Illustration of the ToF setup deployment. The ToF was collisions. Sensors are polled by the gateway using the RF
measured at four different positions ranging in distance from 0 m to radio; this allows us to control, configure, and diagnose the
30 m. How well the distance between the different positions can be network centrally.
estimated using the ToA measurement was evaluated. The node consists of a low-power MODEM and host con-
troller both based on a Cortex-M4 processor. The MODEM
is designed as a software-defined MODEM and uses FSK
modulation with Hamming error correction and ToA estima-
45.00 tion for positioning and time synchronization. Bit decoding
40.00 and bit synchronization are performed using maximum-
Measured distance (m)
35.00 likelihood decoding and ToA estimation is performed using

30.00 cross-correlation with “search and substract” algorithm for
25.00 estimation of first arrival and multipath arrivals. The host
20.00 processor is used for performing networking, reading the
15.00 onboard sensors, and controlling the 433 MHz radio.
10.00 Tests were performed to measure the power consumption
5.00 and performance of the node. In low-power mode, the node
0.00 draws about 8 mW. In a lifetime test we have achieved the
50 baud 100 baud 150 baud
10 m
desired one-month lifetime, using a 5000 mAh@12 V battery
12.11 12.37 11.95
20 m 22.73 20.87 20.79 and a duty cycle of transmitting 2 seconds every 10 minutes.
30 m 42.03 39.95 39.62 Initial communication tests in our test environment show
that the Proteus II is able to achieve a PDR of 30–40% in
Figure 14: Measured distance between Proteus II nodes at different an environment which has concrete walls and floors and
symbol rates and at different distances. is therefore highly reflective. Moreover, communication was
performed over the distance of 140 m.
ToA estimation was also evaluated and the fact that
the MODEM is able to perform ToA estimation with an
In the same environment as the ToA setup, maximum accuracy of about 1 to 2 meters was shown. The ToA based
communication range tests indicate that the node is able to on the modulated FSK signal is still affected by multipath
communicate over a distance of 140 m. Because the matching interference.
of the transducer is very much frequency dependent, we In the future, multitone FSK and perhaps PSK will be
believe the range can be improved by tuning the frequency researched to see if an improvement can be made in com-
of the communication. For our current setup the FSK were munication speed and ToA by using a wider band signal.
fixed in the band of 20–22 kHz. Multitone FSK likely has a positive effect on the delivery
rate, because many multipaths are present within the envi-
5. Conclusion ronment. We would also like to evaluate the performance
of our nodes in less reflective environments. The maximum
For our future research we require an underwater sensor communication range of 140 m can also be improved in
node which can be reconfigured at all networks layers, the the future by tuning the frequency to the matching of
physical layer up to the transport layer. Moreover, we require the transducer. In our experiments the used frequencies
a node which can be extended with different sensors. The were fixed; by allowing configurable frequencies, the use of
underwater node should be able to run on a rechargeable different frequencies can be evaluated.
Conflict of Interests [15] P. Koopman and T. Chakravarty, “Cyclic redundancy code

(CRC) polynomial selection for embedded networks,” in Pro-
The authors declare that there is no conflict of interests ceedings of the International Conference on Dependable Systems
regarding the publication of this paper. and Networks, pp. 145–154, IEEE, June-July 2004.
[16] C. Falsi, D. Dardari, L. Mucchi, and M. Z. Win, “Time of
arrival estimation for UWB localizers in realistic environ-
Acknowledgment ments,” Eurasip Journal on Applied Signal Processing, vol. 1,
Article ID 032082, 2006.
This work was funded through the European FP7 project
SUNRISE, under Grant no. 611449. [17] B. S. Borowski, Application of channel estimation to underwater,
acoustic communication [Ph.D. thesis], Stevens Institute of
Technology, Hoboken, NJ, USA, 2011, AAI3467242.
References
[1] I. F. Akyildiz, D. Pompili, and T. Melodia, “Underwater acoustic
sensor networks: research challenges,” Ad Hoc Networks, vol. 3,
no. 3, pp. 257–279, 2005.
[2] J. Rice, B. Creber, C. Fletcher et al., “Evolution of seaweb under-
water acoustic networking,” in Proceedings of the MTS/IEEE
Conference and Exhibition (OCEANS ’00), vol. 3, pp. 2007–2017,
IEEE, Providence, RI, USA, September 2000.
[3] T. S. Husøy, F. R. Knudsen, B. Gjelstad, and A. Furdal, “Product
development at kongsberg maritime related to underwater
sensor networks,” in Proceedings of the 7th ACM International
Conference on Underwater Networks and Systems (WUWNet
’12), pp. 16:1–16:5, ACM, New York, NY, USA, November 2012.
[4] Evologics, http://www.evologics.de/.
[5] L. Freitag, M. Grund, S. Singh, J. Partan, P. Koski, and K. Ball,
“The WHOI micro-modem: an acoustic communications and
navigation system for multiple platforms,” in Proceedings of the
MTS/IEEE OCEANS Conference, vol. 2, pp. 1086–1092, IEEE,
Washington, DC , USA, September 2005.
[6] R. Bannasch and K. Kebkal, “Method and devices for transmit-
ting and receiving information,” US Patent 6,985,749, 2006.
[7] Y. Yang, Z. Xiaomin, P. Bo, and F. Yujing, “Design of sensor
nodes in underwater sensor networks,” in Proceedings of the
4th IEEE Conference on Industrial Electronics and Applications
(ICIEA ’09), pp. 3978–3982, May 2009.
[8] C. Petrioli, R. Petroccia, and D. Spaccini, “SUNSET version
2.0: enhanced framework for simulation, emulation and real-
life testing of underwater wireless sensor networks,” in Proceed-
ings of the 8th ACM International Conference on Underwater
Networks and Systems (WUWNet ’13), pp. 43:1–43:8, ACM,
Kaohsiung, Taiwan, November 2013.
[9] Gumstix, http://www.gumstix.com/.
[10] W. A. P. van Kleunen, N. A. Moseley, N. Meratnia, and P. J. M.
Havinga, “Experiments with aLS-Coop-Loc cooperative com-
bined localization and time-synchronization,” in Proceedings of
the 12th International Symposium on Modeling and Optimization
in Mobile, Ad Hoc and Wireless Networks (WiOpt ’14), pp. 86–91,
IEEE Computer Society, Hammamet, Tunisia, May 2014.
[11] J. Mitola, “The software radio architecture,” IEEE Communica-
tions Magazine, vol. 33, no. 5, pp. 26–38, 1995.
[12] FreeRTOS, A free open source rtos for small embedded real
time systems, http://www.freertos.org/.
[13] P. Levis, S. Madden, J. Polastre et al., “Tinyos: an operating
system for sensor networks,” in Ambient Intelligence, pp. 115–
148, Springer, Berlin, Germany, 2005.
[14] S. W. Golomb and R. A. Scholtz, “Generalized barker sequen-
ces,” IEEE Transactions on Information Theory, vol. 11, no. 4, pp.
533–537, 1965.

Underwater Wireless Sensor Networks 2015

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Underwater Wireless Sensor Networks 2015

Hochgeladen von

Copyright:

Verfügbare Formate

International Journal of Distributed Sensor Networks

Underwater Wireless Sensor

Underwater Wireless Sensor Networks 2015

Guest Editors: Dongkyun Kim, Juan C. Cano, Wei Wang,

Underwater Sensor Network Applications: A Comprehensive Survey, Emad Felemban,

Correspondence should be addressed to Dongkyun Kim; dongkyun@knu.ac.kr

Received 23 November 2015; Accepted 23 November 2015

En Cheng, Shengli Chen, and Fei Yuan

Correspondence should be addressed to Fei Yuan; yuanfei@xmu.edu.cn

Received 2 February 2015; Revised 30 April 2015; Accepted 4 May 2015

Academic Editor: Juan Carlos Cano

1. Introduction been used in underwater wireless communication (UWC)

Taking V = √𝐵/(𝑀 + 1)𝑇ℎ 𝑡, 𝑢 = √𝐵/(𝑀 + 1)𝑇ℎ (𝑡−𝑇ℎ /2),

Figure 3: Receiver block diagram for MLC-CDMA.

xlocal(t) 3.2. Fractional Fourier Transform (FrFT). FrFT, which can

𝑟 (𝑡) = 𝐴 0 cos (2𝜋𝑓𝑐 𝑡 + 𝜋𝑢𝑡2 ) + 𝑛 (𝑡) , 0 ≤ 𝑡 ≤ 𝑇ℎ , (13)

Bit “1,” first branch

After MCR, we get

(𝑀 + 1 − 2𝑚) 𝐵𝑇 − 4𝑚𝐵𝜏𝑖 𝑇 (𝑀 + 1 − 2𝑚) 𝐵 𝑇 2 (𝑀 + 1 − 𝑚) 𝑚𝐵 (21)

The signals of different nodes are detected according to to (5), we get

𝑀−3 𝑇 2 4𝑡 − 3𝑇ℎ Assuming 0 < 𝑡1 < 𝑇ℎ /2 < 𝑡2 < 𝑇ℎ , and 0 ≤ 𝑡 ≤ 𝑇ℎ , after

𝑀−5 𝑇 2 4𝑡 − 3𝑇ℎ 𝑇ℎ 2 4𝑡 − 3𝑇ℎ

Bit “1,” user 1 Bit “1,” user 1, the first branch

Bit “1,” user 1 Bit “1,” user 1

The received signal The received signal

but only when 𝑝 = 𝑝1 = 1.0309 at the first branch we get

The received signal

User 1, bit “1”

The received signal

User 1, bit “1”

Signal chirp signal 4 users, MLC 8 users, MLC

Chan-Ho Hwang,1 Ki-Man Kim,1 Seung-Yong Chun,2 and Sang-Kook Lee2

Correspondence should be addressed to Ki-Man Kim; kimkim@hhu.ac.kr

Received 2 March 2015; Accepted 15 April 2015

Academic Editor: Juan Carlos Cano

1. Introduction the Doppler shift frequency is obtained according to the

Correlator 1 The first recursive data frame is demodulated using the

Recursive frame Recursive frame Recursive frame

Recursive frame Recursive frame Recursive frame

Recursive frame Recursive frame Recursive frame

Figure 2: The proposed remodulation method.

Symbol spaced sampling

Figure 3: Receiver structure.

Table 1: Parameters for simulation.

Doppler frequency (Hz)

1 Figure 8: Scattering function.

4 Cumulative uncoded BER (%)

1000 symbols 250 symbols

Figure 6: Uncoded bit error rate second Doppler estimation (a)

Acknowledgments [15] J. K. Cavers, “An analysis of pilot symbol assisted modulation

Chang-Uk Baek and Ji-Won Jung

Correspondence should be addressed to Ji-Won Jung; jwjung@kmou.ac.kr

Received 24 April 2015; Accepted 28 June 2015

Academic Editor: Dongkyun Kim

1. Introduction only transmit small packets, which reduces the throughput.

Convolutional Symbol Dk Packet

Figure 1: Baseband model of transmission system with turbo equalization.

0.5 s 0.5 s 0.5 s 0.5 s

0.2 s 1000 0.2 s 0.2 s 1000 0.2 s

Figure 3: Frequency and phase offset compensation.

0.5 s 0.5 s 0.5 s 0.5 s

0.2 s 500 0.2 s 0.2 s 500 0.2 s

Figure 4: Packet structure.