See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272564433

Acoustic Communication System for Underwater Wireless Sensor Networks

Conference Paper · July 2014

DOI: 10.13140/2.1.2014.0325

CITATIONS READS

0 567

10 authors, including:

Raúl Sáez-Cañete Ricardo Mercado

Universitat Politècnica de València Universitat Politècnica de València
4 PUBLICATIONS   4 CITATIONS    6 PUBLICATIONS   59 CITATIONS   

SEE PROFILE SEE PROFILE

Angel Perles Pedro Yuste

Universitat Politècnica de València Universitat Politècnica de València
54 PUBLICATIONS   568 CITATIONS    35 PUBLICATIONS   275 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Underwater Acoustic Sensor Networks View project

Choreography and IoT View project

All content following this page was uploaded by Raúl Sáez-Cañete on 21 February 2015.

The user has requested enhancement of the downloaded file.

 Acoustic Communication System for
Underwater Wireless Sensor Networks
Underwater communication

Raúl Sáez, Antonio Sánchez, Ricardo Mercado, Àngel Perles, Pedro Yuste,
Juan Vicente Capella, Sara Blanc, Vicente Busquets, Juan José Serrano, and
Rafa Ors

Instituto ITACA, Universitat Politècnica de València, Camino de Vera s/n, 46022

Valencia, Spain;
e-mail: [email protected];
website: http://ucert.upv.es

Abstract. Underwater wireless sensor Networks (UWSN) provide a va-

riety of attractive working fields such as aquaculture, offshore exploita-
tion and underwater monitoring. These applications require spreading
an important number of nodes which requires to develop simple, reliable
and energy efficient modem architectures to reduce the economical cost,
the developing time and the power consumption, still being efficient and
robust.
This paper overviews research efforts of our research team conducted
during the last year to develop an acoustic communication system for
UWSN. To address the cost and power consumption limitations, the de-
sign and the implementation of a low cost and energy efficient underwater
modem and a new acoustic-triggered wake-up system were researched.
The result of combining both researches resulted in the development of
ITACA-S1000 modem, a solution with a very low power consumption and
with a very advanced remote wake-up system. Since this modem was the
first coherent FSK solution presented to date, it was necessary to reach
a new optimal multipath and Doppler-shift correction algorithm for co-
herent FSK modem architectures with resources and power limitations
to archive an efficient and robust communication.

Keywords: underwater wireless sensor networks; acoustic modems;

uwsn; underwater sensor networks; wake-up

1 Introduction

Underwater Wireless Sensor Network (UWSN) development is of current inter-

est due to the potential variety of attractive working fields such as aquaculture,
offshore exploitation, biological monitoring as well as water and seafloor pollu-
tion, seismic activity and ocean currents. Applications of these fields may require
one or more of the following: short range underwater communications between
collaborative devices; short/medium range communications between sea bottom

82 C.Fernandez-Llatas and A. Bonastre (Ed.):

Workshop on Innovation on
Information and Communication Technologies
(ITACA-WIICT 2014)

ISBN 978-84-697-1166-8
2

and surface; and longer range communications between the site and land. In
these applications, it is often desirable to eliminate or reduce the number of
wires and connectors to a minimum to reduce cost and maintenance and in-
crease reliability. Practical solutions for these applications mainly use floating
buoys that recollect data from a large number of submerged sensor nodes via
wireless acoustic communication[1] and communicate them to a base station
using ISM frequency bands.
Since sensor networks are formed by a large number of nodes, sensor nodes
must be low cost. Moreover, nodes must be self-powered because replacement is
difficult once deployed. So, they must be equipped with a battery which discharge
determines the node lifetime. In order to prolong nodes’ life, it is recommendable
to halt node activity when no data are needed to be transmitted or received.
During these non-activity periods a node can operate in a low-power mode, but
can’t receive any message. The solution to this issue is asynchronously to wake
up a slept node through an acoustic signal[2]. For this, an acoustic-triggered
wake-up system must be include within its architecture. It is desirable that the
wake up system can recognize predefined patterns in order to individually wake
up network nodes. Thus, the solution is more flexible and can be adapted to
many scenarios.
This paper describes research efforts of our research team conducted during
the last year to develop an acoustic communication system for UWSN. In 2011, a
flexible, energy efficient and low-cost underwater modem was designed[3] due to
difficulties in finding an adequate underwater acoustic modem to address the cost
and power consumption limitations of UWSN. Besides, a new acoustic-triggered
wake-up system for low power nodes with predefined pattern recognition was
designed[4] because of problems to find a wake-up solution useful for low-power
nodes. In 2012, the combination of both researches and the addition of several
improves resulted in a flexible, low cost and energy efficient acoustic modem ar-
chitecture[5]. The result of this architecture was born the ITACA-S1000 modem,
the solution with the less power consumption and with the remote wake-up sys-
tem most advanced to date. However, since this new modem was the first coher-
ent FSK solution presented so far, there were no efficient correction algorithms
designed to improve the efficiency and the robustness of modem communica-
tions. Hence, a new algorithm for both Doppler-shift and multipath corrections
for coherent FSK modem was researched[6].
This paper is organized as follows. Section 2 describes the research on an
acoustic-triggered wake-up system. Section 3 describes the research on the low
cost and energy efficient acoustic modem architecture. Section 4 describes the
optimal multipath and Doppler-shift correction algorithm for coherent FSK mo-
dem. Finally, section 5 concludes this paper.

2 An acoustic-triggered wake-up system

In 2011, a new flexible, energy efficient and low-cost underwater modem was
designed and implemented for UWSN[3]. With the purpose of prolonging nodes’

83
 3

life, a search was conducted in order to find an underwater wake up system with
a power consumption as low as possible and with an order of magnitude similar
to the power consumption of the new modem. However, previous solutions[7][2]
resulted not adequate. Therefore, a new wake-up solution adapted to these re-
quirements was researched. The result was a new acoustic-triggered wake-up
system for low power nodes presented in [4].
The new acoustic-triggered wake-up system waits listening to the channel
to detect possible incoming wake-up signals. If a signal is received, the system
processes it, waking up the modem MCU just in case of a positive recognition.
The wake-up signal is On-Off-Keying modulated in order to save energy[8].
On the receiver side, a specific hardware is required to always listen the chan-
nel, waiting for a wake-up signal. The design contains a commercial IC and a
specific matching net between piezoelectric and IC input designed to mach both
impedances and suitably couples the acoustic incoming signal. The commercial
IC is the AS3933 of Austria Microsystems[9] which includes false wake-up de-
tection and has a low power consumption suitable for modem requirements. The
matching net is formed by a 3-stage T-structure band-pass filter with passive
inductors and capacitors. This structure was found the most suitable since ca-
pacitors avoid any circuit bias modification and inductors mimic coil-antenna
magnetic coupling.
On the transmitter side, since the modem transmits data with FSK modula-
tion, only a software adjustment on modulation is required. Hence, no hardware
modifications are needed to produce the OOK modulation. The modem simply
switches power amplifier output on and off according to transmitted symbol and
the bit synchronism is archived by Manchester coding.

Fig. 1. Asynchronous wake-up system

In 2012, an improve to the acoustic-triggered wake-up system and a simulink

model to acoustic-triggered wake-up systems were presented in [10]. The simulink
model is designed to accurately and quickly evaluate network performance with
acoustic-triggered wake-up capability in a predefined scenario. With this model,
it is expected evaluate future extensions of the system. The improve of the
wake-up system consisted in an extension of pattern recognition for recognizing
multiple patterns. The multiple pattern recognition is based in extending the

84
4

wake-up signal and performing a detection algorithm. The wake-up signal is

extended with an m additional bits pattern. This new pattern is OOK modulated
too.

Received acoustic signal

1

0.5

0
Carrier Preamble AS3933-pattern Multiple Pattern

Fig. 2. Wake-up signal extension

The detection algorithm uses this additional pattern to be detected by the

modem MCU. First, the modem remains in sleep mode while the wake-up periph-
eral listens to the channel. After detecting a valid AS3933-pattern, the wake-up
peripheral wakes up the modem MCU. Then, the MCU checks the m additional
bits. If the additional bits aren’t a valid pattern, the MCU sleeps again. To min-
imize the power consumption, the MCU works at a frequency 60 times slower
than the frequency in the transmission and reception modes while checking the
additional pattern.

3 A low cost and energy efficient acoustic modem

architecture

The first description of the ITACA modem was given in [3]. In 2012, the design
was improved with enhanced features such as a triggered wake-up system, CCA,
and RSSI measurement. The result of this improve was a low cost and energy
efficient acoustic modem architecture presented in [5].
The acoustic modem architecture is built around a low power microcon-
troller (MCU) and includes several blocks designed under cost, complexity and
consumption constraints. The figure 3 shows the modem architecture block dia-
gram.
The transducer chosen for this architecture is a commercial piezoelectric-
based transducer, the HUMMINBIRD XP 9 20[11]. The use of this element is a
key factor that must be considered when a low cost underwater node is wanted
since its price is significantly lower than hydrophones. The transducer has two
identified bands around 85 kHz and 200 kHz. However, due to water attenuation,
it is more worthwhile to use the lowest possible frequency band: 85 KHz.

85
 5

Fig. 3. ITACA acoustic modem architecture

The power amplifier is specially designed to reach the maximum efficiency

with the minimum consumption. The designed solution, named Push-Pull class-
B-based Digitally controlled amplifier (D-PP-B), is a custom HW-SW power
amplifier stage that maximizes the power that is transferred to the load. The
amplifier is based on an AB-class amplifier[12] but push-pull stage transistors
are totally and independently activated or de-activated using two different out-
put pins of the modem MCU. In this way, a simultaneous activation of both
output push-pull stage transistors is avoided which improves significantly the
low efficiency of AB-class amplifiers.
Instead of a simple pre-amplier[13], the analog reception amplifier is based
on a pre-amplifier with a pass-band filter. The design introduces digital poten-
tiometers and operational amplifiers that work correctly with signals around 85
kHz. The modem MCU can vary the resistance of each potentiometer indepen-
dently to modify the overall reception filter gain. So, the received signal is filtered
and amplified by a two-stage pass-band filter based on single-supply operational
amplifiers. Besides, to filter as much noise as possible before signal amplifica-
tion, an addition tree-stage L-C band pass passive filter was inserted next to
the piezoelectric element input. This filtering network is especially designed to
match the transducer and modem impedances. As a result, the final filter has a
more selective band and a higher attenuation slope.
Transmissions in this architecture are based in a coherent-FSK modulation
which provides 1 kbps with 1 kHz bandwidth using an 85 kHz carrier frequency
(84.5 kHz frequency is used for symbol ‘0’ and 85.5 kHz for symbol ‘1’). The
coherent binary FSK modulation algorithm is very simple. The frequency of
the transmitted signal changes according to the bit to be transmitted. So, this
task is performed using an internal peripheral counter of the modem MCU.
However, demodulation algorithm is more complicated because a phase-locked-
loop (PLL) is needed. The demodulation system is formed of three main blocks:
a phase detector, a low pass filter and a controlled oscillator. The phase detector

86
6

is based on an XOR gate and filtering on a lead-lag low-pass filter while the
controlled oscillator is implemented by the MCU. It is an energy-efficient HW-
SW solution because both the external phase detector and the low pass filter
relieve the MCU from the most consuming tasks in demodulation.
The triggered wake-up system is designed to work both a synchronous and
asynchronous manner. The synchronous wake-up is implemented through a real
time clock (RTC) with a 32.768 Khz crystal which can be programmed by MCU
via serial interface. The RTC has several outputs that are activated when any
alarm condition are met. These outputs are used to wake-up the MCU from
its deeper low-power modes. Although this implementation does not any real
innovation, it remains very useful and was included in the modem architecture to
enable the implementation of upper layer protocols. The asynchronous wake-up
is implemented through the asynchronous wake-up system described previously
in section 2.
The CCA feature is implemented as a physical-layer mechanism and an MCU
software routine. Physically, CCA requires carrier frequency (CF) detection or
RSSI observation, both of which are supported by the already implemented
wake-up circuit without additional hardware. The wake-up peripheral integrates
a frequency detector based on a zero crossing counter. After CF detection, the
gain of the amplifier is set to the maximum and the AGC reduces this level
according to the received signal input level. After 1.06 ms, the AGC algorithm
is complete and a stable RSSI value can be read from the corresponding register
of the peripheral. Then, the MCU routine can read the RSSI value for eight
symbols to obtain an average value above 2 dB.
As a result of this architecture was born the ITACA-S1000 modem, the so-
lution with a very low power consumption and with a very advanced remote
wake-up system. Figure 3 shows a comparative analysis between commercial
acoustic modems with modem ITACA-S1000 using the concept of normalized
power for each working state: idle, reception and transmission. This parameter,
shown in (1), describes the relationship between the average power consumption
(P̄modem ), the maximum distance reached (Dmax ) and the spectral efficiency in
a concise and quantitative way. The lower the value of this parameter, the higher
the efficiency of the modem in each state.

P̄modem (W )
Pnormalized = (1)
Dmax (m) · Spectralef f iciency (bps/Hz)

The ITACA-S1000 presents the lowest normalized power in idle state because
of its revolutionary idle consumption, only 11µW. In transmission, the efficiency
is the highest due to the power stage designed specifically. Finally, in reception,
its efficiency is similar to the best commercial options, ATM modems. Therefore,
considering the normalized power, ITACA-S1000 is the most optimal solution in
its operating range.

87
 7

Fig. 4. Comparison between commercial acoustics modem with modem ITACA-S1000

4 An optimal multipath and Doppler-shift correction

algorithm for coherent FSK modem

To reach a reliable acoustic communication, the multipath and Doppler effects

must be addressed in the communication[14], specially in shallow water. On the
one hand, multipath distortion is produced by echoes from the sea-bottom, the
surface or submerged structures. On the other hand, Doppler-shift is produced
as an effect of the relative movement between transmitter and receiver, and also
can be produced by sound speed changes. In 2011, to improve the reliability and
robustness of modem communications and to adapt the ITACA-S1000 modem
to tough underwater environments, the incorporation in the modem architecture
of correction algorithms to address both effects was considered. Unfortunately,
neither optimal multipath nor Doppler-shift correction algorithms were found for
coherent-FSK modulation in the literature so far. Therefore, a new algorithm
for both Doppler-shift and multipath corrections for coherent FSK modem was
researched and was presented in [6].
This new algorithm was optimized to be run in low resources and low power
modem architectures. The algorithm is based in a closed feedback loop and reuses
the existing modem FSK decoder. Two main blocks, a Doppler estimator and
an adaptive filter, are used to compensate both Doppler and multipath effects
with an initial training step.

88
8

Fig. 5. Block diagram of training algorithm

The algorithm needs that same extra information is added at start of frames
in order to perform the training step. This extra information is a training se-
quence with specific characteristics which can be obtained using primitive poly-
nomials[15] to generate Pseudo Noise (PN) sequences.

Fig. 6. Frame format with extra training information

An experimental testbed was deployed in a small craft marina where frames

were sent to a 10 meter distance with a relative movement ranging from -1 to
+1 m/s. On the one hand, the error probability was analyzed. Table 1 shows
the error probability both with and without corrections. Results revealed that
the more positive the speed is, the higher error probability is obtained. How-
ever, correction algorithms were able to correct all errors. On the other hand,
algorithms convergence and training process were also analyzed. Results showed
that the Doppler correction algorithm converges with 25 symbols - @ 1kbps -
in the worst case, the multipath compensation algorithm converges in only 4
symbols even in the worst multipath scenario and a 31 bit PN training sequence
is enough to train the receiver.

89
 9

Table 1. Experimental mean error rate probability

-1 m/s 0 m/s +1 m/s

No correction 6.5% 15% 29%
Multipath & Doppler correction 0% 0% 0%

5 Conclusions

This paper describes our contributions for underwater wireless sensor networks
through the development of a low cost, energy efficient and reliable acoustic
communication system.
The developed acoustic modem is an ultra-low power and low-cost modem
architecture based in piezoelectric transducers, that combines a typical ultra-low
power microcontroller-based core with energy-efficient mechanisms. This archi-
tecture uses coherent-FSK modulation and the demodulation has been cleverly
divided into a dual hardware-software solution, which reduces processing require-
ments as well as the consumed power. Data rate reached with this architecture
is 1 kbps, enough for UWSN deployment and comparable to other scientific
research solutions.
The developed acoustic-triggered wake-up system is capable of detect pre-
defined patterns and perform selective wake-up. With this feature, modems re-
main in sleep mode as long as possible reducing the overall power consumption.
Besides, the wake-up signal is generated without additional hardware, being the
modem data interface also used to both generate and receive the acoustic wake-
up signal without being necessary a double acoustic transducer. Finally, since
the wake-up system is based on a commercial IC, it can operate in different
frequency bands and it can be easily integrated in several platforms.
To improve the efficiency and the robustness of modem communications, a
novel algorithm for multipath and Doppler-shift effects correction for coherent-
FSK acoustic modem was developed. The algorithm is simple and needs very
low extra resources. This enables its applicability in low power modems to build
long life underwater networks.
In conclusion, our contributions would help in the definition and evaluation
of energy-efficient network communication policies based on asynchronous wake
up medium access control, enable the deployment of wide underwater sensor
networks to perform various applications and improve the efficiency and the
robustness of underwater communications based in coherent-FSK acoustic plat-
forms.

Acknowledgments

This work was supported by the Technical University of Valencia Research

Project SP20120889 and Spanish government under projects CTM2011-29691-
C02-01 and TIN2011-28435-C03-0.

90
 10

References
1. Yang Xiao. Underwater Acoustic Sensor Networks. Auerbach Publications, Boston,
MA, USA, 1 edition, 2009.
2. Jack Wills, Wei Ye, and John Heidemann. Low-power acoustic modem for dense
underwater sensor networks. In Proceedings of the 1st ACM International Work-
shop on Underwater Networks, WUWNet ’06, pages 79–85, New York, NY, USA,
2006. ACM.
3. A. Sanchez, S. Blanc, P. Yuste, and J.J. Serrano. A low cost and high efficient
acoustic modem for underwater sensor networks. In OCEANS, 2011 IEEE - Spain,
pages 1–10, June 2011.
4. A. Sanchez, S. Blanc, P. Yuste, and J.J. Serrano. Rfid based acoustic wake-up
system for underwater sensor networks. In Mobile Adhoc and Sensor Systems
(MASS), 2011 IEEE 8th International Conference on, pages 873–878, Oct 2011.
5. Antonio Sánchez, Sara Blanc, Pedro Yuste, Angel Perles, and Juan José Serrano.
An ultra-low power and flexible acoustic modem design to develop energy-efficient
underwater sensor networks. Sensors, 12(6):6837–6856, 2012.
6. A. Sanchez, S. Climent, P. Yuste, A. Perles-Ivars, and J. J. Serrano. Low com-
plexity multipath and doppler-shift correction algorithm for reliable underwater
coherent-fsk acoustic modems: Short paper. In Proceedings of the Seventh ACM
International Conference on Underwater Networks and Systems, WUWNet ’12,
pages 21:1–21:5, New York, NY, USA, 2012. ACM.
7. Albert F. Harris, III, Milica Stojanovic, and Michele Zorzi. When underwater
acoustic nodes should sleep with one eye open: Idle-time power management in
underwater sensor networks. In Proceedings of the 1st ACM International Work-
shop on Underwater Networks, WUWNet ’06, pages 105–108, New York, NY, USA,
2006. ACM.
8. M.S. Durante and S. Mahlknecht. An ultra low power wakeup receiver for wireless
sensor nodes. In Sensor Technologies and Applications, 2009. SENSORCOMM ’09.
Third International Conference on, pages 167–170, June 2009.
9. Data sheet, “as3933 – 3d low frequency wakeup receiver”. Austria Microsystems.
v1.0 Jul 01, 2010.
10. A. Sanchez, S. Blanc, P. Yuste, I. Piqueras, and J. J. Serrano. Advanced acoustic
wake-up system for underwater sensor networks. Communications in Information
Science and Management Engineering(CISME), 2(2):1–10, 2012.
11. Humminbird: Fishfinder, depth finder, gps chartplotters, side imaging sonar tech-
nology and marine fish finders. Available online: http://ww.humminbird.com (ac-
cessed on 3 April 2012).
12. R. Jurdak, C.V. Lopes, and P. Baldi. Software acoustic modems for short range
mote-based underwater sensor networks. In OCEANS 2006 - Asia Pacific, pages
1–7, May 2006.
13. Saunvit Pandya, J. Engel, J. Chen, Zhifang Fan, and Chang Liu. Coral: minia-
ture acoustic communication subsystem architecture for underwater wireless sensor
networks. In Sensors, 2005 IEEE, pages 4 pp.–, Oct 2005.
14. Jim Partan, Jim Kurose, and Brian Neil Levine. A survey of practical issues in
underwater networks. In Proceedings of the 1st ACM International Workshop on
Underwater Networks, WUWNet ’06, pages 17–24, New York, NY, USA, 2006.
ACM.
15. Frank Ehlers, Warren Fox, Dirk Maiwald, Martin Ulmke, and Gary Wood. Ad-
vances in signal processing for maritime applications. EURASIP Journal on Ad-
vances in Signal Processing, 2010(1):512767, 2010.

91

View publication stats