research article an energy-efficient routing algorithm for...
TRANSCRIPT
Research ArticleAn Energy-Efficient Routing Algorithm for Underwater WirelessSensor Networks Inspired by Ultrasonic Frogs
Ming Xu12 Guangzhong Liu1 and Huafeng Wu3
1 College of Information Engineering Shanghai Maritime University 1550 Haigang Avenue Shanghai 201306 China2 Shanghai Key Lab of Intelligent Information Processing Fudan University 220 Handan Road Shanghai 200433 China3Merchant Marine College Shanghai Maritime University 1550 Haigang Avenue Shanghai 201306 China
Correspondence should be addressed to Ming Xu mingxushmtueducn
Received 15 August 2013 Accepted 24 December 2013 Published 13 February 2014
Academic Editor Kun Hua
Copyright copy 2014 Ming Xu et alThis is an open access article distributed under the Creative Commons Attribution License whichpermits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
The area of three-dimensional (3D) underwater wireless sensor networks (UWSNs) has attracted significant attention recentlydue to its applications in detecting and observing phenomena that cannot be adequately observed by means of two-dimensionalUWSNsHowever designing routing protocols for 3DUWSNs is a challenging task due to stringent constraints imposed by acousticcommunications and high energy consumption in acoustic modems In this paper we present an ultrasonic frog calling algorithm(UFCA) that aims to achieve energy-efficient routing under harsh underwater conditions of UWSNs In UFCA the process ofselecting relay nodes to forward the data packet is similar to that of calling behavior of ultrasonic frog for mating We define thegravity function to represent the attractiveness from one sensor node to another In order to save energy different sensor nodesadopt different transmission radius and the values can be tuned dynamically according to their residual energy Moreover thesensor nodes that own less energy or locate in worse places choose to enter sleep mode for the purpose of saving energy Simulationresults show the performance improvement in metrics of packet delivery ratio energy consumption throughput and end-to-enddelay as compared to existing state-of-the-art routing protocols
1 Introduction
Underwater wireless sensor networks (UWSNs) have a lotof potential application areas such as oceanographic datacollection disaster prevention pollution monitoring off-shore exploration and military surveillance [1ndash3] Radiofrequency (RF) signals suffer from severe attenuation inwater and have been successfully deployed only at very lowfrequencies involving large antenna and high transmissionpower Hence acoustic signals have been used for wirelesscommunication in current underwater physical layer whichhas challenges to be overcome such as long propagationdelay resulting from low speed of sound propagation severelylimited bandwidth and time-varying multipath propagation[4] All the above distinct features of UWSNs give birthto new challenge areas for every level of the network pro-tocol suite UWSNs mainly consist of two communicationarchitectures two-dimensional and three-dimensional (3D)underwater networks [1] 3D UWSNs are used to detect and
observe phenomena that cannot be adequately observed bymeans of ocean bottom underwater sensor nodes that is toperform cooperative sampling of the 3D ocean environment[5 6] In 3D UWSNs sensors float at different depths toobserve a given phenomenon Many problems arise withUWSNs that need to be solved in order to enable underwatermonitoring in the new environment Among them providingefficient routing is a very challenging task due to the uniquecharacteristics of UWSNs In UWSNs battery power is a vitalresource for each wireless sensor because of the difficultyand cost in recharging sensor batteries once the network isdeployed Manymethods for energy conservation at differentlayers have been investigated in terrestrial wireless sensornetworks However these methods are not applicable toUWSNs According to their architectures the routing proto-cols of UWSNs can be divided into three categories location-based routing flat routing and hierarchical routing [7]Location-based routing has good scalability but it requires
Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2014 Article ID 351520 12 pageshttpdxdoiorg1011552014351520
2 International Journal of Distributed Sensor Networks
a positioning system or positioning algorithm to help thenodes to calculate the location information Flat routing pro-tocols have better robustness but the excessive overhead formaintaining routing information restricts their application tosmall-scale underwater acoustic circumstances Hierarchicalrouting also has good scalability but the cluster maintenanceoverhead and the failure of key nodes will affect the routingefficiency
Frog calling andhearing have been shown to be importantfor species recognition mate assessment and localizationMost interestingly an unusual species called concave-earedtorrent frog (Amolops tormotus) lives near the noisy YellowMountain in eastern China and produces diverse bird-like melodic calls that often contain spectral energy in theultrasonic range (frequencies greater than 20KHz) [8] Itis demonstrated that the male frogs emit advertisementcalls using ultrasound to avoid masking by the widebandbackground noise of local fast-flowing streams Althoughthe female frogs exhibit no ultrasonic sensitivity they emitcourtship calls that evoke extraordinarily precise phonotaxisof the male frogs rivalling that of vertebrates with thehighest localization acuity (barn owls dolphins elephantsand humans) [9 10] Calling is linked to physical size andfemales may be attracted to more vigorous calls The smallestfrogs must consume lots of energy to produce calls In male-male competition some male frogs may stop calling orremain in chorus (each frog calls in turn) for longer periods oftime based on a comparison between the benefit of obtaininga higher mating probability and the cost of losing moreenergy In this paper we present an ultrasonic frog callingalgorithm (UFCA) for routing in UWSNs which has beenethologically inspired by the calling behavior of concave-eared torrent frogUFCAdoes not require fixed routing tablesor periodic flooding messages for the routing path discoveryTherefore it is resistant to node mobility and temporary lossof connectivity which are prevalent in UWSNs In UFCAdifferent sensor nodes adopt different transmission radiuswhich can be tuned dynamically according to their residualenergy Moreover the sensor nodes that own less energy orlocate in worse places choose to enter sleep mode for thepurpose of saving energy As a distributed routing algorithmno topology information needs to be exchanged amongneighboring nodes and only a small fraction of the sensornodes are involved in routing to ensure energy-efficientoperations for surveillance and monitoring applications
The remainder of the paper is organized as followsSection 2 presents a brief overview of related work whileSection 3 introduces the proposed scheme in detail Per-formance evaluation is described in Section 4 Finally weconclude the paper in Section 5
2 Related Work
The underwater environment introduces difficulties indesigning efficient routing protocols not experienced terres-trially such as transmission loss due to geometric spreadingand absorption by the ocean [11 12] Tan et al [13] proposeda new protocol based on hop-by-hop hybrid implicitexplicitacknowledgment scheme which is proposed for a multihop
UWSN In the protocol data packets forwarded by down-stream nodes can work as implicit ACKs for previoustransmitted data packets Vahdat and Becker [14] proposedepidemic routing (ER) protocol where each node replicatesa packet to every encountered node ER can utilize everyopportunity to deliver a packet to the destination and max-imize successful delivery ratio and minimize average end-to-end delay in unconstrained networks However this routingprotocol consumes too many resources that make it notdesirable in resource constrained networks such as UWSNsPompili et al [15] introduced two distributed routing algo-rithms for delay-insensitive and delay-sensitive applicationsrespectively with the objective of minimizing the energyconsumption taking the varying condition of the underwaterchannel and the different application requirements intoaccount
Vector-based forwarding (VBF) [16] is a geographicapproach which allows the nodes to weigh the benefit to for-ward packets and reduce energy consumption by discardinglow benefit packetsTherefore over a multihop path only thenodes that are located within a pipe of given width betweenthe source and the destination are considered for relayingHowever in the areas of low density of nodes VBF may notfind the path close to the routing vector Similarly Jornet etal proposed focused-beam routing (FBR) [17] protocol thatis suitable for networks containing both static and mobilenodes The objective of FBR is to determine which nodes arecandidates for relaying Candidate nodes are those that liewithin a cone of angle plusmn1205792 emanating from the transmittertowards the final destination An RTSCTS handshake is setup to isolate closer nodes within this cone If a node deter-mines that it is within the transmitterrsquos cone it will respondto the RTS Those nodes that are outside the cone will notrespond A theoretical argument supporting geographic rout-ing has been discussed in [18] based on simple propagationand energy consumption models for underwater networksThe study shows that an optimal number of hops along a pathexist and that increasing the number of hops by choosingcloser relays is preferable with respect to keeping the routeshorter In view of this several position-based routing algo-rithms are proposed and compared results show that select-ing relays closer than a givenmaximum distance before seek-ing farther ones achieves in fact optimal energy consumption
Depth-based routing (DBR) [19] can handle networkdynamics efficiently without the assistance of a localizationservice DBR forwards data packets greedily towards thewater surface (ie the plane of data sinks) In DBR a datapacket has a field that records the depth information of itsrecent forwarder and is updated at every hop But DBR hasonly greedy forwarding mode which alone is not able toachieve high delivery ratios in sparse areas Wahid et al[20] proposed an energy-efficient routing protocol calledERP2R (energy-efficient routing protocol based on physicaldistance and residual energy) based on the idea of utilizingthe physical distances of the sensor nodes towards the sinknode ERP2Ralso takes into account the residual energy of thesensor nodes in order to extend the network life-time How-ever ERP2Rmay lead to longer routing path with the growth
International Journal of Distributed Sensor Networks 3
of network density depending on the physical distancestowards the sink node which in turn consumes additionalenergy Moreover the characteristics of node mobility inUWSNs often make the problem worse Ayaz and Abdullah[21] proposed a hop-by-hop dynamic addressing based (H2-DAB) routing protocol to provide scalable and time-efficientrouting for UWSN The H2-DAB routing protocol does notrequire any dimensional location information or any extraspecialized hardware compared with many other routingprotocols in the same areaHowever the problemofmultihoprouting still exists as it is based on multihop architecturewhere nodes near the sinks drain more energy because theyare used more frequently
Packet redundancy and multiple paths can be exploitedin order to increase the reliability of UWSNs Ayaz et al[22] provided a two-hop acknowledgment reliabilitymodel inorder to insure the reliable data deliveries to the surface sinkswhere two copies of the same data packet are maintained inthe network without extra burden on the available resourcesA relay node that has data packets to forward will not replywith the acknowledgment until it cannot find the next hoptowards the destination But if a node is unable to find thenext hop due to any failure or even if it is lost then packetsin the buffer are not considered lost All the nodes that sendthe data packets towards this node will wait for a certainamount of time before trying again for the next hop Xu et al[23] proposed a multiple-path forward error correction (M-FEC) approach that integrated multiple-path communica-tions and Hamming coding to eliminate retransmission andenhance reliability in underwater sensor networksMoreovera Markov model and a dynamical decision and feedbackscheme were developed to decrease the number of the pathsin order to save energy and ensure the desirable packet errorrate However M-FEC may cause much long delay becauseof additional process of encoding and decoding the datapackets
3 Proposed Scheme
31 Network Model and Energy Consumption We considerthat 3D UWSNs are composed of a certain number ofsensor nodes uniformly scattered in monitoring fields Wepresent a generic model for a 3D UWSN that is representedby 119866 = (119881 119864) with 119899 sensor nodes Each sensor nodeis assigned with a triplet of coordinates (119909 119910 119911) We alsoassume that all sensor nodes know their own locationsthrough a certain localization service [24] Such assumptionis justified in underwater systems where fixed bottom-mounted nodes have location information upon deploymentIn fact the underwater localization is a nontrivial taskfor which relatively very few options are available Manyresearchers have proposed a variety of localization schemesand techniques to address this issue specially [25 26] It isnot always feasible to deploy anchor nodes at the sea floor fordeep water environment In this case mobile beacon nodessuch as autonomous underwater vehicles (AUVs) whichare equipped with internal navigation systems are exploitedas reference nodes to assist in corresponding distributed
localization algorithms This paper takes advantage of theseresearch results as existing preconditions
Definition 1 The function 120575(119906 V) defines the distancebetween two nodes 119904
119906and 119904V in a 3D Euclidean space as
120575 119873 times 119873 997888rarr Γ 120575 (119906 V)
120575 (119906 V) = radic(119906119909minus V119909)2+ (119906119910minus V119910)2+ (119906119911minus V119911)2
(1)
Underwater wireless sensor nodes are equipped withsensing devices They collect data from the external environ-ment and transmit these data by one or multihop to the sinknode Sink node is the node that generates data aggregationresults and also the target location of the data transmissionEach sensor node can either transmit or receive data packetsAll sensor nodes can tune their transmission radius rangedfrom 119903min (minimum transmission radius) to 119903max (maximaltransmission radius)
Consider two sensor nodes at minimum hop distance ℎthere exist two values 119906(ℎ) and V(ℎ) such that the Euclideandistance 120575(119906 V) between the two nodes is bounded that is119906(ℎ) le 120575(119906 V) le V(ℎ) The quality of the bounds depends onthe network density 120588 In particular for each ℎ gt 0 holds
lim120588rarrinfin
V (ℎ) minus 119906 (ℎ) = 119903min (2)
where 119903min is the minimum transmission range of the sensornodes
Sensing devices generally have widely different theoret-ical and physical characteristics Thus numerous models ofvarying complexity can be constructed based on applicationneeds and device features However for most kinds ofsensors the sensing ability diminishes as distance increases
Definition 2 For a sensor 119904 the general sensingmodel 119878 at anarbitrary point 119901 is expressed as
119878 (119904 119901) =120582
[119889(119904 119901)]119896 (3)
where 119889(119904 119901) is the Euclidean distance between the sensor 119904and the point 119901 and positive constants 120582 and 119896 are sensortechnology-dependent parameters [27]
We assume that all sensor nodes are equipped withlimited battery resources without recharging or replacingnode batteries after deployment The network lifetime isdefined as the time until the first sensor node in the networkdepletes its energy The energy consumption model is thesame as that in [28] where the attenuation and the energyspreading factor (1 is for cylindrical 15 is for practical and2 is for spherical spreading) are taken into consideration
Acoustic signal has different transmission modes inshallow water (where the depth of the water is lower than100 meters) and deep water (where the depth of the wateris above 100 meters) In shallow water the transmissionof the acoustic signal is limited to a cylindrical area frombottom to the surface while in deepwater the transmission of
4 International Journal of Distributed Sensor Networks
the acoustic signal is mainly with spherical diffusion andthe energy consumption is caused by spherical diffusion andwater absorption This paper concentrates on the shallowwater scenario
The passive sonar equation [29] characterizes the signal-to-noise ratio (SNR) of an emitted underwater signal at thereceiver which is presented by
SNR = SL minus TL minusNL + DI (4)
where SL is the target source level or noise generated bythe target TL is the transmission loss NL is the noise leveland DI is the directivity index (a function of the receiverrsquosdirectional sensitivity)
The transmission loss TL can be defined as the accumu-lated decrease in acoustic intensity as an acoustic pressurewave propagates outwards from a source The transmissionloss for cylindrically spread signals is calculated as
TL = 10 log2120575 (119906 V) + 120572120575 (119906 V) times 10minus3 (5)
where 120575(119906 V) denotes the Euler distance between the trans-mitter and the receiver in meters and 120572 is the frequencydependent medium absorption coefficient in dBKm ItfollowsThorprsquos formula [30] empirically as
120572 =011119891
2
1 + 1198912+
441198912
4100 + 1198912+ 275 times 10
minus41198912+ 0003 (6)
where 119891 is in KHz and 120572 is in dBKmThe noise level NL in shallow water is mainly affected by
waves shipping traffic wind level and the activities of largemammals For simplicity we consider an average value for thenoise level NL to be 70 dB as a representative shallow watercase [30]
SL can be defined as the intensity of the radiated sound indecibels related to the transmitted signal intensity at 1 meterfrom the source according to the following expression
SL = 10 log2
119868119905
1 120583Pa (7)
where 119868119905is in 120583Pa Solving for 119868
119905yields
119868119905= 10
SL10times 067 times 10
minus18 (8)
As a result the transmitter power 119875119905that achieves inten-
sity 119868119905at a distance of 1 meter from the transmitter in the
direction to the receiver is calculated as
119875119905= 2120587 times 119867 times 119868
119905 (9)
where 119875119905is in watts and119867 is the water depth in meters
32 Ultrasonic Frog Calling Strategy UFCA is inspired fromthe calling behavior of concave-eared torrent frog Maleconcave-eared torrent frogs can produce diverse bird-likemelodic advertisement calls with pronounced frequencymodulations that often contain spectral energy in the ultra-sonic range Although female concave-eared torrent frogsexhibit no ultrasonic sensitivity their courtship calls can
5
7
8
6
4zzz 9
zzz
Male
Female
fpfk
fh
fjfi
fq
ri
rj
Figure 1 Ultrasonic frog calling strategy
evoke extraordinarily precise phonotaxis of the male frogswith high localization acuity
Suppose there are six concave-eared torrent frogs ran-domly distributed in a space as shown in Figure 1 Frog 119891
119894
is a gravid female frog (with tone bursts frequency range 1ndash14KHz [10]) Others are male frogs that can emit ultrasonicsound and have ultrasonic hearing capacity in response totone bursts at frequency ranged from 1KHz to 35KHz [10]At first 119891
119894emits a courtship call in order to attract some
nearby male frogs The solid circle with radius 119903119894represents
the covering space of 119891119894rsquos courtship call The number in each
frog denotes its body size As 119891119895is the nearest male frog to
119891119894 it will emit an advertisement call at frequencies ranged
fromnormal sound to ultrasonic sound immediately after thereception of 119891
119894rsquos courtship call The dashed circle with radius
119903119895represents the covering space of 119891
119895rsquos advertisement call
which is bigger than the covering space of of 119891119894rsquos courtship
call After the male frog 119891119896receives 119891
119894rsquos courtship call and 119891
119895rsquos
advertisement call it extracts the body size information fromthese calls As 119891
119896rsquos body size is smaller than that of 119891
119895rsquos it will
not broadcast any advertisement call in order to save energyThe male frog 119891
ℎcan also hear 119891
119895rsquos advertisement call but it
still keeps silent since 119891ℎis located outside of the covering
space of 119891119894rsquos courtship call Both the male frogs 119891
119901and 119891
119902
locatewithin the covering space of119891119894rsquos courtship call Suppose
119891119901and 119891119902receive 119891
119895rsquos advertisement call simultaneously they
compare their body sizes and conclude that the probability ofwinning the competition is high Therefore both 119891
119901and 119891
119902
directly replywith advertisement calls to119891119894 which include the
information of their body sizes and locations Judging fromadvertisement calls of different male frogs 119891
119894selects 119891
119902as its
mate because 119891119902owns the biggest body size among the three
mating candidates 119891119895 119891119901 and 119891
119902 At last 119891
119894calculates 119891
119902rsquos
position and leaps to 119891119902
International Journal of Distributed Sensor Networks 5
5
7
8
6
4zzz 9
zzz
zzz
Sink
Receiver
Transmitter
st
rmin A
shsp sk
sjsi
sq
Gij
r minltr jlt2r min
Figure 2 Candidate discovery phase
33 Routing Algorithm UFCA consists of two phases candi-date discovery phase and relay node selection phase Figure 2illustrates the candidate discovery phase Each frog denotesa sensor node and each number in the frog denotes theresidual energy of local sensor node Sink nodes do not haveany energy constraints because they are equipped with bothradio-frequency (RF) and acousticmodems and are deployedat the water surface As for static sink nodes they only needto broadcast their positions to the whole network one time atthe initial stage of the network operation which would notproduce significant energy dissipation [31] The sensor nodethat holds the data packet is the transmitter which is similarto the gravid female frog in Figure 1 Each data packet carriesthe positions of the source node the sink node and therelay node (ie the node that transmits this packet) Suppose119904119894is a transmitter as shown in Figure 2 then other sensor
nodes are receivers before the data packet is forwarded Atfirst 119904
119894transmits a courtship packet with radius 119903min which
includes the positions of 119904119894and the sink node 119904
119905 As 119904119895is the
nearest receiver to 119904119894 it will calculate the cosine of the angle
between the direction from 119904119894to 119904119895and the direction from 119904
119894
to 119904119905(denoted by 119860 in Figure 2) upon receipt of 119904
119894rsquos courtship
packet If the cosine value is not below zero 119904119895will transmit
an advertisement packet with radius 119903119895 which is calculated as
119903119895= MIN(1 +
120576res119895
120576max119895
) sdot 119903min 119903max (10)
where 120576res119895
denotes the residual energy of sensor node 119904119895and
120576max119895
denotes the maximum energy of sensor node 119904119895 Thus
119903119895ranges from 119903min to 2119903min In the best case the residual
energy of 119904119895is full and 119903max gt 2119903min and 119903
119895equals 2119903min
according to formula (10) which is enough to cover 119904119894rsquos
transmission circle In the worst case the residual energy of119904119895is almost exhausted it will only transmit an advertisement
packet with radius 119903min in order to reach the position of119904119894 Moreover 119904
119895rsquos position and residual energy information
is included in its advertisement packet After 119904119896receives 119904
119894rsquos
courtship packet and 119904119895rsquos advertisement packet it extracts the
position and the residual energy information from thesepackets As 119904
119896rsquos residual energy is less than that of 119904
119895rsquos it
chooses to enter sleep mode in order to save energy withouttransmitting any advertisement packet Another receiver 119904
119902
can also receive 119904119894rsquos courtship packet and 119904
119895rsquos advertisement
packet But 119904119902will choose to enter sleep mode because the
cosine of the angle between the direction from 119904119894to 119904119902and the
direction from 119904119894to 119904119905is below zero In other words 119904
119902locates
in a worse place compared with other receivers Althoughthe receiver 119904
ℎlocates within the transmission radius of 119904
119895rsquos
advertisement packet it still keeps sleep mode since 119904ℎcannot
receive 119904119894rsquos courtship packet After 119904
119901receives 119904
119894rsquos courtship
packet and 119904119895rsquos advertisement packet it extracts the position
and the residual energy information from these packets As119904119901rsquos residual energy is more than that of 119904
119895rsquos and the cosine
of the angle between the direction from 119904119894to 119904119901and the
direction from 119904119894to 119904119905is not below zero it concludes that the
probability of winning the competition is high Therefore 119904119901
will transmit an advertisement packet with radius 119903119901 which
includes the information of its location and residual energyAt last 119904
119894will add 119904
119895and 119904119901to its candidate set after the receipt
of their advertisement packets The sensor node that goes tosleep mode will wake up immediately after another sensornode broadcasts a courtship packet and the sleep sensor nodelocates exactly within its transmission range
The process of selecting a candidate as the relay node toforward the data packet is illustrated in Figure 3 After thetransmitter 119904
119894rsquos candidate set is constructed it will select the
most attractive candidate as the relay node according to acertain standard which is described as the gravity functionin this paper
Definition 3 Given a sensor node 119904119894and its neighbor node 119904
119895
the gravity function from 119904119894to 119904119895is defined as119866
119894119895and its value
is calculated as
10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816=
120576res119894sdot 120576
res119895sdot cos119860
120575(119894 119895)2
(11)
where 120576res119894
and 120576res119895
denote the residual energy of sensor nodes119904119894and 119904119895 119860 is the intersection angle between the direction
from 119904119894to 119904119895and the direction from 119904
119894to the sink node 119904
119905 and
120575(119894 119895) is the Euclidean distance from 119904119894to 119904119895
At last the transmitter 119904119894computes the gravity values
with every sensor node in its candidate set and chooses thecandidatewithmaximal gravity value to be the relay node thatis in charge of forwarding the data packet
Algorithm 1 describes the process of building the routingpath with ultrasonic frog calling algorithm in detail
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
2 International Journal of Distributed Sensor Networks
a positioning system or positioning algorithm to help thenodes to calculate the location information Flat routing pro-tocols have better robustness but the excessive overhead formaintaining routing information restricts their application tosmall-scale underwater acoustic circumstances Hierarchicalrouting also has good scalability but the cluster maintenanceoverhead and the failure of key nodes will affect the routingefficiency
Frog calling andhearing have been shown to be importantfor species recognition mate assessment and localizationMost interestingly an unusual species called concave-earedtorrent frog (Amolops tormotus) lives near the noisy YellowMountain in eastern China and produces diverse bird-like melodic calls that often contain spectral energy in theultrasonic range (frequencies greater than 20KHz) [8] Itis demonstrated that the male frogs emit advertisementcalls using ultrasound to avoid masking by the widebandbackground noise of local fast-flowing streams Althoughthe female frogs exhibit no ultrasonic sensitivity they emitcourtship calls that evoke extraordinarily precise phonotaxisof the male frogs rivalling that of vertebrates with thehighest localization acuity (barn owls dolphins elephantsand humans) [9 10] Calling is linked to physical size andfemales may be attracted to more vigorous calls The smallestfrogs must consume lots of energy to produce calls In male-male competition some male frogs may stop calling orremain in chorus (each frog calls in turn) for longer periods oftime based on a comparison between the benefit of obtaininga higher mating probability and the cost of losing moreenergy In this paper we present an ultrasonic frog callingalgorithm (UFCA) for routing in UWSNs which has beenethologically inspired by the calling behavior of concave-eared torrent frogUFCAdoes not require fixed routing tablesor periodic flooding messages for the routing path discoveryTherefore it is resistant to node mobility and temporary lossof connectivity which are prevalent in UWSNs In UFCAdifferent sensor nodes adopt different transmission radiuswhich can be tuned dynamically according to their residualenergy Moreover the sensor nodes that own less energy orlocate in worse places choose to enter sleep mode for thepurpose of saving energy As a distributed routing algorithmno topology information needs to be exchanged amongneighboring nodes and only a small fraction of the sensornodes are involved in routing to ensure energy-efficientoperations for surveillance and monitoring applications
The remainder of the paper is organized as followsSection 2 presents a brief overview of related work whileSection 3 introduces the proposed scheme in detail Per-formance evaluation is described in Section 4 Finally weconclude the paper in Section 5
2 Related Work
The underwater environment introduces difficulties indesigning efficient routing protocols not experienced terres-trially such as transmission loss due to geometric spreadingand absorption by the ocean [11 12] Tan et al [13] proposeda new protocol based on hop-by-hop hybrid implicitexplicitacknowledgment scheme which is proposed for a multihop
UWSN In the protocol data packets forwarded by down-stream nodes can work as implicit ACKs for previoustransmitted data packets Vahdat and Becker [14] proposedepidemic routing (ER) protocol where each node replicatesa packet to every encountered node ER can utilize everyopportunity to deliver a packet to the destination and max-imize successful delivery ratio and minimize average end-to-end delay in unconstrained networks However this routingprotocol consumes too many resources that make it notdesirable in resource constrained networks such as UWSNsPompili et al [15] introduced two distributed routing algo-rithms for delay-insensitive and delay-sensitive applicationsrespectively with the objective of minimizing the energyconsumption taking the varying condition of the underwaterchannel and the different application requirements intoaccount
Vector-based forwarding (VBF) [16] is a geographicapproach which allows the nodes to weigh the benefit to for-ward packets and reduce energy consumption by discardinglow benefit packetsTherefore over a multihop path only thenodes that are located within a pipe of given width betweenthe source and the destination are considered for relayingHowever in the areas of low density of nodes VBF may notfind the path close to the routing vector Similarly Jornet etal proposed focused-beam routing (FBR) [17] protocol thatis suitable for networks containing both static and mobilenodes The objective of FBR is to determine which nodes arecandidates for relaying Candidate nodes are those that liewithin a cone of angle plusmn1205792 emanating from the transmittertowards the final destination An RTSCTS handshake is setup to isolate closer nodes within this cone If a node deter-mines that it is within the transmitterrsquos cone it will respondto the RTS Those nodes that are outside the cone will notrespond A theoretical argument supporting geographic rout-ing has been discussed in [18] based on simple propagationand energy consumption models for underwater networksThe study shows that an optimal number of hops along a pathexist and that increasing the number of hops by choosingcloser relays is preferable with respect to keeping the routeshorter In view of this several position-based routing algo-rithms are proposed and compared results show that select-ing relays closer than a givenmaximum distance before seek-ing farther ones achieves in fact optimal energy consumption
Depth-based routing (DBR) [19] can handle networkdynamics efficiently without the assistance of a localizationservice DBR forwards data packets greedily towards thewater surface (ie the plane of data sinks) In DBR a datapacket has a field that records the depth information of itsrecent forwarder and is updated at every hop But DBR hasonly greedy forwarding mode which alone is not able toachieve high delivery ratios in sparse areas Wahid et al[20] proposed an energy-efficient routing protocol calledERP2R (energy-efficient routing protocol based on physicaldistance and residual energy) based on the idea of utilizingthe physical distances of the sensor nodes towards the sinknode ERP2Ralso takes into account the residual energy of thesensor nodes in order to extend the network life-time How-ever ERP2Rmay lead to longer routing path with the growth
International Journal of Distributed Sensor Networks 3
of network density depending on the physical distancestowards the sink node which in turn consumes additionalenergy Moreover the characteristics of node mobility inUWSNs often make the problem worse Ayaz and Abdullah[21] proposed a hop-by-hop dynamic addressing based (H2-DAB) routing protocol to provide scalable and time-efficientrouting for UWSN The H2-DAB routing protocol does notrequire any dimensional location information or any extraspecialized hardware compared with many other routingprotocols in the same areaHowever the problemofmultihoprouting still exists as it is based on multihop architecturewhere nodes near the sinks drain more energy because theyare used more frequently
Packet redundancy and multiple paths can be exploitedin order to increase the reliability of UWSNs Ayaz et al[22] provided a two-hop acknowledgment reliabilitymodel inorder to insure the reliable data deliveries to the surface sinkswhere two copies of the same data packet are maintained inthe network without extra burden on the available resourcesA relay node that has data packets to forward will not replywith the acknowledgment until it cannot find the next hoptowards the destination But if a node is unable to find thenext hop due to any failure or even if it is lost then packetsin the buffer are not considered lost All the nodes that sendthe data packets towards this node will wait for a certainamount of time before trying again for the next hop Xu et al[23] proposed a multiple-path forward error correction (M-FEC) approach that integrated multiple-path communica-tions and Hamming coding to eliminate retransmission andenhance reliability in underwater sensor networksMoreovera Markov model and a dynamical decision and feedbackscheme were developed to decrease the number of the pathsin order to save energy and ensure the desirable packet errorrate However M-FEC may cause much long delay becauseof additional process of encoding and decoding the datapackets
3 Proposed Scheme
31 Network Model and Energy Consumption We considerthat 3D UWSNs are composed of a certain number ofsensor nodes uniformly scattered in monitoring fields Wepresent a generic model for a 3D UWSN that is representedby 119866 = (119881 119864) with 119899 sensor nodes Each sensor nodeis assigned with a triplet of coordinates (119909 119910 119911) We alsoassume that all sensor nodes know their own locationsthrough a certain localization service [24] Such assumptionis justified in underwater systems where fixed bottom-mounted nodes have location information upon deploymentIn fact the underwater localization is a nontrivial taskfor which relatively very few options are available Manyresearchers have proposed a variety of localization schemesand techniques to address this issue specially [25 26] It isnot always feasible to deploy anchor nodes at the sea floor fordeep water environment In this case mobile beacon nodessuch as autonomous underwater vehicles (AUVs) whichare equipped with internal navigation systems are exploitedas reference nodes to assist in corresponding distributed
localization algorithms This paper takes advantage of theseresearch results as existing preconditions
Definition 1 The function 120575(119906 V) defines the distancebetween two nodes 119904
119906and 119904V in a 3D Euclidean space as
120575 119873 times 119873 997888rarr Γ 120575 (119906 V)
120575 (119906 V) = radic(119906119909minus V119909)2+ (119906119910minus V119910)2+ (119906119911minus V119911)2
(1)
Underwater wireless sensor nodes are equipped withsensing devices They collect data from the external environ-ment and transmit these data by one or multihop to the sinknode Sink node is the node that generates data aggregationresults and also the target location of the data transmissionEach sensor node can either transmit or receive data packetsAll sensor nodes can tune their transmission radius rangedfrom 119903min (minimum transmission radius) to 119903max (maximaltransmission radius)
Consider two sensor nodes at minimum hop distance ℎthere exist two values 119906(ℎ) and V(ℎ) such that the Euclideandistance 120575(119906 V) between the two nodes is bounded that is119906(ℎ) le 120575(119906 V) le V(ℎ) The quality of the bounds depends onthe network density 120588 In particular for each ℎ gt 0 holds
lim120588rarrinfin
V (ℎ) minus 119906 (ℎ) = 119903min (2)
where 119903min is the minimum transmission range of the sensornodes
Sensing devices generally have widely different theoret-ical and physical characteristics Thus numerous models ofvarying complexity can be constructed based on applicationneeds and device features However for most kinds ofsensors the sensing ability diminishes as distance increases
Definition 2 For a sensor 119904 the general sensingmodel 119878 at anarbitrary point 119901 is expressed as
119878 (119904 119901) =120582
[119889(119904 119901)]119896 (3)
where 119889(119904 119901) is the Euclidean distance between the sensor 119904and the point 119901 and positive constants 120582 and 119896 are sensortechnology-dependent parameters [27]
We assume that all sensor nodes are equipped withlimited battery resources without recharging or replacingnode batteries after deployment The network lifetime isdefined as the time until the first sensor node in the networkdepletes its energy The energy consumption model is thesame as that in [28] where the attenuation and the energyspreading factor (1 is for cylindrical 15 is for practical and2 is for spherical spreading) are taken into consideration
Acoustic signal has different transmission modes inshallow water (where the depth of the water is lower than100 meters) and deep water (where the depth of the wateris above 100 meters) In shallow water the transmissionof the acoustic signal is limited to a cylindrical area frombottom to the surface while in deepwater the transmission of
4 International Journal of Distributed Sensor Networks
the acoustic signal is mainly with spherical diffusion andthe energy consumption is caused by spherical diffusion andwater absorption This paper concentrates on the shallowwater scenario
The passive sonar equation [29] characterizes the signal-to-noise ratio (SNR) of an emitted underwater signal at thereceiver which is presented by
SNR = SL minus TL minusNL + DI (4)
where SL is the target source level or noise generated bythe target TL is the transmission loss NL is the noise leveland DI is the directivity index (a function of the receiverrsquosdirectional sensitivity)
The transmission loss TL can be defined as the accumu-lated decrease in acoustic intensity as an acoustic pressurewave propagates outwards from a source The transmissionloss for cylindrically spread signals is calculated as
TL = 10 log2120575 (119906 V) + 120572120575 (119906 V) times 10minus3 (5)
where 120575(119906 V) denotes the Euler distance between the trans-mitter and the receiver in meters and 120572 is the frequencydependent medium absorption coefficient in dBKm ItfollowsThorprsquos formula [30] empirically as
120572 =011119891
2
1 + 1198912+
441198912
4100 + 1198912+ 275 times 10
minus41198912+ 0003 (6)
where 119891 is in KHz and 120572 is in dBKmThe noise level NL in shallow water is mainly affected by
waves shipping traffic wind level and the activities of largemammals For simplicity we consider an average value for thenoise level NL to be 70 dB as a representative shallow watercase [30]
SL can be defined as the intensity of the radiated sound indecibels related to the transmitted signal intensity at 1 meterfrom the source according to the following expression
SL = 10 log2
119868119905
1 120583Pa (7)
where 119868119905is in 120583Pa Solving for 119868
119905yields
119868119905= 10
SL10times 067 times 10
minus18 (8)
As a result the transmitter power 119875119905that achieves inten-
sity 119868119905at a distance of 1 meter from the transmitter in the
direction to the receiver is calculated as
119875119905= 2120587 times 119867 times 119868
119905 (9)
where 119875119905is in watts and119867 is the water depth in meters
32 Ultrasonic Frog Calling Strategy UFCA is inspired fromthe calling behavior of concave-eared torrent frog Maleconcave-eared torrent frogs can produce diverse bird-likemelodic advertisement calls with pronounced frequencymodulations that often contain spectral energy in the ultra-sonic range Although female concave-eared torrent frogsexhibit no ultrasonic sensitivity their courtship calls can
5
7
8
6
4zzz 9
zzz
Male
Female
fpfk
fh
fjfi
fq
ri
rj
Figure 1 Ultrasonic frog calling strategy
evoke extraordinarily precise phonotaxis of the male frogswith high localization acuity
Suppose there are six concave-eared torrent frogs ran-domly distributed in a space as shown in Figure 1 Frog 119891
119894
is a gravid female frog (with tone bursts frequency range 1ndash14KHz [10]) Others are male frogs that can emit ultrasonicsound and have ultrasonic hearing capacity in response totone bursts at frequency ranged from 1KHz to 35KHz [10]At first 119891
119894emits a courtship call in order to attract some
nearby male frogs The solid circle with radius 119903119894represents
the covering space of 119891119894rsquos courtship call The number in each
frog denotes its body size As 119891119895is the nearest male frog to
119891119894 it will emit an advertisement call at frequencies ranged
fromnormal sound to ultrasonic sound immediately after thereception of 119891
119894rsquos courtship call The dashed circle with radius
119903119895represents the covering space of 119891
119895rsquos advertisement call
which is bigger than the covering space of of 119891119894rsquos courtship
call After the male frog 119891119896receives 119891
119894rsquos courtship call and 119891
119895rsquos
advertisement call it extracts the body size information fromthese calls As 119891
119896rsquos body size is smaller than that of 119891
119895rsquos it will
not broadcast any advertisement call in order to save energyThe male frog 119891
ℎcan also hear 119891
119895rsquos advertisement call but it
still keeps silent since 119891ℎis located outside of the covering
space of 119891119894rsquos courtship call Both the male frogs 119891
119901and 119891
119902
locatewithin the covering space of119891119894rsquos courtship call Suppose
119891119901and 119891119902receive 119891
119895rsquos advertisement call simultaneously they
compare their body sizes and conclude that the probability ofwinning the competition is high Therefore both 119891
119901and 119891
119902
directly replywith advertisement calls to119891119894 which include the
information of their body sizes and locations Judging fromadvertisement calls of different male frogs 119891
119894selects 119891
119902as its
mate because 119891119902owns the biggest body size among the three
mating candidates 119891119895 119891119901 and 119891
119902 At last 119891
119894calculates 119891
119902rsquos
position and leaps to 119891119902
International Journal of Distributed Sensor Networks 5
5
7
8
6
4zzz 9
zzz
zzz
Sink
Receiver
Transmitter
st
rmin A
shsp sk
sjsi
sq
Gij
r minltr jlt2r min
Figure 2 Candidate discovery phase
33 Routing Algorithm UFCA consists of two phases candi-date discovery phase and relay node selection phase Figure 2illustrates the candidate discovery phase Each frog denotesa sensor node and each number in the frog denotes theresidual energy of local sensor node Sink nodes do not haveany energy constraints because they are equipped with bothradio-frequency (RF) and acousticmodems and are deployedat the water surface As for static sink nodes they only needto broadcast their positions to the whole network one time atthe initial stage of the network operation which would notproduce significant energy dissipation [31] The sensor nodethat holds the data packet is the transmitter which is similarto the gravid female frog in Figure 1 Each data packet carriesthe positions of the source node the sink node and therelay node (ie the node that transmits this packet) Suppose119904119894is a transmitter as shown in Figure 2 then other sensor
nodes are receivers before the data packet is forwarded Atfirst 119904
119894transmits a courtship packet with radius 119903min which
includes the positions of 119904119894and the sink node 119904
119905 As 119904119895is the
nearest receiver to 119904119894 it will calculate the cosine of the angle
between the direction from 119904119894to 119904119895and the direction from 119904
119894
to 119904119905(denoted by 119860 in Figure 2) upon receipt of 119904
119894rsquos courtship
packet If the cosine value is not below zero 119904119895will transmit
an advertisement packet with radius 119903119895 which is calculated as
119903119895= MIN(1 +
120576res119895
120576max119895
) sdot 119903min 119903max (10)
where 120576res119895
denotes the residual energy of sensor node 119904119895and
120576max119895
denotes the maximum energy of sensor node 119904119895 Thus
119903119895ranges from 119903min to 2119903min In the best case the residual
energy of 119904119895is full and 119903max gt 2119903min and 119903
119895equals 2119903min
according to formula (10) which is enough to cover 119904119894rsquos
transmission circle In the worst case the residual energy of119904119895is almost exhausted it will only transmit an advertisement
packet with radius 119903min in order to reach the position of119904119894 Moreover 119904
119895rsquos position and residual energy information
is included in its advertisement packet After 119904119896receives 119904
119894rsquos
courtship packet and 119904119895rsquos advertisement packet it extracts the
position and the residual energy information from thesepackets As 119904
119896rsquos residual energy is less than that of 119904
119895rsquos it
chooses to enter sleep mode in order to save energy withouttransmitting any advertisement packet Another receiver 119904
119902
can also receive 119904119894rsquos courtship packet and 119904
119895rsquos advertisement
packet But 119904119902will choose to enter sleep mode because the
cosine of the angle between the direction from 119904119894to 119904119902and the
direction from 119904119894to 119904119905is below zero In other words 119904
119902locates
in a worse place compared with other receivers Althoughthe receiver 119904
ℎlocates within the transmission radius of 119904
119895rsquos
advertisement packet it still keeps sleep mode since 119904ℎcannot
receive 119904119894rsquos courtship packet After 119904
119901receives 119904
119894rsquos courtship
packet and 119904119895rsquos advertisement packet it extracts the position
and the residual energy information from these packets As119904119901rsquos residual energy is more than that of 119904
119895rsquos and the cosine
of the angle between the direction from 119904119894to 119904119901and the
direction from 119904119894to 119904119905is not below zero it concludes that the
probability of winning the competition is high Therefore 119904119901
will transmit an advertisement packet with radius 119903119901 which
includes the information of its location and residual energyAt last 119904
119894will add 119904
119895and 119904119901to its candidate set after the receipt
of their advertisement packets The sensor node that goes tosleep mode will wake up immediately after another sensornode broadcasts a courtship packet and the sleep sensor nodelocates exactly within its transmission range
The process of selecting a candidate as the relay node toforward the data packet is illustrated in Figure 3 After thetransmitter 119904
119894rsquos candidate set is constructed it will select the
most attractive candidate as the relay node according to acertain standard which is described as the gravity functionin this paper
Definition 3 Given a sensor node 119904119894and its neighbor node 119904
119895
the gravity function from 119904119894to 119904119895is defined as119866
119894119895and its value
is calculated as
10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816=
120576res119894sdot 120576
res119895sdot cos119860
120575(119894 119895)2
(11)
where 120576res119894
and 120576res119895
denote the residual energy of sensor nodes119904119894and 119904119895 119860 is the intersection angle between the direction
from 119904119894to 119904119895and the direction from 119904
119894to the sink node 119904
119905 and
120575(119894 119895) is the Euclidean distance from 119904119894to 119904119895
At last the transmitter 119904119894computes the gravity values
with every sensor node in its candidate set and chooses thecandidatewithmaximal gravity value to be the relay node thatis in charge of forwarding the data packet
Algorithm 1 describes the process of building the routingpath with ultrasonic frog calling algorithm in detail
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Distributed Sensor Networks 3
of network density depending on the physical distancestowards the sink node which in turn consumes additionalenergy Moreover the characteristics of node mobility inUWSNs often make the problem worse Ayaz and Abdullah[21] proposed a hop-by-hop dynamic addressing based (H2-DAB) routing protocol to provide scalable and time-efficientrouting for UWSN The H2-DAB routing protocol does notrequire any dimensional location information or any extraspecialized hardware compared with many other routingprotocols in the same areaHowever the problemofmultihoprouting still exists as it is based on multihop architecturewhere nodes near the sinks drain more energy because theyare used more frequently
Packet redundancy and multiple paths can be exploitedin order to increase the reliability of UWSNs Ayaz et al[22] provided a two-hop acknowledgment reliabilitymodel inorder to insure the reliable data deliveries to the surface sinkswhere two copies of the same data packet are maintained inthe network without extra burden on the available resourcesA relay node that has data packets to forward will not replywith the acknowledgment until it cannot find the next hoptowards the destination But if a node is unable to find thenext hop due to any failure or even if it is lost then packetsin the buffer are not considered lost All the nodes that sendthe data packets towards this node will wait for a certainamount of time before trying again for the next hop Xu et al[23] proposed a multiple-path forward error correction (M-FEC) approach that integrated multiple-path communica-tions and Hamming coding to eliminate retransmission andenhance reliability in underwater sensor networksMoreovera Markov model and a dynamical decision and feedbackscheme were developed to decrease the number of the pathsin order to save energy and ensure the desirable packet errorrate However M-FEC may cause much long delay becauseof additional process of encoding and decoding the datapackets
3 Proposed Scheme
31 Network Model and Energy Consumption We considerthat 3D UWSNs are composed of a certain number ofsensor nodes uniformly scattered in monitoring fields Wepresent a generic model for a 3D UWSN that is representedby 119866 = (119881 119864) with 119899 sensor nodes Each sensor nodeis assigned with a triplet of coordinates (119909 119910 119911) We alsoassume that all sensor nodes know their own locationsthrough a certain localization service [24] Such assumptionis justified in underwater systems where fixed bottom-mounted nodes have location information upon deploymentIn fact the underwater localization is a nontrivial taskfor which relatively very few options are available Manyresearchers have proposed a variety of localization schemesand techniques to address this issue specially [25 26] It isnot always feasible to deploy anchor nodes at the sea floor fordeep water environment In this case mobile beacon nodessuch as autonomous underwater vehicles (AUVs) whichare equipped with internal navigation systems are exploitedas reference nodes to assist in corresponding distributed
localization algorithms This paper takes advantage of theseresearch results as existing preconditions
Definition 1 The function 120575(119906 V) defines the distancebetween two nodes 119904
119906and 119904V in a 3D Euclidean space as
120575 119873 times 119873 997888rarr Γ 120575 (119906 V)
120575 (119906 V) = radic(119906119909minus V119909)2+ (119906119910minus V119910)2+ (119906119911minus V119911)2
(1)
Underwater wireless sensor nodes are equipped withsensing devices They collect data from the external environ-ment and transmit these data by one or multihop to the sinknode Sink node is the node that generates data aggregationresults and also the target location of the data transmissionEach sensor node can either transmit or receive data packetsAll sensor nodes can tune their transmission radius rangedfrom 119903min (minimum transmission radius) to 119903max (maximaltransmission radius)
Consider two sensor nodes at minimum hop distance ℎthere exist two values 119906(ℎ) and V(ℎ) such that the Euclideandistance 120575(119906 V) between the two nodes is bounded that is119906(ℎ) le 120575(119906 V) le V(ℎ) The quality of the bounds depends onthe network density 120588 In particular for each ℎ gt 0 holds
lim120588rarrinfin
V (ℎ) minus 119906 (ℎ) = 119903min (2)
where 119903min is the minimum transmission range of the sensornodes
Sensing devices generally have widely different theoret-ical and physical characteristics Thus numerous models ofvarying complexity can be constructed based on applicationneeds and device features However for most kinds ofsensors the sensing ability diminishes as distance increases
Definition 2 For a sensor 119904 the general sensingmodel 119878 at anarbitrary point 119901 is expressed as
119878 (119904 119901) =120582
[119889(119904 119901)]119896 (3)
where 119889(119904 119901) is the Euclidean distance between the sensor 119904and the point 119901 and positive constants 120582 and 119896 are sensortechnology-dependent parameters [27]
We assume that all sensor nodes are equipped withlimited battery resources without recharging or replacingnode batteries after deployment The network lifetime isdefined as the time until the first sensor node in the networkdepletes its energy The energy consumption model is thesame as that in [28] where the attenuation and the energyspreading factor (1 is for cylindrical 15 is for practical and2 is for spherical spreading) are taken into consideration
Acoustic signal has different transmission modes inshallow water (where the depth of the water is lower than100 meters) and deep water (where the depth of the wateris above 100 meters) In shallow water the transmissionof the acoustic signal is limited to a cylindrical area frombottom to the surface while in deepwater the transmission of
4 International Journal of Distributed Sensor Networks
the acoustic signal is mainly with spherical diffusion andthe energy consumption is caused by spherical diffusion andwater absorption This paper concentrates on the shallowwater scenario
The passive sonar equation [29] characterizes the signal-to-noise ratio (SNR) of an emitted underwater signal at thereceiver which is presented by
SNR = SL minus TL minusNL + DI (4)
where SL is the target source level or noise generated bythe target TL is the transmission loss NL is the noise leveland DI is the directivity index (a function of the receiverrsquosdirectional sensitivity)
The transmission loss TL can be defined as the accumu-lated decrease in acoustic intensity as an acoustic pressurewave propagates outwards from a source The transmissionloss for cylindrically spread signals is calculated as
TL = 10 log2120575 (119906 V) + 120572120575 (119906 V) times 10minus3 (5)
where 120575(119906 V) denotes the Euler distance between the trans-mitter and the receiver in meters and 120572 is the frequencydependent medium absorption coefficient in dBKm ItfollowsThorprsquos formula [30] empirically as
120572 =011119891
2
1 + 1198912+
441198912
4100 + 1198912+ 275 times 10
minus41198912+ 0003 (6)
where 119891 is in KHz and 120572 is in dBKmThe noise level NL in shallow water is mainly affected by
waves shipping traffic wind level and the activities of largemammals For simplicity we consider an average value for thenoise level NL to be 70 dB as a representative shallow watercase [30]
SL can be defined as the intensity of the radiated sound indecibels related to the transmitted signal intensity at 1 meterfrom the source according to the following expression
SL = 10 log2
119868119905
1 120583Pa (7)
where 119868119905is in 120583Pa Solving for 119868
119905yields
119868119905= 10
SL10times 067 times 10
minus18 (8)
As a result the transmitter power 119875119905that achieves inten-
sity 119868119905at a distance of 1 meter from the transmitter in the
direction to the receiver is calculated as
119875119905= 2120587 times 119867 times 119868
119905 (9)
where 119875119905is in watts and119867 is the water depth in meters
32 Ultrasonic Frog Calling Strategy UFCA is inspired fromthe calling behavior of concave-eared torrent frog Maleconcave-eared torrent frogs can produce diverse bird-likemelodic advertisement calls with pronounced frequencymodulations that often contain spectral energy in the ultra-sonic range Although female concave-eared torrent frogsexhibit no ultrasonic sensitivity their courtship calls can
5
7
8
6
4zzz 9
zzz
Male
Female
fpfk
fh
fjfi
fq
ri
rj
Figure 1 Ultrasonic frog calling strategy
evoke extraordinarily precise phonotaxis of the male frogswith high localization acuity
Suppose there are six concave-eared torrent frogs ran-domly distributed in a space as shown in Figure 1 Frog 119891
119894
is a gravid female frog (with tone bursts frequency range 1ndash14KHz [10]) Others are male frogs that can emit ultrasonicsound and have ultrasonic hearing capacity in response totone bursts at frequency ranged from 1KHz to 35KHz [10]At first 119891
119894emits a courtship call in order to attract some
nearby male frogs The solid circle with radius 119903119894represents
the covering space of 119891119894rsquos courtship call The number in each
frog denotes its body size As 119891119895is the nearest male frog to
119891119894 it will emit an advertisement call at frequencies ranged
fromnormal sound to ultrasonic sound immediately after thereception of 119891
119894rsquos courtship call The dashed circle with radius
119903119895represents the covering space of 119891
119895rsquos advertisement call
which is bigger than the covering space of of 119891119894rsquos courtship
call After the male frog 119891119896receives 119891
119894rsquos courtship call and 119891
119895rsquos
advertisement call it extracts the body size information fromthese calls As 119891
119896rsquos body size is smaller than that of 119891
119895rsquos it will
not broadcast any advertisement call in order to save energyThe male frog 119891
ℎcan also hear 119891
119895rsquos advertisement call but it
still keeps silent since 119891ℎis located outside of the covering
space of 119891119894rsquos courtship call Both the male frogs 119891
119901and 119891
119902
locatewithin the covering space of119891119894rsquos courtship call Suppose
119891119901and 119891119902receive 119891
119895rsquos advertisement call simultaneously they
compare their body sizes and conclude that the probability ofwinning the competition is high Therefore both 119891
119901and 119891
119902
directly replywith advertisement calls to119891119894 which include the
information of their body sizes and locations Judging fromadvertisement calls of different male frogs 119891
119894selects 119891
119902as its
mate because 119891119902owns the biggest body size among the three
mating candidates 119891119895 119891119901 and 119891
119902 At last 119891
119894calculates 119891
119902rsquos
position and leaps to 119891119902
International Journal of Distributed Sensor Networks 5
5
7
8
6
4zzz 9
zzz
zzz
Sink
Receiver
Transmitter
st
rmin A
shsp sk
sjsi
sq
Gij
r minltr jlt2r min
Figure 2 Candidate discovery phase
33 Routing Algorithm UFCA consists of two phases candi-date discovery phase and relay node selection phase Figure 2illustrates the candidate discovery phase Each frog denotesa sensor node and each number in the frog denotes theresidual energy of local sensor node Sink nodes do not haveany energy constraints because they are equipped with bothradio-frequency (RF) and acousticmodems and are deployedat the water surface As for static sink nodes they only needto broadcast their positions to the whole network one time atthe initial stage of the network operation which would notproduce significant energy dissipation [31] The sensor nodethat holds the data packet is the transmitter which is similarto the gravid female frog in Figure 1 Each data packet carriesthe positions of the source node the sink node and therelay node (ie the node that transmits this packet) Suppose119904119894is a transmitter as shown in Figure 2 then other sensor
nodes are receivers before the data packet is forwarded Atfirst 119904
119894transmits a courtship packet with radius 119903min which
includes the positions of 119904119894and the sink node 119904
119905 As 119904119895is the
nearest receiver to 119904119894 it will calculate the cosine of the angle
between the direction from 119904119894to 119904119895and the direction from 119904
119894
to 119904119905(denoted by 119860 in Figure 2) upon receipt of 119904
119894rsquos courtship
packet If the cosine value is not below zero 119904119895will transmit
an advertisement packet with radius 119903119895 which is calculated as
119903119895= MIN(1 +
120576res119895
120576max119895
) sdot 119903min 119903max (10)
where 120576res119895
denotes the residual energy of sensor node 119904119895and
120576max119895
denotes the maximum energy of sensor node 119904119895 Thus
119903119895ranges from 119903min to 2119903min In the best case the residual
energy of 119904119895is full and 119903max gt 2119903min and 119903
119895equals 2119903min
according to formula (10) which is enough to cover 119904119894rsquos
transmission circle In the worst case the residual energy of119904119895is almost exhausted it will only transmit an advertisement
packet with radius 119903min in order to reach the position of119904119894 Moreover 119904
119895rsquos position and residual energy information
is included in its advertisement packet After 119904119896receives 119904
119894rsquos
courtship packet and 119904119895rsquos advertisement packet it extracts the
position and the residual energy information from thesepackets As 119904
119896rsquos residual energy is less than that of 119904
119895rsquos it
chooses to enter sleep mode in order to save energy withouttransmitting any advertisement packet Another receiver 119904
119902
can also receive 119904119894rsquos courtship packet and 119904
119895rsquos advertisement
packet But 119904119902will choose to enter sleep mode because the
cosine of the angle between the direction from 119904119894to 119904119902and the
direction from 119904119894to 119904119905is below zero In other words 119904
119902locates
in a worse place compared with other receivers Althoughthe receiver 119904
ℎlocates within the transmission radius of 119904
119895rsquos
advertisement packet it still keeps sleep mode since 119904ℎcannot
receive 119904119894rsquos courtship packet After 119904
119901receives 119904
119894rsquos courtship
packet and 119904119895rsquos advertisement packet it extracts the position
and the residual energy information from these packets As119904119901rsquos residual energy is more than that of 119904
119895rsquos and the cosine
of the angle between the direction from 119904119894to 119904119901and the
direction from 119904119894to 119904119905is not below zero it concludes that the
probability of winning the competition is high Therefore 119904119901
will transmit an advertisement packet with radius 119903119901 which
includes the information of its location and residual energyAt last 119904
119894will add 119904
119895and 119904119901to its candidate set after the receipt
of their advertisement packets The sensor node that goes tosleep mode will wake up immediately after another sensornode broadcasts a courtship packet and the sleep sensor nodelocates exactly within its transmission range
The process of selecting a candidate as the relay node toforward the data packet is illustrated in Figure 3 After thetransmitter 119904
119894rsquos candidate set is constructed it will select the
most attractive candidate as the relay node according to acertain standard which is described as the gravity functionin this paper
Definition 3 Given a sensor node 119904119894and its neighbor node 119904
119895
the gravity function from 119904119894to 119904119895is defined as119866
119894119895and its value
is calculated as
10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816=
120576res119894sdot 120576
res119895sdot cos119860
120575(119894 119895)2
(11)
where 120576res119894
and 120576res119895
denote the residual energy of sensor nodes119904119894and 119904119895 119860 is the intersection angle between the direction
from 119904119894to 119904119895and the direction from 119904
119894to the sink node 119904
119905 and
120575(119894 119895) is the Euclidean distance from 119904119894to 119904119895
At last the transmitter 119904119894computes the gravity values
with every sensor node in its candidate set and chooses thecandidatewithmaximal gravity value to be the relay node thatis in charge of forwarding the data packet
Algorithm 1 describes the process of building the routingpath with ultrasonic frog calling algorithm in detail
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
4 International Journal of Distributed Sensor Networks
the acoustic signal is mainly with spherical diffusion andthe energy consumption is caused by spherical diffusion andwater absorption This paper concentrates on the shallowwater scenario
The passive sonar equation [29] characterizes the signal-to-noise ratio (SNR) of an emitted underwater signal at thereceiver which is presented by
SNR = SL minus TL minusNL + DI (4)
where SL is the target source level or noise generated bythe target TL is the transmission loss NL is the noise leveland DI is the directivity index (a function of the receiverrsquosdirectional sensitivity)
The transmission loss TL can be defined as the accumu-lated decrease in acoustic intensity as an acoustic pressurewave propagates outwards from a source The transmissionloss for cylindrically spread signals is calculated as
TL = 10 log2120575 (119906 V) + 120572120575 (119906 V) times 10minus3 (5)
where 120575(119906 V) denotes the Euler distance between the trans-mitter and the receiver in meters and 120572 is the frequencydependent medium absorption coefficient in dBKm ItfollowsThorprsquos formula [30] empirically as
120572 =011119891
2
1 + 1198912+
441198912
4100 + 1198912+ 275 times 10
minus41198912+ 0003 (6)
where 119891 is in KHz and 120572 is in dBKmThe noise level NL in shallow water is mainly affected by
waves shipping traffic wind level and the activities of largemammals For simplicity we consider an average value for thenoise level NL to be 70 dB as a representative shallow watercase [30]
SL can be defined as the intensity of the radiated sound indecibels related to the transmitted signal intensity at 1 meterfrom the source according to the following expression
SL = 10 log2
119868119905
1 120583Pa (7)
where 119868119905is in 120583Pa Solving for 119868
119905yields
119868119905= 10
SL10times 067 times 10
minus18 (8)
As a result the transmitter power 119875119905that achieves inten-
sity 119868119905at a distance of 1 meter from the transmitter in the
direction to the receiver is calculated as
119875119905= 2120587 times 119867 times 119868
119905 (9)
where 119875119905is in watts and119867 is the water depth in meters
32 Ultrasonic Frog Calling Strategy UFCA is inspired fromthe calling behavior of concave-eared torrent frog Maleconcave-eared torrent frogs can produce diverse bird-likemelodic advertisement calls with pronounced frequencymodulations that often contain spectral energy in the ultra-sonic range Although female concave-eared torrent frogsexhibit no ultrasonic sensitivity their courtship calls can
5
7
8
6
4zzz 9
zzz
Male
Female
fpfk
fh
fjfi
fq
ri
rj
Figure 1 Ultrasonic frog calling strategy
evoke extraordinarily precise phonotaxis of the male frogswith high localization acuity
Suppose there are six concave-eared torrent frogs ran-domly distributed in a space as shown in Figure 1 Frog 119891
119894
is a gravid female frog (with tone bursts frequency range 1ndash14KHz [10]) Others are male frogs that can emit ultrasonicsound and have ultrasonic hearing capacity in response totone bursts at frequency ranged from 1KHz to 35KHz [10]At first 119891
119894emits a courtship call in order to attract some
nearby male frogs The solid circle with radius 119903119894represents
the covering space of 119891119894rsquos courtship call The number in each
frog denotes its body size As 119891119895is the nearest male frog to
119891119894 it will emit an advertisement call at frequencies ranged
fromnormal sound to ultrasonic sound immediately after thereception of 119891
119894rsquos courtship call The dashed circle with radius
119903119895represents the covering space of 119891
119895rsquos advertisement call
which is bigger than the covering space of of 119891119894rsquos courtship
call After the male frog 119891119896receives 119891
119894rsquos courtship call and 119891
119895rsquos
advertisement call it extracts the body size information fromthese calls As 119891
119896rsquos body size is smaller than that of 119891
119895rsquos it will
not broadcast any advertisement call in order to save energyThe male frog 119891
ℎcan also hear 119891
119895rsquos advertisement call but it
still keeps silent since 119891ℎis located outside of the covering
space of 119891119894rsquos courtship call Both the male frogs 119891
119901and 119891
119902
locatewithin the covering space of119891119894rsquos courtship call Suppose
119891119901and 119891119902receive 119891
119895rsquos advertisement call simultaneously they
compare their body sizes and conclude that the probability ofwinning the competition is high Therefore both 119891
119901and 119891
119902
directly replywith advertisement calls to119891119894 which include the
information of their body sizes and locations Judging fromadvertisement calls of different male frogs 119891
119894selects 119891
119902as its
mate because 119891119902owns the biggest body size among the three
mating candidates 119891119895 119891119901 and 119891
119902 At last 119891
119894calculates 119891
119902rsquos
position and leaps to 119891119902
International Journal of Distributed Sensor Networks 5
5
7
8
6
4zzz 9
zzz
zzz
Sink
Receiver
Transmitter
st
rmin A
shsp sk
sjsi
sq
Gij
r minltr jlt2r min
Figure 2 Candidate discovery phase
33 Routing Algorithm UFCA consists of two phases candi-date discovery phase and relay node selection phase Figure 2illustrates the candidate discovery phase Each frog denotesa sensor node and each number in the frog denotes theresidual energy of local sensor node Sink nodes do not haveany energy constraints because they are equipped with bothradio-frequency (RF) and acousticmodems and are deployedat the water surface As for static sink nodes they only needto broadcast their positions to the whole network one time atthe initial stage of the network operation which would notproduce significant energy dissipation [31] The sensor nodethat holds the data packet is the transmitter which is similarto the gravid female frog in Figure 1 Each data packet carriesthe positions of the source node the sink node and therelay node (ie the node that transmits this packet) Suppose119904119894is a transmitter as shown in Figure 2 then other sensor
nodes are receivers before the data packet is forwarded Atfirst 119904
119894transmits a courtship packet with radius 119903min which
includes the positions of 119904119894and the sink node 119904
119905 As 119904119895is the
nearest receiver to 119904119894 it will calculate the cosine of the angle
between the direction from 119904119894to 119904119895and the direction from 119904
119894
to 119904119905(denoted by 119860 in Figure 2) upon receipt of 119904
119894rsquos courtship
packet If the cosine value is not below zero 119904119895will transmit
an advertisement packet with radius 119903119895 which is calculated as
119903119895= MIN(1 +
120576res119895
120576max119895
) sdot 119903min 119903max (10)
where 120576res119895
denotes the residual energy of sensor node 119904119895and
120576max119895
denotes the maximum energy of sensor node 119904119895 Thus
119903119895ranges from 119903min to 2119903min In the best case the residual
energy of 119904119895is full and 119903max gt 2119903min and 119903
119895equals 2119903min
according to formula (10) which is enough to cover 119904119894rsquos
transmission circle In the worst case the residual energy of119904119895is almost exhausted it will only transmit an advertisement
packet with radius 119903min in order to reach the position of119904119894 Moreover 119904
119895rsquos position and residual energy information
is included in its advertisement packet After 119904119896receives 119904
119894rsquos
courtship packet and 119904119895rsquos advertisement packet it extracts the
position and the residual energy information from thesepackets As 119904
119896rsquos residual energy is less than that of 119904
119895rsquos it
chooses to enter sleep mode in order to save energy withouttransmitting any advertisement packet Another receiver 119904
119902
can also receive 119904119894rsquos courtship packet and 119904
119895rsquos advertisement
packet But 119904119902will choose to enter sleep mode because the
cosine of the angle between the direction from 119904119894to 119904119902and the
direction from 119904119894to 119904119905is below zero In other words 119904
119902locates
in a worse place compared with other receivers Althoughthe receiver 119904
ℎlocates within the transmission radius of 119904
119895rsquos
advertisement packet it still keeps sleep mode since 119904ℎcannot
receive 119904119894rsquos courtship packet After 119904
119901receives 119904
119894rsquos courtship
packet and 119904119895rsquos advertisement packet it extracts the position
and the residual energy information from these packets As119904119901rsquos residual energy is more than that of 119904
119895rsquos and the cosine
of the angle between the direction from 119904119894to 119904119901and the
direction from 119904119894to 119904119905is not below zero it concludes that the
probability of winning the competition is high Therefore 119904119901
will transmit an advertisement packet with radius 119903119901 which
includes the information of its location and residual energyAt last 119904
119894will add 119904
119895and 119904119901to its candidate set after the receipt
of their advertisement packets The sensor node that goes tosleep mode will wake up immediately after another sensornode broadcasts a courtship packet and the sleep sensor nodelocates exactly within its transmission range
The process of selecting a candidate as the relay node toforward the data packet is illustrated in Figure 3 After thetransmitter 119904
119894rsquos candidate set is constructed it will select the
most attractive candidate as the relay node according to acertain standard which is described as the gravity functionin this paper
Definition 3 Given a sensor node 119904119894and its neighbor node 119904
119895
the gravity function from 119904119894to 119904119895is defined as119866
119894119895and its value
is calculated as
10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816=
120576res119894sdot 120576
res119895sdot cos119860
120575(119894 119895)2
(11)
where 120576res119894
and 120576res119895
denote the residual energy of sensor nodes119904119894and 119904119895 119860 is the intersection angle between the direction
from 119904119894to 119904119895and the direction from 119904
119894to the sink node 119904
119905 and
120575(119894 119895) is the Euclidean distance from 119904119894to 119904119895
At last the transmitter 119904119894computes the gravity values
with every sensor node in its candidate set and chooses thecandidatewithmaximal gravity value to be the relay node thatis in charge of forwarding the data packet
Algorithm 1 describes the process of building the routingpath with ultrasonic frog calling algorithm in detail
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Distributed Sensor Networks 5
5
7
8
6
4zzz 9
zzz
zzz
Sink
Receiver
Transmitter
st
rmin A
shsp sk
sjsi
sq
Gij
r minltr jlt2r min
Figure 2 Candidate discovery phase
33 Routing Algorithm UFCA consists of two phases candi-date discovery phase and relay node selection phase Figure 2illustrates the candidate discovery phase Each frog denotesa sensor node and each number in the frog denotes theresidual energy of local sensor node Sink nodes do not haveany energy constraints because they are equipped with bothradio-frequency (RF) and acousticmodems and are deployedat the water surface As for static sink nodes they only needto broadcast their positions to the whole network one time atthe initial stage of the network operation which would notproduce significant energy dissipation [31] The sensor nodethat holds the data packet is the transmitter which is similarto the gravid female frog in Figure 1 Each data packet carriesthe positions of the source node the sink node and therelay node (ie the node that transmits this packet) Suppose119904119894is a transmitter as shown in Figure 2 then other sensor
nodes are receivers before the data packet is forwarded Atfirst 119904
119894transmits a courtship packet with radius 119903min which
includes the positions of 119904119894and the sink node 119904
119905 As 119904119895is the
nearest receiver to 119904119894 it will calculate the cosine of the angle
between the direction from 119904119894to 119904119895and the direction from 119904
119894
to 119904119905(denoted by 119860 in Figure 2) upon receipt of 119904
119894rsquos courtship
packet If the cosine value is not below zero 119904119895will transmit
an advertisement packet with radius 119903119895 which is calculated as
119903119895= MIN(1 +
120576res119895
120576max119895
) sdot 119903min 119903max (10)
where 120576res119895
denotes the residual energy of sensor node 119904119895and
120576max119895
denotes the maximum energy of sensor node 119904119895 Thus
119903119895ranges from 119903min to 2119903min In the best case the residual
energy of 119904119895is full and 119903max gt 2119903min and 119903
119895equals 2119903min
according to formula (10) which is enough to cover 119904119894rsquos
transmission circle In the worst case the residual energy of119904119895is almost exhausted it will only transmit an advertisement
packet with radius 119903min in order to reach the position of119904119894 Moreover 119904
119895rsquos position and residual energy information
is included in its advertisement packet After 119904119896receives 119904
119894rsquos
courtship packet and 119904119895rsquos advertisement packet it extracts the
position and the residual energy information from thesepackets As 119904
119896rsquos residual energy is less than that of 119904
119895rsquos it
chooses to enter sleep mode in order to save energy withouttransmitting any advertisement packet Another receiver 119904
119902
can also receive 119904119894rsquos courtship packet and 119904
119895rsquos advertisement
packet But 119904119902will choose to enter sleep mode because the
cosine of the angle between the direction from 119904119894to 119904119902and the
direction from 119904119894to 119904119905is below zero In other words 119904
119902locates
in a worse place compared with other receivers Althoughthe receiver 119904
ℎlocates within the transmission radius of 119904
119895rsquos
advertisement packet it still keeps sleep mode since 119904ℎcannot
receive 119904119894rsquos courtship packet After 119904
119901receives 119904
119894rsquos courtship
packet and 119904119895rsquos advertisement packet it extracts the position
and the residual energy information from these packets As119904119901rsquos residual energy is more than that of 119904
119895rsquos and the cosine
of the angle between the direction from 119904119894to 119904119901and the
direction from 119904119894to 119904119905is not below zero it concludes that the
probability of winning the competition is high Therefore 119904119901
will transmit an advertisement packet with radius 119903119901 which
includes the information of its location and residual energyAt last 119904
119894will add 119904
119895and 119904119901to its candidate set after the receipt
of their advertisement packets The sensor node that goes tosleep mode will wake up immediately after another sensornode broadcasts a courtship packet and the sleep sensor nodelocates exactly within its transmission range
The process of selecting a candidate as the relay node toforward the data packet is illustrated in Figure 3 After thetransmitter 119904
119894rsquos candidate set is constructed it will select the
most attractive candidate as the relay node according to acertain standard which is described as the gravity functionin this paper
Definition 3 Given a sensor node 119904119894and its neighbor node 119904
119895
the gravity function from 119904119894to 119904119895is defined as119866
119894119895and its value
is calculated as
10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816=
120576res119894sdot 120576
res119895sdot cos119860
120575(119894 119895)2
(11)
where 120576res119894
and 120576res119895
denote the residual energy of sensor nodes119904119894and 119904119895 119860 is the intersection angle between the direction
from 119904119894to 119904119895and the direction from 119904
119894to the sink node 119904
119905 and
120575(119894 119895) is the Euclidean distance from 119904119894to 119904119895
At last the transmitter 119904119894computes the gravity values
with every sensor node in its candidate set and chooses thecandidatewithmaximal gravity value to be the relay node thatis in charge of forwarding the data packet
Algorithm 1 describes the process of building the routingpath with ultrasonic frog calling algorithm in detail
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
6 International Journal of Distributed Sensor Networks
5
7
6
Sink
Receiver
Transmitter
st
sp
Gip
rminsi
Gij
sj
A
Figure 3 Relay node selection phase
All data packets at relay nodes should have limited life-time which are controlled by TTL (time-to-live) informationcarried in the packet header At first the routing path 119901 iscreated as an empty queue structure after initialization asdescribed in line 1 While TTL value is bigger than zero andthe sink node is not reached the process of building therouting path is repeatedly executed And then the sourcenode 119904
119894resets its candidate set and transmits a courtship
packetwith theminimum transmission radius 119903min in order tofind some candidates as described from line 3 to line 4 Afterthat all sensor nodes that locate within the covering spaceof 119904119894rsquos transmission radius will check their positions Suppose
119904119895is the first receiver with 120575(119894 119895) lt 119903min If the cosine of the
angle between the direction from 119904119894to 119904119895and the direction
from 119904119894to 119904119905is below zero then 119904
119895chooses to enter sleep
mode for saving energy Otherwise 119904119894adds 119904
119895to its candidate
set and 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10) And then all sensor nodes thatlocate within the covering space of 119904
119895rsquos transmission radius
will compare their residual energy with that of 119904119895rsquos Suppose
119904119896is a sensor node that receives 119904
119894rsquos courtship packet and 119904
119895rsquos
advertisement packet If 119904119896rsquos residual energy is less than that of
119904119895rsquos it will choose to enter sleepmodewithout competingwith
119904119895 Otherwise 119904
119894will add 119904
119895to its candidate setThe operation
is iterated until all candidates are discovered as describedfrom line 5 to line 17 During the phase of relay node selection119904119894calculates the gravity values with every sensor node in its
candidate set Suppose 119904119895is the candidate with the maximal
gravity value among all candidates As a result 119904119895is selected as
the relay node and is added to the routing path 119901 as describedfrom line 18 to line 19 Hereafter 119904
119895becomes the transmitter
and continues to find the relay node of next hop Meanwhilethe TTL value is decreased by 1 so as to control the lifetimeof the data packet as described from line 20 to line 23 At lastif the sink node 119904
119905is found within the given TTL value an
optimized routing path 119901 is returned Otherwise all elementswill be removed from 119901 which means no sink node is foundas described from line 24 to line 28
In many proactive routing protocols the active sensornodesmust send periodic update packets to other nodes evenwhen the routing information is similar to the previous oneMoreover the storage overhead for routing tablemaintenancealso grows quickly as the size of the network increasesAlthough some reactive routing protocols can avoid theoverhead incurred by routing tablemaintenance the periodicflooding messages for the routing path discovery is anotherdeadly cost in resource-constraint underwaterwireless sensornetworks In UFCA the update of candidate set is evokedonly when this sensor node is selected as a transmitter Afterthat it can determine where to forward a data packet withoutthe need of routing table maintenance or any floodingmechanism
4 Performance Evaluation
41 Simulation Settings We use Aqua-Sim [32] as simulationframework to evaluate our approach Aqua-Sim is an 119899119904-2based underwater sensor network simulator developed byunderwater sensor network lab at University of ConnecticutTo simulate acoustic channels we extend Aqua-Sim withspherical path loss andThorp attenuationWe use a 3D regionwith size 1000m times 1000m times 1000m and different numberof sensor nodes varied from 100 to 600 Six sink nodes arerandomly deployed at the water surface which are assumedstationary in all simulations The sensor nodes follow therandom-walk mobility pattern Each sensor node randomlyselects a direction and moves to the new position with arandom speed between the minimal speed and maximalspeed which are 0ms and 4ms respectively The datagenerating rate varies fromone packet per second to 6 packetsper second with a packet size of 50 bytes (ie from 400 bpsto 24 kbps) The communication parameters are similar tothose on a commercial acoustic modem and the bit rate is10 kbps TTL (time-to-live) value is set to 30 hops for eachdata packet Each result is obtained from the average run of40 times
As the long propagation delay and limited bandwidth ofacoustic channels make the existing MAC protocols widelyused in radio networks unpractical for UWSNs this paperadopts R-MAC [32] protocol as the underlyingMACprotocolin order to avoid data packet collision R-MAC schedules thetransmission of control packets and data packets at both thesender and the receiver to avoid data packet collisionsThere-fore we donot distinguish courtship packet and advertisementpacket from each other inMAC layer In fact we only need tomake certain that which node is the sender and which nodeis the receiver in this session
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Distributed Sensor Networks 7
Input source node 119904119894 sink node 119904
119905 TTL
Output routing path p(1) Queue 119901 larr Φ routing path initialization(2) while (TTL gt 0) and (119904
119894= 119904119905) do
(3) 119904119894119862119886119899119889119894119878119890119905 larr Φ
(4) 119904119894transmits a courtship packet with radius 119903min
(5) for all 119904119895with 120575(119894 119895) lt 119903min do
(6) if cos (ang119879119868119869) lt 0 then(7) 119904
119895sleep()
(8) else 119904119894CandiSetadd(119904
119895)
(9) 119904119895transmits an advertisement packet with radius 119903
119895
according to formula (10)(10) for all 119904
119896with 120575(119895 119896) lt 119903
119895do
(11) if (120576res119896
lt 120576res119895) then
(12) 119904119896sleep()
(13) else 119904119894CandiSetadd(119904
119896)
(14) endif(15) endfor(16) endif(17) endfor(18) if 10038161003816100381610038161003816119866119894119895
10038161003816100381610038161003816= MAX119866
119894119896(119904119896isin 119904119894CandiSet) then
(19) penqueue(119904119895)
(20) 119904119894larr 119904119895
(21) TTLminusminus(22) endif(23) endwhile(24) if 119904
119894= 119904119905then
(25) pclear()(26) return Φ(27) else return p(28) endif
Algorithm 1 Building the routing path with UFCA
We use the followingmetrics to evaluate the performanceof different routing protocols
(1) Packet delivery ratio is defined as the ratio of thenumber of distinct data packets received successfullyat the sinks to the total number of data packetsgenerated at the source node
(2) Energy consumption takes into account the totalenergy consumed in packet delivery including trans-mitting receiving and idling energy consumption ofall nodes in the network
(3) Throughput equals the total data bits received at thesink nodes divided by the simulation time
(4) Average end-to-end delay represents the average timetaken by a data packet that travels from a source nodeto any sink node
We compared the performance of ultrasonic frog callingalgorithm (UFCA) with that of vector-based forwarding(VBF) andERP2R (energy-efficient routing protocol based onphysical distance and residual energy) In the simulations ofUFCA the minimal and maximal transmission range is setto 50 meters and 100 meters respectively in all directionswhile the transmission range in VBF and ERP2R is fixed at
100 meters Moreover the routing pipe radius in VBF is set to20 meters which is a default value in [16]
42 Simulation Results In the first set of simulations wecompared the packet delivery ratio with the number of nodesin different routing protocols The average speed of nodesis set to 2ms As shown in Figure 4 the packet deliveryratio of three routing protocols is proportional to the numberof nodes UFCA performs best among the three routingprotocols in the same circumstances andVBF achieves higherpacket delivery ratio than that of ERP2RMoreover the curveof VBF rises faster than other protocols This is because withthe growth of network density more sensor nodes will fall inthe routing pipe of VBF with fixed radius as the transmissionrange The packet delivery ration of UFCA is significantlyimproved over other protocols especially when the network issparse as UFCA can findmore routing paths for data deliveryin sparse networks Specifically UFCA improves 343 of thepacket delivery ratio than that of ERP2R and 119 of thepacket delivery ratio than that of VBF on average
Figure 5 illustrates the comparison of the packet deliv-ery ratio with average speed of nodes in different routingprotocols The number of sensor nodes is set to 400 foreach protocol Overall the packet delivery ratio of three
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
8 International Journal of Distributed Sensor Networks
100 200 300 400 500 60003
04
05
06
07
08
09
10
Pack
et d
eliv
ery
ratio
Number of nodes ()
VBFERP2RUFCA
Figure 4 Packet delivery ratio versus number of nodes
routing protocols is inversely proportional to average speedof nodes UFCA achieves higher packet delivery ratio thanthat of ERP2R and VBF when their speeds of nodes are thesame The packet delivery ratio of ERP2R decreases rapidlywith the growth of node mobility This is because the rate ofupdating routing information in ERP2R cannot catch up withthe increase of node mobility Specifically UFCA improves325 of the packet delivery ratio than that of ERP2R and64 of the packet delivery ratio than that of VBF on average
In the second set of simulations we compared the energyconsumption with the number of nodes in different routingprotocols The average speed of nodes is set to 2ms Asshown in Figure 6 the energy consumption of three routingprotocols is proportional to the number of nodes UFCAperforms better than other routing protocols in the samecircumstances Moreover the curve of UFCA has a gentlerslope compared with that of ERP2R and VBF This is mainlydue to more sensor nodes entering the sleep mode with theincrease in sensor nodes in UFCA ERP2R consumes lessenergy than VBF because energy factor is not given in therouting determination of VBF As a result UFCA decreases261 of the energy consumption than ERP2R and 415 ofthe energy consumption than VBF on average
Figure 7 illustrates the comparison of the energy con-sumption with average speed of nodes in different routingprotocols The number of nodes is set to 400 for eachprotocol The energy consumption of three routing protocolsis proportional to the TTL value UFCA consumes less energythan ERP2R and VBF when their speeds of nodes are thesame Moreover the curve slopes of UFCA and VBF arerather gentle compared with that of ERP2R which means
0 1 2 3 4
060
065
070
075
080
085
090
Pack
et d
eliv
ery
ratio
Average speed of nodes (ms)
VBFERP2RUFCA
Figure 5 Packet delivery ratio versus average speed of nodes
100 200 300 400 500 60006
08
10
12
14
16
18
20
Number of nodes ()
VBFERP2RUFCA
Ener
gy co
nsum
ptio
n (104
mJ)
Figure 6 Energy consumption versus number of nodes
that the factor of node mobility has slight influence onenergy consumption of UFCA and VBF ERP2R consumesless energy than VBF except when average speed of nodesreaches 4ms On average UFCA decreases 257 of theenergy consumption than ERP2R and 362 of the energyconsumption than VBF
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Distributed Sensor Networks 9
0 1 2 3 406
08
10
12
14
16
18
20
Average speed of nodes (ms)
Ener
gy co
nsum
ptio
n (104
mJ)
VBFERP2RUFCA
Figure 7 Energy consumption versus average speed of nodes
In the third set of simulations we compared the through-put with the number of nodes in different routing protocolsThe average speed of nodes is set to 2ms for each protocol Asshown in Figure 8 the throughput of three routing protocolsis proportional to the number of nodes The front parts ofcurves indicate rapid increases in throughput while the rearparts of curves show slow growth rates after the number ofnodes has reached high value The reason is that with thegrowth of network density the routing paths become morecrowded and downstream nodes cannot receive data packetsfrom several of its upstream nodes simultaneously OverallUFCA performs better than other routing protocols in thesame circumstances VBF achieves higher throughput thanERP2R On average UFCA improves 215 of the throughputthan ERP2R and 93 of the throughput than VBF
Figure 9 depicts the comparison of the throughput withaverage speed of nodes in different routing protocols Thenumber of nodes is set to 400 for each protocolThe through-put of three routing protocols is inversely proportional toaverage speed of nodes UFCA achieves higher throughputthan that of ERP2R and VBF when their average speeds ofnodes are the same Noticeably the throughput of ERP2Rdecreases sharply when average speed of nodes is more than2ms This is because more routing cost and residual energyof the nodes as well as their neighbors along routing pathshave to be recalculated with the increase in average speedof nodes in ERP2R On average UFCA improves 154 ofthe throughput than ERP2R and 65 of the throughput thanVBF
In the last set of simulations we compared the averageend-to-end delay with the number of nodes in different
100 200 300 400 500 600
Thro
ughp
ut (b
itss
)
Number of nodes ()
12k
11k
10k
9k
8k
7k
6k
5k
4k
3k
VBFERP2RUFCA
Figure 8 Throughput versus number of nodes
0 1 2 3 4
Thro
ughp
ut (b
itss
)
Average speed of nodes (ms)
105 k
100 k
95k
90k
85 k
80 k
VBFERP2RUFCA
Figure 9 Throughput versus average speed of nodes
routing protocols The average speed of nodes is set to 2msfor each protocol As shown in Figure 10 the average end-to-end delay of three routing protocols is inversely proportionalto the number of nodes UFCA achieves less end-to-enddelay than ERP2R and VBF when the number of nodes isthe same The reason is that UFCA introduces less control
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
10 International Journal of Distributed Sensor Networks
100 200 300 400 500 600
400
450
500
550
600
650
700
750
800
850
900
950
Aver
age e
nd-to
-end
del
ay (m
s)
Number of nodes ()
VBFERP2RUFCA
Figure 10 Average end-to-end delay versus number of nodes
packets than other protocols for communicating with therelated sensor nodes during the process of routing The costfor the computation of residual energy and gravity valuesin UFCA is far less than that in network communicationERP2Rperforms better thanVBF because the highest prioritynode in ERP2R has a holding time of zero which can reducethe end-to-end delay to a certain degree On average UFCAdecreases 112 of the average end-to-end delay than ERP2Rand 312 of the average end-to-end delay than VBF
Figure 11 shows the comparison of the average end-to-end delay with the average speed of nodes in differentrouting protocols The number of nodes is set to 400 foreach protocol Overall the average end-to-end delay of threerouting protocols is inversely proportional to the averagespeed of nodes UFCA achieves less end-to-end delay thanERP2R and VBF when their average speeds of nodes are thesame It is worth noting that ERP2R owns a curve with rapidincreasing trendThis is becausemore sensor nodes in ERP2Rneed to reevaluate their distances to the sink node with thegrowth of node mobility Specifically UFCA decreases 81of the average end-to-end delay than ERP2R and 263 of theaverage end-to-end delay than VBF on average
43 Discussion Compared to algorithms such as VBF andERP2R UFCA is totally a different approach In VBF onlythe sensor nodes located in a predefined routing pipe areeligible for packet forwarding and those which are not closeto the routing pipe do not forward the packets no matterwhether they are suitable for building a shorter routingpath Therefore the routing performance in VBF mainlydepends on the node density and it cannot benefit from the
0 1 2 3 4250
300
350
400
450
500
550
600
650
700
750
800
Average speed of nodes (ms)
Aver
age e
nd-to
-end
del
ay (m
s)
VBFERP2RUFCA
Figure 11 Average end-to-end delay versus speed of nodes
deployment ofmultiple sink nodes if they are not close to eachother In ERP2R forwarding nodes are selected based on thephysical distance of the sensor nodes Each sender selects thenodes nearer to the sink node for routing decision which isnot always helpful when the node density is sparse AlthoughERP2R can balance the energy consumption using a residualenergy-based timer its performance decreases dramaticallywith the growth of node mobility UFCA is inspired by thecalling behavior of concave-eared torrent frog In UFCA theprocess of finding an optimal routing path is similar to theprocess of mating with an appropriate frog with character-istics of accurate and energy-efficient Consequently UFCAachieves better routing performance than VBF and ERP2Rregardless of node density and mobility Moreover differentsensor nodes adopt different transmission radius accordingto their residual energy in UFCA and the sensor nodes thatown less energy or locate in worse places choose to entersleep mode for the purpose of saving energy Through thesemeans the energy consumption is somehow equalized onthe whole and the network lifetime is prolonged Thus theinherent adaptive nature of such algorithm is one of the mainattractions in biologically inspired approaches
5 Conclusion
Finding an optimal routing path in adverse underwaterenvironment in 3D UWSNs has always been a challengingtask especially when the factor of energy consumption istaken into consideration Inspired by the calling behaviorof ultrasonic frog in mating this paper proposed an ultra-sonic frog calling algorithm (UFCA) that aims to achieveenergy-efficient routing under harsh underwater conditions
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of Distributed Sensor Networks 11
of UWSNs UFCA does not require fixed routing tables orperiodic flooding messages for the discovery of routing pathInUFCA different sensor nodes adopt different transmissionradius which can be tuned dynamically according to theirresidual energy Moreover the sensor nodes that own lessenergy or locate in worse places choose to enter sleep modefor the purpose of saving energy Simulation results show theperformance improvement inmetrics of packet delivery ratioenergy consumption throughput and end-to-end delay ascompared to existing state-of-the-art routing protocols
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This work was sponsored by the National Nature ScienceFoundation of China (61202370 51279099) the Innova-tion Program of Shanghai Municipal Education Commis-sion (14YZ110) the Shanghai Pujiang Program from Sci-ence and Technology Commission of Shanghai Municipality(11PJ1404300) and the Open Program of Shanghai KeyLaboratory of Intelligent Information Processing (IIPL-2011-008)
References
[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005
[2] C Detweiler M Doniec I Vasilescu and D Rus ldquoAutonomousdepth adjustment for underwater sensor networks design andapplicationsrdquo IEEEASME Transactions onMechatronics vol 17no 1 pp 16ndash24 2012
[3] S Basagni C Petrioli R Petroccia and M Stojanovic ldquoOpti-mized packet size selection in underwater wireless sensor net-work communicationsrdquo IEEE Journal of Oceanic Engineeringvol 37 no 3 pp 321ndash337 2012
[4] J M Jornet M Stojanovic and M Zorzi ldquoOn joint frequencyand power allocation in a cross-layer protocol for underwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 35no 4 pp 936ndash947 2010
[5] G Isbitiren and O B Akan ldquoThree-dimensional underwatertarget tracking with acoustic sensor networksrdquo IEEE Transac-tions onVehicular Technology vol 60 no 8 pp 3897ndash3906 2011
[6] D Pompili T Melodia and I F Akyildiz ldquoThree-dimensionaland two-dimensional deployment analysis for underwateracoustic sensor networksrdquo Ad Hoc Networks vol 7 no 4 pp778ndash790 2009
[7] M Ayaz I Baig A Abdullah and I Faye ldquoA survey on routingtechniques in underwater wireless sensor networksrdquo Journal ofNetwork and Computer Applications vol 34 no 6 pp 1908ndash1927 2011
[8] A S Feng P M Narins C-H Xu et al ldquoUltrasonic communi-cation in frogsrdquo Nature vol 440 no 7082 pp 333ndash336 2006
[9] J-X Shen A S Feng Z-M Xu et al ldquoUltrasonic frogs showhyperacute phonotaxis to female courtship callsrdquo Nature vol453 no 7197 pp 914ndash916 2008
[10] J-X Shen Z-M Xu Z-L Yu S Wang D-Z Zheng and S-C Fan ldquoUltrasonic frogs show extraordinary sex differences inauditory frequency sensitivityrdquo Nature Communications vol 2no 1 article 342 2011
[11] J-H Cui J Kong M Gerla and S Zhou ldquoThe challenges ofbuilding scalable mobile underwater wireless sensor networksfor aquatic applicationsrdquo IEEE Network vol 20 no 3 pp 12ndash182006
[12] I F Akyildiz D Pompili and T Melodia ldquoState-of-the-art inprotocol research for underwater acoustic sensor networksrdquo inProceedings of the 1st ACM International Workshop on Under-water Networks pp 7ndash16 Los Angeles Calif USA September2006
[13] H-P Tan W K G Seah and L Doyle ldquoA multi-hop ARQprotocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007
[14] A Vahdat and D Becker ldquoEpidemic routing for partiallyconnected Ad Hoc networksrdquo Tech Rep TR CS-200006 2000
[15] D Pompili T Melodia and I F Akyildiz ldquoRouting algo-rithms for delay-insensitive and delay-sensitive applicationsin underwater sensor networksrdquo in Proceedings of the 12thAnnual International Conference on Mobile Computing andNetworking (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006
[16] P Xie J-H Cui and L Lao ldquoVBF vector-based forwardingprotocol for underwater sensor networksrdquo in NETWORKING2006 Networking Technologies Services and Protocols Perfor-mance of Computer and Communication Networks Mobile andWireless Communications Systems vol 3976 of Lecture Notes inComputer Science pp 1216ndash1221 2006
[17] J M Jornet M Stojanovic and M Zorzi ldquoFocused beamrouting protocol for underwater acoustic networksrdquo in Pro-ceedings of the 3rd International Workshop on UnderwaterNetworks (WUWNet rsquo08) pp 75ndash82 San Francisco Calif USASeptember 2008
[18] M Zorzi P Casari N Baldo and A F Harris III ldquoEnergy-efficient routing schemes for underwater acoustic networksrdquoIEEE Journal on Selected Areas in Communications vol 26 no9 pp 1754ndash1766 2008
[19] H Yan Z J Shi and J-H Cui ldquoDBR depth-based routingfor underwater sensor networksrdquo in NETWORKING 2008 AdHoc and Sensor Networks Wireless Networks Next GenerationInternet vol 4982 of Lecture Notes in Computer Science pp 72ndash86 Springer 2008
[20] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residualenergyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011
[21] M Ayaz and A Abdullah ldquoHop-by-hop dynamic addressingbased (H2-DAB) routing protocol for underwater wirelesssensor networksrdquo in Proceedings of the International Conferenceon Information and Multimedia Technology (ICIMT rsquo09) pp436ndash441 Jeju Island South Korea December 2009
[22] M Ayaz A Abdullah and I Faye ldquoHop-by-hop reliabledata deliveries for underwater wireless sensor networksrdquo inProceedings of the 5th International Conference on BroadbandWireless Computing Communication andApplications (BWCCArsquo10) pp 363ndash368 November 2010
[23] J Xu K Li and G Min ldquoReliable and energy-efficient mul-tipath communications in underwater sensor networksrdquo IEEETransactions on Parallel and Distributed Systems vol 23 no 7pp 1326ndash1335 2012
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
12 International Journal of Distributed Sensor Networks
[24] Z Zhou Z Peng J-H Cui Z Shi and A Bagtzoglou ldquoScalablelocalization with mobility prediction for underwater sensornetworksrdquo IEEE Transactions on Mobile Computing vol 10 no3 pp 335ndash348 2011
[25] W Cheng A Y Teymorian L Ma X Cheng X Lu andZ Lu ldquoUnderwater localization in sparse 3D acoustic sensornetworksrdquo in Proceedings of the 27th IEEE CommunicationsSociety Conference on Computer Communications (INFOCOMrsquo08) pp 798ndash806 Phoenix Ariz USA April 2008
[26] H-P Tan Z A Eu and W K G Seah ldquoAn enhancedunderwater positioning system to support deepwater installa-tionsrdquo in Proceedings of the MTSIEEE OCEANS 2009 MarineTechnology for Our Future Global and Local Challenges pp 1ndash8Biloxi Miss USA October 2009
[27] S Meguerdichian F Koushanfar G Qu and M PotkonjakldquoExposure in wireless ad-hoc sensor networksrdquo in Proceedingsof the 7th Annual International Conference onMobile Computingand Networking pp 139ndash150 Rome Italy July 2001
[28] E M Sozer M Stojanovic and J G Proakis ldquoUnderwateracoustic networksrdquo IEEE Journal of Oceanic Engineering vol 25no 1 pp 72ndash83 2000
[29] R J Urick Principles of Underwater Sound McGraw-Hill 1983[30] R J Urick Principles of Underwater Sound Peninsula Publish-
ing 3rd edition 1996[31] S Shen A Zhan P Yang and G Chen ldquoExploiting sink
mobility to maximize lifetime in 3D underwater sensor net-worksrdquo in Proceedings of the IEEE International Conference onCommunications (ICC rsquo10) pp 1ndash5 Cape Town South AfricaMay 2010
[32] P Xie Z Zhou Z Peng et al ldquoAqua-sim an NS-2 basedsimulator for underwater sensor networksrdquo inProceedings of theMTSIEEE OCEANS 2009 Marine Technology for Our FutureGlobal and Local Challenges pp 1ndash7 Biloxi Miss USA October2009
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of