Research ArticleBalance Transmission Mechanism in Underwater AcousticSensor Networks

Jiabao Cao Jinfeng Dou and Shunle Dong

College of Information Science and Engineering Ocean University of China Qingdao 266100 China

Correspondence should be addressed to Jinfeng Dou jinfengdououceducn

Received 25 April 2014 Revised 10 September 2014 Accepted 10 September 2014

Academic Editor Lei Shu

Copyright copy 2015 Jiabao Cao et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

With the rapid development of underwater acoustic modem technology underwater acoustic sensor networks (UWASNs) havemore applications in long-term monitoring of the deployment area In the underwater environment the sensors are costly withlimited energy And acoustic communication medium poses new challenges including high path loss low bandwidth and highenergy consumption Therefore designing transmission mechanism to decrease energy consumption and to optimize the lifetimeof UWASN becomes a significant task This paper proposes a balance transmission mechanism and divides the data transmissionprocess into two phases In the routing set-up phase an efficient routing algorithm based on the optimum transmission distance ispresent to optimize the energy consumption of the UWASN And then a data balance transmission algorithm is introduced in thestable data transmission phase The algorithm determines one-hop or multihop data transmission of the node to underwater sinkaccording to the current energy level of adjacent nodes Furthermore detailed theoretical analysis evaluates the optimum energylevels in the UWASNs with different scales The simulation results prove the efficiency of the BTM

1 Introduction

Recent advances in acoustic communication and wirelesssensor networks have motivated the development of under-water acoustic sensor networks (UWASNs) which consist ofunderwater acoustic sensor nodes Compared with the tradi-tional approach to ocean-bottom or ocean-column monitor-ing UWASNs can be used to collect real-time informationfrom a given area and increase the efficiency of manyapplications such as real-time warship monitoring oceano-graphic data collection environmental monitoring off-shore exploration maritime rescue disaster prevention andautonomous underwater vehicles (AUVs) management [1]

Due to the high attenuation of seawater wireless com-munications in underwater environment are usually basedon acoustic links Underwater acoustic communication chan-nels are significantly different from radio channels in theair many energy-efficient schemes developed for terrestrialwireless sensor networks perform poorly in the underwaterenvironment [2] The specific characteristics of acousticchannels include high energy consumption limited availablebandwidth limited battery power low transmission speed

severely attenuated channel long propagation delay and highbit error ratesThe new protocols are expected to be designedin terms of the unique character in underwater acousticchannels [3]

Since it is difficult to accomplish battery replacementand the sensor nodes in underwater scenarios are expensiveit is extremely important to achieve lifetime prolonging ofUWASNs When some sensor nodes run out of their energyand die early the remaining nodes cannot cover the sensingarea efficiently and UWASN cannot receive enough dataso that the network will collapse in the meantime somesensors still keep a lot of energy and the costly underwateracoustic sensor nodes will be wasted Consequently themain problem of lifetime of UWASNs is that the systemwill collapse for a few nodes dying out even if there maystill be significant amount of energy in the other nodesThis problem can be described as the unbalance of energyconsumption among sensors nodes According to the energyconsumption model of underwater acoustic sensor theenergy consumption is related to the transmission distancepath loss data load and so forth [4]The transmissionmodesof the sensor nodes include direct transmission andmultihop

Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2015 Article ID 429340 12 pageshttpdxdoiorg1011552015429340

2 International Journal of Distributed Sensor Networks

transmission in UWASN In the direct transmission modesensor nodes directly transmit the data to the sink by singlehop Thus the farthest sensors away from the sink run out oftheir energy earlier because of longer transmission distanceSensor nodes transmit the data to the sink by multihopforwarding transmission in the multihop transmissionmode which is usually employed in the underwater acousticcommunication because of the short transmission distancewith less energy consumption [2] However in the multihoptransmission the sensor nodes closer to sink tend to runout of their energy earlier because of more data load whilethey pass all the data collected from the sensor nodes to thesink [5] As a result the unbalance of energy consumptionamong sensor nodes is an important problem existing inboth transmission modes that is when some sensor nodesrun out of their energy and die early UWASN will collapsein the meantime the other sensor nodes still keep a lot ofenergy and will be wasted To avoid some sensor nodesdying earlier and the huge energy being wasted because ofthe sensors unbalancing employment in the UWASN theswitching of communication modes is necessary betweendirect transmission and multihop communications

This paper elaborates balance transmission mechanism(BTM) Necessary acoustic direct transmission is adoptedwhen excessive local energy consumption occurs inmultihoptransmission BTM decides the data transmission mode ofone-hop or multihop to the sink based on the current energylevel (EL) of underwater sensor nodes (uw-sensor nodes) Atfirst BTM needs to build the fundamental route for packettransmission This study gives the efficient route algorithmin the route set-up phase In the previous research theoptimum transmission distance was obtained by minimizingthe energy consumption in the whole process of each uw-sensor node sending packets to the uw-sink in the multihopcommunication This paper selects the relay node where thepoint-point distance between the sending node and the relaynode is closest to the optimum transmission distance andestablishes the trees-route in the efficient route algorithmAll nodes transmit the packets by multihop forwardingnext hop towards the underwater sink (uw-sink) along thetree-route When the load in a node is so heavy that theEL of the node decreases to certain degree it informs itspredecessor nodes obtained in the routing tree to transmitthe packet directly to the uw-sink And then the data willbe transmitted by multihop forwarding transmission againwhen this predecessor nodes EL decreases to some degreeDuring data transmission a node can decide whether toforward data to next hop towards the uw-sink or to send datadirectly to the uw-sink by comparing its EL to its adjacentneighbors Furthermore through theoretical analysis we getthe optimum number of ELs classification in both the smallscale UWASN and the large scale one and use it in the BTMThe UWASNs lifetime can be further prolonged

The contributions of this paper are as follows (1) Thedata balance transmission algorithm with EL is proposedwhich prolongs the lifetime of the whole UWASN by solvingthe unbalance of energy consumption among sensors nodes(2) The optimum energy levels (OELs) are evaluated from

theoretical analysis in both the small scale UWASN and thelarge scale one

The subsequence of this paper is organized as followingRelated works are discussed in Section 2 Section 3 illustratesthe network model and energy consumption model Thedesign of BTM is described in detail in Section 4 Simulationsare presented in Section 5 Conclusions are given in Section 6

2 Related Work

Because underwater acoustic communication channels aresignificantly different from radio channels in the air manyenergy-efficient transmission schemes developed for terres-trial wireless sensor networks perform poorly in the under-water environment In the underwater environment somefactors increase the designing cost of the sensors such aswaterproof corrosion protection and more difficult charg-ing And underwater acoustic wave communication posesnew challenges including high path loss low bandwidth andhigh energy consumption Consequently the sensors and theenergy become more significant resources in the underwaterenvironment and it also introduces difficulties in designingtransmission mechanism to use the sensors energy efficientlyand prolong the network lifetime [6]

Tan et al [7] proposed a protocol based on hop-by-hop hybrid acknowledgment scheme which was for a mul-tihop UWASN In the protocol data packets forwarded bydownstream nodes could work as implicit ACKs for previoustransmitted data packets Ayaz and Abdullah [8] proposed adynamic addressing based hop-by-hop routing protocol toprovide scalable and time-efficient routing for UWSN Therouting protocol did not require any dimensional locationinformation or any extra specialized hardware comparedwith many other routing protocols in the same area Vector-based forwarding [9] was a geographic approach whichallowed the nodes to weigh the benefit to forward packetsand reduce energy consumption by discarding low benefitpackets Therefore over a multihop path only the nodesthat were located within a pipe of given width between thesource and the destination were considered for relayingIn the areas with low density of nodes this approach maynot find the path close to the routing vector SimilarlyJornet et al proposed focused-beam routing [10] protocolthat was suitable for networks containing both static andmobile nodes The objective was to determine which nodesare candidates for relaying Candidate nodes were those thatlie within a cone of angle emanating from the transmittertowards the final destination Those nodes outside the conewould not respondMTE [11] was a classical representation ofenergy-efficient multihop transmission in which each sensornode always forwarded the data packets to its neighboringnode towards the sink until the data packets reached thedestination The problem of multihop routing still exists as itis based onmultihop architecture where nodes near the sinksdrain more energy because they are used more frequently

Pompili et al [12] introduced two distributed routingalgorithms for delay-insensitive and delay-sensitive appli-cations respectively they aimed at minimizing the energy

International Journal of Distributed Sensor Networks 3

consumption taking the varying condition of the under-water channel and the different application requirementsinto account A theoretical argument supporting geographicrouting was discussed [13] based on simple propagation andenergy consumption models for underwater networks Thestudy showed that an optimal number of hops along a pathexist and showed that increasing the number of hops bychoosing closer relays is preferred with respect to keepingthe route shorter Ponnavaikkoy et al [14] studied the energyoptimization problem and delay constraints The authors[15] developed a relay selection criterion called cooperativebest relay assessment for underwater cooperative acousticnetworks tominimize the one-way packet transmission timeThe criterion took into account both the spectral efficiencyand the underwater long propagation delay to improve theoverall throughput performance of the network with energyconstraint A novel layered multipath power control scheme[16] took noise attenuation in deep water areas into accountand proved that its optimization problem was NP completeThe authors solved the key problems including establish-ment of the energy-efficient tree and management of energydistribution and further developed a heuristic algorithm toachieve the feasible solution of the optimization problemAn ultrasonic frog calling algorithm [17] was present aimingto achieve energy-efficient routing under harsh underwaterconditions of UWASNs The process of selecting relay nodesto forward the data packet was similar to that of callingbehavior of ultrasonic frog formating Different sensor nodesadopted different transmission radius and the values can betuned dynamically according to their residual energy Thesensor nodes that owned less energy or were located inworse places chose sleep mode for the purpose of savingenergyWahid et al [18] proposed an energy-efficient routingprotocol based on physical distance and residual energy Theprotocol also took into account the residual energy of thesensor nodes in order to extend the network lifetime It mightoften make the problem worse in terms of the characteristicsof node mobility in UWASNs

This paper analyzes the characteristic of energy consump-tion in data transmission and receiving The optimal trans-mission range of acoustic communication can be decidedon such analysis An efficient route algorithm is proposedconsidering the factor of the optimal transmission range todecrease the energy consumption And then balance trans-mission method based on the EL is present The one-hoptransmission and multihop transmission are decided bythe current EL of adjacent nodes The OELs are evaluatedthrough theoretical analysis in terms of different scale net-works Through this mechanism the lifetime of the wholeUWASN can be prolonged

3 Network Model Assumptions and EnergyConsumption Model

31 Network Model and Assumptions The architecture ofUWASNs consists of sensor nodes AUVs and uw-sinkswhich are connected by acoustic communications Severalcommunication architectures of UWASNs are introduced

Surface station

Uw-sink

Uw-sensorVertical linkHorizontalmultihop link

Figure 1 Network model

including two-dimensional and three-dimensional underwa-ter networks [6] AUVs are usually employed to gather datadirectly from the underwater sensor by direct transmissionwhich can use optical or acoustic communication [19 20]Nevertheless AUVs are more costly The reference architec-ture for two-dimensional underwater networks is shown inFigure 1 A group of sensor nodes are anchored to the bottomof the ocean with deep ocean anchors Uw-sensor nodes areinterconnected to one or more uw-sinks by means of wirelessacoustic links Sensors can be connected to uw-sinks viadirect transmission links or through multihop transmissionpaths Uw-sinks as shown in Figure 1 are network devicesin charge of relaying data from the ocean-bottom network toa surface station The typical application of UWASNs is themonitoring of a remote environmentThis paper only studiesthe local area with uw-sensors and uw-sink The uw-sink isresponsible for collecting data from sensor nodes Once theuw-sink has all the data from the nodes it transmits the datato the surface station by acoustic communication

For the simplification we consider a sensor networkconsisting of119872 sensor nodes uniformly deployed over a vastfield to continuously monitor the environment and an uw-sink away from the nodes with a constant power supplythrough which the end user can access data from theUWASN Some reasonable assumptions are made about thesensor nodes

(1) Uw-sink is stationary after deployment and sensornodes move slowly and rarely move

(2) All sensor nodes are homogeneous and have the samecapabilities

(3) All sensor nodes can adjust the amount of transmitpower with the transmission distance Energy sensornode has its maximum transmission power that isto say it has its maximum transmission range Thesink is within the maximum transmission range of allsensor nodes that is all sensor nodes can transmitdata with enough transmission power to reach theuw-sink Each node has the computational power to

4 International Journal of Distributed Sensor Networks

Table 1 An example of the neighbor node table

Current sensor ID Energy level Successor ID Predecessor ID Number of predecessors11 27318 0 14 15 21 314 27318 11 31 34 215 27318 11 36 41 221 27318 11 42 1

support different MAC protocols and perform signalprocessing functions

(4) Each sensor node is location aware by some exist-ing positioning methods The positioning methodsbelong to another research area which are notinvolved in this study The uw-sink has strong com-putation power

(5) Each sensor node senses and generates data evenly ineach round

(6) To avoid interference CDMA technology can be usedto achieve multiple simultaneous wireless transmis-sions [21]

These assumptions are reasonable due to technologicaladvances in acoustic communication hardware and low-power computing

32 Energy ConsumptionModel To quantify the energy con-sumption of the UWASNs we adopt the energy dissipationmodel based on underwater acoustic communication princi-ple To transmit a data packet from one node to another overa transmission distance 119909 each node needs to transmit at apower level 119875

1= 1198750119860(119909) where 119875

0is a power level needed

at the input to the receiver and 119860(119909) is the attenuation Theattenuation [4] is given as

119860 (119909) = 119909119896119886119909 (1)

where 119896 is the energy spreading factor (1 for cylindrical 15for practical and 2 for spherical spreading) and

119886 = 10120572(119891)10 (2)

is a frequency-dependent term obtained from the absorptioncoefficientThe absorption coefficient for the frequency rangeof interest is calculated according toThorprsquos expression [4] as

120572 (119891) = 0111198912

1 + 1198912+ 44

1198912

4100 + 1198912

+ 275 times 10minus41198912+ 0003

(3)

in dBkm for 119891 in kHzTo transmit a 119897-bit packet over a transmission distance 119909

the transmitter expends

119864119879(119897 119909) = 119897119875

0119909119896119886119909 (4)

and to receive this packet the receiver expends119864119877(119897) = 119897119875

119903 (5)

where 119875119903is a constant parameter which depends on the

receiver devices

4 Balance Transmission Mechanism

The BTM includes the route set-up phase and the stable datatransmission phase A route set-up phase is executed whenthe UWASN begins to work and then goes to a stable datatransmission phasewhen data are transferred from the sensornodes to the uw-sink BTM presents an energy-efficient rout-ing algorithm based on the optimum transmission distance[22] in the route set-up phase and a data transmission algo-rithm based on energy level in the stable data transmissionphase to balance network energy consumption

41 Efficient Routing Algorithm Based on Optimum Transmis-sion Range (ERA) Obviously if the energy consumption inthe whole process of each uw-sensor node sending packetsto the uw-sink is minimized the energy consumption of thenetwork will be minimized The study minimized the totalenergy consumed in the network and obtained the optimumtransmission distance [22] in the multihop communicationIn order to optimize the energy consumption this paper pro-poses the ERA to establish the route in the route set-up phase

Definition 1 Let 119903opt be the optimum transmission distancewhich can be obtained in the previous research [22]

The goal of ERA is to design an algorithm that selectsensor nodes as more as possible near the position whichis 119903opt far from the sending node as relays Assume that thecoordinate positions of sensor nodes and the uw-sink in theUWASN are known

At the route set-up phase the detailed steps of the ERAalgorithm are as follows

(1) The sensor node 119878119894 in the network sends a query

packet including its location information and thelocation information of the uw-sink All the nodesreceiving the query packet 119878

119895| 119895 = 119894 compute

119876119895= 120579|119889(119878

119895 119878119894)minus119903opt|+(1minus120579)119889(119878119895 119880119878) where119889(119878119895 119878119894)

is the distance between the node 119878119895and the sending

node 119878119894and119889(119878

119895 119880119878) is the distance between the node

119878119895and the uw-sink 120579 is a system parameter defined in

this paper here let 120579 be 05 initially By adjusting thevalue of 120579 only one sensor node toward the sink andclosest to the optimum transmission distance will beselected as the relay node of the sensor node 119878

119894

(2) Select the nodewith theminimal119876119895as the relay node

Then record the position of the relay into the neighbornode table (eg Table 1) of the sending node 119878

119894as a

successor and record the position of the sending node119878119894into the neighbor node table of the relay node as

International Journal of Distributed Sensor Networks 5

0

11

34

42 36

4115

2114

31

Uw-sinkUw-sensormiddot middot middot

middot middot middot

Figure 2 Trees-route at the route set-up phase

a predecessor If the minimal 119876119895values of several

sensor nodes are the same let 120579 get larger Repeat (1)If the minimal 119876

119895values of several sensor nodes are

still the same look up their neighbor table and selectthe node with the least predecessors as the relay nodeNote the successor must be only one node

(3) Repeat (1) and (2) until the uw-sink is within the opti-mum transmission distance of the sensor node send-ing a query packet When the uw-sink is within theoptimum transmission distance of the sensor nodethe sensor node can transmit the data to the sinkwithout relaying

(4) Obviously if all the nodes repeat (1) (2) and (3) atrees-route will be set up

Figure 2 shows the route result containing some trees atthe route set-up phase

Note once the route is set up the data can be transmittedin the stable data transmission phase Because the uw-sensornodes are mobile in the real case the route set-up phase willbe operated again when the uw-sensor nodes move far awayfrom their origin position so as not to find relay Thereforeour algorithm is suitable for the case that the move of theuw-sensor nodes is not frequent Otherwise the route set-up phase must be executed constantly which induces moreenergy consumption

42 Data Balance Transmission Algorithm (DBT) By a routeset-up phase a set of trees-route is obtained Because theforwarding data is not aggregated in this paper thereforethe nodes close to the uw-sink die out much faster thanthose far away from the uw-sink In order to balance theenergy consumption we introduce an EL that is we dividethe nodersquos initial energy into 119898 equal parts and the ELof a node is 119894 (1 le 119894 le 119898) if its current energy is

more than (119894 minus 1) parts and less than or equal to 119894 partsDuring stable data transmission phase each uw-sensor hastwo transmission modes forwarding the packets towards theuw-sink by multihop transmission (MT) or transmitting thepackets directly to the uw-sink (DT) A node can decidewhether to forward data towards the uw-sink via MT modeor to send data directly to the uw-sink via DT by comparingits EL with that of its successor node The DBT algorithm isdescribed in detail as follows

Initially every node forwards the packet to its successortowards the uw-sink by MT mode and their ELs is m Somerounds later the node which is closest to the uw-sink willdissipate more energy because of more traffic load so itsenergy drops into the next energy level that is119898minus1 Then itbroadcasts an EL notice packet (ELNP) to their predecessornodes the predecessor nodes will transmit data to the uw-sink by DT mode since their ELs are larger than that of thesuccessor node An ELNP can be set as 1 bit for saving energySeveral rounds again these predecessor nodes will dissipatemore energy because of DTmodeTherefore when the EL ofa predecessor node decreases the node broadcasts the ELNPto its predecessor nodes similarly and if the EL of the nodeis equal to or less than its successor node it will transmitdata by MT mode again Thus all uw-sensors transmitdata ultimately to the uw-sink by similar operation to theabove process To avoid interference CDMA technologycan be used to achieve multiple simultaneous transmissionsBy alternately changing uw-sensorsrsquo transmission mode thenodes can consume their energy evenly accordingly thenetwork lifetime is prolonged

Obviously the magnitude of the EL affects the energydissipated in the network For all sensor nodes if their initialnumbers of the EL are 1 their transmission mode becomesmultihop transmission If the initial EL of each uw-sensoris very large the transmission mode is similar to the directtransmission Therefore an optimal EL will be found in thefollowing section

43 Analysis of Optimum Energy Level (OEL) When somesensor nodes run out of their energy the network mightcollapse because it is not able to collect and to transmitdata efficiently while other sensor nodes still have muchremaining energy which causes energy waste In order tobalance energy consumption in some cases it is an ideal casethat all the nodes in the network run out of their energy at thesame timeTherefore this section analyzes the wasted energyminimizes it and studies theOELWe can virtually divide thenetwork disk section into 119899 ring sections or slices the radiusof slice is the optimum transmission distance

Definition 2 Starting from the center of the disk where theuw-sink is located we define the set of slices We denote thefirst slice containing the uw-sink by Slice 1 which has a radius119903 and use Slice 119894 (2 le 119894 le 119873) to represent the slice next to Slice119894 minus 1 from the opposite side of the uw-sink which is definedby two successive circles one of radius 119894 times 119903 and the other ofradius (119894minus1)times119903 119903 is the optimum transmission distance 119903opt

The area of Slice 119894 is 119860119894= int119894119903

(119894minus1)119903int2120587

0119903119889119903119889120579 = (2119894 minus 1)120587119903

2

6 International Journal of Distributed Sensor Networks

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

transmission in UWASN In the direct transmission modesensor nodes directly transmit the data to the sink by singlehop Thus the farthest sensors away from the sink run out oftheir energy earlier because of longer transmission distanceSensor nodes transmit the data to the sink by multihopforwarding transmission in the multihop transmissionmode which is usually employed in the underwater acousticcommunication because of the short transmission distancewith less energy consumption [2] However in the multihoptransmission the sensor nodes closer to sink tend to runout of their energy earlier because of more data load whilethey pass all the data collected from the sensor nodes to thesink [5] As a result the unbalance of energy consumptionamong sensor nodes is an important problem existing inboth transmission modes that is when some sensor nodesrun out of their energy and die early UWASN will collapsein the meantime the other sensor nodes still keep a lot ofenergy and will be wasted To avoid some sensor nodesdying earlier and the huge energy being wasted because ofthe sensors unbalancing employment in the UWASN theswitching of communication modes is necessary betweendirect transmission and multihop communications

This paper elaborates balance transmission mechanism(BTM) Necessary acoustic direct transmission is adoptedwhen excessive local energy consumption occurs inmultihoptransmission BTM decides the data transmission mode ofone-hop or multihop to the sink based on the current energylevel (EL) of underwater sensor nodes (uw-sensor nodes) Atfirst BTM needs to build the fundamental route for packettransmission This study gives the efficient route algorithmin the route set-up phase In the previous research theoptimum transmission distance was obtained by minimizingthe energy consumption in the whole process of each uw-sensor node sending packets to the uw-sink in the multihopcommunication This paper selects the relay node where thepoint-point distance between the sending node and the relaynode is closest to the optimum transmission distance andestablishes the trees-route in the efficient route algorithmAll nodes transmit the packets by multihop forwardingnext hop towards the underwater sink (uw-sink) along thetree-route When the load in a node is so heavy that theEL of the node decreases to certain degree it informs itspredecessor nodes obtained in the routing tree to transmitthe packet directly to the uw-sink And then the data willbe transmitted by multihop forwarding transmission againwhen this predecessor nodes EL decreases to some degreeDuring data transmission a node can decide whether toforward data to next hop towards the uw-sink or to send datadirectly to the uw-sink by comparing its EL to its adjacentneighbors Furthermore through theoretical analysis we getthe optimum number of ELs classification in both the smallscale UWASN and the large scale one and use it in the BTMThe UWASNs lifetime can be further prolonged

The contributions of this paper are as follows (1) Thedata balance transmission algorithm with EL is proposedwhich prolongs the lifetime of the whole UWASN by solvingthe unbalance of energy consumption among sensors nodes(2) The optimum energy levels (OELs) are evaluated from

theoretical analysis in both the small scale UWASN and thelarge scale one

The subsequence of this paper is organized as followingRelated works are discussed in Section 2 Section 3 illustratesthe network model and energy consumption model Thedesign of BTM is described in detail in Section 4 Simulationsare presented in Section 5 Conclusions are given in Section 6

2 Related Work

Because underwater acoustic communication channels aresignificantly different from radio channels in the air manyenergy-efficient transmission schemes developed for terres-trial wireless sensor networks perform poorly in the under-water environment In the underwater environment somefactors increase the designing cost of the sensors such aswaterproof corrosion protection and more difficult charg-ing And underwater acoustic wave communication posesnew challenges including high path loss low bandwidth andhigh energy consumption Consequently the sensors and theenergy become more significant resources in the underwaterenvironment and it also introduces difficulties in designingtransmission mechanism to use the sensors energy efficientlyand prolong the network lifetime [6]

Tan et al [7] proposed a protocol based on hop-by-hop hybrid acknowledgment scheme which was for a mul-tihop UWASN In the protocol data packets forwarded bydownstream nodes could work as implicit ACKs for previoustransmitted data packets Ayaz and Abdullah [8] proposed adynamic addressing based hop-by-hop routing protocol toprovide scalable and time-efficient routing for UWSN Therouting protocol did not require any dimensional locationinformation or any extra specialized hardware comparedwith many other routing protocols in the same area Vector-based forwarding [9] was a geographic approach whichallowed the nodes to weigh the benefit to forward packetsand reduce energy consumption by discarding low benefitpackets Therefore over a multihop path only the nodesthat were located within a pipe of given width between thesource and the destination were considered for relayingIn the areas with low density of nodes this approach maynot find the path close to the routing vector SimilarlyJornet et al proposed focused-beam routing [10] protocolthat was suitable for networks containing both static andmobile nodes The objective was to determine which nodesare candidates for relaying Candidate nodes were those thatlie within a cone of angle emanating from the transmittertowards the final destination Those nodes outside the conewould not respondMTE [11] was a classical representation ofenergy-efficient multihop transmission in which each sensornode always forwarded the data packets to its neighboringnode towards the sink until the data packets reached thedestination The problem of multihop routing still exists as itis based onmultihop architecture where nodes near the sinksdrain more energy because they are used more frequently

Pompili et al [12] introduced two distributed routingalgorithms for delay-insensitive and delay-sensitive appli-cations respectively they aimed at minimizing the energy

International Journal of Distributed Sensor Networks 3

consumption taking the varying condition of the under-water channel and the different application requirementsinto account A theoretical argument supporting geographicrouting was discussed [13] based on simple propagation andenergy consumption models for underwater networks Thestudy showed that an optimal number of hops along a pathexist and showed that increasing the number of hops bychoosing closer relays is preferred with respect to keepingthe route shorter Ponnavaikkoy et al [14] studied the energyoptimization problem and delay constraints The authors[15] developed a relay selection criterion called cooperativebest relay assessment for underwater cooperative acousticnetworks tominimize the one-way packet transmission timeThe criterion took into account both the spectral efficiencyand the underwater long propagation delay to improve theoverall throughput performance of the network with energyconstraint A novel layered multipath power control scheme[16] took noise attenuation in deep water areas into accountand proved that its optimization problem was NP completeThe authors solved the key problems including establish-ment of the energy-efficient tree and management of energydistribution and further developed a heuristic algorithm toachieve the feasible solution of the optimization problemAn ultrasonic frog calling algorithm [17] was present aimingto achieve energy-efficient routing under harsh underwaterconditions of UWASNs The process of selecting relay nodesto forward the data packet was similar to that of callingbehavior of ultrasonic frog formating Different sensor nodesadopted different transmission radius and the values can betuned dynamically according to their residual energy Thesensor nodes that owned less energy or were located inworse places chose sleep mode for the purpose of savingenergyWahid et al [18] proposed an energy-efficient routingprotocol based on physical distance and residual energy Theprotocol also took into account the residual energy of thesensor nodes in order to extend the network lifetime It mightoften make the problem worse in terms of the characteristicsof node mobility in UWASNs

This paper analyzes the characteristic of energy consump-tion in data transmission and receiving The optimal trans-mission range of acoustic communication can be decidedon such analysis An efficient route algorithm is proposedconsidering the factor of the optimal transmission range todecrease the energy consumption And then balance trans-mission method based on the EL is present The one-hoptransmission and multihop transmission are decided bythe current EL of adjacent nodes The OELs are evaluatedthrough theoretical analysis in terms of different scale net-works Through this mechanism the lifetime of the wholeUWASN can be prolonged

3 Network Model Assumptions and EnergyConsumption Model

31 Network Model and Assumptions The architecture ofUWASNs consists of sensor nodes AUVs and uw-sinkswhich are connected by acoustic communications Severalcommunication architectures of UWASNs are introduced

Surface station

Uw-sink

Uw-sensorVertical linkHorizontalmultihop link

Figure 1 Network model

including two-dimensional and three-dimensional underwa-ter networks [6] AUVs are usually employed to gather datadirectly from the underwater sensor by direct transmissionwhich can use optical or acoustic communication [19 20]Nevertheless AUVs are more costly The reference architec-ture for two-dimensional underwater networks is shown inFigure 1 A group of sensor nodes are anchored to the bottomof the ocean with deep ocean anchors Uw-sensor nodes areinterconnected to one or more uw-sinks by means of wirelessacoustic links Sensors can be connected to uw-sinks viadirect transmission links or through multihop transmissionpaths Uw-sinks as shown in Figure 1 are network devicesin charge of relaying data from the ocean-bottom network toa surface station The typical application of UWASNs is themonitoring of a remote environmentThis paper only studiesthe local area with uw-sensors and uw-sink The uw-sink isresponsible for collecting data from sensor nodes Once theuw-sink has all the data from the nodes it transmits the datato the surface station by acoustic communication

For the simplification we consider a sensor networkconsisting of119872 sensor nodes uniformly deployed over a vastfield to continuously monitor the environment and an uw-sink away from the nodes with a constant power supplythrough which the end user can access data from theUWASN Some reasonable assumptions are made about thesensor nodes

(1) Uw-sink is stationary after deployment and sensornodes move slowly and rarely move

(2) All sensor nodes are homogeneous and have the samecapabilities

(3) All sensor nodes can adjust the amount of transmitpower with the transmission distance Energy sensornode has its maximum transmission power that isto say it has its maximum transmission range Thesink is within the maximum transmission range of allsensor nodes that is all sensor nodes can transmitdata with enough transmission power to reach theuw-sink Each node has the computational power to

4 International Journal of Distributed Sensor Networks

Table 1 An example of the neighbor node table

Current sensor ID Energy level Successor ID Predecessor ID Number of predecessors11 27318 0 14 15 21 314 27318 11 31 34 215 27318 11 36 41 221 27318 11 42 1

support different MAC protocols and perform signalprocessing functions

(4) Each sensor node is location aware by some exist-ing positioning methods The positioning methodsbelong to another research area which are notinvolved in this study The uw-sink has strong com-putation power

(5) Each sensor node senses and generates data evenly ineach round

(6) To avoid interference CDMA technology can be usedto achieve multiple simultaneous wireless transmis-sions [21]

These assumptions are reasonable due to technologicaladvances in acoustic communication hardware and low-power computing

32 Energy ConsumptionModel To quantify the energy con-sumption of the UWASNs we adopt the energy dissipationmodel based on underwater acoustic communication princi-ple To transmit a data packet from one node to another overa transmission distance 119909 each node needs to transmit at apower level 119875

1= 1198750119860(119909) where 119875

0is a power level needed

at the input to the receiver and 119860(119909) is the attenuation Theattenuation [4] is given as

119860 (119909) = 119909119896119886119909 (1)

where 119896 is the energy spreading factor (1 for cylindrical 15for practical and 2 for spherical spreading) and

119886 = 10120572(119891)10 (2)

is a frequency-dependent term obtained from the absorptioncoefficientThe absorption coefficient for the frequency rangeof interest is calculated according toThorprsquos expression [4] as

120572 (119891) = 0111198912

1 + 1198912+ 44

1198912

4100 + 1198912

+ 275 times 10minus41198912+ 0003

(3)

in dBkm for 119891 in kHzTo transmit a 119897-bit packet over a transmission distance 119909

the transmitter expends

119864119879(119897 119909) = 119897119875

0119909119896119886119909 (4)

and to receive this packet the receiver expends119864119877(119897) = 119897119875

119903 (5)

where 119875119903is a constant parameter which depends on the

receiver devices

4 Balance Transmission Mechanism

The BTM includes the route set-up phase and the stable datatransmission phase A route set-up phase is executed whenthe UWASN begins to work and then goes to a stable datatransmission phasewhen data are transferred from the sensornodes to the uw-sink BTM presents an energy-efficient rout-ing algorithm based on the optimum transmission distance[22] in the route set-up phase and a data transmission algo-rithm based on energy level in the stable data transmissionphase to balance network energy consumption

41 Efficient Routing Algorithm Based on Optimum Transmis-sion Range (ERA) Obviously if the energy consumption inthe whole process of each uw-sensor node sending packetsto the uw-sink is minimized the energy consumption of thenetwork will be minimized The study minimized the totalenergy consumed in the network and obtained the optimumtransmission distance [22] in the multihop communicationIn order to optimize the energy consumption this paper pro-poses the ERA to establish the route in the route set-up phase

Definition 1 Let 119903opt be the optimum transmission distancewhich can be obtained in the previous research [22]

The goal of ERA is to design an algorithm that selectsensor nodes as more as possible near the position whichis 119903opt far from the sending node as relays Assume that thecoordinate positions of sensor nodes and the uw-sink in theUWASN are known

At the route set-up phase the detailed steps of the ERAalgorithm are as follows

(1) The sensor node 119878119894 in the network sends a query

packet including its location information and thelocation information of the uw-sink All the nodesreceiving the query packet 119878

119895| 119895 = 119894 compute

119876119895= 120579|119889(119878

119895 119878119894)minus119903opt|+(1minus120579)119889(119878119895 119880119878) where119889(119878119895 119878119894)

is the distance between the node 119878119895and the sending

node 119878119894and119889(119878

119895 119880119878) is the distance between the node

119878119895and the uw-sink 120579 is a system parameter defined in

this paper here let 120579 be 05 initially By adjusting thevalue of 120579 only one sensor node toward the sink andclosest to the optimum transmission distance will beselected as the relay node of the sensor node 119878

119894

(2) Select the nodewith theminimal119876119895as the relay node

Then record the position of the relay into the neighbornode table (eg Table 1) of the sending node 119878

119894as a

successor and record the position of the sending node119878119894into the neighbor node table of the relay node as

International Journal of Distributed Sensor Networks 5

0

11

34

42 36

4115

2114

31

Uw-sinkUw-sensormiddot middot middot

middot middot middot

Figure 2 Trees-route at the route set-up phase

a predecessor If the minimal 119876119895values of several

sensor nodes are the same let 120579 get larger Repeat (1)If the minimal 119876

119895values of several sensor nodes are

still the same look up their neighbor table and selectthe node with the least predecessors as the relay nodeNote the successor must be only one node

(3) Repeat (1) and (2) until the uw-sink is within the opti-mum transmission distance of the sensor node send-ing a query packet When the uw-sink is within theoptimum transmission distance of the sensor nodethe sensor node can transmit the data to the sinkwithout relaying

(4) Obviously if all the nodes repeat (1) (2) and (3) atrees-route will be set up

Figure 2 shows the route result containing some trees atthe route set-up phase

Note once the route is set up the data can be transmittedin the stable data transmission phase Because the uw-sensornodes are mobile in the real case the route set-up phase willbe operated again when the uw-sensor nodes move far awayfrom their origin position so as not to find relay Thereforeour algorithm is suitable for the case that the move of theuw-sensor nodes is not frequent Otherwise the route set-up phase must be executed constantly which induces moreenergy consumption

42 Data Balance Transmission Algorithm (DBT) By a routeset-up phase a set of trees-route is obtained Because theforwarding data is not aggregated in this paper thereforethe nodes close to the uw-sink die out much faster thanthose far away from the uw-sink In order to balance theenergy consumption we introduce an EL that is we dividethe nodersquos initial energy into 119898 equal parts and the ELof a node is 119894 (1 le 119894 le 119898) if its current energy is

more than (119894 minus 1) parts and less than or equal to 119894 partsDuring stable data transmission phase each uw-sensor hastwo transmission modes forwarding the packets towards theuw-sink by multihop transmission (MT) or transmitting thepackets directly to the uw-sink (DT) A node can decidewhether to forward data towards the uw-sink via MT modeor to send data directly to the uw-sink via DT by comparingits EL with that of its successor node The DBT algorithm isdescribed in detail as follows

Initially every node forwards the packet to its successortowards the uw-sink by MT mode and their ELs is m Somerounds later the node which is closest to the uw-sink willdissipate more energy because of more traffic load so itsenergy drops into the next energy level that is119898minus1 Then itbroadcasts an EL notice packet (ELNP) to their predecessornodes the predecessor nodes will transmit data to the uw-sink by DT mode since their ELs are larger than that of thesuccessor node An ELNP can be set as 1 bit for saving energySeveral rounds again these predecessor nodes will dissipatemore energy because of DTmodeTherefore when the EL ofa predecessor node decreases the node broadcasts the ELNPto its predecessor nodes similarly and if the EL of the nodeis equal to or less than its successor node it will transmitdata by MT mode again Thus all uw-sensors transmitdata ultimately to the uw-sink by similar operation to theabove process To avoid interference CDMA technologycan be used to achieve multiple simultaneous transmissionsBy alternately changing uw-sensorsrsquo transmission mode thenodes can consume their energy evenly accordingly thenetwork lifetime is prolonged

Obviously the magnitude of the EL affects the energydissipated in the network For all sensor nodes if their initialnumbers of the EL are 1 their transmission mode becomesmultihop transmission If the initial EL of each uw-sensoris very large the transmission mode is similar to the directtransmission Therefore an optimal EL will be found in thefollowing section

43 Analysis of Optimum Energy Level (OEL) When somesensor nodes run out of their energy the network mightcollapse because it is not able to collect and to transmitdata efficiently while other sensor nodes still have muchremaining energy which causes energy waste In order tobalance energy consumption in some cases it is an ideal casethat all the nodes in the network run out of their energy at thesame timeTherefore this section analyzes the wasted energyminimizes it and studies theOELWe can virtually divide thenetwork disk section into 119899 ring sections or slices the radiusof slice is the optimum transmission distance

Definition 2 Starting from the center of the disk where theuw-sink is located we define the set of slices We denote thefirst slice containing the uw-sink by Slice 1 which has a radius119903 and use Slice 119894 (2 le 119894 le 119873) to represent the slice next to Slice119894 minus 1 from the opposite side of the uw-sink which is definedby two successive circles one of radius 119894 times 119903 and the other ofradius (119894minus1)times119903 119903 is the optimum transmission distance 119903opt

The area of Slice 119894 is 119860119894= int119894119903

(119894minus1)119903int2120587

0119903119889119903119889120579 = (2119894 minus 1)120587119903

2

6 International Journal of Distributed Sensor Networks

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

consumption taking the varying condition of the under-water channel and the different application requirementsinto account A theoretical argument supporting geographicrouting was discussed [13] based on simple propagation andenergy consumption models for underwater networks Thestudy showed that an optimal number of hops along a pathexist and showed that increasing the number of hops bychoosing closer relays is preferred with respect to keepingthe route shorter Ponnavaikkoy et al [14] studied the energyoptimization problem and delay constraints The authors[15] developed a relay selection criterion called cooperativebest relay assessment for underwater cooperative acousticnetworks tominimize the one-way packet transmission timeThe criterion took into account both the spectral efficiencyand the underwater long propagation delay to improve theoverall throughput performance of the network with energyconstraint A novel layered multipath power control scheme[16] took noise attenuation in deep water areas into accountand proved that its optimization problem was NP completeThe authors solved the key problems including establish-ment of the energy-efficient tree and management of energydistribution and further developed a heuristic algorithm toachieve the feasible solution of the optimization problemAn ultrasonic frog calling algorithm [17] was present aimingto achieve energy-efficient routing under harsh underwaterconditions of UWASNs The process of selecting relay nodesto forward the data packet was similar to that of callingbehavior of ultrasonic frog formating Different sensor nodesadopted different transmission radius and the values can betuned dynamically according to their residual energy Thesensor nodes that owned less energy or were located inworse places chose sleep mode for the purpose of savingenergyWahid et al [18] proposed an energy-efficient routingprotocol based on physical distance and residual energy Theprotocol also took into account the residual energy of thesensor nodes in order to extend the network lifetime It mightoften make the problem worse in terms of the characteristicsof node mobility in UWASNs

This paper analyzes the characteristic of energy consump-tion in data transmission and receiving The optimal trans-mission range of acoustic communication can be decidedon such analysis An efficient route algorithm is proposedconsidering the factor of the optimal transmission range todecrease the energy consumption And then balance trans-mission method based on the EL is present The one-hoptransmission and multihop transmission are decided bythe current EL of adjacent nodes The OELs are evaluatedthrough theoretical analysis in terms of different scale net-works Through this mechanism the lifetime of the wholeUWASN can be prolonged

3 Network Model Assumptions and EnergyConsumption Model

31 Network Model and Assumptions The architecture ofUWASNs consists of sensor nodes AUVs and uw-sinkswhich are connected by acoustic communications Severalcommunication architectures of UWASNs are introduced

Surface station

Uw-sink

Uw-sensorVertical linkHorizontalmultihop link

Figure 1 Network model

including two-dimensional and three-dimensional underwa-ter networks [6] AUVs are usually employed to gather datadirectly from the underwater sensor by direct transmissionwhich can use optical or acoustic communication [19 20]Nevertheless AUVs are more costly The reference architec-ture for two-dimensional underwater networks is shown inFigure 1 A group of sensor nodes are anchored to the bottomof the ocean with deep ocean anchors Uw-sensor nodes areinterconnected to one or more uw-sinks by means of wirelessacoustic links Sensors can be connected to uw-sinks viadirect transmission links or through multihop transmissionpaths Uw-sinks as shown in Figure 1 are network devicesin charge of relaying data from the ocean-bottom network toa surface station The typical application of UWASNs is themonitoring of a remote environmentThis paper only studiesthe local area with uw-sensors and uw-sink The uw-sink isresponsible for collecting data from sensor nodes Once theuw-sink has all the data from the nodes it transmits the datato the surface station by acoustic communication

For the simplification we consider a sensor networkconsisting of119872 sensor nodes uniformly deployed over a vastfield to continuously monitor the environment and an uw-sink away from the nodes with a constant power supplythrough which the end user can access data from theUWASN Some reasonable assumptions are made about thesensor nodes

(1) Uw-sink is stationary after deployment and sensornodes move slowly and rarely move

(2) All sensor nodes are homogeneous and have the samecapabilities

(3) All sensor nodes can adjust the amount of transmitpower with the transmission distance Energy sensornode has its maximum transmission power that isto say it has its maximum transmission range Thesink is within the maximum transmission range of allsensor nodes that is all sensor nodes can transmitdata with enough transmission power to reach theuw-sink Each node has the computational power to

4 International Journal of Distributed Sensor Networks

Table 1 An example of the neighbor node table

Current sensor ID Energy level Successor ID Predecessor ID Number of predecessors11 27318 0 14 15 21 314 27318 11 31 34 215 27318 11 36 41 221 27318 11 42 1

support different MAC protocols and perform signalprocessing functions

(4) Each sensor node is location aware by some exist-ing positioning methods The positioning methodsbelong to another research area which are notinvolved in this study The uw-sink has strong com-putation power

(5) Each sensor node senses and generates data evenly ineach round

(6) To avoid interference CDMA technology can be usedto achieve multiple simultaneous wireless transmis-sions [21]

These assumptions are reasonable due to technologicaladvances in acoustic communication hardware and low-power computing

32 Energy ConsumptionModel To quantify the energy con-sumption of the UWASNs we adopt the energy dissipationmodel based on underwater acoustic communication princi-ple To transmit a data packet from one node to another overa transmission distance 119909 each node needs to transmit at apower level 119875

1= 1198750119860(119909) where 119875

0is a power level needed

at the input to the receiver and 119860(119909) is the attenuation Theattenuation [4] is given as

119860 (119909) = 119909119896119886119909 (1)

where 119896 is the energy spreading factor (1 for cylindrical 15for practical and 2 for spherical spreading) and

119886 = 10120572(119891)10 (2)

is a frequency-dependent term obtained from the absorptioncoefficientThe absorption coefficient for the frequency rangeof interest is calculated according toThorprsquos expression [4] as

120572 (119891) = 0111198912

1 + 1198912+ 44

1198912

4100 + 1198912

+ 275 times 10minus41198912+ 0003

(3)

in dBkm for 119891 in kHzTo transmit a 119897-bit packet over a transmission distance 119909

the transmitter expends

119864119879(119897 119909) = 119897119875

0119909119896119886119909 (4)

and to receive this packet the receiver expends119864119877(119897) = 119897119875

119903 (5)

where 119875119903is a constant parameter which depends on the

receiver devices

4 Balance Transmission Mechanism

The BTM includes the route set-up phase and the stable datatransmission phase A route set-up phase is executed whenthe UWASN begins to work and then goes to a stable datatransmission phasewhen data are transferred from the sensornodes to the uw-sink BTM presents an energy-efficient rout-ing algorithm based on the optimum transmission distance[22] in the route set-up phase and a data transmission algo-rithm based on energy level in the stable data transmissionphase to balance network energy consumption

41 Efficient Routing Algorithm Based on Optimum Transmis-sion Range (ERA) Obviously if the energy consumption inthe whole process of each uw-sensor node sending packetsto the uw-sink is minimized the energy consumption of thenetwork will be minimized The study minimized the totalenergy consumed in the network and obtained the optimumtransmission distance [22] in the multihop communicationIn order to optimize the energy consumption this paper pro-poses the ERA to establish the route in the route set-up phase

Definition 1 Let 119903opt be the optimum transmission distancewhich can be obtained in the previous research [22]

The goal of ERA is to design an algorithm that selectsensor nodes as more as possible near the position whichis 119903opt far from the sending node as relays Assume that thecoordinate positions of sensor nodes and the uw-sink in theUWASN are known

At the route set-up phase the detailed steps of the ERAalgorithm are as follows

(1) The sensor node 119878119894 in the network sends a query

packet including its location information and thelocation information of the uw-sink All the nodesreceiving the query packet 119878

119895| 119895 = 119894 compute

119876119895= 120579|119889(119878

119895 119878119894)minus119903opt|+(1minus120579)119889(119878119895 119880119878) where119889(119878119895 119878119894)

is the distance between the node 119878119895and the sending

node 119878119894and119889(119878

119895 119880119878) is the distance between the node

119878119895and the uw-sink 120579 is a system parameter defined in

this paper here let 120579 be 05 initially By adjusting thevalue of 120579 only one sensor node toward the sink andclosest to the optimum transmission distance will beselected as the relay node of the sensor node 119878

119894

(2) Select the nodewith theminimal119876119895as the relay node

Then record the position of the relay into the neighbornode table (eg Table 1) of the sending node 119878

119894as a

successor and record the position of the sending node119878119894into the neighbor node table of the relay node as

International Journal of Distributed Sensor Networks 5

0

11

34

42 36

4115

2114

31

Uw-sinkUw-sensormiddot middot middot

middot middot middot

Figure 2 Trees-route at the route set-up phase

a predecessor If the minimal 119876119895values of several

sensor nodes are the same let 120579 get larger Repeat (1)If the minimal 119876

119895values of several sensor nodes are

still the same look up their neighbor table and selectthe node with the least predecessors as the relay nodeNote the successor must be only one node

(3) Repeat (1) and (2) until the uw-sink is within the opti-mum transmission distance of the sensor node send-ing a query packet When the uw-sink is within theoptimum transmission distance of the sensor nodethe sensor node can transmit the data to the sinkwithout relaying

(4) Obviously if all the nodes repeat (1) (2) and (3) atrees-route will be set up

Figure 2 shows the route result containing some trees atthe route set-up phase

Note once the route is set up the data can be transmittedin the stable data transmission phase Because the uw-sensornodes are mobile in the real case the route set-up phase willbe operated again when the uw-sensor nodes move far awayfrom their origin position so as not to find relay Thereforeour algorithm is suitable for the case that the move of theuw-sensor nodes is not frequent Otherwise the route set-up phase must be executed constantly which induces moreenergy consumption

42 Data Balance Transmission Algorithm (DBT) By a routeset-up phase a set of trees-route is obtained Because theforwarding data is not aggregated in this paper thereforethe nodes close to the uw-sink die out much faster thanthose far away from the uw-sink In order to balance theenergy consumption we introduce an EL that is we dividethe nodersquos initial energy into 119898 equal parts and the ELof a node is 119894 (1 le 119894 le 119898) if its current energy is

more than (119894 minus 1) parts and less than or equal to 119894 partsDuring stable data transmission phase each uw-sensor hastwo transmission modes forwarding the packets towards theuw-sink by multihop transmission (MT) or transmitting thepackets directly to the uw-sink (DT) A node can decidewhether to forward data towards the uw-sink via MT modeor to send data directly to the uw-sink via DT by comparingits EL with that of its successor node The DBT algorithm isdescribed in detail as follows

Initially every node forwards the packet to its successortowards the uw-sink by MT mode and their ELs is m Somerounds later the node which is closest to the uw-sink willdissipate more energy because of more traffic load so itsenergy drops into the next energy level that is119898minus1 Then itbroadcasts an EL notice packet (ELNP) to their predecessornodes the predecessor nodes will transmit data to the uw-sink by DT mode since their ELs are larger than that of thesuccessor node An ELNP can be set as 1 bit for saving energySeveral rounds again these predecessor nodes will dissipatemore energy because of DTmodeTherefore when the EL ofa predecessor node decreases the node broadcasts the ELNPto its predecessor nodes similarly and if the EL of the nodeis equal to or less than its successor node it will transmitdata by MT mode again Thus all uw-sensors transmitdata ultimately to the uw-sink by similar operation to theabove process To avoid interference CDMA technologycan be used to achieve multiple simultaneous transmissionsBy alternately changing uw-sensorsrsquo transmission mode thenodes can consume their energy evenly accordingly thenetwork lifetime is prolonged

Obviously the magnitude of the EL affects the energydissipated in the network For all sensor nodes if their initialnumbers of the EL are 1 their transmission mode becomesmultihop transmission If the initial EL of each uw-sensoris very large the transmission mode is similar to the directtransmission Therefore an optimal EL will be found in thefollowing section

43 Analysis of Optimum Energy Level (OEL) When somesensor nodes run out of their energy the network mightcollapse because it is not able to collect and to transmitdata efficiently while other sensor nodes still have muchremaining energy which causes energy waste In order tobalance energy consumption in some cases it is an ideal casethat all the nodes in the network run out of their energy at thesame timeTherefore this section analyzes the wasted energyminimizes it and studies theOELWe can virtually divide thenetwork disk section into 119899 ring sections or slices the radiusof slice is the optimum transmission distance

Definition 2 Starting from the center of the disk where theuw-sink is located we define the set of slices We denote thefirst slice containing the uw-sink by Slice 1 which has a radius119903 and use Slice 119894 (2 le 119894 le 119873) to represent the slice next to Slice119894 minus 1 from the opposite side of the uw-sink which is definedby two successive circles one of radius 119894 times 119903 and the other ofradius (119894minus1)times119903 119903 is the optimum transmission distance 119903opt

The area of Slice 119894 is 119860119894= int119894119903

(119894minus1)119903int2120587

0119903119889119903119889120579 = (2119894 minus 1)120587119903

2

6 International Journal of Distributed Sensor Networks

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

Table 1 An example of the neighbor node table

Current sensor ID Energy level Successor ID Predecessor ID Number of predecessors11 27318 0 14 15 21 314 27318 11 31 34 215 27318 11 36 41 221 27318 11 42 1

support different MAC protocols and perform signalprocessing functions

(4) Each sensor node is location aware by some exist-ing positioning methods The positioning methodsbelong to another research area which are notinvolved in this study The uw-sink has strong com-putation power

(5) Each sensor node senses and generates data evenly ineach round

(6) To avoid interference CDMA technology can be usedto achieve multiple simultaneous wireless transmis-sions [21]

These assumptions are reasonable due to technologicaladvances in acoustic communication hardware and low-power computing

32 Energy ConsumptionModel To quantify the energy con-sumption of the UWASNs we adopt the energy dissipationmodel based on underwater acoustic communication princi-ple To transmit a data packet from one node to another overa transmission distance 119909 each node needs to transmit at apower level 119875

1= 1198750119860(119909) where 119875

0is a power level needed

at the input to the receiver and 119860(119909) is the attenuation Theattenuation [4] is given as

119860 (119909) = 119909119896119886119909 (1)

where 119896 is the energy spreading factor (1 for cylindrical 15for practical and 2 for spherical spreading) and

119886 = 10120572(119891)10 (2)

is a frequency-dependent term obtained from the absorptioncoefficientThe absorption coefficient for the frequency rangeof interest is calculated according toThorprsquos expression [4] as

120572 (119891) = 0111198912

1 + 1198912+ 44

1198912

4100 + 1198912

+ 275 times 10minus41198912+ 0003

(3)

in dBkm for 119891 in kHzTo transmit a 119897-bit packet over a transmission distance 119909

the transmitter expends

119864119879(119897 119909) = 119897119875

0119909119896119886119909 (4)

and to receive this packet the receiver expends119864119877(119897) = 119897119875

119903 (5)

where 119875119903is a constant parameter which depends on the

receiver devices

4 Balance Transmission Mechanism

The BTM includes the route set-up phase and the stable datatransmission phase A route set-up phase is executed whenthe UWASN begins to work and then goes to a stable datatransmission phasewhen data are transferred from the sensornodes to the uw-sink BTM presents an energy-efficient rout-ing algorithm based on the optimum transmission distance[22] in the route set-up phase and a data transmission algo-rithm based on energy level in the stable data transmissionphase to balance network energy consumption

41 Efficient Routing Algorithm Based on Optimum Transmis-sion Range (ERA) Obviously if the energy consumption inthe whole process of each uw-sensor node sending packetsto the uw-sink is minimized the energy consumption of thenetwork will be minimized The study minimized the totalenergy consumed in the network and obtained the optimumtransmission distance [22] in the multihop communicationIn order to optimize the energy consumption this paper pro-poses the ERA to establish the route in the route set-up phase

Definition 1 Let 119903opt be the optimum transmission distancewhich can be obtained in the previous research [22]

The goal of ERA is to design an algorithm that selectsensor nodes as more as possible near the position whichis 119903opt far from the sending node as relays Assume that thecoordinate positions of sensor nodes and the uw-sink in theUWASN are known

At the route set-up phase the detailed steps of the ERAalgorithm are as follows

(1) The sensor node 119878119894 in the network sends a query

packet including its location information and thelocation information of the uw-sink All the nodesreceiving the query packet 119878

119895| 119895 = 119894 compute

119876119895= 120579|119889(119878

119895 119878119894)minus119903opt|+(1minus120579)119889(119878119895 119880119878) where119889(119878119895 119878119894)

is the distance between the node 119878119895and the sending

node 119878119894and119889(119878

119895 119880119878) is the distance between the node

119878119895and the uw-sink 120579 is a system parameter defined in

this paper here let 120579 be 05 initially By adjusting thevalue of 120579 only one sensor node toward the sink andclosest to the optimum transmission distance will beselected as the relay node of the sensor node 119878

119894

(2) Select the nodewith theminimal119876119895as the relay node

Then record the position of the relay into the neighbornode table (eg Table 1) of the sending node 119878

119894as a

successor and record the position of the sending node119878119894into the neighbor node table of the relay node as

International Journal of Distributed Sensor Networks 5

0

11

34

42 36

4115

2114

31

Uw-sinkUw-sensormiddot middot middot

middot middot middot

Figure 2 Trees-route at the route set-up phase

a predecessor If the minimal 119876119895values of several

sensor nodes are the same let 120579 get larger Repeat (1)If the minimal 119876

119895values of several sensor nodes are

still the same look up their neighbor table and selectthe node with the least predecessors as the relay nodeNote the successor must be only one node

(3) Repeat (1) and (2) until the uw-sink is within the opti-mum transmission distance of the sensor node send-ing a query packet When the uw-sink is within theoptimum transmission distance of the sensor nodethe sensor node can transmit the data to the sinkwithout relaying

(4) Obviously if all the nodes repeat (1) (2) and (3) atrees-route will be set up

Figure 2 shows the route result containing some trees atthe route set-up phase

Note once the route is set up the data can be transmittedin the stable data transmission phase Because the uw-sensornodes are mobile in the real case the route set-up phase willbe operated again when the uw-sensor nodes move far awayfrom their origin position so as not to find relay Thereforeour algorithm is suitable for the case that the move of theuw-sensor nodes is not frequent Otherwise the route set-up phase must be executed constantly which induces moreenergy consumption

42 Data Balance Transmission Algorithm (DBT) By a routeset-up phase a set of trees-route is obtained Because theforwarding data is not aggregated in this paper thereforethe nodes close to the uw-sink die out much faster thanthose far away from the uw-sink In order to balance theenergy consumption we introduce an EL that is we dividethe nodersquos initial energy into 119898 equal parts and the ELof a node is 119894 (1 le 119894 le 119898) if its current energy is

more than (119894 minus 1) parts and less than or equal to 119894 partsDuring stable data transmission phase each uw-sensor hastwo transmission modes forwarding the packets towards theuw-sink by multihop transmission (MT) or transmitting thepackets directly to the uw-sink (DT) A node can decidewhether to forward data towards the uw-sink via MT modeor to send data directly to the uw-sink via DT by comparingits EL with that of its successor node The DBT algorithm isdescribed in detail as follows

Initially every node forwards the packet to its successortowards the uw-sink by MT mode and their ELs is m Somerounds later the node which is closest to the uw-sink willdissipate more energy because of more traffic load so itsenergy drops into the next energy level that is119898minus1 Then itbroadcasts an EL notice packet (ELNP) to their predecessornodes the predecessor nodes will transmit data to the uw-sink by DT mode since their ELs are larger than that of thesuccessor node An ELNP can be set as 1 bit for saving energySeveral rounds again these predecessor nodes will dissipatemore energy because of DTmodeTherefore when the EL ofa predecessor node decreases the node broadcasts the ELNPto its predecessor nodes similarly and if the EL of the nodeis equal to or less than its successor node it will transmitdata by MT mode again Thus all uw-sensors transmitdata ultimately to the uw-sink by similar operation to theabove process To avoid interference CDMA technologycan be used to achieve multiple simultaneous transmissionsBy alternately changing uw-sensorsrsquo transmission mode thenodes can consume their energy evenly accordingly thenetwork lifetime is prolonged

Obviously the magnitude of the EL affects the energydissipated in the network For all sensor nodes if their initialnumbers of the EL are 1 their transmission mode becomesmultihop transmission If the initial EL of each uw-sensoris very large the transmission mode is similar to the directtransmission Therefore an optimal EL will be found in thefollowing section

43 Analysis of Optimum Energy Level (OEL) When somesensor nodes run out of their energy the network mightcollapse because it is not able to collect and to transmitdata efficiently while other sensor nodes still have muchremaining energy which causes energy waste In order tobalance energy consumption in some cases it is an ideal casethat all the nodes in the network run out of their energy at thesame timeTherefore this section analyzes the wasted energyminimizes it and studies theOELWe can virtually divide thenetwork disk section into 119899 ring sections or slices the radiusof slice is the optimum transmission distance

Definition 2 Starting from the center of the disk where theuw-sink is located we define the set of slices We denote thefirst slice containing the uw-sink by Slice 1 which has a radius119903 and use Slice 119894 (2 le 119894 le 119873) to represent the slice next to Slice119894 minus 1 from the opposite side of the uw-sink which is definedby two successive circles one of radius 119894 times 119903 and the other ofradius (119894minus1)times119903 119903 is the optimum transmission distance 119903opt

The area of Slice 119894 is 119860119894= int119894119903

(119894minus1)119903int2120587

0119903119889119903119889120579 = (2119894 minus 1)120587119903

2

6 International Journal of Distributed Sensor Networks

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

0

11

34

42 36

4115

2114

31

Uw-sinkUw-sensormiddot middot middot

middot middot middot

Figure 2 Trees-route at the route set-up phase

a predecessor If the minimal 119876119895values of several

sensor nodes are the same let 120579 get larger Repeat (1)If the minimal 119876

119895values of several sensor nodes are

still the same look up their neighbor table and selectthe node with the least predecessors as the relay nodeNote the successor must be only one node

(3) Repeat (1) and (2) until the uw-sink is within the opti-mum transmission distance of the sensor node send-ing a query packet When the uw-sink is within theoptimum transmission distance of the sensor nodethe sensor node can transmit the data to the sinkwithout relaying

(4) Obviously if all the nodes repeat (1) (2) and (3) atrees-route will be set up

Figure 2 shows the route result containing some trees atthe route set-up phase

Note once the route is set up the data can be transmittedin the stable data transmission phase Because the uw-sensornodes are mobile in the real case the route set-up phase willbe operated again when the uw-sensor nodes move far awayfrom their origin position so as not to find relay Thereforeour algorithm is suitable for the case that the move of theuw-sensor nodes is not frequent Otherwise the route set-up phase must be executed constantly which induces moreenergy consumption

42 Data Balance Transmission Algorithm (DBT) By a routeset-up phase a set of trees-route is obtained Because theforwarding data is not aggregated in this paper thereforethe nodes close to the uw-sink die out much faster thanthose far away from the uw-sink In order to balance theenergy consumption we introduce an EL that is we dividethe nodersquos initial energy into 119898 equal parts and the ELof a node is 119894 (1 le 119894 le 119898) if its current energy is

more than (119894 minus 1) parts and less than or equal to 119894 partsDuring stable data transmission phase each uw-sensor hastwo transmission modes forwarding the packets towards theuw-sink by multihop transmission (MT) or transmitting thepackets directly to the uw-sink (DT) A node can decidewhether to forward data towards the uw-sink via MT modeor to send data directly to the uw-sink via DT by comparingits EL with that of its successor node The DBT algorithm isdescribed in detail as follows

Initially every node forwards the packet to its successortowards the uw-sink by MT mode and their ELs is m Somerounds later the node which is closest to the uw-sink willdissipate more energy because of more traffic load so itsenergy drops into the next energy level that is119898minus1 Then itbroadcasts an EL notice packet (ELNP) to their predecessornodes the predecessor nodes will transmit data to the uw-sink by DT mode since their ELs are larger than that of thesuccessor node An ELNP can be set as 1 bit for saving energySeveral rounds again these predecessor nodes will dissipatemore energy because of DTmodeTherefore when the EL ofa predecessor node decreases the node broadcasts the ELNPto its predecessor nodes similarly and if the EL of the nodeis equal to or less than its successor node it will transmitdata by MT mode again Thus all uw-sensors transmitdata ultimately to the uw-sink by similar operation to theabove process To avoid interference CDMA technologycan be used to achieve multiple simultaneous transmissionsBy alternately changing uw-sensorsrsquo transmission mode thenodes can consume their energy evenly accordingly thenetwork lifetime is prolonged

Obviously the magnitude of the EL affects the energydissipated in the network For all sensor nodes if their initialnumbers of the EL are 1 their transmission mode becomesmultihop transmission If the initial EL of each uw-sensoris very large the transmission mode is similar to the directtransmission Therefore an optimal EL will be found in thefollowing section

43 Analysis of Optimum Energy Level (OEL) When somesensor nodes run out of their energy the network mightcollapse because it is not able to collect and to transmitdata efficiently while other sensor nodes still have muchremaining energy which causes energy waste In order tobalance energy consumption in some cases it is an ideal casethat all the nodes in the network run out of their energy at thesame timeTherefore this section analyzes the wasted energyminimizes it and studies theOELWe can virtually divide thenetwork disk section into 119899 ring sections or slices the radiusof slice is the optimum transmission distance

Definition 2 Starting from the center of the disk where theuw-sink is located we define the set of slices We denote thefirst slice containing the uw-sink by Slice 1 which has a radius119903 and use Slice 119894 (2 le 119894 le 119873) to represent the slice next to Slice119894 minus 1 from the opposite side of the uw-sink which is definedby two successive circles one of radius 119894 times 119903 and the other ofradius (119894minus1)times119903 119903 is the optimum transmission distance 119903opt

The area of Slice 119894 is 119860119894= int119894119903

(119894minus1)119903int2120587

0119903119889119903119889120579 = (2119894 minus 1)120587119903

2

6 International Journal of Distributed Sensor Networks

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

Definition 3 Select one sensor node in Slice 1 as one rootnode a tree of the trees-route can be obtained starting fromSlice 1 Let the tree be single tree As a result the trees-routecan be composed of many single trees

From Figure 2 the trees-route consists of many singletrees whose root node is one node near the uw-sink

Lemma4 Theaverage number of uw-sensors in one single treeis 2119894 minus 1 in Slice 119894

Proof If the uw-sensors density is 120588 the number of sensors inthe Slice 119894 is 120588119860

119894 According to Definition 1 the proportion of

the sensors number in two successive slices Slice 119894 and Slice119894 + 1 is

120588119860119894

120588119860119894+1

=(2119894 minus 1)

(2119894 + 1) (6)

To make it simple we assume an ideal single tree that isthe number of nodes in each layer of the tree is the averagenumber of nodes of all trees in each layer and all nodesof each layer in this single tree are in one slice because ofRTB algorithm in Section 3 the relay is selected near theoptimum transmission distance Furthermore the number ofroot nodes in the single tree is 1 in Slice 1 and the averagesensors number of one tree in the Slice 119894 is 2119894minus1 from (6)

Definition 5 Let119898 be the energy level and let1198640be the initial

energy of each node thus the unit energy per level is 1198640119898

Definition 6 Let the direct transmission energy consumptionof one packet for the uw-sensor node farthest from theuw-sink be 119864

119891 that is 119864

119891= 1198750119877119896119886119877119897 let the multihop

transmission energy consumption of one packet for the uw-sensor node be 119864

119898 that is 119864

119898= 1198750119903119896119886119903119897 where 119897 is the length

of each packet

Definition 7 Let the wasted energy be 119864119908

The wasted energy includes remaining energy of all uw-sensors when the networks collapse and all energy used tobroadcast backwards ELNP We can minimize the wastedenergy 119864

119908to find an OEL

We discuss two cases considering the size of UWSAN (1)We assume 119864

119891le 1198640119898 in the small scale UWASN (2) We

assume 119864119891gt 1198640119898 in the large scale UWASN

Case I We assume 119864119891le 1198640119898 in the small scale UWASN

Because the maximum energy consumption of transmittinga packet is the direct transmission energy consumption ofthe uw-sensor node farthest from the uw-sink 119864

119891 and 119864

119891le

1198640119898 the energy consumption of transmitting a packet is

not more than 1198640119898 The case avoids the ELs of sensor

nodes decrease frequently On the contrary if the energyconsumption of transmitting a packet is more than 119864

0119898

sensor nodersquos EL will decrease after it transmits a packetThatis to say

1198750119877119896119886119877119897 le

1198640

119898 (7)

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=

119873minus1

sum

119894=1

1198750119903119896

119894119886119903119894119898(2119894 minus 1) (8)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then 119898(2119894 minus 1) stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903

119894is the

average transmission distance of sending backward the ELNPin Slice 119894 Here we can have 119903

119894= 119903

Definition 8 Let 119864119894119903be the remaining energy of the nodes in

Slice 119894 and let EL119894be the energy level of the nodes in Slice 119894

Lemma 9 If 119864119891le 1198640119898 when the network dies out the EL

difference of the nodes in the successive slices is not more than1 that is 119864119871

119894+1minus 119864119871119894le 1

Proof We assume that each node generates a 119897-bit packet ineach round and there are a node A in Slice 119894 and one nodeB of its predecessor nodes in Slice 119894 + 1 whose ELs are all 119898at first Initially the transmission mode of B is MT mode Inthat event EL

119894+1minus EL119894= 0 Because node A transmits its

own data and relays the data of node B node A consumesmore energy and its EL decreases firstly after several roundsBecause the energy consumption of transmitting a packet isnot more than 119864

0119898 that is 119864

119891le 1198640119898 the EL of A is119898minus 1

and EL119894+1

minus EL119894= 1 Then A will inform B to transmit the

data directly to uw-sink by DT mode and A only transmitsits own data In addition B is farther than A from the uw-sink Therefore 119864119894+1

119903will decrease more quickly than 119864119894

119903and

the energy difference between A and B is diminishing Afterseveral rounds again the EL of B decreases to119898minus1 It informsits predecessor nodes to transmit the data directly to uw-sinkNode B only transmits its own data Because B is farther thanA from the uw-sink 119864119894+1

119903will still decrease more quickly than

119864119894

119903 When the EL of node B is not more than that of node A

it changes its transmission mode to MT mode Going roundand round we can prove EL

119894+1minus EL119894le 1

As a result when the node closest to the uw-sink runsout of its energy and its remaining energy is 0 the maximalremaining energy of other uw-sensors will be obtained

Definition 10 We can determine the maximum remainingenergy of one node in Slice 119894 as (119894minus1)119864

0119898 by Lemma 9 when

the network collapses Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873

119894=2(2119894 minus 1) (119894 minus 1) 119864

0

119898 (9)

Then the maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903

= 1198750119903119896119886119903119898

119873minus1

sum

119894=1

(2119894 minus 1) +1198640

119898

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1)

(10)

International Journal of Distributed Sensor Networks 7

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

Table 2 Optimal values of 119898 with different power and networkradius 119877 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2

1198750= 1 times 10

minus4 Jbit 119875119903= 02 times 10

minus4 Jbit119877 (km) 119898

04 2434406 2434408 253381 27318

Table 3 Remaining energy of the other nodes in one tree when thefirst node dies out in Figure 3

Node ID Remaining energy (J)011 014 001585815 00185913121 003190131 000972034 000972036 0009290541 0009290042 00091901

Theorem 11 If 119864119891le 1198640119898 the optimal energy level can be

obtained from

119898

=radic1198640 [119873 (119873 + 1) (2119873 + 1) 3 minus 2 minus (31198732 + 2) (119873 minus 1)]

1198750119903119896119886119903 (119873 minus 1)

2

(11)

Proof From (10) we let the derivative of 119864119908with respect to

119898 be 0 that is

1198641015840

119908= 1198750119903119896119886119903

119873minus1

sum

119894=1

(2119894 minus 1) minus1198640

1198982

119873

sum

119894=2

(2119894 minus 1) (119894 minus 1) = 0 (12)

we minimize 119864119908and the OEL can be obtained

We can see that the OEL is decided by the initial energy1198640 the transmitting power 119875

0 the receiving power 119875

119903 the

frequency 119891 and the slice number119873In Table 2 the OELs are given with different network

radius 119877 the transmitting power 1198750 and the receiving power

119875119903 where 119864

0= 3000 J 119891 = 20 kHz and 119896 = 2 The

transmitting power 1198750 the receiving power 119875

119903 and frequency

119891 decide the transmission distance 119903Figure 3 shows that the network lifetime of UWASN is

longest when the EL is varying near the OEL where networkradius = 1 km In Table 3 we can observe the remainingenergy of the other nodes when the first node dies out in onetree-route (see Figure 2)

Case II Then we assume 119864119891gt 1198640119898 discuss the OEL in

detail for the large scale UWASN

0 1 2 3 4 515476

15478

1548

15482

15484

15486

15488

Energy level

Net

wor

k lif

etim

e (ro

unds

)

times105

times104

Figure 3 Network lifetime of UWASN versus EL

Because the maximum energy consumption of transmit-ting a packet is the direct transmission energy consumptionof the uw-sensor node farthest from the uw-sink the energyconsumption of transmitting a packet is not more than 119864

119891

We assume that 119864119898lt 1198640119898 It is reasonable otherwise the

transmission is nearly the DT mode

Definition 12 Let the maximum energy consumption oftransmitting a packet in terms of each uw-sensor node for thelarge scale UWASNs 119875

0119877119896119886119877119897 be 119899119864

0119898 That is to say

119864119891=1198991198640

119898 (13)

The direct transmission energy of a packet for the nodefarthest from the uw-sink is large so that it runs out of itsenergy quickly

Lemma 13 The times of broadcasting backward the ELNP foreach uw-sensor is not more than119898119899

Proof From Definitions 6 and 12 we can know that if theenergy consumption of the uw-sensor nodes farthest from theuw-sink for the multihop forwarding is not considered theirenergy will run out via119898119899 times of the direct transmissionOf course the network collapses in that event Thereforeaccording to our routing algorithm the times of broadcastingbackward the ELNP for each uw-sensor is notmore than119898119899

When the network collapses the maximum energy con-sumption of all nodes for broadcasting backwards the ELNPis

119864119887119887=sum119873minus1

119894=1(2119894 minus 1) 119875

0119903119896119886119903119898

119899 (14)

where (2119894 minus 1) is the number of the nodes in Slice 119894 fromLemma 4 and then119898(2119894minus1)119899 stands for the total maximumtimes of broadcasting backwards ELNPs in Slice 119894 119903 is theaverage distance of sending backward the ELNP in Slice 119894

8 International Journal of Distributed Sensor Networks

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

Lemma 14 When the network collapses the EL difference ofthe nodes in two successive slices is not more than 119899

Proof We assume that there are a node A in Slice 119894 and anode B of its predecessor nodes in Slice 119894 + 1 which are allat 119898 energy level at first Initially the transmission mode ofB is MT mode In that event the EL difference is 0 Becausenode A transmits its own data and relays the data of nodeB node A consumes more energy and its EL decreases firstlyafter several rounds Furthermore the energy consumptionof transmitting a packet is not more than 119864

0119898 the EL of A

is 119898-1 and the EL difference is 1 Then A will inform B totransmit the data directly to uw-sink byDTmode andA onlytransmits its own data If the direct transmission energy of apacket for node B is 119899

1198611198640119898 the EL of B decreases 119899

119861level

after a round In that event (1) if the transmission mode of Ais MT mode we can obtain the EL difference between A andB (119898minus1)minus(119898minus119899

119861) = 119899119861minus1 le 119899

119861le 119899 (2) if the transmission

mode ofA isDTmode and the direct transmission energy of apacket for node A is 119899

1198601198640119898 we can obtain the EL difference

betweenA and B (119898minus1minus119899119860)minus(119898minus119899

119861) = 119899119861minus1minus119899

119860le 119899119861le 119899

because 119899119860lt 119899119861 Then node B informs its predecessor node

to transmit data by DT mode and change its mode into MTmode if its EL is not more than that of node A Going roundand round we can prove the Lemma

Definition 15 We can determine the maximum remainingenergy of one node in Slice 119894 as (119873 minus 119894)119899119864

0119898 by Lemma 14

when the network collapses (the node farthest from the uw-sink dies out earlier) Then the total pessimistic remainingenergy of all the nodes is

119864119901119903=sum119873minus1

119894=1(2119894 minus 1) (119873 minus 119894) 119899119864

0

119898 (15)

Theorem 16 If 119864119891

= 1198991198640119898 for large scale UWASNs the

optimum energy level can be obtained from

119898 = 119899((1198640[119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

times (1198750119903119896120572119903(119873 minus 1)

2)minus1

)

12

(16)

Proof The maximum wasted energy dissipated in thisscheme is

119864119908= 119864119887119887+ 119864119901119903 (17)

From (14) (15) and (17) we let the derivative of 119864119908with

respect to119898 be 0 then119864119908can beminimized and theOEL can

be obtained

From (3) and (16) we can see that the OEL for largeUWSNs is also decided by the initial energy 119864

0 the transmit-

ting power 1198750 the transmission distance 119903 the frequency 119891

and the network size 119877 or119873

Table 4 Optimum values with different 119903

119903 (km) 119877 (km) 119898

02 157 2252105 368 86471 732 403815 1098 2514

From (13) and (16) we can obtain

1198750119877119896119886119877119897 = ( (119864

01198750119903119896119886119903(119873 minus 1)

2)

times ([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)])

minus1

)

12

(18)

Equation (18) can be simplified to

([119873 (119873 minus 1) (2119873 + 1)

2minus 119873 (119873 minus 1)

minus 2 (119873 (119873 + 1) (2119873 + 1)

6minus 1)]119875

011987721198961198862119877minus119903

1198972)

times (1198640119903119896(119873 minus 1)

2)minus1

= 1

(19)

Because 119873 can be decided by the network radius 119877 andthe transmission distance 119903 we can obtain the relationshipbetween the network size and the transmission distancewhen1198640 1198750 119897 and 119891 are given from (19)

The relationship between119891 and the transmission distanceis given in [1] In Table 4 the transmission distance 119903 thenetwork radius 119877 and ELs are given according to (13) and(16) with different 119903 where 119875

0= 1 times 10

minus3 Jbit 1198640= 3000 J

119891 = 10 kHZ and 119896 = 2 We can obtain 119873 from 119877 and119903 We can observe that 119877 is increasing with the increasingof 119903 Furthermore 119898 is decreased with the increasing of 119903When 119877 becomes larger the node farthest from the uw-sinkconsumed more energy at the DT mode therefore less DTmode can savemore energy and prolong the network lifetimeThe decreasing of 119898 enhances the probability of MT modeand decreases the probability of DT mode for the node thusthe results in Table 4 are reasonable

5 Simulations

We made some simulations and employed the methods in alab experiment system The performance evaluation is doneusing the data from the experiment andMatLab simulator Inthe simulations the sensor nodes are uniformly distributedin a disk area the uw-sink is in the center of the disk Thedisk radius is 119877 The initial energy of each node is 3000 Jand in each round every sensor node generates a 200-bitpacket that will be propagated to the uw-sink ultimately Inthe simulation network lifetime is denoted by identical unit

International Journal of Distributed Sensor Networks 9

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

02 03 04 05 06 07 08 09 1Network radius (km)

BTMDirectMTE

0

05

1

15

2

25

3

35

Net

wor

k lif

etim

e (ro

unds

)

times106

Figure 4 Network lifetime varies from different network radius 119877when the network is small

02 03 04 05

05

06 07 08 09 10

1

15

2

25

Network radius (km)

Ener

gy co

nsum

ptio

n of

net

wor

k (J

)

BTMDirectMTE

Figure 5 Energy consumption varies from different network radius119877 when the network is small

rounds Network lifetime is defined as the time when thefirst node is depleted of its battery power In simulationswe compare the performance for direct transmission by one-hop communication MTE (minimum transmission energyalgorithm) [11] by multihop communication and BTM inboth the small scaleUWASNand the large scale oneMTE is aclassical and knownmultihopmethod so this study used thatas the communication protocol in the multihop transmissionfor comparing in the simulation

51 Small Scale UWASN The simulation employs 80 nodesdistributed uniformly in the small scale UWASN We showthe network lifetime and energy consumption with different119877 in Figures 4 and 5 We can see the network lifetimein BTM is longest in Figure 4 Network lifetime is defined

Time (rounds)

01

02

03

04

05

06

07

08

09

1

Prop

ortio

n of

sens

ors s

till a

live

times105

1 150 05 2 25 3 35 4 45 5

BTMDirectMTE

Figure 6 Proportion of nodes that remain alive over the simulationtime where the network radius = 1 km

as the time when the first node is depleted of its batterypower The network lifetime of the direct transmission islonger than that of MTE because the network size is smallso that the transmission energy consumption is not toolarge in the direct transmission and the receiving energyconsumption is contrastively more in MTEThis is obviouslyobserved from Figure 5 Furthermore we also obtain othersimulation results under different sensor densities whereasthe node densities do not have too distinct impact on thecurrent Therefore the simulation is enough to demonstratethe validity of BTM in which the network lifetime can beprolonged and the energy can be saved better which isextremely critical for UWASNs Note that when networkradius is small enough the network lifetime and the energyconsumption in BTM are nearly equal to that of directtransmission

Figure 6 shows the proportion of nodes that remain aliveover the simulation time While the first node dies out theoperating time in BTM is longer than other schemes MTE isthe worst The reason is that the more hops generate moreloads on the nodes closer to the uw-sink With more andmore nodes running out of their energy our scheme is stillbetter whereas when the proportion of nodes still alive issmall enough direct transmission is better than others Thisis because BTM prefers energy balance to the ending time ofthe last nodesThe nodes closest to the uw-sink have so shortdistance so that the energy spent in direct transmission isvery small therefore the ending time of those nodes is longerNote the network maybe already collapses in that event

To evaluate the effectiveness of energy balancing in BTMunder real network scenarios we compare the average energyconsumption per selected annulus for different networksettings in the 2500th round Here the selected annulus is aspecified area in simulations which is not the slice specifiedin this paper For achieving fair simulations the selectedannulus width is given as 02 km In Figure 7 compared to theother schemes the average energy consumption per selected

10 International Journal of Distributed Sensor Networks

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

1 2 3 4 50

Aver

age e

nerg

y co

nsum

ptio

n pe

r sele

cted

annu

lus (

J)

010203040506070809

1

Selected annulus i

BTMDirectMTE

Figure 7 Average energy consumption per selected annulus in the2500th roundwhere the network radius = 1 km and selected annuluswidth = 02 km

0 10 20 30 40 50 60 70 800

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 30000Rounds = 60000Rounds = 90000

Rounds = 120000Rounds = 150000

Figure 8 Average remaining energy per node for different roundswhere the network radius = 1 km

annulus is best balanced in this paper The critical maximumenergy consumption in MTE is larger than that in the directtransmission because of the small network size That is tosay the average spent energy of the selected annulus closestto the UW-sink is the largest in MTE and the average spentenergy of the selected annulus farthest from the UW-sink isthe largest in direct transmission which is in agreement withthe above simulations As a result BTM improves the networklifetime

For further describing the balance character of BTM weanalyze the average remaining energy per sensor node inBTM under different roundsThe results satisfy us and verifythe efficient balance of BTM Figure 8 shows the averageremaining energy per sensor node under different roundsin BTM All remaining energy has no apparent difference

1 15 2 25 3 35 4 45 50

1000

2000

3000

4000

5000

6000

7000

8000

9000

Network radius (km)

Net

wor

k lif

etim

e (ro

unds

)

BTMDirectMTE

Figure 9 Network lifetime varies from different network radius 119877where the nodes density is 100 nodeskm2

in the same round The results prove that BTM can indeedbalance the energy consumption of each sensor node becauseFigure 7 compares the simulation results in the randomround to verify that BTM is better than the direct transmis-sion and MTE Of course simulations show that the directtransmission and MTE cannot balance energy consumptionand they have their own obvious energy consumption distri-bution as we know To avoid being complex and unclear wedo not depict the results of the direct transmission and MTEin Figure 8

52 Large ScaleUWASN Figure 9 shows the network lifetimewith different 119877 in the large scale UWASN We can see thenetwork lifetime in BTM is longer The network lifetimeof the direct transmission is longer than that of MTE atthe earlier time because the smaller the network radiusis the less the transmission energy consumption is in thedirect transmission and the receiving energy consumption iscontrastively more in MTEWith the network radius becom-ing larger MTE is not worse than the direct transmissionbecause of the increasing of the transmission energy in thedirect transmission Therefore the simulation demonstratesthe validity of BTM in which the network lifetime can beprolonged which is extremely critical for UWASNs

Figure 10 shows the proportion of nodes that remain aliveover the simulation time where the network radius 119877 = 2 kmand 119903 = 03 km BTM is still better

Figure 11 compares the average energy consumption perselected annulus for different network settings in 2000throundThe average energy consumption per selected annulusis best balanced in BTM

Considering different rounds we simulate the averageremaining energy per sensor node of BTM inFigure 12 where119877 = 2 km and the nodes density = 25 nodeskm2 Theremaining energy of each sensor node still keeps balance andhas no apparent difference under different roundsThe results

International Journal of Distributed Sensor Networks 11

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

0 1000 2000 3000 4000 5000 6000 7000 800001

02

03

04

05

06

07

08

09

1

Time (rounds)

Prop

ortio

n of

sens

ors s

till a

live

BTMDirectMTE

Figure 10 Proportion of nodes that remain alive over the simulationtime

1 2 3 4 5 6 70

05

1

15

2

25

3

35

4

45

Selected annulus i

Aver

age e

nerg

y co

nsum

ptio

npe

r sele

cted

annu

lus (

J)

BTMDirectMTE

Figure 11 Average energy consumption per selected annulus where119877 = 2 km and 119903 = 03 km

prove further that BTM can balance the energy consumptionof each sensor node efficiently We also obtain the averageremaining energy per sensor node under different rounds indifferent methods The direct transmission and MTE havethe obvious energy consumption distribution as Figure 11described As a result we do not include them in Figure 12so that the results of BTM can be clearly depicted

6 Conclusions

Because of severe energy constraints of the nodes it is impor-tant to design protocols for UWASNs with long lifetimeThis paper described the efficient routing algorithm in theroute set-up phase The routing algorithm selects relay nearthe optimum transmission Then the energy level based data

0 50 100 150 200 250 3000

500

1000

1500

2000

2500

3000

3500

4000

Sensor node ID

Rem

aini

ng en

ergy

of e

very

nod

e (J)

Rounds = 300Rounds = 600Rounds = 900

Rounds = 1200Rounds = 1500

Figure 12 Average remaining energy per node for different roundswhere 119877 = 2 km and the nodes density = 25 nodeskm2

balance transmission algorithm was used to balance energyconsumed The uw-sensor nodes decided their transmissionmode by the current EL of successor nodes The optimalclassification number of EL has been evaluated throughtheoretical analysis to balance the energy consumption betterThrough the balance transmissionmechanism the lifetime ofthe whole UWASNs can be prolongedThe research model inthis paper is two-dimensional UWASNsmodel It can also beemployed in three-dimensional UWASNs as is not discussedrepeatedly in this paper

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

This work was financially supported by the National NaturalScience Foundation (61202403) and National ScholarshipFund

References

[1] I F Akyildiz D Pompili and TMelodia ldquoUnderwater acousticsensor networks research challengesrdquo Ad Hoc Networks vol 3no 3 pp 257ndash279 2005

[2] JHeidemannM Stojanovic andMZorzi ldquoUnderwater sensornetworks applications advances and challengesrdquo PhilosophicalTransactions of the Royal Society A Mathematical Physical andEngineering Sciences vol 370 no 1958 pp 158ndash175 2012

[3] P Casari and M Zorzi ldquoProtocol design issues in underwateracoustic networksrdquo Computer Communications vol 34 no 17pp 2013ndash2025 2011

[4] L Berkhovskikh and Y Lysanov Fundamentals of Ocean Acous-tics Springer 1982

12 International Journal of Distributed Sensor Networks

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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

[5] SOlariu and I Stojmenovic ldquoDesign guidelines formaximizinglifetime and avoiding energy holes in sensor networks withuniform distribution and uniform reportingrdquo in Proceedings ofthe 25th IEEE International Conference on Computer Communi-cations (INFOCOM rsquo06) pp 1ndash12 Barcelona Spain April 2006

[6] D Pompili T Melodia and I F Akyildiz ldquoDistributed routingalgorithms for underwater acoustic sensor networksrdquo IEEETransactions on Wireless Communications vol 9 no 9 pp2934ndash2944 2010

[7] H P Tan W K G Seah and L Doyle ldquoA multi-hop ARQ pro-tocol for underwater acoustic networksrdquo in Proceeding of theOCEANS rsquo07 pp 1ndash6 Aberdeen Scotland June 2007

[8] 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 Republic of Korea December 2009

[9] 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 Notesin Computer Science pp 1216ndash1221 Springer Berlin Germany2006

[10] 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

[11] M Ettus ldquoSystem capacity latency and power consumption inMulti hop routed SS-CDMA wireless networksrdquo in Proceedingsof the Radio and Wireless Conference pp 55ndash58 August 1998

[12] 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 onMobile Computing and Net-working (MOBICOM rsquo06) pp 298ndash309 Los Angeles CalifUSA September 2006

[13] 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

[14] P Ponnavaikkoy K Yassiny S K Wilsony M Stojanoviczand J Holliday ldquoEnergy optimization with delay constraintsin underwater acoustic networksrdquo in Proceedings of the IEEEGlobal Communications Conference (GLOBECOM rsquo13) pp 551ndash556 Atlanta Ga USA December 2013

[15] Y Luo L Pu Z Peng Z Zhou J-H Cui and Z ZhangldquoEffective relay selection for underwater cooperative acousticnetworksrdquo in Proceedings of the 10th IEEE International Con-ference on Mobile Ad-Hoc and Sensor Systems (MASS rsquo13) pp104ndash112 October 2013

[16] J Xu K Li G Min K Lin and W Qu ldquoEnergy-efficient tree-based multipath power control for underwater sensor net-worksrdquo IEEE Transactions on Parallel and Distributed Systemsvol 23 no 11 pp 2107ndash2116 2012

[17] M Xu G Z Liu and H F Wu ldquoAn energy-efficient routingalgorithm for underwater wireless sensor networks inspiredby ultrasonic frogsrdquo International Journal of Distributed SensorNetworks vol 2014 Article ID 351520 12 pages 2014

[18] A Wahid S Lee and D Kim ldquoAn energy-efficient routingprotocol for UWSNs using physical distance and residual

energyrdquo in Proceedings of the OCEANS rsquo11 pp 1ndash6 SantanderSpain June 2011

[19] S Basagni L Boloni P Gjanci C Petrioli C A Phillipsand D Turgut ldquoMaximizing the value of sensed informationin underwater wireless sensor networks via an autono mousunderwater vehiclerdquo in Proceedings of the IEEE InternationalConference on Computer Communications (INFOCOM rsquo14) pp988ndash996 Toronto Canada AprilndashMay 2014

[20] G A Hollinger S Choudhary P Qarabaqi et al ldquoUnderwaterdata collection using robotic sensor networksrdquo IEEE Journal onSelected Areas in Communications vol 30 no 5 pp 899ndash9112012

[21] S De C Qiao D A Pados M Chatterjee and S J PhilipldquoAn integrated cross-layer study of wireless CDMA sensor net-worksrdquo IEEE Journal on Selected Areas in Communications vol22 no 7 pp 1271ndash1285 2004

[22] J F Dou G X Zhang Z W Guo and J B Cao ldquoPAS prob-ability and sub-optimal distance-based lifetime prolongingstrategy for underwater acoustic sensor networksrdquo WirelessCommunications and Mobile Computing vol 8 no 8 pp 1061ndash1073 2008

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