Research ArticleA Swarm Intelligent Algorithm Based Route MaintainingProtocol for Mobile Sink Wireless Sensor Networks
Xiaoming Wu,1,2,3 Yinglong Wang,1,2,3 and Yifan Hu2,3
1College of Information Science and Engineering, Shandong University of Science and Technology, Qingdao 266590, China2Shandong Computer Science Center (National Supercomputer Center in Jinan), Jinan 250014, China3Shandong Provincial Key Laboratory of Computer Network, Jinan 250014, China
Correspondence should be addressed to Yinglong Wang; [email protected]
Received 2 June 2015; Revised 5 September 2015; Accepted 30 September 2015
Recent studies have shown that mobile sink can be a solution to solve the problem that energy consumption of sensor nodes isnot balanced in wireless sensor networks (WSNs). Caused by the sink mobility, the paths between the sensor nodes and the sinkchange frequently and have profound influence on the lifetime of WSN. It is necessary to design a protocol that can find efficientroutings between the mobile sink and nodes but does not consume toomany network resources. In this paper, we propose a swarmintelligent algorithm based route maintaining protocol to resolve this issue.The protocol utilizes the concentric ring mechanism toguide the route researching direction and adopts the optimal routing selection to maintain the data delivery route in mobile sinkWSN. Using the immune based artificial bee colony (IABC) algorithm to optimize the forwarding path, the routing maintainingprotocol could find an alternative routing path quickly and efficiently when the coordinate of sink is changed in WSN. The resultsof our extensive experiments demonstrate that our proposed route maintaining protocol is able to balance the network traffic loadand prolong the network lifetime.
1. Introduction and Related Work
Wireless sensor network (WSN) is an intelligent monitor-ing self-organized network consisting of many microsensornodes with the capabilities of communication, sensing, andcomputing deployed inside or around the monitoring area. Ithas broad application prospects and great application value inindustrial and agricultural control, urbanmanagement, envi-ronmental testing, hazardous area remote control, and otherfields. WSN involves multiple frontier research fields; it isconsidered as one of the top-tenworld-changing technologiesin the future [1].
A typical WSN is composed of many sensor nodes andone or several sinks. The sink collects data from the sensingenvironment. However, if the sink involved in collecting datais static, the sensors connected directly to a sink deplete theirenergy much faster than the rest of the network since theycarry all the data gathered by the sensors [2], which is called“the crowded center effect” [3] or “energy hole problem”[4, 5]. One method to avoid the formation of energy holes
is to use sink mobility. The sensor nodes can take turns tobecome the neighbors of the sink due to the sinkmobility in acontrolled manner, so the energy is consumed evenly amongthe nodes. It has been demonstrated that a mobile sink canpotentially prolong the network’s lifetime as the mobile sinkwould cause the sensor nodes to consume less energy [6].
Mobile sink can be a solution to solve the problem thatenergy consumption of nodes is not balanced in WSN; wecall it mobile sink WSN (MSWSN); the network architectureis illustrated in Figure 1; the source nodes constantly deliverdata through multihops relay node path to the mobile sink.But caused by the sink mobility, the paths between the sensornodes and the sink would change with time [7]. Routingprotocol in MSWSN is a great challenge due to the followingreasons. Firstly, it is not easy to grasp the whole networktopology and it is hard to find a routing path. Secondly,sensor nodes are tightly constrained in terms of energy,processing, and storage capacities. The unpredictable andconstant changes in the sink’s location form the obstacleof designing the route maintaining protocols in the energy
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2015, Article ID 823909, 10 pageshttp://dx.doi.org/10.1155/2015/823909
2 Mathematical Problems in Engineering
Source node
Relay node
Mobile sink
Date delivery pathSink moving track
Figure 1: Mobile sink wireless sensor network model.
constrained WSN [8]. It is necessary to design a protocolthat can find efficient routes between the mobile sink andsensor nodes but does not consume too many networkresources.
Various routing protocols have been proposed to addressthe routing problem of mobile sink. In the literature [9],the TTDD concentrates on efficient data delivery to mobilesinks by clustering nodes into cells. Mobile sinks broadcastrequest packets only in their local cell, and an overlay routingscheme keeps track of the current cells of the sinks for routingdata to them. While effective in high mobility scenarios, theoverhead to build and maintain the overlay is significant,especially in periodic reporting scenarios, which are moretraffic intensive than event-based reporting. Thus, TTDDis better suited to event-detecting sensor networks withsporadic rather than continuous traffic.
SEAD [10] and its developed protocol DEED [11] attemptto optimize routes from a single source to mobile sinks byallowing each sink to select an access sensor node. A datadelivery tree is built between the source and all access nodesbased on a geographic location heuristic. When the sinkmoves, a path between its current nearest neighbor and theaccess node ismaintained, eliminating the need to rebuild thetree. However, if the sink moves far away, a new access nodewould be selected and the tree is rebuilt.
An enhanced real time with load distribution (ERTLD)routing protocol for mobile WSN is proposed in literature[12]. ERTLD utilizes corona mechanism and optimal for-wardingmetrics to forward the data packet inmobileWSN. It
ensures high packet delivery ratio and experiences minimumend-to-end delay and enhances the total performance, relia-bility, and flexibility of data forwardingmechanism inmobileWSN.
However, in these classical routing protocols, the routemaintaining scheme for the large scale sensor nodes hasrarely been considered. Moreover, as the routing optimiza-tion problem to find the optimal routing is a NP-hardnessproblem, the heuristic deterministic methods always fall intolocal optimum and only get the approximate optimal result.In this paper, we propose a swarm intelligent algorithmoptimized routemaintaining protocol to optimize the routingpath of the MSWSN, and this optimization algorithm wouldconverge to the optimal resolution of the path. Our proposedscheme is developed on the basis of the corona mechanismin ERTLD and differs from the above works. The maincontributions of this paper are as follows:
(1) We utilize the concentric ring mechanism to guidethe route researching direction and adopt self-adaptfeature of our swarm intelligent routing protocol todeal with the dynamic route maintaining problemof MSWSN, which results in saving energy andmaximizing packet delivery ratio.
(2) We analyze and compare our protocol with TTDD[9] and ERTLD [12] in terms of packet deliveryand energy consumption. Simulation results showsuperiority of the protocol.
Mathematical Problems in Engineering 3
CR_N = 1
Source node Relay node
Mobile sink
CR_N = 0CR_N = 2
CR_N = 3CR_N = 3
R
CR_N = 4�2s
�1s �1
�2�3
�4
�5 �6�7
�sink
(a)
(b)
Figure 2: The route maintaining model of MSWSN. (a) The process of delivering route maintaining packet through possible paths. (b) Theconcentric ring number of each node on the path.
The contributions of this paper are different from theliteratures [8, 13], though the network models of them are allthe WSNs with mobile sink:
(1) The routing mechanism is different, we provide con-centric ring mechanism to detect the relative distanceand direction of each node to themobile sink, in orderto efficiently guide the route researching directionfrom source node to the sink, comparing to therouting researching mechanism without directionguidance in [13] and another different routing mech-anism in [8].
(2) The kernel optimization algorithm of routing proto-col is different; our protocol adopts the artificial beecolony (ABC) algorithm to optimize the forwardingpath, comparing to the particle swarm optimizationalgorithm in literature [13] and machine learning-based approach in literature [8]. The tests for severalstandard functions by [14] have shown that the per-formance of ABC is better than the other population-based algorithms with the advantage of employing
fewer control parameters, such as PSO and otheralgorithms.
The rest of this paper is organized as follows. Section 2 for-mulates our network model and method. Section 3 describesthe proposed algorithm in detail, and the routing protocol isanalyzed. Section 4 evaluates the performance of the IABCby comparing it with other routing protocols. Finally, theconclusion is presented in Section 5.
2. Description of Network Model
2.1. Concentric Ring Mechanism. The MSWSN is modeledas a time-dependent connected graph 𝐺(𝑉, 𝐸), where 𝑉is a finite set of sensor nodes and 𝐸 is the set of edgesrepresenting connection between these nodes. The mobilesink is a rendezvous point and moves in the network areafor the purpose of collecting information of nodes. In ourmodel, the regular moving track of the mobile sink is therectangular polyline track, as shown in Figure 1, which canefficiently traverse the whole network. The similar traversetrack method has been used in the literature [15]. Figure 2
4 Mathematical Problems in Engineering
represents a certain time 𝑇0when the mobile sink has moved
to place A in the network model. V𝑖(0 < 𝑖 < 𝑁 − 2) is the
relay sensor node, V𝑠is the source node, and Vsink is themobile
sink. Suppose there exist 𝑚 (0 < 𝑚 < 𝑁 − 1) source nodesand onemobile sink; these source nodes can route the packetsthrough multihops relay node path to the mobile sink, asillustrated in the right blue dash-line box (b) of Figure 2.After deploying in themiddle area of the network, the mobilesink would broadcast the concentric ring (CR) packets toits one-hop neighbors, in order to determine the concentricring number for all sensor nodes in the network. The mainfields of CR are CR N (the ring number; initial number iszero) and CR ID (the packet ID sent by sink). Then, thenodes forward the packet to the next-hop neighbors. Thesink would not rebroadcast the CR packet until it moves toa new coordinate which is more than 𝐷 meters away fromthe previous coordinate. In our assumption, if the movingdistance of sink is less than 𝐷 meters, it is still within thecommunication range of the closest relay node.
As shown in Figure 2, all circles are concentric at the sinkand can form the coordinate system, and the width of eachconcentric ring is assumed to be equal to the sensor node’stransmission range 𝑅. Therefore, each sensor node wouldbelong to exactly one concentric ring. The data would bedelivered from node V
𝑖with high value of CR N to node V
𝑗
with low value of CR N and at last to the mobile sink. If thereis not any candidatewith lower number ofCR N than V
𝑖in the
neighbor table of V𝑖, datawould be forwarded to the nodewith
the same number of CR N.The right blue box (b) of Figure 2shows the CR N of each node on the multihop path.
When the mobile sink moves to the random position, thecoordinate system is also changed. Once the sinkmoves up to𝑅meters away, it would rebroadcast the CR N to the network,and each node’s CR N would be changed. The packet CRwould be delivered to all one-hop neighbors. If one relaynode receives CR, it would check CR ID; if it is the sameas the node’s previous CR ID, the node would discard thepacket; if not, the node would increase CR N in CR and savethis value as its new concentric ring number. Then, the nodewould broadcast CR to its neighbors, until the nodes of thewhole network are upgraded. With this method, the networkcan automatically respond to the dynamic topology. Oncethe sink moves to a certain distance, the previous scenariowould be repeated. Using the concentric ringmechanism, therelative distance and direction of each node to the mobilesink can be detected, and the following routing researchfrom source node can be guided toward the direction ofsink.
Our network routing model is based on the followingassumptions: (1) The area is covered by a large number ofhomogeneous sensor nodes. Sensor nodes are stationary, butsink moves and changes position constantly with a relativelyfixed speed. (2) Data is sensed and transmitted from eachsource node to the mobile sink every 𝑇 time period.
2.2. RouteMaintainingAlgorithm. Theconcentric ringmech-anism can calculate the sensor node’s ring number accordingto its distance to the sink. The following routing algorithm
can calculate and decide the optimal forwarding path fromthe source nodes to the mobile sink, in order to update andmaintain the dynamic route.
As assumed, the path generated from a given sourcenode V1
𝑠to the mobile sink Vsink is denoted by 𝑝
𝑗(𝑗 ∈ 1,
2, . . . , 𝑛). The 𝑘th relay node on path 𝑝𝑗is denoted by V𝑘
𝑗, {𝑘 ∈
1, 2, . . . , ℎ𝑗}, in which ℎ
𝑗is the hop count on path 𝑝
𝑗. 𝑒𝑚𝑗rep-
resents the 𝑚th direct edge between two neighbor nodes on𝑝𝑗. Let 𝑁(𝑝
𝑗) = {V𝑆
𝑗, V1𝑗, V2𝑗, . . . , V𝑘
𝑗, . . . , V𝑛
𝑗} ⊂ 𝑁(V) be
the set of the sensor nodes existing along path 𝑝𝑗, where 𝑘
represents the distance from the sink to the node on a hopscale. As shown in the right blue box (b) of Figure 2, path 𝑝
1
connects source V1𝑠with the sink and contains 6 relay nodes
Each node V owns its neighbor table, which stores itssurrounding neighbor node’s ID V
𝑖and other information
(node signal strength RSSI(V𝑖) from the next-hop node V
𝑖,
end-to-end delay Delay(V𝑖) between the node and V
𝑖).
The movement of mobile sink would firstly upgradeall the nodes’ CN R. Once the sink has moved out of thecommunication distance of relay node V𝑛
𝑗and the data packet
cannot be delivered to the sink, the routing failed informationpacket would be sent backward to the source node V𝑠
𝑗through
path 𝑝𝑗; then, the route maintaining algorithm would be
implemented.V𝑠𝑗would broadcast the routemaintaining (RM)packets to
one-hop neighbors, and the packet contains the informationof battery voltage and source’s CR N (called CR N S). Ourrule is that if the relay node’s CR N is not higher than thesource’s CR N S, it can relay RM packet, so RM can bedelivered forward of the mobile sink. If the relay node V
𝑖
receives RM, it will check its own CR N; if its CR N is higherthan CR N S in RM, V
𝑖will discard the packet; if not, V
𝑖will
decrease and save CR N S in RM and relay RM to its one-hop neighbors; the packet would contain information of itsRSSI(V
𝑖), Delay(V
𝑖), and its battery voltage 𝑉bat(V𝑖).Then, the
node will relay RM to the other neighbors, until the datapacket RM is delivered through any multihop path 𝑝
𝑗(0 <
𝑗 < 𝑛) to themobile sink.Thus, the sink would receive severalpackets RM from these paths, and all the paths deliveringRM to the sink are considered as the possible alternativepaths set 𝑃all. The process of delivering packet RM throughmultihop paths is illustrated in the left red dash-line box (a) ofFigure 2.
With this method, the optimal path 𝑝op with thesuitable relay node sequence {V𝑆op, V
1
op, V2
op, . . . , V𝑘
op, . . . ,
V𝑛op} is among these possible alternative paths set 𝑃all. Thesink collects all the information and extracts RSSI(V
𝑖),
Delay(V𝑖), and 𝑉bat(V𝑖) of these nodes on the possible
alternative paths 𝑃all and calculates the fitness value ofpath and selects one optimal alternative path 𝑝op withoptimal fitness fitness(𝑝op) (e.g., path {V1
𝑆, V1, V2, V3, V4,
V5, V6, V7} in left red box (a) of Figure 2). The nodes
sequence of a possible alternative path 𝑝𝑗is {V𝑆𝑗, V1𝑗, V2𝑗, . . . , V𝑘
𝑗,
. . . , V𝑛𝑗}, and the factors affecting the choice of 𝑝
𝑗include the
Mathematical Problems in Engineering 5
nodes’ signal strength, end-to-end delay, and battery voltageon the path. These parameters can determine the fitnessfunction of 𝑝
𝑗, fitness(𝑝
𝑗):
fitness (𝑝𝑗) = max(𝜆
1
𝑛
∑
𝑖=1
RSSIthRSSI (V𝑖
𝑗)
+ 𝜆2
𝑛
∑
𝑖=1
Delayth − Delay (V𝑖
𝑗)
Delayth+ 𝜆3
𝑛
∑
𝑖=1
𝑉bat (V𝑖
𝑗)
𝑉batth) ,
(1)
where RSSI(V𝑖𝑗) is the signal strength of V𝑖
𝑗from the next-
hop node V𝑖−1𝑗
and its value can indirectly reflect the distancebetween V𝑖
𝑗and V𝑖−1
𝑗. RSSIth is the signal strength value at
threshold point 1m which is −45 dBm. Delay(V𝑖𝑗) is the end-
to-end delay between the node and next-hop node; Delayth isthe end-to-end delay thresholdwhich is set to 250ms.𝑉bat(V
𝑖
𝑗)
is the node battery voltage; its value can indirectly reflectthe remaining energy of (V𝑖
𝑗). 𝑉batth is the node threshold
battery voltage and equals 3.5 v. 𝜆1, 𝜆2, and 𝜆
3are the weights
of remaining energy, delay, and distance constraints in thefitness function, respectively, and 𝜆
1+ 𝜆2+ 𝜆3= 1. We set
𝜆1= 0.4, 𝜆
2= 0.2, and 𝜆
3= 0.4. The higher fitness value
indicates the more suitable routing path; therefore, the bestpath 𝑝
𝑏with the optimal fitness fitness(𝑝
𝑏) will be selected.
After that if the number of possible alternative paths inthe set 𝑃all is more than 5, the selection of calculated possiblealternative paths would be so complex, and we would usethe swarm intelligent algorithm to optimize the selection ofoptimal path, or else the optimal path would be calculatedand selected directly. We consider each possible alternativepath 𝑝
𝑗from source V𝑖
𝑠to Vsink as a solution, and the number
of solutions in the population is 𝑚. The details of swarmintelligent algorithm are described in the next section.
3. Design of the Protocol in MSWSN
3.1. The Swarm Intelligent Algorithm of Route MaintainingProtocol. The kernel optimization algorithm of our proposedroute maintaining protocol is a swarm intelligent optimiza-tion algorithm. This optimization algorithm would computeand select the optimized path from source node to thesink among all the possible paths, in which each possiblepath represents a solution. As new avenues in the field ofoptimization, swarm intelligence was defined by Bonabeauas the attempt to design algorithms or distributed problem-solving devices inspired by collective behavior of social insectcolonies and other animal societies [16]. Common examplesrange from flock of birds to colony of bees. The main essenceof swarm intelligence is including the functional behavior ofthese group agents into a practical computational model [13],such as artificial bee colony (ABC) algorithm.
The population-based artificial bee colony (ABC) algo-rithm proposed by Okdem et al. is based on the minimalforaging model of honey bees used in nectar collectionfrom the adjoining environment of their honeycomb [17],with the behavior like self-organization, task allocation,
and communication among the individuals. Its advantagesof fewer parameter settings, faster convergence speed, andhigher convergence precision have attracted the attention ofmany scholars since it was proposed.The algorithm has beensuccessfully applied in the function optimization problem,WSN, and other areas [18, 19]. The tests for several standardfunctions by [14] have demonstrated that the performance ofABC is better than the mainstream optimization algorithms,such as GA, PSO, and DE algorithm.
In the ABC, bees are categorized into three groups:employed bees, onlooker bees, and scout bees. The employedbees search food sources and share the information to recruitthe onlooker bees. The onlooker bees make decision tochoose a food source from those found by the employed beesand search food around it. The food source that has morenectar amount (fitness value) would have a higher probabilityto be selected by onlooker bees. The scout bees are translatedfrom a few employed bees, which discard their food sourcesand randomly search new ones. For a search problem in a𝐷-dimensional space, the position of a food source representsa potential solution. The nectar amount of a food source isthe fitness value of the associate solution. Each food source isexploited by only one employed bee.Thenumber of employedor onlooker bees is equal to the number of solutions in thepopulation.
The ABC can use the positive feedback mechanisms ofoptimized search between bees to effectively speed up theprocess of global optimization and set fewer parameters. Butwhen searching in the near global optimal solution, the searchspeedwould slowdown and the diversity of populationwouldbe reduced. Therefore, we draw on outstanding diversitycharacteristic of immune mechanism and develop the IABCalgorithm. Each solutionwould be considered as an antibody.After the antibody clone and selection step, the better oneis reserved and the worse one is discarded, which wouldincrease its diversity.
is the 𝑖th food source (solution) in the population, and 𝐷 isthe problem dimension size. Each employed bee generates anew food source 𝑉
𝑖around the neighborhood of its previous
food source position as follows:
V𝑖𝑗= 𝑥𝑖𝑗+ 𝜙𝑖𝑗(𝑥𝑖𝑗− 𝑥𝑘𝑗) , (2)
where 𝑖 = 1, 2, 3, . . . , SN, SN is the population size, 𝑗 =
1, 2, . . . , 𝐷 is a random index, and 𝑥𝑘is a randomly selected
solution in the current population (𝑘 = 𝑖). 𝜙𝑖𝑗is a random
number in the range [−1, 1]. If the new Vi is better than itsparent𝑋
𝑖, then 𝑉
𝑖replaces𝑋
𝑖.
After employed bees phase, the probability value Pro𝑖
is calculated according to the food sources fitness as thefollowing:
Pro𝑖=
fitness (𝑥𝑖)
∑SN𝑖=1
fitness (𝑥𝑖)
, (3)
where fitness(𝑥𝑖) is the fitness value of the 𝑖th solution in the
population and Pro𝑖is proportional to fitness(𝑥
𝑖). A better
food source has higher probability to be selected.
6 Mathematical Problems in Engineering
If a food source cannot be improved further over apredefined number of rounds, the food source is abandoned.Assume that the abandoned source is 𝑥
𝑖; the scout bee
randomly searches a new food source to be replaced with 𝑥𝑖.
The operation is defined as follows:
𝑥𝑖𝑗= 𝑥
min𝑗+ Rand (0, 1) (𝑥max
𝑗− 𝑥
min𝑗) , (4)
where Rand(0, 1) is uniformly distributed in the range [0, 1]and [𝑥
min𝑗, 𝑥
max𝑗] is the boundary constraint for the 𝑗th
variable.Then, if the fitness value of the solution is the optimalfitness or the number of iterations increases from zero toGen,the optimal path 𝑝op (solution) would be output, or else go tothe immunization step.
3.1.2. Immunization Mechanism. In this step, each solutionwould be considered as an antibody. The sequence numberSN of solution 𝑥
𝑖arranged in the optimal solution set 𝑋
𝑏is
considered as the affinity of solution SA𝑖, and SA
𝑖= SN.Then,
the clone number CN𝑖is calculated as follows:
CN𝑖= SA𝑖×
𝑁𝑝
⌈∑𝑚
𝑗=1𝑗⌉
, (5)
where 𝑁𝑝is the solution number, m is the size of optimal
solution set 𝑋𝑏, and the total cloned solution number is
Sum = ∑𝑚
𝑗=1CN𝑖. Thus, the cloned solution number is
proportional to the fitness. Then, the solution mutation isused in the clone populations and the mutation rule is asfollows:
CS𝑖= 𝑥𝑖+ 𝛾Rand (0, 1) , (6)
where 𝑥𝑖is the original antibody, 𝛾 represents mutation
factors and 𝛾 = 0.5, and CS𝑖is the clone solution individual.
In the solution restrain rule, we calculate the antigen stimulusdegree of solution in 𝑁(𝑋
𝑏) and the mutation solution. The
Euclidean distance between solution 𝑁(CS𝑡) and antigen
(fitness) fitness(𝑥𝑡) is𝐷(𝑖, 𝑗) = √∑𝑛
𝑖=1(CS𝑖𝑡− fitness(𝑥
𝑖𝑡))2.
Therefore, the stimulus degree of antibody (solution) is
SD (𝑖, 𝑗) = 1
𝐷 (𝑖, 𝑗)=
1
√∑𝑛
𝑖=1(CS𝑖𝑡− fitness (𝑥
𝑖𝑡))2
. (7)
After comparing each solution with the stimulus thresh-old Th, the better solution (SD(𝑖, 𝑗) > Th) would be reservedin the memory cell, and the worse one is discarded. Thedetailed process of the proposed IABC algorithm is shownin Figure 3.
The IABC optimization algorithm is the kernel algorithmof our route maintaining protocol of the MSWSN. To dealwith the path maintaining problem due to the movement ofsink, the proposed algorithm would optimize the path fitnessfunction to provide the fast routing recovery mechanismwith an alternative optimal-fitness path. Apparently, themoresuitable alternative path selected from source to the mobilesink would contain nodes with stronger signal strength, lessend-to-end delay, and higher battery voltage.
3.2. Assumption of Energy Model. The calculation of min-imum energy consumption emphasizing the effect of dis-tances will be as in (8) and (9) expressing sum of theenergy consumptions of network [9]. The abbreviations 𝑗,Ene𝑗, EneTX, EneRX, Eneelec, Eneamp, 𝑘, and 𝑑 in (9) are
node index, energy consumption of the 𝑗th node, transmitenergy, receive energy, radio electronics parameter, transmitamplifier parameter, number of bits of the transmitting data,and distance value between 𝑗th node and next-hop node,respectively:
Ene𝑗≥ (EneRX + EneTX) (8)
EneRX = Eneelec ⋅ 𝑘,
EneTX = (Eneelec + Eneamp⋅ 𝑑2
𝑗) ⋅ 𝑘.
(9)
Using this method, the total energy consumption of thedata transmission and executing the proposed IABC perround can be calculated in the simulation.
4.1. Model and Assumption. Our system uses MATLAB2008a to simulate and evaluate the performance of the pro-tocol. The experimental hardware environments are Intel i7-4600M, 2.90GHz CPU, and 4GBmemory, and the operatingsystem is MS windows 7. The whole MSWSN is simulatedin the area of 3500m × 3500m. The field is static and200∼300 sensor nodes are deployed uniformly in which10% sensor nodes are source nodes. The sensor nodes arehomogeneous and have the same initial energy of 120 J. Theircommunication radius is 300m. This experiment comparesour protocol with TTDD, ERTLD routing protocols forMSWSN. The purpose of the simulation is to illustrate thatour protocol could provide a more robust and efficienttransmission environment.
The other network environment parameters are as fol-lows: One mobile sink moves in the network and its speedis 3∼6m/s, the source node delivers packets at the rate of20 data packets per round, with 10 KB of each packet size,and the simulation lasts for 600 rounds. In the energy model,Eneelec = 40 nJ/bit and Eneamp
= 60 nJ/bit. The sink isassumed to provide sufficient energy to receive data fromnodes and operate our protocol. The values of parametersused for the IABC are function dimension 𝐷 = 20 anditerated generation Gen = 150.
A snapshot from the source node to the mobile sink dur-ing the network simulation is shown in Figure 4. We can seethat when the sink moves from A to a new coordinate B, thesource node immediately establishes an optimal alternativepath (path 1-2-3-8-9-10-11-B in Figure 4) to reach the sink, soas to replace the previous broken path (path 1-2-3-4-5-6-7-Ain Figure 4).
4.2. Evaluation of the Experimental Results. Theperformancemetrics used for the comparison are packet delivery ratio
Mathematical Problems in Engineering 7
T < Gen?
Calculate the individual fitnessvalue, the better half as employed
bees, the other half as onlooker bees
Onlooker bees search and generatenew onlooker bee population
Ends
Combine employed and onlookerbees to form iterative population
Solution immune clone, mutation
Initialize parameter and generate initialpopulation
Routing maintaining operation
Employed bees search and generatenew employed bee population
Generate scout bees?Yes
Scout bees search andupdate iterative population
No
No
Output optimal solution (optimal path)
Solution selection
Solution restrain
Possible paths number >5?No
Yes
Calculate fitness ofpossible paths
Yes
T = T + 1
Figure 3: The flowchart of the IABC algorithm for route maintaining.
Figure 5: Comparison of packet delivery ratio between the three protocols with different node number: (a) 200 sensor nodes and (b) 300sensor nodes.
(the ratio between the successfully received data packetsat the sink and the successfully sent data packets by thesource node), energy expenditure ratio (the ratio betweenthe consumed energy of nodes and initial energy of thesenodes), and average end-to-end delay (the difference betweenthe time a packet is received by the sink and the time it wasoriginally sent by the source node). The results of packetdelivery ratio and energy expenditure ratio are normalized,which is helpful to compare their performances.
In terms of packet delivery ratio in Figures 5(a) and 5(b),it can be clearly seen that the delivery ratio for all protocolsdrops as the number of simulation rounds is increased.This isbecause the mobility of sink affects the quality of the selectedlinks of the path; the process of links repair and loss ofpacket would reduce the packet delivery ratio. Among theselected protocols, IABC has the highest packet delivery rateand ERTLD achieves the second one. This is due to severalfactors: data traffic is routed along shorter paths by usingconcentric ring mechanism and IABC optimizing algorithm,whichwould reduce the packet loss rate; the intelligent swarmoptimization mechanism keeps the route update faster andmore efficiently, with its fast responsiveness to the changingsink position.
In our second experiment presented in Figures 6(a) and6(b), we vary the velocity of a mobile sink from 3m/s to6m/s. The energy expenditure ratios of all protocols areincreased with an increasing number of simulation rounds.This ratio also increases as the mobile sink moves faster,because the change of the frequent topology will incurheavier communication overhead. IABC has lower energyexpenditure, followed by ERTLD and finally TTDD. IABCperforms slightly better due to its concentric ringmechanism,
which can guide the route researching direction and reducecommunication overhead of nodes, and also due to theintelligent routing optimization mechanism which enablesfaster recovery of routes and reduces the energy consumptionof protocol. The delivery rate trend in the case of highervelocity of mobile sink is also as expected, dropping withhigher velocities. This is due to the fact that nodes wouldconsume more energy to search the route to the sink whenthey moves faster away from their transmission radius.Notably, the appropriate speed of the sink will be needed forall protocols, which can better reflect the performance of therouting protocols.
In the third experiment, the smaller average end-to-enddelay means the faster data transmission. We can observein Figures 7(a) and 7(b) that the IABC outperforms theERTLD and TTDD in terms of average delay. It can beexplained by the faster node communication routing andshorter alternative path selection of the proposed IABC forroute maintaining, which can balance the network trafficload and prolong the network lifetime. The delay curvevalue of IABC almost has not changed as the number ofsimulation rounds increases, because our protocol wouldalways choose the optimal path with the stable routing delay.Comparing with Figures 7(a) and 7(b), as the more sensornodes number indicates the longer delivery length of the path,thus the advantage of our IABC with the better alternativepath selection is demonstrated more obviously, which meansthat our routing protocol is more suitable for deploying in thelarge scale networks with mobile sink.
In summary, these experiments demonstrate clearly therouting optimization ability of IABC and its intelligentoptimization mechanism to quickly identify routes to mobile
Mathematical Problems in Engineering 9
0.25
0.2
0.15
0.1
0.05
0120 240 360 480 600
Number of rounds
Ener
gy ex
pend
iture
ratio
(nor
mal
ized
)
IABCERTLDTTDD
(a) Sink speed 3m/s
0.25
0.2
0.15
0.1
0.05
0120 240 360 480 600
Number of rounds
Ener
gy ex
pend
iture
ratio
(nor
mal
ized
)
IABCERTLDTTDD
(b) Sink speed 6m/s
Figure 6: Comparison of energy expenditure between the three protocols as a function of different node speeds: (a) sink speed 3m/s and (b)sink speed 6m/s.
150 200 250 300 350 400 450 500 550 6000.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
Number of rounds
End-
to-e
nd d
elay
(s)
IABCERTLDTTDD
(a) 200 sensor nodes
150 200 250 300 350 400 450 500 550 6000.6
End-
to-e
nd d
elay
(s)
0.8
1
1.2
1.4
1.6
1.8
2
2.2
Number of rounds
IABCERTLDTTDD
(b) 300 sensor nodes
Figure 7: Comparison of end-to-end delay between the three protocols with different node number: (a) 200 sensor nodes and (b) 300 sensornodes.
sink. Compared to the other routing protocols, it consumessignificantly less energy, needs less end-to-end delay, andachieves considerably higher delivery rates.
5. Conclusion
This study presents a novel route maintaining protocol basedon the IABC for the MSWSN. In the proposed protocol,
the concentric ring mechanism is utilized to guide the routeresearching direction, and the optimal routing selection isadopted to preserve the data delivery route in the network.Using the immune based artificial bee colony (IABC) algo-rithm to optimize the forwarding path, the protocol couldfind an alternative routing path quickly and efficiently whenthe coordinate of sink is changed in MSWSN. More impor-tantly, this paper demonstrated the applicability and thepotential of IABC algorithm for solving routing optimization
10 Mathematical Problems in Engineering
problems. The results of our extending performance arecompared to the other aforementioned routing protocols interms of energy, packet delivery, and delay. Our proposedroute maintaining protocol could efficiently solve the energyhole problem, balance the network traffic load, and maintainthe network robustness against topology changes.
In the future we will focus on improving the convergenceperformance, reducing the computational complexity of theIABC algorithm, and validating the proposed protocol ondifferent scenarios with various movement trajectories ofmobile sink, and the most important optimization objectiveis maximizing the network lifetime. In addition, the nodeswould have GPS to locate themselves, and the sink maybroadcast its position instead of CR packets to build theconcentric ring mechanism.
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper.
Acknowledgments
This work was supported in part by the National Nat-ural Science Foundation of China (nos. 61401257 and61501282), Shandong Province Young and Middle-AgedScientists Research Awards Fund (nos. BS2013DX021 andBS2013DX019), Shandong Academy Young Scientists FundProject (no. 2013QN037), Shandong Academy of ScienceDoctoral Fund (no. [2012] 58), Shandong Academy BasicResearch Fund, and Shandong Academy of Sciences PilotProject for Science and Technology.
References
[1] H. W. Ding, D. F. Zhao, Y. S. Gao, and N. S. Zhao, “Researchprogress of wireless sensor network,” Experiment Science &Technology, vol. 9, no. 6, pp. 26–29, 2011.
[2] Y. S. Jiang, W. R. Shi, X. G. Wang, and H. B. Li, “A distributedrouting for wireless sensor networks with mobile sink based onthe greedy embedding,” Ad Hoc Networks, vol. 20, pp. 150–162,2014.
[3] A. Jawahar, S. Radha, and S. Vadivelan, “Connectivity-guaranteed hybrid topology management scheme for improv-ing the operational lifetime of 3-dimensional wireless sensornetworks,” International Journal of Distributed Sensor Networks,vol. 2010, Article ID 547368, 11 pages, 2010.
[4] Y. Xue, X. M. Chang, S. M. Zhong, and Y. Zhuang, “An efficientenergy hole alleviating algorithm for wireless sensor networks,”IEEE Transactions on Consumer Electronics, vol. 60, no. 3, pp.347–355, 2014.
[5] J. Wang, X. Q. Yang, Z. Q. Zhang, B. Li, and J.-U. Kim, “Asurvey about routing protocols with mobile sink for wirelesssensor network,” International Journal of Future GenerationCommunication and Networking, vol. 7, no. 5, pp. 221–228, 2014.
[6] K. Karenos and V. Kalogeraki, “Traffic management in sensornetworks with amobile sink,” IEEE Transactions on Parallel andDistributed Systems, vol. 21, no. 10, pp. 1515–1530, 2010.
[7] Y.-F. Hu, Y.-S. Ding, L.-H. Ren, K.-R. Hao, and H. Han, “Anendocrine cooperative particle swarm optimization algorithm
for routing recovery problem of wireless sensor networks withmultiple mobile sinks,” Information Sciences, vol. 300, pp. 100–113, 2015.
[8] A. Forster and A. L. Murphy, “Froms: a failure tolerant andmobility enabled multicast routing paradigm with reinforce-ment learning for WSNs,” Ad Hoc Networks, vol. 9, no. 5, pp.940–965, 2011.
[9] H. Luo, F. Ye, J. Cheng, S. Lu, and L. Zhang, “TTDD: two-tier data dissemination in large-scale wireless sensor networks,”Wireless Networks, vol. 11, no. 1-2, pp. 161–175, 2005.
[10] S. Tian, S. M. Shatz, Y. Yu, and J. Li, “Querying sensornetworks using AdHocmobile devices: a two-layer networkingapproach,” Ad Hoc Networks, vol. 7, no. 5, pp. 1014–1034, 2009.
[11] H. S. Kim, T. F. Abdelzaher, and W. H. Kwon, “Dynamic delay-constrained minimum-energy dissemination in wireless sensornetworks,”ACMTransactions on EmbeddedComputing Systems,vol. 4, no. 3, pp. 679–706, 2005.
[12] A. Ali Ahmed, “An enhanced real-time routing protocol withload distribution for mobile wireless sensor networks,” Com-puter Networks, vol. 57, no. 6, pp. 1459–1473, 2013.
[13] Y. F. Hu, Y. S. Ding, K. R. Hao, L. H. Ren, and H. Han,“An immune orthogonal learning particle swarm optimisationalgorithm for routing recovery of wireless sensor networks withmobile sink,” International Journal of Systems Science, vol. 45,no. 3, pp. 337–350, 2014.
[14] D. Karaboga and B. Akay, “A comparative study of artificial beecolony algorithm,” Applied Mathematics and Computation, vol.214, no. 1, pp. 108–132, 2009.
[15] D.-T. Ho, J. Park, and S. Shimamoto, “Performance evaluationof the PFSC based MAC protocol for WSN employing UAVin rician fading,” in Proceedings of the IEEE Wireless Commu-nications and Networking Conference (WCNC ’11), pp. 55–60,Cancun, Mexico, March 2011.
[16] G. J. Zhang and J. W. Hu, “Adaptive distributed gatewaydiscovery with swarm intelligence in hybrid wireless networks,”International Journal of Future Generation Communication andNetworking, vol. 7, no. 5, pp. 137–150, 2014.
[17] S. Okdem, D. Karaboga, and C. Ozturk, “An application ofwireless sensor network routing based on Artificial Bee Colonyalgorithm,” in Proceedings of the IEEE Congress of EvolutionaryComputation (CEC ’11), pp. 326–330, New Orleans, La, USA,June 2011.
[18] R.-C. Chen, W.-L. Chang, C.-F. Shieh, and C. C. Zou, “Usinghybrid artificial bee colony algorithm to extend wireless sensornetwork lifetime,” in Proceedings of the 3rd International Confer-ence on Innovations in Bio-Inspired Computing and Applications(IBICA ’12), pp. 156–161, IEEE, Kaohsiung, Taiwan, September2012.
[19] M. Saleem, G. A. D. Caro, and M. Farooq, “Swarm intelligencebased routing protocol for wireless sensor networks: surveyand future directions,” Information Sciences, vol. 181, no. 20, pp.4597–4624, 2011.
