QUALITY OF SERVICE OPTIMIZATION OF WIRELESS SENSOR

NETWORKS USING A MULTI-OBJECTIVE IMMUNE

CO-EVOLUTIONARY ALGORITHM

Xing-Jia Lu, Zhi-Rong Chen

School of Sciences, Ningbo University of Technology, Ningbo 315016, China

Lin Guo, Yong-Sheng Ding

School of Trafﬁc and Transportation , Ningbo University of Technology, Ningbo 315016, China

College of Information Sciences and Technology, Donghua University, Shanghai 201620, China

Keywords:

Wireless sensor networks, Quality of service, Multi-objective immune co-evolutionary algorithm.

Abstract:

Quality of Service(QoS) is the the performance level of a service offered by the wireless sensor networks

(WSNs) to user, which is an important topic of WSNs. The goal of QoS is to achieve a more deterministic

network behavior. QoS of WSNs is an extension of the multi-objective optimization problem, which is mod-

elled as a optimal model with constraint of network connection. The QoS must satisfy the multi-objectives

such as energy consumption, bandwidth, delay jitter, packer loss rate. In order to search the optimal solution

of the QoS of WSNs, we propose a multi-objective immune co-evolutionary algorithm (MOICEA) for QoS

of WSNs. The MOICEA is inspired from the biological mechanisms of immune systems including clonal

proliferation, hypermutation, co-evolution, immune elimination, and memory mechanism. The afﬁnity be-

tween antibody and antigen is used to measure the optimal set of QoS, and the afﬁnity between antibodies and

antibodies is used to evaluate the diversity of population and to instruct the population evolution process. In

order to examine the effectiveness of the MOICEA, we compare its performance with that of genetic algorithm

(GA) in terms of four objectives while maintaining network connectivity. The experiment results show that

the MOICEA could obtain promising performance in efﬁciently searching optimal solution by comparing with

other approaches

1 INTRODUCTION

Wireless sensor networks (WSNs) have recently at-

tracted much research due to their ability of collecting

data from the environmentand reporting them back to

a sink. Although they were originally driven by mil-

itary applications (I. F. Akyildiz and Melodia, 2005),

WSNs are being applied in many different civilian

applications, such as habitat monitoring, earthquake

observationc (I. F. Akyildiz and Vuran, 2009), ve-

hicle tracking system, and healthcare applications

(I. F.Akyildiz and Wang, 2005). WSNs are composed

of a number of small, autonomous, and energy lim-

ited sensor nodes(P. Baronti, 2007) (Yang and Cao,

2008)(Y. P. Aneja, 2010). Due to the limitation of

battery size and weight, and recharging sensor’s bat-

tery is not easy, many strategies have been proposed

to reduce the energy consumption to prolong the life-

time as long as possible(Chou, 2010)(Y. B. Trkoullari,

2010)(J. Bahi and Mostefaoui, 2008). Energy sav-

ing techniques can generally be classiﬁed into two

categories: Sensor schedule and adjusting sensing

ranges(Marta and Cardei, 2009).

Much research has been carried out on the energy-

awareness of WSNs, especially from the perspec-

tive of energy-efﬁcient routing whose focus is to

ﬁnd the most energy-efﬁcient route given the cur-

rent energy status of each node in the networks with

one objective being prolonging the network’s life-

time(R.Rajagopalan and Varshney, 2009)(W. Xue,

2010). The study purpose of energy-efﬁcient route is

to minimize the global power consumption of WSNs

while keeping its connectivity constraint. In this pa-

per, the Quality of Service problem is modelled by

the optimization constraint problem, which have been

intensively researched by graph theory or set theory,

such as dominating set and connected dominating set.

The dominating set must satisfy the multi-objectives

such as energy consumption, delay jitter, loss packet

rate, and trafﬁc ﬂow. It is also termed as the better

160

Lu X., Chen Z., Guo L. and Ding Y..

QUALITY OF SERVICE OPTIMIZATION OF WIRELESS SENSOR NETWORKS USING A MULTI-OBJECTIVE IMMUNE CO-EVOLUTIONARY

ALGORITHM.

DOI: 10.5220/0003510601600164

In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 160-164

ISBN: 978-989-8425-56-0

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

Quality of Service problem by Reddy(Reddy, 2006)

and Cardei (Y. Yang and Cardei, 2010), and has been

proved to be a NP complete problem. In order to

search the optimal solution of the Quality of Service

problem, we propose a multi-objective immune co-

evolutionary algorithm (MOICEA) for Quality of Ser-

vice optimization of the WSNs. In the MOICEA, the

immune operators, which includes Antibody initial-

ization, Clonal selection, Clonal proliferation, Hyper-

mutation, Immune selection, Recruitment, Immune

update and Termination criterion(R. L. King, 2001).

Antibody initizlization process generate the initial so-

lution of feasible set of population, Clonal selection

is used to select the parent’s population by roulette

method, Clonal proliferation is used to generate a new

population with offsprings. Hypermutation is used to

diversity the search process. Immune selection is con-

sidered as the domain knowledge of Quality of Ser-

vice and to eliminate the inferior ones to keep the

stable population. Immune update is used to store

the feasible solutions and update the population. Ter-

mination criterion is used to judge whether meet the

exit condition. In MOICEA, The afﬁnity between an-

tibody and antigen is used to measure the objective

of quality of networks, and the afﬁnity between an-

tibodies and antibodies is used to evaluate the diver-

sity of population and to instruct the population evo-

lution process. The MOICEA employs an improve-

ment procedure to further minimize the overall en-

ergy consumption, bandwidth allocation, and delay

jitter of the network as much as possible. The main

contributions of this study lie in the follows: Firstly,

the energy consumption, bandwidth, and delay jitter

are regarded as the objective functions of the WSNs,

and the solution of the connection set would be meet

with the constraint of sensor node’s battery capac-

ity and network connectivity. Secondly, an encoding

method of Quality of Service and route information

for each node into an antibody is proposed. Thirdly,

the MOICEA is proposed for solving optimal solution

of Quality of Service in WSNs, and demonstrated its

out-performance over the existing heuristic solutions.

The rest of the paper is organized as follows. Sec-

tion 2 brieﬂy describes the related work in Quality of

Service for the WSNs. The proposed MOICEA for

Quality of Service in the WSNs is presented in Sec-

tion 3. Simulation results of performance compari-

son between the MOICEA, Genetic Algorithm (GA)

in terms of four objectives while maintaining network

connectivity are provided in Section 4. Finally, Sec-

tion 5 presents the conclusion of the whole paper.

2 RELATED WORKS

The general Quality of Service of WSNs is introduced

and pointed out by Reddy, Quality of Service is a

measure of the WSNs of the sensing function and

is subject to a wide range of interpretations due to a

large variety of sensors and applications. The goal is

to have each location in the physical space of inter-

est within the sensing range of at least one sensor. A

survey on Quality of Service in WSNs presented by

Reddy, and the Quality of Service can be classiﬁed in

the following(S. Chen, 1999):

Quality of Service is the performance level of a

service offered by the network to the user. The goal of

Quality of Service provision ing is to achieve a more

deterministic network behavior, so that information

carried by the network can be better deliveredand net-

work resources can be better utilized. A network or a

service provider can offer different kinds of services

to the users. Here, a service can be characterized by

a set of measurable prespeciﬁed service requirements

such as minimum bandwidth, maximum delay, max-

imum delay variance (jitter), and maximum packet

loss rate. After accepting a service request from the

user, the network has to ensure that service require-

ments of the user,As ﬂow are met, as per the agree-

ment, throughout the duration of the ﬂow (a packet

stream from the source to the destination). In other

words, the network has to provide a set of service

guarantees while transporting a ﬂow. After receiving

a service request from the user, the ﬁrst task is to ﬁnd

a suitable loop-free path from the source to the desti-

nation that will have the necessary resources available

to meet the Quality of Service requirements of the de-

sired service. This process is known as Quality of Ser-

vice routing. After ﬁnding a suitable path, a resource

reservation protocol is employed to reserve necessary

resources along that path. Quality of Service guaran-

tees can be provided only with appropriate resource

reservation techniques.

3 MULTI-OBJECTIVE IMMUNE

CO-EVOLUTIONARY

ALGORITHM FOR QUALITY

OF SERVICE

3.1 Network Assumptions

We consider the WSNs investigated here have the fol-

lowing features: The sensor nodes are located in a

two-dimensional space, and the location of each sen-

sor node can be obtained after the deployment. The

QUALITY OF SERVICE OPTIMIZATION OF WIRELESS SENSOR NETWORKS USING A MULTI-OBJECTIVE

IMMUNE CO-EVOLUTIONARY ALGORITHM

161

location information is used for calculating the dis-

tance between two sensor nodes. The sensor uses the

omnidirectional antenna, which means that a sensor

radiates and receives equally in all directions. If a

sensor transmits with a power level: p

= ζ× d

, then

any sensor within the distance d and the power thresh-

old ζ can receive the signal. The path loss exponent

a is between 2 and 4. Suppose there are two nodes

and n

, then the distance between the two nodes

can be calculated by using the Euclidean distance for-

mula k x

− x

k , where and is the location vector of

node x

and x

, respectively. The power threshold ζ is

considered as a constant and can be ignored since the

receivers in the network have the same power thresh-

old. The MOICEA uses transmission power in en-

ergy calculation without considering the transmission

time. Sensor nodes can operate in different initial

power levels, with a lower and an upper bound. This

consequently leads to asymmetric wireless links and

a directed graph. The asymmetry of the communi-

cation links combined with a request for a different

initial power level makes the problem more complex

and renders the topology control problem more chal-

lenging.

3.2 Quality of Service

The Quality of Service problem is the optimization

problem of ﬁnding a optimal route in a given graph.

In this paper, we describe the Quality of Service opti-

mization as follows: Let’s denote V as the set of wire-

less sensor nodes, denote C as route nodes set, and

denote G(V, E) as the hypergraph on V that contains

all possible edges if each node transmits signal. The

edge set E of G is constructed in such a manner that

there is a directed edge from u to v if and only if u

can reach v. In Quality of Service problem, Graph G

is an instance, and the smallest number k such that is

a nodes set C for G of size k. The optimal Quality of

Service can be formulated as the following, Assume

that every vertex has an associated cost of c(v) ,

min

∑

v∈V

c(v)x

s.t. x

+ x

≥ 1 for all {u, v} ∈ E

∈ {0, 1} for all v ∈ E

(1)

Constraint condition 1 denotes covering every

edge of the graph, and Constraint condition 2 denotes

every sensor is either in the route set or not. Graph G

sets a lower bound on the connectivity that a wireless

network can have. The algorithm returns a topology

graph T constructed from G , i.e., T is a hypergraph

of T on V . WSNs should fulﬁll the following con-

nectivity requirement: For any pair of nodes u and v,

if there is a path from u to v in G then there is also a

path from u to v in T.

Since our study is the extension of the Quality of

Service, the route set must satisfy the multi-objective

such as energy consumption, delay jitter, loss packet

rate, and trafﬁc ﬂow, and additional constraint condi-

tion should include network connectivity. The formal

deﬁnition of the problem is given as follows: Given

a set of wireless nodes s, a set of route nodes t, the

sensing ranges r, and corresponding energy consump-

tion e, the route nodes set is to determine the power

assignment of the nodes such that: (1) The induced

directed graph T is strongly connected. (2) The to-

tal energy consumption, bandwidth, and time delay

of the network

∑

i=1

is minimized, where p

denotes

the power assigned to node s

. The optimal multi-

objective as follow:

min : E

min

+ ...c

) (2)

min : Delay jitter

min

+ ...c

) (3)

min : Losspacketsrate

min

+ ...c

) (4)

min : Traf ficflow

min

+ ...c

) (5)

3.3 Multi-objective Immune

Co-evolutionary Algorithm for

Quality of Service

This section starts with a presentation as to how

the Quality of Service optimization is represented by

MOICEA. Then it gives a detailed presentation as to

how each step is designed and implemented for the

Quality of Service following the MOICEA ﬂow in

Fig. 1.

Multi-objective immune co-evolutionary algo-

rithm designed for optimal route selection incorpo-

rates the main immune strategies as follows: (1)

clonal proliferation, (2) hypermutation, (3) afﬁnity

measures, (4) co-evolution, and (5) immune mem-

ory mechanism. These strategies are implemented

as operators, procedures or memory mechanism on

the antibody vector structure to generate new antibod-

ies with diversity and evolve the superiors for opti-

mization. An overview framework of the algorithm

is shown in Algorithm 1. The detailed concepts and

procedures are studied in the following sub sections.

4 SIMULATION RESULTS

In order to examine the effectivenessof the MOICEA,

we compare its performance with that of GA in terms

of total four objectives while maintaining network

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

162

Figure 1: The diagram of the multi-objective immune co-

evolutionary algorithm.

connectivity. The GA is selected because it is an evo-

lutionary algorithm to the Quality of Service and also

popularly selected as a benchmark in Quality of Ser-

vice. To evaluate the performance of the MOICEA,

we design a corresponding simulation scenario upon

the Linux platform. The simulation experiment is

constructed on Red Hat 9.0 operating system with In-

tel Pentium 4 processor (2.4 GHz) and 512 MB RAM.

The simulator is NS2-allonine-2.29,and scripting lan-

guage is TCL. The Awk and GNU Plot are used to

present the simulation results. Before conducting

simulation experiments, we need to conﬁgure the sim-

ulation environment of NS2: 10 to 40 sensor nodes

are randomly deployed on a two-dimensional plane

(1000 × 1000m

). We assume the sensors are ho-

mogeneous and have the same energy (1200mAh),

whose sensing radius are 300 meters and communi-

cation radius are 600 meters. The sensor nodes are

deployed randomly and the largest bandwidth of the

WSNs is 2Mb/s. The nodes of WSNs are located at

coordinates (1000, 1000), which receives the data of

source node. The size of WSNs is the same for dif-

ferent algorithms and the afﬁnity functions (four ob-

jectives) are then measured. Fig. 2 presents the ﬁ-

nal result of the MOICEA how to ﬁnd the optimal

route set and build the route to the sink after 300 runs.

It depicts the route building information with N = 7

and R = 200m. When the source nodes set out data

packets during the ﬁrst run, the pheromone has ac-

cumulated on any nodes. On the route set building,

In Fig.2. The attributes of each link are shown in a

tuple hBandwidth, Delayi, Where Bandwidth repre-

Figure 2: Seven nodes WSNs.

Table 1: Available paths from node 2 to node 5.

No Path Hop Mbps Delay(ms)

1 2–6–5 2 2 9

2 2–6–4–5 3 2 11

3 2–3–4–5 3 4 15

4 2–3–4–6–5 4 3 19

5 2–1–7–6–5 4 2 23

6 2–1–7–6–4–5 5 2 25

sent available bandwidth in Mbps and Delay repre-

sent transmission delay,propagation delay and queu-

ing delay. Suppose a packet-ﬂow from node 2 to node

5 requires a bandwidth guarantee of 4Mbps. Quality

of Service routing searches for a path that has sufﬁ-

cient bandwidth to meet the bandwidth requirement

of the ﬂow.Here 6 paths are available between nodes

2 to 5 are shown in Table 1,Quality of Service routing

selects path 3 (2–3–4–5).

in Fig.3.It depicts the route building information

with N = 200 and R = 200m.The MOICEA can ﬁnd

suitable paths In table 2, as a summery, the WSNs

on the four main objectives are statistically studied

Figure 3: 200 nodes WSNs.

QUALITY OF SERVICE OPTIMIZATION OF WIRELESS SENSOR NETWORKS USING A MULTI-OBJECTIVE

IMMUNE CO-EVOLUTIONARY ALGORITHM

163

Table 2: Four main component comparision with GA,

MOICEA methods.

Network Size (50) GA MOICEA

Energy of consumption (mAh) 1.8 1.8

delay jitter(s) 2.7 2.6

Loss packages 18 13

Trafﬁc ﬂow (Kb/s) 14.1 13.9

in 200 nodes. We develop the energy consump-

tion model to compute the battery consumption in

an hour, each node has one Alkaline Battery with

1200(mAh) capacity.The experiment present results

in energy consumption, delay jitter, Loss packages,

and trafﬁc ﬂow in the two methods.

5 CONCLUSIONS

In this paper, we propose the MOICEA for the Qual-

ity of Service of WSNs. The MOICEA is stud-

ied with the immune operators of clonal prolifera-

tion, hypermutation, co-evolution, immune elimina-

tion, and afﬁnity measure. Based on real scenario,

the experiment presents the promising ability of the

MOICEA. Simulation results have shown that bet-

ter solutions can be obtained by the MOICEA than

GA. The MOICEA also demonstrates its strength in

generating initial route information, fault tolerance,

and robustness. As for future suggestion, the fol-

lowing directions are under the way: Firstly, the ex-

periment’s scale needs to be enriched. The current

scale is difﬁcult to study adequate size of affections.

Secondly, a guided mutation based on the feature of

the optimal objectives is to be investigated, and how

to reduce the computational complexity is also the

next-step when accommodating the above future re-

search plans. Thirdly, it is valuable to incorporate the

MOICEA into the Zigbee protocol and 802.15.4 pro-

tocol, so that the optimal multi-objective can be tried

virtually.

ACKNOWLEDGEMENTS

This paper was sponsored by the NingBo University

of Technology, 2009 research project ”Wireless Sen-

sor Networks on NingBo harbor emergency manage-

ment”; National Natural Science Foundation of China

(Funding No.40901241); Natural Science Foundation

of Zhejiang Provincial(Funding No.Y5090377)

REFERENCES

Chou, C. C. (2010). An efﬁcient algorithm for relay place-

ment in a ring sensor networks. Expert Systems with

Applications.

Akyildiz, I. F. and Melodia, T. (2005). Underwater acoustic

sensor networks: research challenges. Ad Hoc Net-

works.

Akyildiz, I. F. and Vuran, M. C. (2009). Signal propagation

techniques for wireless underground communication

networks. Physical Communication.

Akyildiz, I. F. and Wang, W. (2005). Wireless mesh net-

works: a survey, computer networks. Computer Net-

works.

Bahi, A. M. and Mostefaoui, A. (2008). Hilbert mobile bea-

con for localisation and coverage in sensor networks.

International Journal of Systems Science.

Marta, M. and Cardei, M. (2009). Improved sensor net-

work lifetime with multiple mobile sinks. Pervasive

and Mobile Computing.

Baronti, P. and Pillai, P. and et al (2007). Wireless sen-

sor networks: A survey on the state of the art and the

802.15.4 and zigbee standards. Computer Communi-

cations.

King, R. L. and Russ, S. H. and A. B. L. e. a. (2001). An

artiﬁcial immune system model for intelligent agents.

Future Generation Computer Systems.

Reddy, T. B. (2006). Quality of service provisioning in ad

hoc wireless networks: a survey of issues and solu-

tions. Ad Hoc Networks.

Rajagopalan, R. and Varshney, P. (2009). Connectivity anal-

ysis of wireless sensor networks with regular topolo-

gies in the presence of channel fading. Wireless Com-

munications, IEEE Transactions on.

Chen, S. K. N. (1999). Distributed quality-of-service rout-

ing in ad hoc networks. IEEE Journal on Selected Ar-

eas in Communications.

Xue, W. and Junjie, M. and W. S. e. a. (2010). Distributed

energy optimization for target tracking in wireless

sensor networks. Mobile Computing, IEEE Transac-

tiXueons on.

Trkoullari, Y. B. and Aras, N. and I. K. A. e. a. (2010). An

efﬁcient heuristic for placement, scheduling and rout-

ing in wireless sensor networks. Ad Hoc Networks.

Aneja, Y. P. and Chandrasekaran, R. and X. L. e. a. (2010).

A branch-and-cut algorithm for the strong minimum

energy topology in wireless sensor networks. Euro-

pean Journal of Operational Research.

Yang, Y. and Cardei, M. (2010). Improving network life-

time with mobile wireless sensor networks. Computer

Communications.

Yang, S. and Cao, Y. (2008). Networked control systems

and wireless sensor networks: theories and applica-

tions. International Journal of Systems Science.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

164