Network

Channel Assignment in Multi-Radio Wireless Mesh Network UsingElite PSO with varying inertia and Bee Algorithm

Dhamayandhi N*, Dr Visalakshi P***PG Scholar,** Associate Professor,

Department of Computer Science and Engineering,PSG College of TechnologyCoimbatore, India

*[email protected],**[email protected];+919751313566

Abstract

The Wireless Mesh Network (WMN) is an emerging technology to provide broadband

wireless connection for the end users. The performance of WMN can be increased

by exploiting the advantage of multiple radios and multiple channels. Having

multiple radios and multiple channels the capacity of network can be increased.

But the deployment of multiple channels can cause interference in the networks.

The improper assignment of channel leads to performance degradation and also

causes network partition and link failure. The channel assignment problem

involves the assignment of channel to the links in order to reduce the network

interference and also the efficient utilization of channel resources. This

paper aims at assigning orthogonal (non-overlapping) channels to mesh nodes

with available channel and to minimize the co-channel interference in the

network. Heuristic approaches such as Elite PSO and bees algorithm is

formulated to find the optimal solution.

Keywords: Wireless Mesh Network; Channel assignment; Bees algorithm; ElitePSO; Heuristic algorithm

INTRODUCTION

The Wireless Mesh Network (WMN) is a wireless technology for broadband

communication, enterprise networking and other applications. The communication

in WMN takes place in the form of multi-hop fashion. Having multi-hop

1

communication the network achieves higher throughput and use less transmission

power. Unlike ad hoc networks, the nodes in wireless mesh network are

stationary.

A WMN is dynamically self-organized and self-configured, with the nodes in the

network automatically establishing and maintaining mesh connectivity among

themselves. This feature brings many advantages to WMNs such as low up-front

cost, easy network maintenance, robustness, and reliable service coverage. The

nodes are equipped with wireless network interface cards (NICs) which is used

for communication. The radio or NIC are interfaces used for communication. Co-

channel interference is the major factor which affects the performance of WMN.

Even when communication takes place in two different channel interference will

occur this factor is said to be co-channel interference. Because of

interference the capacity reduction of network also happens. Hence, the co-

channel interference is an important issue while assigning channels to the mesh

nodes. It is proposed to use multiple radios as well as multiple channels in

the network in order to reduce network interference and to increase the

throughput in the network [1,5].

The objective of the channel assignment problem is to assign channels to the

radios in the mesh nodes. When two communications takes place simultaneously

and it use orthogonal channels then they get interfere only if they are in the

interference range. But using overlapping channels, the communications get

interfere even if they are not in interference range of each other. In IEEE

802.11 b/a there are only 3 orthogonal channels out of 12 channels. This paper

aims to use orthogonal channel for channel assignment.

Another issue while assigning channel to the radios is connectivity guarantee.

One of the method to assign channel is to assign common channel for all radios.

In this strategy the network connectivity is maximum but the channel

interference will be more. Another method is to assign different channel to

radios. In this case, the interference of channel will get reduce but it may

2

cause network partition or link failure. In order to make sure of connection

guarantee the edge graph coloring technique is used so that the nodes always

have common channels. The channel assignment problem is proved to be NP-hard

and to find the optimized solution in polynomial time is impossible. This paper

aims to use Elite Particle swarm optimization(EPSO) and bees algorithm to solve

channel assignment problem.

LITERATURE SURVEY

The channel assignment problem is proved to be NP-hard[3]. There are three type

of strategies for channel assignment namely static, dynamic and hybrid. In

dynamic channel assignment scheme the channel assigned to radio are changed

frequently. During each packet transmission the channel assigned to radio will

get varied. The delay is of the order of millisecond which is greater than the

packet transmission time(microsecond). Due to frequent switching the dynamic

schemes become unsuitable for mesh network where nodes are sensitive to delay.

Another scheme is static channel assignment. In this scheme the channels are

assigned statically. The channel assigned to radio get varied only when new

nodes are added or when node fails. The hybrid channel assignment approach is

the combination of both static and dynamic channel assignment approach. In this

approach some of the radios are assigned with fixed channel and in some radios

channels are changed frequently. Still there is some delay in this approach.

Hence the more suitable scheme for WMN is static scheme.

Das et al. [1] addressed the various metrics used in static channel assignment.

This paper focus on on minimization of the average and maximum collision domain

sizes.

Raniwala et al. [3] presented algorithm on static channel assignment and

proposed two algorithm namely neighbor partitioning and load-aware channel

assignment. Both algorithm consider traffic load but it was hard to realize in

3

real network. Also this algorithms assign channels greedily and cause ripple

effect.

Das et al. [4] proposed two linear programming models for static channel

assignment scheme but the complexity is more.

Cheng et al. [6] also proposed two algorithms like linear programming and

heuristic algorithm for the static channel assignment scheme. This paper aims

to preserve network topology but may cause decrease in throughput.

Skall et al.[7] presented MesTic algorithm for static channel assignment

scheme. The channels are assigned based on rank assigned to the node.

Cheng et al. [8] presented DPSO algorithm for channel assignment problem but

the algorithm performs only neighbour search and takes long time to reach

solution.

Tang et al. [9] proposed BAR algorithm which considers dynamic traffic. It

explore path which satisfy bandwidth requirement. This paper aims to assign

channel efficiently and efficient routing.

Ramachandran et al.[11] presented BFS channel assignment algorithm on dynamic

strategy. This paper tried to minimize the drawbacks of dynamic strategy but

could not reduce channel switching time.

Marina et al. [13] presented Connected Low Interference Channel Assignment

(CLICA) which is heuristic approach. This algorithm assign channel based on

priority of nodes but the disadvantage is it did not consider different traffic

pattern.

Sridha et al. [16] used genetic algorithm to solve channel assignment problem.

This algorithm did not consider about broadcasting characteristics caused the

algorithm inefficient.

Some of the related works concentrate on both channel assignment and routing

algorithm. But this cause overhead to the network and it cannot contribute to

the application scenario. Hence this paper concentrates on efficient channel

assignment to the networks.

4

MATHEMATICAL MODEL

Network model assumption

1) The wireless Mesh Network contains stationary nodes in which each node is

equipped with Network interface cards.

2) The number of radios, transmission range and interference range of the

network are fixed.

3) The WMN is constructed as undirected graph G=(V,E) where V represent the

nodes and E represent the communication link in the network. The figure 1

represent example network containing 6 nodes.

4) The two links in the network communicate with each other if they are in

the transmission range of each other.

5) The channels used are assumed to be orthogonal, by using orthogonal

channel the communication can take place simultaneously even if the two

links are in interference range.

Figure 1 Network Graph

6) The given graph is converted into matrix M. The matrix M is represented

in the form of Nnode X Nnode

Where Nnode denotes number of nodes in the network. The value in the matrix

is represented as

5

Interference model

The interference model checks whether two links are interfere with each other

or not. It is stated as the two links (i,j) and (u,v) interfere with each other

if any of the following distances di,u , di,v, dj,u, dj,v are less than the

interference range.

Conflict graph model

The conflict graph Gc =( Vc , Ec) can be constructed from network graph where Vc

represents the link in the network graph and Ec ,the edge represents that the

two links are in the interference range of each other. The network graph in the

figure 1 is converted into conflict graph as shown in the figure 2. In the

network graph, the link (a,b) interfere with link (b,f), (b,c), (c,d) hence

there is edges between (ab,bc), (ab,bf), (ab,cd) in the conflict graph.

Figure 2 Conflict Graph

The matrix CG is constructed by using M matrix and interference model. The

size of matrix is Nlink Χ Nlink where link represents the number of link in the

network graph. The value in the CG matrix is represented

6

The link (a,b) in the network becomes node in the conflict graph.

Interface constraint

The interface constraint is stated as the number of channel assigned to each

node should always be less than or equal to the number of radios in it.

Sometimes the channel assignment problem cause greater number of channels to

assign than the number of radios in the node. This is illustrated in the figure

3 where the number of radios is fixed as 2. The number of channel assigned to

node B is 3.

Figure 3 Node B having interface constraint

Channel Assignment Problem

The aim of channel assignment problem is to minimize the interference number in

the network. The interference number indicates the conflict in the network

which causes the decrease in network performance. The problem should satisfy

7

the interface constraint. The channel assignment problem can be formulated

using the equation

Where Ec represents edges in the conflict graph.

ELITE PSO WITH VARIED INERTIA

Particle swarm optimization (PSO) is a population based optimization technique

inspired by social behavior of bird flocking or fish schooling. Each single

solution represents particle. The particles are evaluated using fitness

function. The position of particle is updated using two best values. One is

pbest value, it is the best value of the particle achieved so far. Another one

is gbest value, it is best value achieved by any of the particle. In Elite PSO

the best particle is selected and is duplicated (replaced with worst particle)

to form the next generation. By this the performance of PSO can be increased.

The performance of EPSO can be further increased by using variable inertia. In

this algorithm, instead of assigning constant value to inertia, the dynamic

value is assigned. The equation for varying inertia is described below

Where wmax =0.9 , wmin=0.4, i represent current iteration, imax represent maximum

number of iteration. During each iteration the value of inertia gets varied.

While iteration in progress, the value of inertia gets reduced from 0.9 to 0.4.

Steps

8

1. Initial population is generated randomly and the length of particle is

based on the communication link in the network.

2. Particles are evaluated using fitness function i.e total number of

interference in the network.

3. Set the pbest value of the particle and gbest value

4. The velocity of the particle is updated using the following formula

Where pbestfit , gbestfit, pcurfit represent fitness value of pbest,

gbest and current particle respectively. Initially vmin and vmax are

generated randomly. The value of w=0.5, c1=c2=1.The value r1, r2 are

generated randomly at each iteration.

5. The position of particle is updated using vmin and vmax value where vmin

represent channel and vmax represent vertex.

6. After updating the particle , new population of particle will be

generated. Select the best particle from that population and replace the

best particle with worst particle.

7. Repeat the step from 4 and 6 until the termination condition is reached.

The algorithm will get terminate when interference number becomes 0 or

when reach maximum iteration.

BEES ALGORITHM

The Bees Algorithm is a new population-based search algorithm, first developed

in 2005 by Pham DT etc. and Karaboga.D independently. The algorithm mimics the

food foraging behaviour of swarms of honey bees. In its basic version, the

algorithm performs a kind of neighbourhood search combined with random search

and can be used for optimization problems.

9

Bees are highly organized social insects. The survival of the entire colony

depends on every individual bee. Bees use systematic task segregation among

them to ensure a continued existence of its colony. A very interesting swarm in

nature is honey bee swarm that allocates the tasks dynamically and adapts

itself in response to changes in the environment in a collective intelligent

manner. The honey bees have photographic memories, space-age sensory and

navigation systems, possibly even insight skills, group decision making process

during selection of their new sites, and they perform tasks such as queen and

brood tending, storing, retrieving and distributing honey and pollen,

communication and foraging.

When a bee finds a new food source, upon returning to its hive, it will perform

a waggle dance. The algorithm performs a kind of neighborhood search combined

with random search and can be used for both combinatorial optimization and

functional optimization. For neighborhood selection, the highest fitness bees

are chosen as selected bees. For recruitment, bees are assigned randomly and

their fitness has to be evaluated.

Encoding:The bee is encoded using integer representation. The bee represents

the feasible solution in the channel assignment problem. The feasible solution

should satisfy radio number constraint. The bee i at time t can be represented

by xit = (x1

t, x2t… xm

t), where xkt denotes one of available channels and m is the

number of nodes in the conflict graph Gc. The length is determined by the

communication link in the network. The example for the bee is shown in the

figure 4.

2 1 3 1 2 2 1 3

Figure 4 Encoding

10

Population Intialization: The algorithm needs the following parameter to be

set. Number of scout bees, number of sites selected m out of n visited sites,

number of best sites e out of m selected sites, number of bees recruited for

best e sites nep, number of bees recruited for the other (m-e) selected sites

nsp, number of iteration.

Fitness Function: The main objective of our channel assignment problem is to

minimize the total network interference so that network throughput is

significantly improved. Therefore, the bees algorithm uses the network

interference number as the fitness value directly. Fitness Function can be

denoted as I (f).

Neighbour generation:The neighbour for bee can be generated by randomly picking

any of the vertex and changes its channel to other randomly selected channel.

Steps: The bees algorithm involves the following steps

Step 1. Initialise population with random solutions.

Step 2. Evaluate fitness of the population.

Step 3. While (stopping criterion not met)

//Forming new population.

Step 4. Select sites for neighbourhood search.

Step 5 . Recruit bees for selected sites (more bees for best e sites)

and evaluate fitnesses.

Step 6. Select the fittest bee from each patch.

Step 7. Assign remaining bees to search randomly

and evaluate their fitnesses.

Step 8. End While.

EXPERIMENTAL RESULTS

11

The number of nodes is fixed and network is randomly generated. The

transmission range and interference range are made as constant value. Using

Euclidean distance, the communication link in the network is identified. The

interference model is constructed from the network graph. From graph and

interference model, conflict graph is constructed. The figure 5 represent

network with 20 nodes and it is randomly generated.

Figure 5 Input graph

The algorithms are made to run for 10 times and interference number for each

algorithm is identified and compared.

12

Figure 6 comparision between algorithms

The conflict between three algorithm are depicted using bar graph in the figure

7

13

CONCLUSION

In wireless mesh network, channel assignment is the major factor which affects

the performance of network. The interference in the network reduces throughput

value. In this paper, Elite PSO and bees algorithm are proposed to solve the

channel assignment problem. The performance of EPSO is better compared to DPSO

and tabu search. The performance of bees algorithm is better compared with all

three algorithm and produce better results.

References

[1] A.Das, R. Vijayakumar, S. Roy, Static channel assignment in multiradiomulti-channel 802.11 wireless mesh networks: issues, metrics and algorithms,in: Proceedings of IEEE GLOBECOM, 2006.

14

[2] Yong Ding, Li Xiao, Channel allocation in multi-channel wireless meshnetworks, in: Proceedings of communication network, 2011.

[3] A. Raniwala, K. Gopalan, T. Chiueh, Centralized channel assignment androuting algorithms for multi-channel wireless mesh networks, in: Proceedingsof ACM SIGMOBILE

[4] A.K. Das, H. Das, R. Vijayakumar, S. Roy, Optimization models for fixedchannel assignment in wireless mesh networks with multiple radios, in:Proceedings of IEEE SENCON, 2005, pp. 463–474.

[5] A. P. Subramanian, H. Gupta, and S. R. Das. Minimum interference channelassignment in multi-radio wireless mesh networks, in Fourth Annual IEEECommunications Society Conference on Sensor, Mesh, and Ad Hoc Communicationsand Networks (SECON 2007), 2007.

[6] H. Cheng, N. Xiong, et al., Channel ASSIGNMENT with topology preservationfor multi-radio wireless mesh networks, Journal of Communications 5 (1)(2010) 63–70.

[7] H. Skall, S.K. Das, L. Lenzini, M. Conti, Traffic and interference awarechannel assignment for multi-radio wireless mesh networks, Technical Report,IMT Lucca, 2006.

[8] Hongju Cheng, Naixue Xiong, Athanasios V.Vasilakos, Laurence TianruoYang, Guolong Chen,Xiaofang Zhuang, Nodes organization for channelassignment with topology preservation in multi-radio wireless meshnetworks , Ad Hoc Networks, 2011.

[9] J. Tang, G. Xue, W. Zhang, Interference-aware topology control and QoSrouting in multi-channel wireless mesh networks, in: Proceedings of ACMMOBIHOC’05, 2005, pp. 68–77.

[10] K. Jain, J. Padhye, V. Padmanabhan, L. Qiu, Impact of interference onmulti-hop wireless network performance, in: Proceedings of ACM MOBICOM,September, 2003.

[11] K.N. Ramachandran, E.M. Belding, K. Almeroth, M. Buddhikot, Interference-aware channel assignment in multi-radio wireless mesh network, in:Proceedings of IEEE INFOCOM, 2006, pp. 1–12.

15

[12] M. Gastpar, M. Vetterli, On the capacity of wireless networks: the relaycase, in: Proceedings of IEEE INFOCOM, 2002.

[13] M. Marina, S.R. Das, A topology control approach for utilizing multiplechannels in multi-radio wireless mesh networks, in: Proceedings of IEEEBroadnets’05, 2005, pp. 381–390.

[14] P. Gupta, P.R. Kumar, The capacity of wireless networks, IEEETransactions on Information Theory 46 (2) (2000) 388–404.

[15] R. Draves, J. Padhye, B. Zill, Routing in multi-radio multi-hop wirelessmesh networks, Algorithmica (1997) 114–128.

[16] S. Sridha, J. Guo, S. Jha, Channel assignment in multi-radio wirelessmesh networks: a graph-theoretic approach, in: Proceedings of CommunicationSystems and Networks and Workshops, January 5–10, Bangalore, India, 2009,pp. 1–10.

[17] Xiaofang Zhuang, Hongju Cheng, Naixue Xiong, Channel Assignment in Multi-radio Wireless Networks Based on PSO Algorithm, IEEE,2011.

16