PSO-Based Multiple-sink Placement Algorithm for Protracting the ...

5 downloads 1628 Views 667KB Size Report
Sep 5, 2015 - Sign up / Log in .... Springer International Publishing AG, Part of Springer Science+Business Media Privacy Policy, Disclaimer, General Terms & ...
PSO-Based Multiple-sink Placement Algorithm for Protracting the Lifetime of Wireless Sensor Networks C. Srinivasa Rao P., Haider Banka and Prasanta K. Jana

Abstract Optimal deployment of multiple sinks has been proven to be one of the energy efficient techniques for prolonging the lifetime of wireless sensor networks (WSNs). In this paper, we propose a particle swarm optimization (PSO) based algorithm called PSO-MSPA for placement of multiple-sink in WSNs. The algorithm is developed with an efficient scheme of particle encoding and novel fitness function. For the energy efficiency of the PSO-MSPA, we consider various parameters such as Euclidian distance and hop count from the gateways to the sinks. The algorithm is tested extensively on various scenarios of WSNs by varying number of gateways and sensor nodes and the results are analyzed to show the efficacy of the proposed algorithm. Keywords Multiple-sink placement problem optimization Network lifetime



 NP-hard problem  Particle swarm

1 Introduction Wireless sensor networks (WSNs) are networks of spatially distributed sensor nodes which collect and process the data from a region of interest and communicate to a base station [1]. WSNs have paid enormous attention in wide range of applications such as disaster management, environment monitoring, health, and military [2]. Due to battery operated sensors, energy conservation of the sensor C. Srinivasa Rao P. (&)  H. Banka  P.K. Jana Department of Computer Science and Engineering, Indian School of Mines, Dhanbad 826004, India e-mail: [email protected] H. Banka e-mail: [email protected] P.K. Jana e-mail: [email protected] © Springer India 2016 S.C. Satapathy et al. (eds.), Proceedings of the Second International Conference on Computer and Communication Technologies, Advances in Intelligent Systems and Computing 379, DOI 10.1007/978-81-322-2517-1_58

605

606

C. Srinivasa Rao P. et al.

nodes is the main challenging issue in the development of a large-scale WSN. Clustering sensor nodes is one of the most energy efficient techniques that have been studied rigorously. In a cluster based WSNs, sensor nodes are organized into number of groups called clusters and each cluster has a leader called as cluster head (CHs). The CHs collect the data from each member sensor nodes within their clusters, aggregate them and send the aggregate data to a remote base station called sink [3]. To overcome the network deficiencies researchers have introduced special nodes called gateways which are provisioned with extra energy [4]. These gateways are responsible for data collection, aggregation, and relaying the aggregated data to the sink and thus actually serve the role of CHs. However, they are also battery operated and therefore, we should take care the energy conservation of the gateways. Position of a sink has a tremendous impact on the energy consumption of gateways in the process of data flow between the gateways and the sink [5, 6]. It may also have the influence of the other performance metric, such as delay, risk of packet loss, and throughput of the network. One possible solution is to use of multiple-sink so that the gateways can disseminated their collected data to their nearest sink with less communication cost and less network delay. However, determination of multiple sink locations is still challenging as it considerably affects the routes in some critical applications. Multiple-sink placement problem is NP-hard under multiple objectives [7]. Traditional optimization algorithms are inefficient with the increasing size of the network. Particle Swarm optimization (PSO) is one of the swarm intelligence based algorithms, which is effective for such NP-hard problem due to ease of implementation, quick convergence, and high quality of solution [8]. Various solutions have been proposed in the literature to tackle the multiple-sink placement problems. Pan et al. [9] have proposed two heuristic algorithms to find the optimal position of sinks. One is for homogeneous application nodes (ANs) and other is for heterogeneous ANs. Homogeneous ANs have same initial energy, data rate, and data transmission rates, where as heterogeneous ANs have different initial energy, data rate, and data transmission rates. However, their algorithms are applicable for small-scale networks, as the size of the network increases the problem become NP-hard. In [10], the authors have assumed that nodes are uniformly deployed in an agriculture field. The total network is divided into clusters to ensure the scalability of the network. They have proposed three approaches for placement of sinks. They are best sink location (BSL), Finding the Minimum number of sinks while maximizing the network lifetime (MSMNL) and finally Minimizing the number of sink nodes for a predefined number of operation period (MSPOP. In [11], the authors have formulated the problem using linear programming (LP) and solved using CPLEX (an integer LP solver) tool for maximizing the network lifetime by distributing the load of among deployed sensors. Some meta-heuristic algorithms based on swarm intelligence have proposed. In [12], the authors have proposed allocation of multiple base stations based on PSO. The authors have placed the multiple-sink with respect to different ANs to increase the lifetime of ANs. Haider et al. [13], have proposed Discrete PSO (DPSO) using local search approach to

PSO-Based Multiple-sink Placement Algorithm …

607

solve multiple sink placement problem for minimize the maximum worst case delay and maximize the life of the network. Md Nafees et al. [14], have proposed PSO-based approach for optimal placement of single sink nodes. In this paper, we address the following problem. For a given location of gateways, find the optimized positions of fixed number of sinks using PSO algorithm. The rest of the paper is organized as follows. The preliminaries of PSO, energy model, network model, and terminologies are provided in Sect. 2. The proposed approach and PSO-based algorithm called as PSO-MSPA are provided in Sect. 3. The simulation results are explained in Sect. 4 followed by the conclusion in Sect. 5.

2 Preliminaries 2.1

Overview of PSO

PSO consists of a predefined number of particles say NP, called swarm. Each particle provides a potential solution. A particle Pi, 1 ≤ i ≤ NP has position Xi,d and velocity Vi,d, 1 ≤ d ≤ D in the dth dimension of the search space. The dimension D is same for all particles. A fitness function is used to evaluate each particle for verifying the quality of the solution. In the initialization process of PSO, each particle is assigned with a random position and velocity to move in the search space. During each iteration (generation), each particle finds its own best, i.e., personal best called Pbesti and the global best called Gbest. To reach the global best solution, it uses its personal and global best to update the velocity Vi,d and position Xi,d using the following equations.   Vi;d ðt þ 1Þ ¼ x   Vi;d ðtÞ þ c1   v1  XPbest i;d  Xi;d þ c2  v2  XGbest d  Xi;d

ð1Þ

Xi;d ðt þ 1Þ ¼ Xi;d ðtÞ þ Vi;d ðt þ 1Þ

ð2Þ

where 0 < ω < 1 is the inertia weight, c1, c2, 0 ≤ c1, c2 ≤ 2 are the acceleration coefficients and, χ1, χ2, 0 < χ1, χ2 < 1 are the randomly generated values. The updating process is repeated until it is reached to an acceptable value of Gbest. After getting new updated position, the particle evaluates the fitness function and updates Pbesti as well as Gbest as follows  Pbesti ¼  Gbest ¼

if ðFitnessðPi Þ\Fitnessð Pbesti Þ Pi ; Pbesti ; otherwise

ð3Þ

Pbesti ; if ðFitnessðPbesti Þ\FitnessðGbestÞÞ Gbest; otherwise

ð4Þ

608

2.2

C. Srinivasa Rao P. et al.

Energy Model

The energy model used in this paper is based on the same radio model as used in [3]. The energy consumption of the node depends on the amount of the data and distance to be sent. In this model, energy consumption of a node is proportional to d2when the propagation distance (d) less than the threshold distance d0 otherwise it is proportional to d4 [3]. The total energy consumption of each node in the network for transmitting the l-bit data packet is given by the following equations.  ETX ðl; d Þ ¼

l  Eelec þ l  efs  d 2 ; if d\d0 l  Eelec þ l  emp  d 4 ; if d  d0

ð5Þ

where Eelec the energy is dissipated per bit to run the transmitter or receiver circuit, amplification energy for free space model εfs and for multipath model εfs depends on the transmitter amplifier model and d0 is the threshold transmission distance. In the same way to receive l-bit of data the energy consumed by the receiver is given by ERX ðlÞ ¼ l  Eelec

2.3 2.3.1

ð6Þ

Network Model and Terminologies Network Model

The sensors along with some gateway nodes are randomly deployed throughout the sensing field and a node can compute the distance to the other node based on the received signal strength. All the sensor nodes and gateway nodes are assumed to be stationary after deployment. Each node performs sensing periodically and has always data to send to its CH or BS. In the process of transfer of data from gateways to BS, gateways use multi-hop communication. The communication links are wireless, symmetric, and established between the nodes when they are within the communication range of each other.

2.3.2

Terminologies

For better understanding of the proposed algorithm, we first describe some terminologies as follows. S The set of sensor nodes, i.e, S = {s1, s2,s3, … , sn} G The set of gateway nodes, i.e, G = {g1, g2,g3, … , gm} where, m

Suggest Documents