ONCTR: Optimum Number of Clusters and Transmission Range based Clustering in Wireless Sensor Networks

Author Name(s): Sumedha Sirsikar, Manoj Chandak
Author Email: sirsikarsd@gmail.com

Abstract

The field of Wireless Sensor Network (WSN) is striving for devising ways to minimize energy consumption. Clustering reduces energy consumption and increases scalability along with network lifetime. There is a need to identify appropriate number of clusters to balance traffic in network which is a challenging task for energy efficient WSN. Manually it is difficult to decide number of clusters. Finding optimum number of clusters to minimize energy consumption is the major issue in WSN. Existing algorithms find optimum number of clusters but not optimum transmission range. This paper contributes towards the aforesaid issue by proposing a novel method to find optimum number of clusters and a first attempt to find optimum transmission range. We report a new algorithm, where the number of clusters obtained from proposed method is compared with state-of-the-art methods. Extensive experiments are carried out and result comparison with state-of-the-art approaches demonstrate that our method shows significantly better performance. The analysis reveals that optimum number of clusters obtained by proposed method is less than state-of-the-art method. It is especially suitable for clustering in WSN.

Keywords

Optimal Clustering, Optimum Transmission Range, Self-organized Clustering

Introduction

In a sensor network, cost of communication between two nodes is much more than the cost of computation at each node where nodes are deployed randomly in a region. To provide a cost effective solution for minimizing communication overhead, nodes are grouped into various clusters. The challenges of energy efficient clustering are to find optimum number of nodes in a cluster, optimum number of clusters [1] and optimum position of CH [10][11]. In the design process of WSN, energy is the prime constraint. Optimum number of clusters is one of the ways to reduce energy consumption. However it has been observed that selecting random number of clusters gives non optimal results, increase in the number of clusters result into higher overall energy consumption and less number of clusters means more delay and more energy consumption. Hence there is need to address aforesaid issue. With this foothold,

We are proposing a new method called as Optimal Clustering (OC). Here Self-organized (SO) Clustering is used [17]. The main contributions of this paper are summarized as follows: 1. We presented a Square Root Methodology to find optimum number of clusters from number of deployed sensor nodes. 2. We proposed OC method to find optimum number of clusters and optimum transmission range. 3. We conducted extensive experiment on WSN to obtain optimum number of clusters to verify our proposal and also perform empirical comparison with state-of-the-art algorithms.

The rest of the paper is organized as follows: related work gives overview of various methods for optimal clustering and finding optimal number of clusters in WSN. The Square Root Methodology (SRM) for optimum number of clusters is explained. Then the proposed Optimal Clustering algorithm is described. Experimental results are summarized along with analytical findings and future work

Conclusion

An Optimal Clustering algorithm uses SelfOrganized Clustering based on the attributes like node degree, distance and residual energy. Number of clusters for k-means and Elbow method are obtained and compared with OC method. OC method gives Optimum number of clusters and Optimum transmission ranges. Optimum number of clusters obtained from Square Root Method is approximately equal to square root of total number of sensor nodes. k-means gives random number of clusters which is not optimum and there is need to recalculate position of CH which is not required in OC method. OC performs better in terms of energy and network lifetime. Average Energy Consumption in OC is reduced by 1.23% as compared to k-means and 0.31% as compared to Elbow Method. Similarly, Network lifetime is increased by 18.93% as compared to k-means and 4.62% as compared to Elbow Method. In future, mobile nodes and sink can be introduced in the sensor network to have more energy efficient WSN.

1,542 total views, no views today

Download PDF File

About the author: admin