Energy efficient clustering based data gathering using hybrid DB-EMGM in distributed sensor networks

Clustering is one of the most useful methods for data gathering in distributed wireless sensor networks (WSNs). In additions, sensors in such environment are energy constrained and generate the huge amount of data due to redundant data transmission and thus reducing lifetime of networks. Moreover, there is need to use efficient clustering techniques for collecting relevant data from nodes to eliminate redundancy. The data gathering using existing clustering algorithms engage some problems like arbitrary shape of clusters, estimating input parameters, detection of noisy points. In this paper, we propose an improved clustering method known as hybrid DB-EMGM that combines both EMGM and DBSCAN methods. The improved method aims to reduce the data traffic inside the WSNs network by minimizing intra cluster distance and, thus, finding location of cluster heads using calculated cluster covariance and mean when forming different clusters. The concept of mobile sink node is utilized for collecting the relevant data from all nodes by visiting every cluster heads. Since the mobility concept of node tries to reduce wireless data transmission by following the shortest path based on cluster heads location. The results of our method show that it is computationally faster than EMGM by reducing intra and inter cluster communication distance and increases the transmission rate to the sink node. The performance of our improved method is validated through results presented.