Energy-Efficient WSNs by Integrating DQL with LEACH Protocol

Abstract

 Longer life of wireless sensor networks is achieved with energy saving methods for advanced applications. The up gradation in low energy adaptive clustering hierarchy (LEACH) protocol is proposed in this paper with the use of deep Q-learning (DQL) which becomes adaptive network for saving energy and increasing life. The most popular machine learning strategy that make use of Support Vector Machine is compared with the proposed method which shows improved decision making ability of the protocol. The selection of cluster heads is improved with consideration of changing conditions of environment, residual energy, along with topological combinations. For larger area networks, DQL shows improved performance with increasing lifetime of the network. The comparative with state of the art methods shows DQL based approach shows better performance in terms of not only extended lifetime but also optimum throughput, packet delivery ratio and end to end delay.