Network Performance Evaluation of M2M With Self Organising Cluster Head to Sink Mapping

Abstract

In this paper, a machine-to-machine (M2M) networks is arranged hierarchically to support an energy-efficient routing protocol for data transmission from terminal nodes to a sink node via cluster heads in a Wireless Sensor Network (WSN). Network congestion caused by heavy M2M traffic is tackled using load balancing solutions to maintain high levels of network performance. Firstly, a Multilevel Clustering Multiple Sinks (MLCMS) with IPv6 protocol over Low Wireless Personal Area Networks (6LoWPAN) is promoted to prolong network lifetime. Secondly, enhanced network performance is achieved through linear integer-based optimisation. A Self-Organising Cluster Head to Sink Algorithm (SOCHSA) is proposed, hosting Discrete Particle Swarm Optimisation (DPSO) and Genetic Algorithm (GA) as Evolutionary Algorithms (EAs) to solve the network performance optimisation problem. Network Performance is measured based on Key Performance Indicators (KPIs) for load fairness and average residual network energy. The SOCHSA algorithm is tested by two benchmark problems with two and three sinks. DPSO and GA are compared with the Exhaustive Search (ES) algorithm to analyse their performances for each benchmark problem. Both algorithms achieve optimum network performance evaluation values of 108.059 and 108.1686 in the benchmark problems P1 and P2, respectively. Using three sinks under the same simulation settings, the average residual energy is improved by 2% when compared to two sinks. Computational results prove that DPSO outperforms GA regarding complexity and convergence, thus being best suited for a proactive Internet of Things (IoT) network. The proposed mechanism satisfies different network performance requirements of M2M traffic by instant identification and dynamic rerouting.