A Multi-Objective Particle Swarm Optimisation Approach in Interference Mitigation Underlay 5G-Enabled Machine-to-Machine (M2M) Network in Comparsion with Genetic Algorithm (GA) and Simulated Annealing (SA) Algorithms

Abstract

The increase in the number of connected devices has caused a paradigm shift in cellular standards, which has characterized the Long-Term Evolution (LTE) standard. The fifth-generation (5G) standard supports several promising mobile technologies, including Machine-to-Machine (M2M) communication and Device-to-Device (D2D) communication, connecting numerous ubiquitous intelligent devices. M2M communication has found applications across various areas of human activity, such as industrial automation, geological and environmental monitoring, e-health, smart grids, intelligent monitoring, and intelligent transport systems. The deployment of M2M devices underlaid 5G cellular network has introduced new challenges in resource allocation and interference management. Interference due to decreased inter-cell distance and the seamless integration of heterogeneous devices into the 5G cellular network results in degraded Quality of Service (QoS) and network performance. This paper proposes an interference-aware architecture for M2M communication on 5G network to mitigate interference arising from the localization of M2M devices. Furthermore, the proposed interference mitigation scheme will be mathematically modeled as a multi-objective particle swarm optimization (MOPSO) problem, considering the complexity of the fitness function and the trade-off among particles. However, the proposed interference mitigation scheme was bench-marked with Genetic Algorithm (GA) and Simulated Annealing (SA) optimisation algorithms. Simulation results proved that the optimal configuration identification process of the MOPSO has achieved significant improvement in interference reduction from M2M devices relative to GA and SA strategies. Furthermore, the SINR (25.0 dB) and throughput (95.0 Mbps) of the overall network was enhanced with MOPSO in comparison with lower results obtained when GA (SINR 21.5 dB, Throughput 85.0 Mbps) and SA (SINR 18.0 dB, Throughput 78.0) were applied. The improvements achieved by MOPSO shows that the SINR when compared with SA and GA decreased by 28% and 14% respectively. While the throughput achieved by MOPSO when compared with SA and GA showed a decrease by 18% and 10% respectively. Therefore, a balance between optimal M2M separation distance and improved network performance was achieved.