Energy-efficient Neuro-fuzzy-based Multi-node Charging Model for WRSNs using Multiple Mobile Charging Vehicles

Abstract

Mobile charging vehicles (MCVs) have played a significant role in advancing Wireless Rechargeable Sensor Networks (WRSNs). Recent research has primarily focused on on-demand charging in WRSNs using MCVs. However, minimal attention has been given to the simultaneous analysis of multiple MCVs with multi-node energy transfer capabilities. Additionally, when making scheduling decisions, most existing methods still need to consider various network parameters, while a few have overlooked the challenge of sensor nodes’ (SNs) poor charging responses to non-uniform energy expenditure rates. This article addresses the issues above and proposes an innovative charging-scheduling Algorithm based on Neuro-fuzzy with PSO model (ANFIS-PSO) to resolve these challenges. To ensure uniform workload allocation among MCVs and prevent interference during the charging process, we deploy MCVs strategically using the Optimal Fuzzy C-Means clustering method. We then optimize MCV visiting points to recharge energy-critical SNs, implying reduced energy consumption and charging delay. We next adopt the ANFIS-PSO model, which blends various network parameters for creating the charging schedules of the MCVs. We devise an expression to find the SNs’ adaptive charging threshold values based on their remaining lifetime. We also formulate another expression called the average charging weight function to assist the ANFIS-PSO model in selecting the next promising SN for recharging. Experimental simulations validates that our method is effective and outperforms baseline methods, with a substantial 11% reduction in charging delay, a 6% improvement in the life-survival ratio, and a 7% increase in energy utilization efficiency across all performance metrics.