Energy-Efficient Wireless Sensor Networks Optimization using Deep Q-Networks for Smart Energy Applications

Energy efficiency poses a significant difficulty in Wireless Sensor Networks (WSNs), especially within smart energy applications like smart grids, energy monitoring, and predictive maintenance. This study presents an adaptive optimization approach based on reinforcement learning utilizing Deep Q-Networks (DQN) to improve energy efficiency in Wireless Sensor Networks (WSNs) through the dynamic adjustment of transmission power and routing algorithms. The suggested model is trained and assessed utilizing the LEACH (Low-Energy Adaptive Clustering Hierarchy) dataset within a simulated NS-3 and MATLAB environment. The simulation configuration comprises 100 to 500 sensor nodes distributed inside a 100m x 100m region, each with differing energy limitations, utilizing the IEEE 802.15.4(Zigbee) communication protocol. The performance evaluation is carried out under varying traffic loads and network density, contrasting the proposed DQN-based optimization with classic LEACH, GA-based clustering, and Q-learning models. The results indicate a 27.5% enhancement in network longevity relative to LEACH, a 21.8% decrease in energy consumption per node, an 18.3% advancement in packet delivery ratio, and a 16.7% reduction in end-to-end latency, hence providing dependable data transfer. The findings underscore the efficacy of reinforcement learning in prolonging the lifespan of wireless sensor networks and enhancing energy efficiency, rendering it exceptionally appropriate for practical smart energy applications, including power grid monitoring, industrial IoT, and renewable energy management. The proposed DQN-based optimization dynamically adjusts transmission power and routing strategies, resulting in significant energy savings and improved network performance