MITIGATING ENERGY DISSIPATION IN WIRELESS SENSOR NETWORKS VIA REINFORCEMENT LEARNING

Abstract

The Wireless Sensor Networks (WSNs) are widely used in environmental monitoring, healthcare, and smart farming. However, energy consumption remains critical since battery-powered sensor nodes directly affect network lifetime. The conventional clustering and multi-hop routing algorithms are prone to collapse when used in dynamic environments, resulting in poor energy consumption and frequent node failures. This paper proposes a novel reinforcement learning (RL) routing algorithm based on Q-learning to enhance energy savings in WSNs. The algorithm is dynamic in assigning routes based on node energy, communication distance, and packet size, and adapts in real-time to network changes. It ensures that data transfer is efficient and the load distribution throughout the network is even by updating routing decisions with Q-learning. These simulation outcomes indicate that the suggested algorithm can save up to 25% of energy per round relative to traditional protocols and increase network life to 50 times that of the LEACH protocol