ENERGY-EFFICIENT ROUTING PROTOCOL FOR WIRELESS SENSOR NETWORKS USING MACHINE LEARNING

Abstract

Wireless Sensor Networks (WSNs) play an increasingly central role in environmental monitoring, industrial automation, agricultural sensing, and many emerging IoT systems. Due to strict energy limitations of sensor nodes, the design of energy-efficient routing protocols remains one of the most persistent challenges in this field. Traditional routing schemes such as LEACH, HEED, and PEGASIS rely on static rules and often fail to adapt to dynamic network conditions, leading to imbalanced energy consumption and premature node failures. This paper introduces the Machine Learning–based Energy Efficient Routing Protocol (ML-EERP), a hybrid scheme that combines fuzzy-logic-based cluster-head (CH) selection with Q-learning–based multi-hop routing optimization. The fuzzy system incorporates residual energy, node density, link quality, and distance to the base station into a unified CH decision-making process. Once clusters are established, nodes apply Q-learning to gradually learn optimal next-hop routes that reduce transmission energy while maintaining reliability. Through extensive simulations conducted in MATLAB and Python using the first-order radio energy model, ML-EERP demonstrates significantly superior performance compared to LEACH and HEED. The proposed protocol extends the network lifetime, lowers overall energy consumption, and increases packet delivery ratio.