Smart RL-Route: A Reinforcement Learning Framework for Energy-Efficient and Scalable Routing in Mobile Wireless Sensor Networks

Abstract This Research work presents an adaptive routing protocol, Smart RL-Route, which shall be based on Q-Learning for energy-aware scalable communications in wireless sensor networks. The proposed model applies clustering and chaining routing schemes combined with Q-Learning, whereby optimal routing decisions may dynamically be derived based on prevailing network conditions. Further, apart from conventional performance analysis, sensitivity analysis has been carried out to test how important factors like node density, communication range, and mobility of both sensor nodes and the sink affect the efficiency of energy consumption during routing. Contiki Cooja has been exercised for simulation purposes as a network modelling tool whereas Python was used for the integration of reinforcement learning and subsequent data analyses. In comparison with other available models over different network scenarios, Smart RL-Route enhances packet delivery ratio as well as load balancing alongside prolonging network lifetime by up to 28%. The framework also demonstrates robust adaptability with minimal routing overhead, thereby showcasing its effectiveness when compared to the conventional protocols that are used for scalable and energy-aware applications of WSN, especially in IoT as well as high mobility-sensitive environments.