An Energy-Efficient Two-Phase Intelligent BAN Routing Algorithm Utilizing Reinforcement Learning

Introduction: Wireless Sensor Networks (WSNs) and Body Area Networks (BANs) confront major issues with mobility, scalability, topology management, and energy consumption. Because these networks are dynamic, traditional routing techniques frequently find it difficult to adjust, which results in wasteful energy use and a shorter network lifespan. Methods: This study introduces an energy-efficient Reinforcement Learning (RL) routing strategy for Body Area Networks (BANs). The presented approach encompasses two distinct phases. The process commences by transmitting data to a designated sink node. The subsequent phase is transmitting this data to a server for additional surveillance. In the initial phase, the optimization of data transmission from sensors affixed to the body is achieved by cycling through the ON and OFF states of each sensor. The implementation of a dynamic approach to determine the duration of the ON state, taking into account the priority of data, leads to a reduction in inefficient energy consumption. The subsequent module of the proposed model employs Q-learning, a model-free reinforcement learning technique, to determine the optimal routing policy for the transmission of dynamic data. Reinforcement Learning methods excel at adjusting to dynamic situations, making them ideal for networks with changing topologies. Reinforcement learning enables routing algorithms to adapt dynamically to node movement, improving efficiency. Results: By taking node energy, computing power, and hop count into account, the RL-based routing technique greatly extends network lifetime. According to experimental results, energy usage is reduced by 60–65% when compared to conventional approaches. Discussion: The suggested RL-based routing method efficiently tackles the dynamic and energysensitive characteristics of Body Area Networks by utilising Q-learning for adaptive path selection. The technology effectively minimises superfluous energy use by dynamically regulating sensor activity according to data priority. Conclusion: This reinforcement learning methodology improves network durability and efficiency, attaining up to 65% energy conservation compared to conventional routing techniques. Its versatility to evolving topologies renders it a promising alternative for future energy-efficient BAN applications.