A REINFORCEMENT LEARNING-BASED ENERGY EFFICIENT AND QOS SUPPORTING MAC PROTOCOL FOR WIRELESS SENSOR NETWORKS
Seong Cheol Kim Hye Yun Kim, Seong Cheol Kim Hye Yun Kim · Journal of Southwest Jiaotong University · 2023
Wireless sensor networks (WSNs) play a critical role in realizing the Internet of Things (IoT). As the applications using WSNs expand, there is a growing demand for Quality of Service (QoS) support and the existing demand for efficient use of energy in sensor nodes to increase the overall network lifetime. Also recently, many studies have been made to use machine learning techniques to solve these problems in WSNs. In this study, we propose a new protocol that uses reinforcement learning techniques while supporting QoS as a receiver-initiated protocol that uses duty cycling to efficiently use the energy of sensor nodes in WSNs. Our proposed protocol extends the lifetime of the entire network by controlling the wake-up cycle of the receiver node by considering the characteristics of packets generated in the network, the residual energy of the node, and the priority characteristics of the packet. Simulation results showed that the proposed mechanism increased the average network lifetime by 11.4% and decreased the packet transmission delay by 8.8% compared to the method without reinforcement learning. Keywords: Wireless Sensor Networks, Energy Efficiency, Quality of Service, Reinforcement Learning, Duty Cycling DOI： https://doi.org/10.35741/issn.0258-2724.58.1.57