SEES-QL: An Improved Scalable and Energy-Efficient Scheme for WSNs based on Lightweight Q-learning

The energy efficiency of battery-operated sensing devices in IoT is a critical research area that needs further exploration. This paper employs lightweight reinforcement learning to improve energy savings in large-scale heterogeneous WSNs. We introduce SEES-QL (a Scalable and Energy-Efficient Scheme based on Q-Learning), an enhanced version of the zonal SEES protocol, that addresses the issue of frequent data transmission by dynamically adjusting nodes' transmission cycles without the need for a predefined model. In SEES-QL, on/off periods of radio transceivers are regulated based on transmission history and reading importance of each node independently, positively affecting total energy consumption, traffic load, and overall system lifetime. Performance evaluation demonstrates that SEES-QL achieves significant advancements in energy savings and transmission count reduction, leading to a remarkable 41% increase in the overall system lifetime compared to the traditional SEES protocol.