Optimal Slot Utilization in IEEE 802.15.4e TSCH Networks using Reinforcement Learning

This paper presents an innovative MAC scheduling algorithm to achieve energy savings in IEEE 802.15.4e TSCH sensor networks leveraging reinforcement learning to determine an optimal number of slots to keep active in order to maintain a specific packet delivery ratio for the network. The goal is to turn off slots in the 802.15.4e TSCH frame that are not highly utilized. Each node determines the slots that should be deactivated based on a threshold $\mathbf{Q}$ value. This scheduling strategy allows the nodes to conserve energy effectively by finding the optimal active timeslots for both transmission and reception. This algorithm we name it as RAST Reinforcement-based Slot utilization Technique) algorithm. Through extensive simulations and evaluations across various network configurations, the RAST algorithm reveals a significant packet delivery ratio improvement as compared to Orchestra utilizing less active slots.