Relative reward based scheduling (RRS): An adaptive duty cycle based WSN MAC protocol
Avinash Singh, Kiran Kumar Kaki · 2016
This paper introduces a scheduled based Media Access Protocol for wireless sensor network termed RRS, an adaptive MAC protocol employing Improved Reinforcement Learning algorithm. Existing MAC protocol's work on scheduling the sleep-active cycle of wireless nodes transceivers in order to minimize the ideal energy consumption. More recent protocol employs adaptive duty cycle based on individual node's traffic load as a function to further minimize the energy consumption. RRS actively infer the traffic load of neighboring nodes using a Reinforcement learning algorithm to calculate and compare the current rewards and immediate rewards using variation of Q-learning algorithm called Relative reward based Q-learning. This algorithm increases the convergence rate and thus, reaches optimal state in less number of iterations.