Distributed Reinforcement Learning Algorithm for Energy Harvesting Sensor Networks

In this paper, a distributed reinforcement-learning (RL) algorithm is proposed for power control and data scheduling in energy-harvesting (EH) multi-hop wireless-sensor-networks (WSNs). The WSN consists of M EH sensor nodes aiming to transmit their data to a sink node with minimum delay. Each sensor node has a battery of limited capacity to save the harvested energy and a buffer of limited size to store both the sensed and relayed data from neighboring nodes. A state-action-reward-state-action (SARSA) based distributed algorithm is proposed. The proposed distributed-SARSA (D-SARSA) algorithm adaptively changes the transmitted data and power control at each sensor node according to the state information such that the data of all sensor nodes are received at the sink node with minimum delay. Simulation results demonstrate the merits of the proposed algorithm.