An Actor-Critic Approach for Resource Allocation in Energy Harvesting-Powered Wireless Body Area Network

Wireless body area network (WBAN) is a kind of network that provides continuous monitoring of health parameters. One of the critical constraints for satisfying the quality of service (QoS) in WBAN is the limited energy of the sensors implanted in the human body. Energy harvesting (EH)- powered WBAN is a paradigm that collects the power from surroundings to overcome the energy constraint of the sensors. However, the heterogeneous and time-varying nature of the EH states of the sensors need to be optimized while learning the optimal resource allocation strategy. In this paper, we propose an actor-critic (AC) deep reinforcement learning (DRL) that optimizes the transmission mode, relay selection, transmission power, and time slots with the energy constraint of the sensors to improve the energy efficiency of the EH-enabled WBAN. The simulation results show that the proposed AC technique can efficiently learn the optimal resource allocation policy and achieve a performance improvement in average delivery probability and energy efficiency.