Joint optimization of power control and time slot allocation for wireless body area networks via deep reinforcement learning

Abstract

E-healthcare system based on wireless body area network (WBAN) promises to produce potential benefits in health-care industry. A major issue of such an on-body networked system is the energy efficiency, that is, how to improve the reliability and effectiveness of physiological data transmission with the energy constraints of tiny wireless sensors. Motivated by this, we consider an individual WBAN scenario, focusing on finding an adaptive time slot allocation and power control scheme to maximize the average energy efficiency for implementing the task of health monitoring. We formulate the maximization problem with latency and sensors’ energy budget constraints as a markov decision process (MDP). As a solution, we propose a deep reinforcement learning-based scheme to make a sequence decision for the MDP, which jointly optimizes power control and slot allocation. Simulation results show that the proposed scheme is energy efficient and has a good convergence.