Finite Blocklength Relaying Communication With Unitary Beamforming and Energy Harvesting: Fairness Oriented Design

Energy-efficient wireless communications are very important for the future Internet of Things (IoT). In this article, a finite blocklength relaying network with nonlinear energy harvesting for IoT communications is proposed. A base station (BS) is considered that transfers data and energy to a local node that harvests energy. This node further employs the amplify and forward protocol for relaying information to a remote node. Our goal is to maximize the energy efficiency of the BS from the perspective of fairness. A multiagent deep reinforcement learning algorithm is proposed to arrive at near-optimal transmission and precoding parameters in real time. The case where global channel state information (CSI) for the BS and local node is considered as well as when only partial CSI is available. Numerical results are presented to illustrate the design tradeoffs and verify the performance of the proposed approach.