Short-Packet Communications in Multihop Networks With WET: Performance Analysis and Deep Learning-Aided Optimization

In this paper, we study short-packet communications in multi-hop networks with wireless energy transfer, where relay nodes harvest from power beacons to transmit short packets multiple destinations. It is proposed a novel cooperative beamforming selection (CRS) scheme which incorporates partial and distributed multiuser achieve high-reliable transmission two consecutive hops. A closed-form expression for the average block error rate (BLER) of CRS derived, based on an asymptotic analysis also carried out. To optimal channel uses allocation, formulate fairness end-to-end throughput maximization problem generally NP-hard due non-concavity objective function mixed-integer constraints. solve challenging efficiently, first relax be continuous transform relaxed into equivalent non-convex one, but more tractable form. We then develop low-complexity iterative algorithm relying inner approximation framework convexify parts that converges at least locally solution. Towards real-time settings, design efficient deep convolutional neural network (CNN) multiscale-accumulation connections sub-optimal solution via inference processes. Numerical results are presented verify analytical derivations demonstrate performance improvements over benchmark ones terms BLER, reliability, latency, various settings. Moreover, designed CNN provides lowest root-mean-square compared state-of-the-art learning approaches while CNN-aided optimization estimates accurately allocation low execution time.