Optimal Scheduling of Reliability-Constrained Relaying System Under Outdated CSI in the Finite Blocklength Regime

Under the assumption of outdated channel state information at the source, we maximize the finite blocklength throughput of a two-hop relaying system while guaranteeing a reliability constraint. We investigate the tradeoff between the choice of so-called scheduling weights to avoid transmission errors and the resulting coding rate. We show that the corresponding maximization of the throughput can be solved efficiently by iterative algorithms that require a recomputation of the scheduling weights prior to each transmission. Thus, we also study heuristics relying on choosing the scheduling weights only once. Through numerical analysis, we first provide insights to the structure of the throughput under different scheduling weights and channel correlation coefficients. We then turn to the comparison of the optimal scheduling with the heuristic and show that the performance gap between them is only significant for relaying systems with high average signal-to-noise ratios on the backhaul and relaying link. In particular, the optimal scheduling scheme provides most value in case that the data transmission is subject to strict reliability constraints, justifying the significant additional computational burden.