Retransmission–Throughput Rate Tradeoff for Short-Packet Communications in Industrial IoT

Short-packet communication (SPC) has gained significant attention in Industrial Internet of Things. In SPC systems, retransmissions improve reliability but consume time slots and reduce effective throughput, achieving a tradeoff between retransmission frequency and data rate. Therefore, it is urgent to improve the impact of retransmission on SPC systems. This article investigates a model of an SPC system that incorporates incremental redundancy hybrid automatic repeat request mechanisms. The primary objective is to optimize the tradeoff between retransmission frequency and system throughput across varying operational conditions by determining an optimal block length selection strategy. Under the assumption of known channel distribution information (CDI), an offline optimization approach is utilized to derive the tradeoff solution. To address scenarios with unknown CDI, we propose a motivation-guided Q-learning algorithm, informed by motivation-hygiene theory, to derive the optimal block length policy. Extensive simulation results validate the proposed methods, showing that the learned policies significantly enhance the balance between retransmission count and throughput efficiency within the system.