Optimizing UAV Relay Systems for URLLC Under Interference Based on TD3 Algorithm

This paper investigates the scenario where unmanned aerial vehicle (UAV) relay systems transmit ultra-reliable and low-latency communication (URLLC) command packets among devices under the interference from fixed-position sources. Constrained by finite blocklength, we optimize the blocklength allocation and location of UAV to minimize the decoding error rate. To optimize this sparse reward problem, an iterative optimization approach is introduced based on the Twin Delayed DDPG (TD3) deep reinforcement learning (DRL) algorithm. The simulation experiments indicate that the proposed iterative algorithm significantly outperforms the TD3, DDPG, and PPO algorithms when used individually, while also approaching the performance of the exhaustive algorithm. This research provides an effective solution to enhance the communication performance of UAV relay systems for short packet transmission in the presence of interference.