Low-Latency Content Uploading in Spatial-Aerial Integrated Low Altitude Networks

This paper investigates content uploading in spatialaerial integrated low altitude networks (SALAN), where the distributed low altitude terminals (LATs) are mobile and far away from terrestrial communication infrastructures. In the SALAN, LATs can upload their data to a satellite access point through multiple flexibly deployable aerial relays (ARs). To minimize the maximum uploading latency of LATs, we formulate the AR trajectory planning and resource allocation (ARTPRA) problem and solve it by decomposing it into two sub-problems, i.e., an upload rate control (URC) sub-problem, and an AR trajectory planning and frequency band management (ARTPFM) sub-problem. We prove that the optimal solution to the URC sub-problem can be achieved by solving a series of linear programming problems and propose a multi-agent deep reinforcement learning (DRL)-based framework to solve the ARTPFM sub-problem. Extensive experimental results show that our approach achieves very low offline training complexity for the ARTPFM framework and outperforms the benchmark approaches in terms of latency for uploading tasks, empirical success rate, and online computational time for content uploading.