URLLC-Aware Trajectory Plan and Beamforming Design for NOMA-Aided UAV Integrated Sensing, Communication, and Computation Networks
Peng Qin, Yang Fu, Zhi Yu, Jing Zhang, Xiongwen Zhao
Abstract
Next-generation wireless network should complete integrated sensing, communication, and computation (ISCC) functionalities. To unlock the mutual gains of ISCC, non-orthogonal multiple access (NOMA) and unmanned aerial vehicle (UAV) are appealing techniques that can be invoked. In this work, we design a NOMA-aided UAV ISCC network, where UAV is dispatched as a multi-functional aerial node to perform edge computing for users, whilst radiating integrated sensing and communication beam to sense targets and further offload to a fog node. The objective is to maximize the computation throughput by jointly optimizing UAV trajectory planning, beamforming design and computing resource allocation while satisfying sensing quality, power consumption as well as ultra-reliable low latency communications (URLLC) requirements. We then propose a deep reinforcement learning (DRL)-based algorithm named SAC-TPBD to dynamically modify the UAV trajectory and transmission beamformer according to real-time user queue backlogs and URLLC constraint deviations, thereby realizing URLLC awareness. Evaluations demonstrate that compared to the state-of-the-art methods based on convex approximation techniques, the proposed SAC-TPBD achieves comparable computation performance and beampattern gains with higher implementation efficiency, and the queue backlog can be significantly reduced thanks to the URLLC awareness. Moreover, our approach outperforms baseline DRL schemes in terms of convergence and training stability.