NOMA-Enabled Joint Trajectory and Resource Allocation Optimisation in 6G Low-Altitude Networks: A PPO-Based Approach

Recently, the unmanned aerial vehicle (UAV)assisted mobile edge computing (MEC) system has shown good potential advantages to overcome the limitations in the traditional terrestrial MEC system, such as MEC servers' inability to dynamically deploy, etc. In this paper, we consider a non-orthogonal multiple access (NOMA)-enabled multi-UAV-assisted low-altitude MEC internet of things (IoT) system, where the UAV-assisted MEC servers can dynamically deploy and move close to each IoT user to offer computational services. Based on the system, we aim to minimise the total delay and energy consumption among all IoT users while guaranteeing the quality of service. Specifically, the total delay includes the transmission delay, processing delay, and feedback propagation delay. Meanwhile, the total energy consumption includes the tasks' transmission, processing, and the MEC servers' flying and hovering energy consumption. To achieve the target, we propose a proximal policy optimisation (PPO)-based offloading and transmit power allocation optimisation algorithm in conjunction with optimising the communication and computation resource allocation, and the trajectory optimisation of UAV-assisted MEC servers. Simulation results demonstrate that our proposed algorithm achieves a much lower total service delay and energy consumption among all IoT users than the benchmarks.