Space-air-ground integrated network-empowered mobile edge computing: MASAC-based URLLC-aware energy optimization

To fulfill the flexible, efficient, and random access demands as well as quality of service guarantee requirements of diverse terminals in the new power system, it is imperative to tackle the incomplete communication coverage, weak real-time task response, and device energy limitation issues of the current terrestrial network. In this paper, we propose a space-air-ground integrated network-empowered mobile edge computing system model consisting of low-earth orbit satellites, unmanned aerial vehicles (UAVs), and ground devices. We elaborate on the joint optimization problem of task splitting, task offloading, power control, UAV trajectory planning, and edge computational resource allocation, with the aim of minimizing the long-term terminal energy consumption. However, this problem is intractable due to the coupling between short-term optimization decisions and long-term ultra-reliable and low-latency communication (URLLC) constraints. To this end, we employ Lyapunov optimization to transform the original problem into a series of subproblems that can be solved in a slot-by-slot fashion. We then design a multi-agent soft actor-critic (MASAC)-based URLLC-aware energy optimization algorithm, together with improved reparameterization and activation function configuration tricks, to efficiently solve them. Numerical results verify that the proposed approach outperforms other benchmark methods in reducing system energy consumption whilst guaranteeing URLLC constraints.