Joint Dynamic Computation Offloading and Resource Allocation in SCMA-Aided Space–Air–Ground Integrated Network

6G mobile is characterized by space-air-ground integrated network (SAGIN), which consists of satellites, unmanned aerial vehicles (UAVs), and terrestrial nodes. It is envisioned to provide not only wide-ranging wireless connectivity but also extensive computing services. In this paper, we investigate a dynamic sparse code multiple access (SCMA) aided SAGIN and focus on maximizing the energy efficiency of IoT devices through task offloading and resource allocation. The problem involves stochastic optimization with long-term task queue stability constraints. Under the guidance of Lyapunov optimization method, the problem is decomposed into several sub-problems for each time slot and solved separately. To address the coupling between task offloading and resource allocation strategies, we propose the alternating optimization algorithm (TORA-AO). Furthermore, to reduce the number of iterations, we develop a twin delayed deep deterministic policy gradient based task offloading and resource allocation algorithm (TORA-TD3). Simulation results indicate that the proposed algorithms outperform other benchmark solutions in terms of task throughput, energy consumption, energy efficiency, and network stability. The TORA-TD3 algorithm achieved comparable performance compared to the TORA-AO, while reducing the running time by approximately 70%.