Maximizing SAGIN coverage with joint HAP/UAV placement, power, bandwidth, and beamforming via deep actor-critic with a temperature function-based policy

We address a coverage optimization problem in a space-air-ground integrated network (SAGIN) with high-altitude platforms (HAPs) and unmanned aerial vehicles (UAVs), where the joint optimization of aerial node placement, transmit power control, bandwidth allocation, and beamforming results in a high-dimensional and strongly coupled optimization problem. To address this challenge, we propose a deep actor-critic (DAC)-based framework capable of handling joint continuous control in complex SAGIN environments and adapting to dynamic network conditions. To accelerate training convergence, we incorporate a temperature function that identifies the most frequently used actions and enhances the exploration process of the DAC algorithm. The performance of the proposed algorithm is evaluated through simulations conducted in a realistic network scenario on Jeju Island, South Korea. We compare the performance of the proposed algorithm against benchmarks, including a soft actor-critic (SAC), deep deterministic policy gradient (DDPG), deep actor-critic (DAC), genetic algorithm (GA), and a gradient search (GS). The results demonstrate that our proposed algorithm achieves higher coverage performance and provides more efficient bandwidth utilization, faster convergence, and a slightly improved data rate.