Adaptive Beam Resource Optimization for Enhanced Service Coverage in Satellite Mega Constellations

Satellite mega constellations have significant advantages in areas such as the global Internet of Things (IoT) and emergency communications, providing round-the-clock and full time service capabilities. In satellite mega constellations, efficient beam management is a key technology for achieving differentiated user service coverage. However, due to the limited system resources and the extremely uneven and time-varying traffic demands across beams in the coverage area, it has become an urgent technical challenge to enhance the service coverage by designing dynamic beam management for satellite mega constellations, including the bandwidth resources and power of the beams. In this paper, a mathematical model for joint allocation of heterogeneous resources across multiple beams is first established. Due to the non-convexity and non-linearity of the problem, we transform the problem into an adaptive two-stage optimisation process, which effectively simplifies the solution space and allows for more precise convergence towards optimal resource allocation. We have identified a novel pattern in the demand for non-uniform services, leading to a clever design detail in our Beam Resource Management algorithm based on Evolutionary Algorithm for Neural Network (BRMEANN) Optimization. This method achieves service coverage for non-uniform services while ensuring fairness by optimizing the adaptive resource allocation and tuning process. Simulation results show that the proposed BRM-EANN algorithm can improve the user service coverage by at least $11.46 \%$ compared to the baseline approach.