Intelligent Hierarchical Admission Control for Low-Earth Orbit Satellites Based on Deep Reinforcement Learning

Area of Science:

• Satellite Communications
• Artificial Intelligence
• Resource Management

Background:

• Low-Earth orbit (LEO) satellites face challenges with limited resources and dynamic user demands.
• Optimizing resource allocation and ensuring fair service admission control are critical for LEO satellite networks.

Purpose of the Study:

• To propose an intelligent hierarchical admission control (IHAC) strategy for optimizing resource allocation and service admission in LEO satellite networks.
• To enhance fairness in service admission while efficiently allocating limited satellite spectrum resources.

Main Methods:

• Developed an IHAC strategy combining deep deterministic policy gradient (DDPG) and deep Q network (DQN) algorithms.
• Implemented a hierarchical framework with upper and lower controllers for global and detailed resource management.
• Integrated online decision-making with offline learning for adaptive strategy generation.

Main Results:

• The IHAC strategy significantly improved performance metrics.
• Achieved an average 20.36% increase in accepted services.
• Reduced service drop rate by 17.56% and increased resource fairness by 17.16% on average.

Conclusions:

• The proposed IHAC strategy effectively addresses resource allocation and admission control challenges in LEO satellite networks.
• The DRL-based approach demonstrates superior adaptability and performance compared to traditional methods.
• IHAC ensures fair service admission and efficient resource utilization in dynamic satellite environments.