Meta-DRL-Based Joint Beam Hopping and Resource Allocation for Space–Air Cooperative ISAC Systems
Abstract
The implementation of integrated sensing and communication (ISAC) technology in space-air cooperative systems enhances the spectral efficiency and ensures sensing and communication services for different areas. Although numerous studies have explored resource management in ISAC systems, they generally overlook the uneven spatial distributions of service requirements. This paper introduces beam hopping into space-air cooperative ISAC systems to dynamically schedule resources based on the distribution of service requirements, thereby enhancing resource efficiency. To balance sensing and communication performance, we formulate the joint beam hopping and resource allocation design as a multi-objective optimization problem that jointly maximizes the radar mutual information (RMI) and transmission rate. To address the limited adaptability of existing algorithms to diverse sensing and communication requirements across different tasks, we propose a meta-deep reinforcement learning (meta-DRL) based joint beam hopping and resource allocation algorithm, which is capable of learning a universal initial policy enabling rapid adaptation to various task objectives. Numerical results indicate that the proposed algorithm exhibits fast-adaptation capability and outperforms the benchmark algorithms.