Dual‐Timescale–Based Cache and Bandwidth Control Scheme in Integrated Satellite‐Terrestrial Networks

ABSTRACT Recently, the integrated satellite‐terrestrial network (ISTN) system offers a promising solution for extending wireless communication services to remote and underserved areas. To effectively provide a variety of services, it is essential to cache the corresponding services on satellites while dynamically allocating the satellite bandwidth. However, challenges arise due to dynamic ISTN service requests that vary over time and space, as well as limited cache and bandwidth resource capacities on satellites. In this paper, the caching placement and bandwidth allocation algorithms are investigated for satellites. Inspired by the advantages of multiagent proximal policy optimization (MAPPO) and mutual‐interactive Nash bargaining solution (MI‐NBS), we develop a dual‐timescale optimization framework. In the long timescale, the cooperative caching is carried out using the cooperative‐MAPPO method to improve the cache hit rate and reduce the content delivery delay. In the short timescale, the bandwidth allocation algorithm is implemented based on the concept of MI‐NBS to accommodate the dynamic changes in satellite service requirements. Within temporally varying ISTN environments, the major objective of our two‐tier ISTN control approach is to identify effective strategies that enable efficient utilization of satellite cache and bandwidth resources. Simulation results demonstrate that our sequential hierarchical optimization approach can achieve superior ISTN system performance than the reference protocols by exploiting the strengths of the multiagent learning and bargaining methodology.