Resource-Aware On-board Content Caching in Multi-Layer Satellite Edge Networks

Satellite Edge Computing (SEC) is seen as a promising solution to content caching onboard by reducing retrieval latency and improving user experience. Deciding content type, number of copies, and where to cache it in a satellite con-stellation remains challenging, requiring careful consideration of many factors, such as satellite coverage, content popularity, and resource constraints. In this paper, we study resource-aware onboard content caching strategies in a multi-layer hierarchical satellite network that includes Low Earth Orbit (LEO), Medium Earth Orbit (MEO), and Geostationary Orbit (GEO) satellites. The primary objective is to design a cache placement strategy that maximizes resource utilization ratio while determining the optimal number of content copies, all within the context of a time-varying network topology that primarly occurs due to the mobility of non-geostationary (NGSO) satellites. We model a novel proximity-based hierarchical hybrid content popularity model and formulate the problem as an Integer Linear Programming (ILP) problem to utilize the resources in a network. To solve the ILP problem, we propose two algorithms: Greedy based Content Cache Placement (G_CCP) and Simulated Annealing based Content Cache Placement (SA_CCP). Extensive simulations demonstrate that both G_CCP and SA_CCP are near-optimal and outperform the benchmark in terms of cache-hit ratio, resource utilization ratio, and cache fetching duration.