Graph-Aware Temporal Encoder Based Service Migration and Resource Allocation in Satellite Networks

TL;DR

This work tackles joint service migration and resource allocation in dynamic satellite networks by modeling the system as a time-varying graph and solving via reinforcement learning. It introduces GATE-HPPO, a unified framework that combines Graph-Aware Temporal Encoding with Hybrid Proximal Policy Optimization to produce discrete migration decisions and continuous resource allocations from spatio-temporal graph representations. The approach captures both spatial connectivity and short-term queue dynamics through a two-layer GCN and a sliding-window temporal encoder, enabling coordinated decisions across ground and flight users. Extensive simulations on a realistic MEO constellation show that GATE-HPPO achieves higher accumulated reward, lower service failure rates, and fewer migrations than baselines, highlighting its practical potential for global, latency-sensitive satellite edge computing. The results demonstrate the importance of integrating structured graph representations and short-term temporal context in orchestrating in-orbit services under mobility and heterogeneity constraints.

Abstract

The rapid expansion of latency-sensitive applications has sparked renewed interest in deploying edge computing capabilities aboard satellite constellations, aiming to achieve truly global and seamless service coverage. On one hand, it is essential to allocate the limited onboard computational and communication resources efficiently to serve geographically distributed users. On the other hand, the dynamic nature of satellite orbits necessitates effective service migration strategies to maintain service continuity and quality as the coverage areas of satellites evolve. We formulate this problem as a spatio-temporal Markov decision process, where satellites, ground users, and flight users are modeled as nodes in a time-varying graph. The node features incorporate queuing dynamics to characterize packet loss probabilities. To solve this problem, we propose a Graph-Aware Temporal Encoder (GATE) that jointly models spatial correlations and temporal dynamics. GATE uses a two-layer graph convolutional network to extract inter-satellite and user dependencies and a temporal convolutional network to capture their short-term evolution, producing unified spatio-temporal representations. The resulting spatial-temporal representations are passed into a Hybrid Proximal Policy Optimization (HPPO) framework. This framework features a multi-head actor that outputs both discrete service migration decisions and continuous resource allocation ratios, along with a critic for value estimation. We conduct extensive simulations involving both persistent and intermittent users distributed across real-world population centers.

Figures (10)

• Figure 1: Seamless satellite service scenarios.
• Figure 2: Service migration procedures.
• Figure 3: Graph-Aware Temporal Encoder for Reinforcement Learning (GATE-RL)
• Figure 4: Performance of the proposed algorithm and the other baselines.
• Figure 5: Per-user reward and service demand during the period.