STARS: Stability-Aware SFC Orchestration and Associations in LEO Satellite Networks

Low Earth orbit (LEO) satellite networks present critical challenges for security function chain (SFC) orchestra-tion and associations due to rapid topology changes, resource volatility, and heterogeneous service requirements that render conventional SFC optimization approaches ineffective. To tackle this issue, we introduce here STARS, an optimization framework that fundamentally transforms the sequential time-window optimization for SFC orchestration and satellite association through three techniques: 1) Stability-aware regularization that penal-izes configuration changes across time windows, thus reducing handovers by 54% and security function migrations by 33%; 2) Temporal decoupling that leverages solutions from prior time windows as warm-start seeds and dynamic repairing using real-time visibility constraints; and 3) Hierarchical decoupling that separates satellite association and SFC placement into computationally efficient stages, thus reducing time complexity. Through rigorous formulation as a mixed-integer non-linear programming (MINLP) and simulation-based evaluation, STARS achieves a 57% reduction in optimization solution time, a 7% reduction in the load of deployed security function instances, and efficient CPU utilization (9.81% increase) compared to benchmark schemes. STARS delivers these substantial benefits without any degradation in end-to-end delay. Note that the reported performance values are based on our specific system parameter choices and simulation setup and may not be universally representative. The co-design of stability mechanisms and decoupling strategies establishes STARS as a new paradigm for resilient satellite network optimization, balancing optimality, continuity, and computational tractability under high LEO satellite dynamicity.