Quota-driven flow allocation and packing scheduling for high-value data routing in dynamic satellite networks

Large-scale satellite constellation networks achieve efficient and fair resource allocation through resource scheduling, providing critical technical support for global communication services. However, existing integrated precise modeling methods suffer from high computational complexity due to combinatorial explosion, making it difficult to meet real-time requirements; while hierarchical decomposition methods improve efficiency, they sacrifice model fidelity and solution quality, and their performance in dynamic environments is limited by information transmission delays. To address this, we propose a hierarchical approach, which decomposes the original problem into corridor-level macro-flow allocation and micro-level transmission task execution. First, we construct a multi-priority continuous aggregation flow linear programming model to quickly obtain priority-based traffic quotas, thereby reducing computational complexity. Subsequently, we model the execution of transmission tasks as a multi-dimensional multiple-choice knapsack problem with mission-inherited deadline constraints and design a hierarchical hybrid approximation algorithm. This algorithm hierarchically classifies data units based on their value density and applies customized solution strategies for different levels, ensuring high solution efficiency while avoiding information loss and degradation in solution quality. The experimental results show that our proposed method has improved the total completed value and resource utilization by an average of 38.80% and 26.27%, respectively.