Bottleneck Link Identification and Capacity Optimization in Satellite Constellations: A Residual Network-Based Framework

Bottleneck identification and optimization have been key challenges in enhancing system capacity and designing satellite networks. This paper presents an efficient method for identifying and optimizing bottleneck links in satellite networks through residual network analysis and the Edmonds-Karp algorithm. By simulating various scenarios with different capacities of feeder links (FL), inter-satellite links (ISL), and user links, the method identifies which links act as bottlenecks in limiting system throughput. Simulations confirm the accuracy of the method. Once bottlenecks are detected, a multi-objective optimization strategy is proposed to adjust link capacities, enhancing overall network capacity without unnecessary resource allocation. The method leverages binary search and augmenting path analysis to provide a low-complexity solution for real-time bottleneck detection and capacity optimization. Demonstrate how bottleneck shifts occur as FL and ISL capacities increase and how optimized adjustments lead to significant improvements in network performance.