End-to-end latency in large low-Earth-orbit (LEO) constellations is dominated by propagation delay, making total delay roughly proportional to the network diameter, the longest shortest path in hops. Current inter-satellite link (ISL) layouts have rarely been optimized to minimize network diameter while simultaneously satisfying physical and operational constraints, including maximum link distance, line-of-sight, per-satellite hardware limits, and long-term link viability over orbital periods. In this study, the selection and assignment of inter-plane ISLs is formulated as a diameter-minimization problem on a Starlink-inspired Walker-Delta constellation in which each satellite is equipped with two fixed intra-plane links and may activate up to two inter-plane links. Beginning with a feasible baseline, the topology is iteratively refined by a local-search procedure that replaces or reinforces links to shrink the diameter. The resulting ISL configuration meets all geometric and hardware limits, preserves link stability across multiple orbital periods, and yields a sparse, diameter-aware graph with potential for centralized routing capabilities. Simulations demonstrate that the proposed algorithm achieves low worst-case latency without compromising ISL stability, and the trade-off between hop count and long-term link stability is empirically measured for guidance of future LEO network deployments.

翻译：大型低地球轨道（LEO）星座的端到端延迟主要受传播延迟主导，使得总延迟大致与网络直径（即按跳数计算的最长最短路径）成正比。现有的星间链路（ISL）布局很少在满足物理与运行约束的同时优化网络直径的最小化，这些约束包括最大链路距离、视距条件、单星硬件限制以及轨道周期内的长期链路稳定性。本研究将面间星间链路的选择与分配建模为一个直径最小化问题，研究对象为受星链启发的Walker-Delta星座，其中每颗卫星配备两条固定的面内链路，并可激活最多两条面间链路。从一个可行的基线拓扑出发，通过局部搜索过程迭代优化拓扑结构，通过替换或增强链路以缩小网络直径。所得ISL配置满足所有几何与硬件限制，在多个轨道周期内保持链路稳定性，并生成一个稀疏的、具有直径感知特性的图结构，具备支持集中式路由的潜力。仿真结果表明，所提算法在不牺牲ISL稳定性的前提下实现了较低的最坏情况延迟，并通过实证测量了跳数与长期链路稳定性之间的权衡关系，为未来LEO网络部署提供指导。