Capacity Analysis of LEO Mega-Constellation Networks

This paper proposes a complete network capacity analysis framework for low-earth-orbit (LEO) mega-constellations, where a network capacity estimation problem considering the link packet loss rate is formulated with the support of a time-variant network topology model and a task distribution model. This problem is solved in two steps. First, without considering the link packet loss rate, an improved fully polynomial-time approximation (IFPTA) algorithm is proposed to provide a sub-optimal solution to the multi-commodity flow (MCF) problem, in which a simpler definition of a commodity is given and proved to be equivalent to the original problem. Second, a Jackson-network-based capacity fallback approach is proposed to control the link packet loss rate below a given threshold. Numerical results illustrate the superiority of the proposed IFPTA algorithm in terms of accuracy and time complexity compared to existing solutions. In addition, the capacity characteristics of mega-constellations are analyzed by utilizing the proposed capacity analysis framework, including the relationship between constellation size and capacity, network capacity bottlenecks, and the influence of task distributions.