Demand-Aware Inter-Layer Link Selection for Multi-Layer LEO Satellite Networks

Abstract

Multi-layer low Earth orbit (LEO) satellite networks improve connectivity and cross-layer forwarding capability by introducing inter-layer links (ILLs). However, under a given ILL number matrix, selecting the specific satellite pairs for ILL establishment remains insufficiently investigated, especially under uneven ground traffic demand. To address this endpoint selection problem, a demand-aware ILL selection algorithm is proposed. The algorithm first estimates lower-layer satellite access demand from ground population distribution and selects satellites covering high-demand regions as border satellites. It then chooses upper-layer visible satellites by jointly considering moving-direction consistency and angular proximity. Using a border gateway protocol (BGP)-inspired hierarchical abstraction, each satellite layer is treated as a forwarding domain, and selected ILL endpoints serve as border satellites for domain-based crosslayer forwarding. Simulation results show that, under the same ILL number constraint, the proposed method reduces the steadystate average packet loss rate from $\mathbf{1 5. 1 0 \%}$ to $\mathbf{1 2. 8 3 \%}$, satellitesegment delay from 0.1382 s to 0.1263 s, average hop count from 31.20 to 28.04, and average intra-layer hop count from 29.37 to 26.18, compared with random ILL selection.