Time-Continuous Frequency Allocation for Feeder Links of Mega Constellations with Multi-Antenna Gateway Stations

TL;DR

This work addresses interference mitigation in mega LEO feeder links with multi-antenna gateway stations by recasting the problem as a $K$-coloring task using an adaptive interference threshold. It introduces two time-continuous graph-coloring methods, Generalized Global (GG) and Clique-Based Tabu Search (CTS), with TCFA adaptations to maintain feeder-link stability, and it adds mega-constellation decomposition (CCD and GSCD) plus Vacant Subchannel Utilization (VSU) to boost spectrum efficiency. The approach achieves near-optimal capacity with substantially lower complexity than ILP solvers, demonstrated on the Starlink SX1/SX2 constellation (34396 satellites) where feeder-link failures approach zero and capacity gains are realized. Collectively, the methods offer a scalable, practical pathway for interference management in dense mega-LEO systems and inform deployment considerations for future constellations.

Abstract

With the recent rapid advancement of mega low earth orbit (LEO) satellite constellations, multi-antenna gateway station (MAGS) has emerged as a key enabler to support extremely high system capacity via massive feeder links. However, the densification of both space and ground segment leads to reduced spatial separation between links, posing unprecedented challenges of interference exacerbation. This paper investigates graph coloring-based frequency allocation methods for interference mitigation (IM) of mega LEO systems. We first reveal the characteristics of MAGS interference pattern and formulate the IM problem into a $K$-coloring problem using an adaptive threshold method. Then we propose two tailored graph coloring algorithms, namely Generalized Global (GG) and Clique-Based Tabu Search (CTS), to solve this problem. GG employs a low-complexity greedy conflict avoidance strategy, while CTS leverages the unique clique structure brought by MAGSs to enhance IM performance. Subsequently, we innovatively modify them to achieve time-continuous frequency allocation, which is crucial to ensure the stability of feeder links. Moreover, we further devise two mega constellation decomposition methods to alleviate the complexity burden of satellite operators. Finally, we propose a list coloring-based vacant subchannel utilization method to further improve spectrum efficiency and system capacity. Simulation results on Starlink constellation of the first and second generations with 34396 satellites demonstrate the effectiveness and superiority of the proposed methodology.

Figures (14)

• Figure 1: Interference scenario of a SatCom system with mega constellation and MAGSs.
• Figure 2: Interference comparison of MAGSs and SAGSs.
• Figure 3: Aggregate I/N distributions under different $N_{\rm at}$ without channelization. The vertical dashed line stands for the ITU threshold $I_{\rm th}^{\rm R}=-12.2$ dB.
• Figure 4: Adapitve edge construction of satellite $s_2$ and $s_4$ on interference graph $G(t)$, in which each satellite has its own adaptive threshold. If an edge exists between two vertices, they should use different colors to avoid conflict, i.e., their corresponding satellites should communicate in different subchannels to avoid harmful interference.
• Figure 5: Illustration of Lemma 1, solution structure (\ref{['struct']}) and neighboring solution generation strategy (\ref{['change']}) of CTS algorithm. The red and green vertices of clique 2 are randomly selected and their colors are swapped to generate a neighboring solution, which reduces the number of Intra-GSI conflicts from 2 to 1.

Theorems & Definitions (3)

• Lemma 1
• Proposition 1
• proof