On User Association in Large-Scale Heterogeneous LEO Satellite Network

TL;DR

This work develops a stochastic-geometry framework to analyze large-scale heterogeneous LEO satellite networks with satellites deployed across multiple altitude tiers. Satellites in each tier are modeled as independent PPPs and a power-adjusting mechanism equalizes the maximum received power across tiers, enabling tractable analysis. Three two-stage association schemes—DbA, PbA, and RbA—are studied, with closed-form-like expressions for association probabilities, downlink coverage, and spectral efficiency under SR, Rayleigh, and non-fading channels; PbA consistently offers the best coverage and spectral efficiency, while RbA and DbA exhibit tradeoffs as satellite density changes. The results yield practical guidance on optimal satellite counts and highlight how fading, altitude, and interference shape performance, contributing to the design of scalable multi-tier LEO constellations for 6G and IoT applications.

Abstract

In this paper, we investigate the performance of large-scale heterogeneous low Earth orbit (LEO) satellite networks in the context of three association schemes. In contrast to existing studies, where single-tier LEO satellite-based network deployments are considered, the developed framework captures the heterogeneous nature of real-world satellite network deployments. More specifically, we propose an analytical framework to evaluate the performance of multi-tier LEO satellite-based networks, where the locations of LEO satellites are approximated as points of independent Poisson point processes, with different density, transmit power, and altitude. We propose three association schemes for the considered network topology based on: 1) the Euclidean distance, 2) the average received power, and 3) a random selection. By using stochastic geometry tools, analytical expressions for the association probability, the downlink coverage probability, as well as the spectral efficiency are derived for each association scheme, where the interference is considered. Moreover, we assess the achieved network performance under several different fading environments, including low, typical, and severe fading conditions, namely non-fading, shadowed-Rician and Rayleigh fading channels, respectively. Our results reveal the impact of fading channels on the coverage probability, and illustrate that the average power-based association scheme outperforms in terms of achieved coverage and spectral efficiency performance against the other two association policies. Furthermore, we highlight the impact of the proposed association schemes and the network topology on the optimal number of LEO satellites, providing guidance for the planning of multi-tier LEO satellite-based networks in order to enhance network performance.

Figures (6)

• Figure 1: Network topology of a $K$-tier LEO satellites system.
• Figure 2: Coverage probability achieved by each association scheme, in the Shadow Rician, the Rayleigh and the non-fading scenarios.
• Figure 3: Coverage probability achieved by each association scheme versus the SINR threshold for different LEO satellites' altitudes.
• Figure 4: The coverage probability achieved by each association schemes versus the average number of LEO satellites, where $\beta=10$ dB and $N_1=N_2=N_3=N$.
• Figure 5: The spectral efficiency versus the transmit power of LEO satellites, where $P_2=\frac{H_2^{\alpha}G_{m,1}^L}{H_1^{\alpha}G_{m,2}^L}P_1$ and $P_3=\frac{H_3^{\alpha}G_{m,1}^L}{H_1^{\alpha}G_{m,3}^L}P_1$.

Theorems & Definitions (26)

• Lemma 1
• proof
• Lemma 2
• proof
• Lemma 3
• proof
• Lemma 4
• proof
• Lemma 5
• proof