Optimal Handover Strategies in LEO Satellite Networks

TL;DR

This paper introduces persistent capacity, a renewal-theoretic ergodic-capacity framework for LEO satellite networks under general handover strategies. By modelling the constellation with a semi-stochastic NBPP on a sphere and using shadowed-Rician fading, it derives a persistent capacity and establishes upper/lower bounds, linking it to the traditional non-persistent capacity. The optimal handover strategy is obtained via a Dinkelbach-type algorithm, yielding a practical, near-optimal rule that orders candidate satellites by a capacity-time trade-off; a simpler MSC-based strategy can closely approximate optimal performance. Numerical results validate the framework, show that non-persistent capacity is a good proxy for short serving times, and demonstrate substantial gains from informed handover decisions, especially with longer serving times and higher elevation angles. The work offers actionable insights for designing high-throughput, scalable LEO communications and provides a foundation for future NTN and air-to-space integration in 6G contexts.

Abstract

Existing theoretical analyses of satellite mega-constellations often rely on restrictive assumptions, such as short serving times, or lack tractability when evaluating realistic handover strategies. Motivated by these limitations, this paper develops a general analytical framework for accurately characterising the ergodic capacity of low Earth orbit (LEO) satellite networks under arbitrary handover strategies. Specifically, we model the transmission link as shadowed-Rician fading and introduce the persistent satellite channel, wherein the channel process is governed by an i.i.d. renewal process under mild assumptions of uncoordinated handover decisions and knowledge of satellite ephemeris and fading parameters. Within this framework, we derive the ergodic capacity (persistent capacity) of the persistent satellite channel using renewal theory and establish its relation to the non-persistent capacity studied in prior work. To address computational challenges, we present closed-form upper and lower bounds on persistent capacity. The optimal handover problem is formulated as a non-linear fractional program, obtaining an explicit decision rule via a variant of Dinkelbach's algorithm. We further demonstrate that a simpler handover strategy maximising serving capacity closely approximates the optimal strategy, providing practical insights for designing high-throughput LEO satellite communication systems.

Figures (4)

• Figure 1: A satellite on an orbit towards the visibility cap of a ground user.
• Figure 2: Heat maps of the serving capacity $\mathsf{C}(\theta,\phi,1)/\mathsf{N}(\theta,\phi,1)$ for ascending satellites. Light grey is the highest serving capacity, dark green is the lowest serving capacity, and blue (and white) is zero capacity (outside the visibility cap). The black dot is the location of the ground user, the triangle marker is the satellite location that achieves the capacity upper bound $\overline{C}_{\rm pers}$, the thin black lines are example orbit trajectories, and the thick black curve is the boundary of the visibility cap $\mathsf{Cap}$.
• Figure 3: Persistent capacity with cap serving times for a range of transmit SNRs.
• Figure 4: Persistent capacity for a range of fixed serving times and a fixed transmit SNR $\gamma \approx 120$ dB.

Theorems & Definitions (18)

• Definition 1: Handover strategy
• Theorem 1: Persistent capacity
• proof
• Remark 1
• Corollary 1: Non-persistent capacity
• Proposition 1
• proof
• Definition 2: Optimal handover strategy
• Theorem 2
• proof