Answers are grounded in this paper and cite their sources.
This paper develops an on-board, knowledge-driven decision system for autonomous beam management in massive LEO constellations, in which a bounded local knowledge base and a provably objective-aligned inference rule replace the ground-side optimization loop that current constellations depend on. Massive low-Earth-orbit (LEO) constellations have made orbital spectrum a principal bottleneck on the capacity and service quality of next-generation satellite networks. Letting LEO satellites reuse the pre-assigned geostationary-orbit (GEO) downlink spectrum relieves this scarcity but introduces co-channel interference that must be suppressed. Since a ground user is visible to several LEO satellites and its choice of serving satellite reshapes the beam-pointing geometry, interference can be regulated through beam matching alone once beam power and channel are fixed. On this basis, a spatial-adaptive matching learning (SAML) method is proposed. The user–satellite association is cast as a many-to-one matching game whose exact potential function guarantees that distributed local matching converges to a stable association equivalent to a Nash equilibrium. A quasi-static slot upper bound is then derived in closed form from a joint visibility–Doppler criterion and coupled with the beam-relay overlap window, fixing the hard time budget for service continuity. SAML itself uses only O (1) local interaction and O ( N ) per-slot computation; its asymptotic optimality is proved by Gibbs-measure concentration and refined by a finite-temperature convergence-rate bound tied to the single-slot budget. Simulations show that SAML attains satisfaction within a few percent of a full-information best-response reference, itself a heuristic rather than the exact optimum, sustains stable link-establishment time across relayed slots, meets an end-to-end on-board time budget within the relay overlap window, and allocates resources more fairly than throughput-oriented baselines.
No takes yet. Share an insight, caveat, or question.
Xie et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: