A Time‐Slotted Dynamic Graph Model for Handover Management in LEO Satellite Networks

ABSTRACT Frequent handovers, driven by the rapid movement of low Earth orbit (LEO) satellites, pose a significant challenge to satellite network management. While the multi‐coverage nature of LEO constellations provides multiple candidate satellites for handovers, existing directed graph‐based handover algorithms fail to effectively incorporate dynamic network resources. To overcome this limitation, this paper proposes a time‐slotted dynamic handover graph model. The proposed approach segments the entire access process into discrete time slots and constructs a directed graph for each slot. A virtual reservation mechanism is introduced to handle dynamic resources, such as channel capacity, by assigning virtual values for approximate weight calculation. The link weight comprehensively considers the satellite‐ground distance, remaining coverage time, and available channel capacity. The optimal handover sequence within a slot is then determined by finding the best path using the Dijkstra algorithm. Furthermore, the selection of the time slot length is formulated as an optimization problem that comprehensively considers the handover failure rate, handover count, graph construction computational overhead, available channel fairness index, and handover elevation angle. To resolve this one‐dimensional optimization problem with high efficiency and low computational overhead, we employ the Golden Section Search (GSS) algorithm to pinpoint the optimal slot length. Simulation results under both the Walker Delta and Walker Star constellations demonstrate that the proposed algorithm significantly outperforms baseline algorithms. It effectively reduces the satellite handover failure rate and enhances load balancing while maintaining competitive performance in handover overhead and link quality.