Reinforcement Learning Based Signal Quality Aware Handover Scheme for LEO Satellite Communication Networks

With the increasing of Low Earth Orbit (LEO) satellites emission, utilizing existing LEO satellite network systems have lower CAPEX/OPEX than deploying fixed terrestrial network systems in the remote area. Due to the high mobility of LEO satellites, mobility management mechanisms such as handover schemes are the key issue should be settled on leveraging the merits of LEO satellites telecommunications (e.g. lower propagation delay than GEO satellites). In traditional handover schemes, choosing which one is the next-hop satellite for one user is only determined by evaluating some specific criteria in the current state, not guaranteeing long-term and global optimization. To solve this problem, we use the cumulative signal quality that involves the remaining service time and signal quality and propose a Q-Learning based handover scheme. The simulation results show that the proposed scheme improves the overall signal quality and reduce the average handover number of users compared with other handover schemes.