Mobility management for terrestrial users is mostlyconcerned with avoiding radio link failure for the edge users wherethe cell boundaries are defined. The problem becomes interestingfor an aerial user experiencing fragmented coverage in the sky andline-of-sight conditions with multiple ground base stations (BSs).For aerial users, mobility management is not only concerned withavoiding link failures but also avoiding unnecessary handoverswhile maintaining extended service availability, especially inup-link communication. The line of sight conditions from anUnmanned Aerial Vehicle (UAV) to multiple neighboring BSs makeit more prone to frequent handovers, leading to control packetoverheads and delays in the communication service. Depending onthe use cases, UAVs require a certain level of service availability,which makes their mobility management a critical task. Thecurrent mobility robustness optimization (MRO) procedure thatadaptively manages handover parameters to avoid unnecessaryhandovers is optimized only for terrestrial users. It needs tobe updated to capture the unique mobility challenges of aerialusers. In this work, we propose two approaches to accomplishthis: 1) A model based service availability-aware MRO wherehandover control parameters, such as handover margin and timeto trigger are tuned to maintain high service availability witha minimum number of handovers, and, 2) A deep Q-networkbased model free approach for decreasing unnecessary handoverswhile maintaining high service availability. Simulation resultsdemonstrate that both the proposed algorithms converge promptlyand increase the service availability by more than 40% while thenumber of handovers is reduced by more than 50% as comparedto traditional approaches.
