Predictive-Q Learning Based Interference-and-Mobility Aware Dual-Path Routing for UAV Swarm Networks with Mobile Edge Computing

High-mobility Unmanned Aerial Vehicle (UAV) swarm networks suffer from fast-varying connectivity and interference, and therefore routing decisions must jointly account for link instability and topology changes. By leveraging mobile edge computing (MEC) capabilities, each UAV can perform online routing decisions locally without relying on centralized controllers. This paper develops a Predictive-Q learning framework for dynamic routing under interference and mobility, where the Q-value is trained by a multi-factor reward that explicitly models retransmission costs, predicts link lifetime from relative motion, and anticipates forward connectivity and neighbor redundancy. To further enhance reliability under harsh interference, we extend Predictive-Q with a lightweight dual-path forwarding mechanism. Specifically, it conditionally splits a backup routing when the primary next hop becomes unreliable, and terminates the backup early when continued forwarding is unlikely to be beneficial, thereby controlling overhead. Simulation results demonstrate that, compared with existing methods, the proposed Predictive-Q-Dual improves packet delivery ratio by more than 15% over GPSR under strong interference while maintaining low delay and energy consumption across varying interference intensity, node density, and mobility speed.