A Mission-Driven Online Learning Approach for Multi-UAV Cooperative Anti-Jamming Communication

Faced with the contradictory issues of external jamming and insufficient spectrum resource utilization in mission-driven UAV communication scenarios, this paper proposes a cooperative anti-jamming online learning approach. Firstly, a cooperative multi- UAV anti-jamming model based on mission completion is constructed. Next, the proposed problem is modeled as a Markov game, where NE is analyzed. We utilize clustering methods, mission assignments, and online learning technology to enhance the communication/deception method for multi-user independent learning. Based on independent Q-learning, a multi-UAV joint optimization cooperative anti-jamming algorithm based on mission-driven (MD-QL) is designed to effectively avoid external jamming and internal interference in dynamic trajectory while considering heterogeneous mission requirements for all UAV clusters. Simulation results show that, compared with the existing algorithm, the MD-QL algorithm achieves better communication effect and mission completion under continuous multiple jamming.