Abstract: To address the urgent need for efficient information exchange in UAV swarm emergency rescue scenarios, this paper combines Over-the-Air Computation (OAC) with spectral graph theory and proposes a topology optimization strategy for low-latency and high-consistency communication. The relationship among communication topology, time synchronization accuracy, and consensus convergence of state information is analyzed, and a theoretical model is established to characterize their interactions. A bi-level joint optimization framework is then developed. The upper layer optimizes the network topology based on spectral graph criteria to improve synchronization performance and convergence speed, while the lower layer adaptively adjusts the MAC-layer frame length to reduce end-to-end latency. Experimental results show that the proposed method improves information collaboration efficiency by up to 37% compared with a Deep Q-Network (DQN)-based reinforcement learning scheme, while significantly reducing algorithm execution time. In large-scale swarm scenarios, the proposed approach also achieves lower communication latency than conventional methods, demonstrating its effectiveness for time-critical applications such as disaster emergency response.

Key words: UAV swarm, information consensus, over-the-air computation, topology optimization, spectral graph theory