Facing the scenario of cooperative navigation for UAV swarm, a cooperative navigation method combining belief propagation and information filter is proposed. The method can distributively fuse multi-source measurements such as GNSS, inertial measurement and inter-UAV ranging, which effectively improves the absolute positioning performance of UAV swarm. To address the excessive computational complexity of the traditional belief propagation algorithm, information filter is introduced to solve the linear part of the cooperative navigation system. It transforms the multiplication operation of probability distributions into the addition operation of regular parameters, simplifies the sampling and message transmission procedures, and realizes efficient iterative transmission and updating of cooperative messages among clusters. Outdoor flight test results demonstrate that the proposed cooperative navigation method achieves significantly lower computational load than the traditional belief propagation algorithm while maintaining comparable navigation performance.