Contents I. Introduction
In the process of UAV ad hoc network communication, the highly dynamic characteristics of node movement result in frequent establishment or disconnection of links between nodes, leading to frequent changes in the network topologyy[1]. Accurate topology prediction plays a crucial role in enabling protocols to adapt better to these topology changes and improve communication efficiency and reliability[2]. Conventional topology prediction methods typically rely on location prediction, which uses the historical position and speed information of each node in the network to predict its future position at a given time based on knowledge of the communication distance between nodes. The topology prediction result is obtained by comparing the relative distances between nodes with the communication distance. However, specific characteristics of the UAV ad hoc network may undermine the performance of this conventional method for topology prediction[3]. First, due to the multi-hop transmission nature of such a network, acquiring information from non-local nodes incurs different transmission delays resulting in asynchrony of historical position and velocity information for each node during topology prediction. Second, positioning errors inherent to individual nodes can lead to inaccuracies in calculating relative distances, resulting in incorrect predictions.
