A Location and Velocity Prediction-Assisted FANET Networking Scheme for Highly Mobile Scenarios

Abstract

The proliferation of flying ad hoc networks (FANETs) enables multiple applications in various scenarios. In order to construct and maintain an effective hierarchical structure in FANETs where mobile nodes proceed at high mobility, we propose a novel FANET clustering algorithm by using the Kalman-filter-predicted location and velocity information. First, we use the Silhouette coefficient to determine the number of clusters and the k-means++ method is utilized to group nodes into clusters. Regarding the external disturbances in highly mobile scenarios, a Kalman filter is used to predict locations and velocities for all nodes. When clustering, the relative speeds together with relative distances are considered, and the previous selected cluster heads (CHs) are utilized to initialize current centroids. Furthermore, we propose two metrics, including the cluster stability and the ratio of changed edges, to evaluate the network performance. Relevant simulation results reveal that our proposal can yield a cumulative distribution function (CDF) of cluster stability values close to the sensor-measurement-based data. Moreover, it can reduce communication overheads significantly.