Software Define Radio in Realizing the Intruding UAS Group Behavior Prediction

Abstract

With the advancement of unmanned aerial vehicle (UAV) technology, UAV swarm has been showing its great security threats towards the ground facility. With current technologies, it is still challenging in unknown UAV swarm tracking and neutralization. In this paper, we propose an analytical method in predicting drone flying behavior based on the machine learning algorithm, which could be integrated into swarm behavior prediction. Radiofrequency (RF) signals emitted from the UAV are captured by software-defined radio (SDR) to form the time series data. By using conventional short-time Fourier transform (STFT), a time-frequency spectrum revealing the RF data energy distribution is obtained for analyzing the signal variance pattern formed by the two different types of UAV flying trajectory. The transformed time-frequency domain matrix would be applied in multiple machine learning classifier for telling the difference of different flying trajectory. The results present the applicability of using machine learning in predicting the flying features and modes of intruding UAV swarm. It shows the potential application of this method in realizing effective UAV swarm negation.