Micro-UAV Detection and Identification Based on Radio Frequency Signature

This paper mainly focuses on the detection and identification on micro-unmanned aerial vehicles (UAVs) using radio frequency (RF) signature of the signals from UAV downlink communication. To effectively perform detection and identification, feature engineering is carried out to describe the signature of different micro-UAV signals. The approach for feature engineering is based on the division of raw continuous sampled signals into several valid frames in time domain. In each frame, cyclostationarity features as well as kurtosis and spectrum factors are extracted after signal preprocessing. Selected features of UAV signals and ambient noise are fed to support vector machine (SVM) and k-nearest neighbor (KNN) models to obtain a well-trained classifier. Then the classifier is used to detect and identify non-cooperative micro-UAVs. In the detection phase, all detected UAV signals from ambient noise, specifically WiFi signal in this paper, are treated as invading non-cooperative micro-UAVs where the detection scenario is assumed as a no-fly-zone. In the identification phase, the type of micro-UAV is identified based on its downlink communication protocol from the detected UAV signals. In this paper, two kinds of micro-UAV signals and ambient WiFi signal as background interference are tested versus various signal-to-noise ratio (SNR) levels. Experimental results show that the proposed method proves to be feasible to detect micro-UAVs and identify the protocol UAV used in downlink communication. More different types of micro-UAV signals will be sampled into database for the future work.