Independent Vector Analysis Inspired Amateur Drone Detection through Acoustic Signals

Abstract

Detection of amateur drones (AmDrs) is mandatory requirement of various defence organizations and is also required to protect human life. In literature, various researchers contributed in this regard and developed different algorithms utilizing video, thermal, radio frequencies and acoustic signals. However, inefficiency of the existing techniques is reported in different atmospheric conditions. In this paper, acoustic signal processing is performed based on independent vector analysis (IVA) to detect AmDrs in the field. This technique is capable to detect more than one AmDrs in the sensing field at a time in the presence of strong interfering sources. The IVA is a relatively new and practically applicable technique of blind source separation and is more efficient than the independent component analysis technique. In the proposed technique, recorded mixed signals through multiple microphones are first un-mixed through using the IVA technique. Then various features of the separated signals are extracted. These features include Root Mean Square (RMS) values, Power Spectral Density (PSD) and Mel Frequency Cepstral coefficients (MFCC). Finally, signals classification is performed through Support Vector Machines (SVM)and K Nearest Neighbor (KNN) to detect AmDrs in the field. Performance evaluation of the proposed technique is carried out through simulations and observed the superior performance of the proposed technique.