MULTISTATIC RADAR EMITTER IDENTIFICATION USING ENTROPY MAXIMIZATION BASED INDEPENDENT COMPONENT ANALYSIS

Radar emitter identification is state-of-the-art in modern electronic warfare. Presently multistatic architecture is adapted by almost all the radar systems for better tracking performance and accuracy in target detection. Hence, identification and classification of radar emitters operating in the surveillance region are the major problems. To deal with the difficulty of identification of radar emitters in a complex electromagnetic environment, in this work entropy maximization method of Independent Component Analysis (ICA) based on gradient ascent algorithm is proposed. This algorithm separates unknown source signals from the interleaved multi-component radar signals. The discrete source signals are extracted from the multi-component signal by optimizing the entropy where maximum entropy is achieved using a gradient ascent approach through unsupervised learning. As better detection capability and range resolution are achieved by Linear Frequency Modulated (LFM) signals for radar systems here, multicomponent LFM signals with low SNR are considered as the signal mixture from which, the independent sources separated. A mathematical model of the algorithm for entropy maximization is illustrated in this paper. Simulation result validates the effectiveness of the algorithm in terms of time domain separation of the signal, and time-frequency analysis.