Pulse-compression radar signal sorting using the blind source separation algrithms

With the increase of pulse density, signal sorting becomes extremely difficult for modern electronic reconnaissance, especially for pulse-compression radar signals. Blind source separation (BSS) is a new developed technology for separating signals from mixed observed data. In this paper, we propose various instantaneous mixing models of pulse-compression radar signals, including linear frequency modulation, polyphase code, phase-shift keying and frequency-hopping signals. The combinations of fast independent component analysis (FastICA) and joint diagonalization BSS algorithms are presented for radar signals. The performance index (PI) and signal-to-interference ratio (SIR) are adopted to analyze the separation performance at different signal-to-noise ratios. The experiment results demonstrate the validity and correctness of the proposed method.