Underdetermined Blind Source Separation Algorithm Based on Cyclostationary Analysis and Synchrosqueezing Transform for Anti-Mainlobe Jamming

In the context of radar electronic countermeasures, this paper addresses the issue of mainlobe deception jamming signals by proposing a two-step underdetermined blind source separation (UBSS) method. The method combines a mixing matrix estimation algorithm based on the cyclostationary analysis of Linear Frequency Modulation (LFM) signals and a mixing source signals separation algorithm based on the Synchrosqueezing Transform (SST). This approach effectively utilizes the cyclostationary characteristics of radar communication signals and the linear relationship between time and frequency in LFM signals. Compared to traditional time-frequency analysis methods, the SST-based time-frequency analysis method significantly improves energy concentration in the time-frequency plane, while offering advantages such as the absence of cross-terms and flexible selection of kernel functions.