Radar signal sorting based on resemblance coefficient of Welch power spectrum

The accuracy rate of current sorting methods of radar emitter signal is not high and they are sensitive to the signal-noise-rate (SNR). To solve these problems, a novel sorting method is proposed. Firstly, Welch power spectrum of the received signal is analyzed and its energy is normalized. Then a rectangle pulse sequence and a triangle pulse sequence are constructed. Furthermore, resemblance coefficients of its power spectrum with two pulse sequences are extracted, and are used as the sorting parameters. The last sorting is accomplished by support vector machine. According to simulation results, resemblance coefficients of the extraction have excellent separability and stability. Using Welch power spectrum has good sorting performances, which provides an effective method for frequency-domain analysis.