Feature extraction for complicated radar PRI modulation modes based on auto-correlation function

In order to recognize complicated PRI modulation modes, a classify algorithm based on auto-correlation and normalization is proposed. The mathematical model of four PRI modulation modes is given firstly. On the basis of making auto-correlation and normalization to four kinds of PRI modulation modes, it defines three characteristic quantities: density of peak value, intension of monotone and energy of sequence. According to these characteristic quantities, 3 dimensional feature space is set up. By comparing characteristic quantity with threshold value, PRI modulation mode can be recognized automatically. The simulation result shows that the characteristic quantities perform well in distinguishing four kinds of PRI modulation modes and the correct recognition rate is high when pulse sequence length is short and pulse miss is serious.