Classification of radar emitter signals using the characteristics derived from instantaneous frequencies.

A new approach of automatic classification of the radar emitter signals is presented. This method is based upon the characteristics derived from instantaneous frequencies and contains three sequential steps: firstly, the instantaneous frequencies of radar emitter signals are extracted via the improved instantaneous autocorrelation algorithm; secondly, a two-times-normalizations based feature re-extraction approach is used to obtain the classification characteristic vector; finally, the kernelized fuzzy c-means clustering with validity assessment is carried out to categorize the considered signals automatically. The results of simulation experiments show that, the characteristic vector can reflect the differences of signals with different modulation types and has a good ability to resist noises. The proposed scheme of signals classification can achieve the overall success rate of above 95% , even at the signal to noise ratio of 6dB. [ABSTRACT FROM AUTHOR]