A method of subtle feature extraction for navigation radar emitter
Aiming at the low accuracy of recognition of navigation radar emitter by conventional methods,the time-fre-quency characteristics of pulse signals and unintentional modulation on pulse of navigation radar were analyzed and studied,focusing on the instantaneous frequency of pulse signals,especially at the rising and falling edges of pulses.The change rate of instantaneous frequency was characterized by taking a series of sampling point composition vectors and using support vector machine for classification.Experiments based on simulation data showed that,Compared with the whole pulse,this method can distinguish different radar emitter signals better,which not only improves the recognition accuracy,but also re-duces the calculation amount.The method obtains satisfactory results in single carrier frequency signal,linear frequency modulation signal and binary phase-coded signal.Using the actual data collection verification,when the SNR is 20dB,the re-cognition accuracy is more than 90％.
subtle feature extractioninstantaneous frequency featuresupport vector machines
NETL
NSTL
万方数据
