In Electronic Support (ES) which is a division of electronic warfare the main task is extracting features of received radar signals and using them to In this research we analyze radar signal in time frequency domain. We first derive SNR in quadratic time-frequency transform then analyze the kernel influence on signal representation in time-frequency plane and design a method to improve the detection of LFM signals. Our second goal is classification and identification of the same type radars. We use supervised simulation data to focus on feature extraction from radar signals in IF band. These features are defined in time-frequency transform and divided into three grou features along time axis, features along frequency axis and features along time-frequency axes. In the second step extracted features are classified using SVM and KNN classifiers. Finally, results of the proposed method are compared with cumulant and EMD based method. The experimental results in different scenarios show that the classification rate of the proposed method is 7 to 20 percent higher than the other methods. Keywords: radar, classification, feature extraction, time-frequency transform, IF band
