Study On The Methods Of Radar Deception Jamming Recognition

Radar deceptive interference has the characteristics of flexible multi-pattern and strong confrontation,which poses a great threat to conventional radar applications.the detection and type identification of radar deceptive interference signals is an important basis for radar against deceptive interference.This paper takes the feature extraction,detection and type identification of common radar active deceptive interference in multi-dimensional transform domain as the research content,mainly carried out the corresponding research from the following aspects:The signal characteristics extracted by decomposing the signal with a single wavelet cannot obtain a good detection and recognition accuracy for all kinds of signals,which has limitations.A method of combining multiple wavelet features is proposed.The signals are decomposed by different wavelets respectively.After extracting the signal features,the signal features of multiple wavelets are combined to form a new joint feature vector as the signal classification feature.Under the same SNR condition,the detection and recognition accuracy of the signal using the joint multi-wavelet feature is higher than that of using any single wavelet feature.It also uses empirical mode decomposition as a signal processing method to extract better signal characteristics in deceptive signal type recognition.In the detection of traditional radar deceptive interference signals,it is necessary to use a large number of deceptive interference signal samples with markers to train the classifier.But in practical applications,deceptive interference signal samples are difficult to obtain,difficult to label,and difficult to cover all kinds.This paper proposes a method for radar deceptive interference detection under the condition that the deceptive interference signal samples are missing:combined with the single-class-learning,under the condition that only the normal radar echo signal samples are used as the training set.Training single-class support vector machine and setting reasonable parameters,so that the classifier can obtain the ability to detect deceptive interference signals.It is verified by experiments that when the SNR condition is higher than 5dB,the detection warning rate is close to 80%,and the false alarm rate remains at about 20%.Radar anti-jamming work needs to correctly identify the radar deceptive interference signal,but the radar deceptive interference signal sample is scarce,this paper proposes a method for radar deceptive interference signal recognition under small sample conditions:The semi-supervised learning idea uses a large number of radar mixed signal sets without makers to jointly train the classifier,so that the classifier can acquire the ability to identify the type of deceptive interference signal.It is verified by experiments that accuracy can reach 90%when the SNR is higher than 7dB.