Research Of Radar Signal Pattern Recognition

With the continuous development of computer science and technology,the status of electronic countermeasures(ECM)continues to improve,and radar signal pattern recognition has become the precondition of ECM.Radar signal pattern recognition first accept the information from enemy radar,then recognize the pattern and estimate the degree of threat.Radar signal pattern recognition is mainly including complex radar signal sorting,single emitter signals pattern recognition and radar threat level assessment.In recent years,the electromagnetic environment has become worse and worse,the traditional signal recognition methods are not suitable to new radars whose parameter are not variable.This paper applies machine learning methods to deal with dynamic identification of signal pattern.Following is the summary of this paper:(1)Introducing the signal sorting that applied to the complex radar signal and some traditional signal sorting methods,we use clustering to deal with this problem in this paper.(2)Applying adaptive sliding window to split the coming radar signal and extract feature vector from the windows using sparse dictionary etc.(3)Using traditional SVM and multitask SVM to classify window sequence,(4)Dynamic method of radar recognition:applying probabilistic decision fusion method to identify the false points.(5)Due to the increasing number of signal pattern,constantly updating the training model is in need,so that incremental method is applied to radar recognition.It saves storage space and sped up the training speed.(6)Applying C/S pattern to demonstrate the multi-threaded dynamic project.In order to improve the capacity of ECM,further research of dynamic radar signal recognition technology is needed.