Research On Matching And Classification Algorithm Of Specific Emitter Identification

Nowadays,much attention of the study in electronic countermeasure research is paid to the specific emitter identification(SEI)which is a key subject in the field.The individual differences of the emitters are generated due to the subtle differences of the internal components.SEI can distinguish specific emitter the signal comes from and achieve devices tracking,and then be targeted to carry out monitoring,electronic jamming or military striking to radars and communication devices that are vital to enemies.SEI surely will improve our performances of electronic countermeasure devices,change our relative laggard situation in military devices,and lay a solid technology base for our military to win the war under high-tech conditions.SEI mainly includes three researching plates,that is,fingerprint feature extraction,target matching,classification.This paper makes detailed studies on three researching plates.In the study of fingerprint feature extraction,we propose two valid fingerprint features: instantaneous feature and statistical feature.Instantaneous feature enhances the generalization of fingerprint feature;statistical feature reduces the instantaneous feature dimensions while inheriting the generalization of instantaneous feature.In the matching target,dynamic time warping algorithm(DTW),generalized euclidean distance(GED)and template segmenting optimization algorithm(TSO)are proposed.DTW removes the negative effects that brought from the warping phenomenon to measure distance between fingerprint features;GED considers the geometric distribution of fingerprint features synthetically;TSO solves the high time-consuming problem of optimizing templates.In the classification,the algorithm of parameter optimizing of support vector machines based on heuristic algorithms is proposed.The algorithm takes advantage of heuristic algorithms to reduce optimizing time complexity of support vector machines.In this paper,we have proposed the research algorithms at each plate that were all tested by the measured data to guarantee the authenticity and validity of above all of algorithms.