Research Of The Target State Recognition Techniques In Cognitive Electronic Countermeasure

The cognitive technologs is introduced into the cognitive electronic countermeasure(ECM)system.Based on the target's state transition,the ECM system optimizes jamming strategy fastly and schedules jamming resources to enhance the adaptiveness and intelligence of the environment.The target state recognition is an important technology in ECM system.In this paper,target state refers to the condition of the counter target defined by parameters of the received signal in the jammer.Conventional supervised state recognition needs to manually label the collected signal(initial training sample set)in adavance,which makes target recognition lacking timeliness,and manual annotation is much more difficult when facing complicated external environment and a larger amount of data.In addition,the jammer receives radar signal in the form of data stream and the radar anti-jamming measures may be " hidden " in the peacetime,but the jammer may encounter “unknown state” during the later confrontation process,which never appears in the previous training set.Thus,we pay attention to the unsupervised and incremental target state recognition in ECM system.This paper presents unsupervised and incremental target state recognition on the basis of the framework of cognitive electronic countermeasure system.This paper also designs simulation experiments and verifies the feasibility and validity of the method in combination with other key technologies in ECM system.The characteristics of the proposed unsupervised and incremental method of target state recognition are as follows:(1)Utilizing clustering algorithm to divide the initial signal samples,which saves manual annotation of priori class information.(2)Deciding whether the newly received signal sample belongs to an unknown state.(3)The existing recognition model can be updated,including the updating for existing class model and the generation for unknown class model.