Research Of Complexly Modulated Signal Recognition Based On DS Evidence Theory

Modulation signal recognition is a method that can determine the modulation format of a modulation signal when we have prior information.With the rapid development of communication technology,signal modulation system has become more complex.In a relatively wide frequency band,we usually use different modulation methods,how to effectively identify the modulation signal is very important in the software radio and electronic warfare systems.This paper introduce the temporal-spatial fusion model of evidence theory in the modulation signal recognition,and the main topic of this paper is the model of evidence theory,and some improved algorithms of evidence theory,also introducing the evidence theory to the modulation signal recognition,and constructing modulation signal recognition model,discussing and researching the recognition effect of this model.The main research directions of this paper are as follows:First of all,the basic probability assignment function is the basis of information fusion technology,and how to get the function is a key issue.This paper describes the Analytic Hierarchy Process,target type and environmental weighting coefficient method,the method based on Fuzzy Set Theory,and the method of using gray correlation analysis to obtain basic probability assignment function.Secondly,for the processing of uncertain information,this paper introduced the evidence theory,discussed the problems of the evidence theory,and analysed the principles and shortages of some improved algorithms.On the basis of previous studies,an improved algorithm based on evidence distance is proposed,this algorithm corrects the basic probability assignment function by using Mahalanobis distance to calculate the trust of evidences,and introduced the conflict coefficient into fusion formulas.This algorithm increased the value of the targets' trust,and reduced the uncertainty of evidences,and solved the problem of fusing conflict evidences.Thirdly,the modulation schemes of radiation signal and the characteristics extraction of radiation signal are introduced.Establish a template database for emitter signal recognition,the database characteristics include the approximate entropy,singular spectrum entropy,power entropy,and norm entropy of the signal.Analyzed the simulation results of emitter signal recognition with single feature.At last,by integrating previous sections and referencing the evidence theory models.This paper established a modulation signal recognition model,constructed template databaseusing entropy feature of modulation signal.Use the evidence theory to match and recognize modulation signals,through gradually progressive method,from the type of the signal and the SNR are both known to they are both unknown,we use this new model to distinguish signals.The result shows that this model has strong applicability,and it still can get good recognition results when lacking of priori information.