The Application Of Particle Swarm Optimization Algorithm In Communication Modulation Recognition

Communication signal modulation recognition is widely used in the areas of signal patternconfirmation,interference identification,radio listener,electronic warfare and signal detection.Itbelongs to the scope of the pattern recognition which is mainly to study the solutions of theselection and optimization of classification feature, as well as the design of the classifier. Thesetwo issues are also the main content of the automatic modulation recognition of communicationsignals.Modulation of communication signal contains two main categories:the analogmodulation signal and the digital modulated signal.This paper refer to the research of theautomatic modulation recognition of the digital modulation signal.Firstly, the paper introduces the extraction of these signal’s characteristic parameters whichcan distinguish those six digital communication modulation signal,such as2ASK,4ASK,2PSK,4PSK,2FSK,4FSK.The widely used extraction methods include wavelet transform,high ordercumulants statistics and HHT transform.For these different extraction methods of the signalcharacteristics parameters, their function were analyzed and compared.After completing theextraction of the characteristic parameter of digital communications modulation signal,it proceedto design classifier of high-precision, high speed and high stability which can automaticallyrecognize different modulation signal.The classifier design method by the decision treeclassification algorithm,support vector machine (SVM) algorithm and neural network algorithmis introduced in this paper.And these types of major classification algorithm are analyzed andcompared.The common decision tree identification of communication modulation recognitionconstruct too much node.The ordinary multi-class classifier based on support vector machine(SVM) is complex.And neural networks in practical applications is easy to fall into localminima.It particularly introduce the chaotic particle swarm optimization algorithm to selectcritical parameters effectively for those common classifer design algorythm.The chaotic particleswarm optimization has the outstanding advantage in the area of function optimization,datamining.The new algorythm greatly improves the accuracy of these traditional classificationalgorithms and reflects much better.When Signal to Noise Ratio(SNR) equal5%,10%,15%,there’s a simulation test toidentify six digital modulation signal of2ASK4ASK2PSK,4PSK,2FSK,4FSK,respectively,based on decision tree,support vection machine,neural network,support vection machine basedon chaotic partical swarm optimization,decision tree svm based on chaotic partical swarmoptimization and so on in the Matlab.According to the simulation results, the improvedalgorithm perform much better than the traditional algorythm.