Modulation Recognition For QAM Signal Based On Constellation Clustering Analysis

Modulation recognition has become one of the key technologies of intelligent communications,in both military and civilian fields has important value of applications and research.Constellation is not only an important analytical tool can also show a modulated feature especially QAM signal.The cluster analysis is one of the core technologies of artificial intelligence.In this paper,a practical QAM modulation recognition method based on constellation features and clustering analysis is proposed,and used the method to analyzed other modulations.The first step of work is to research relevant theory and algorithm.And then builds the hardware platform to collect the data,analyzes the data with MATLAB software,carries on the simulation verification to each module.Based on the analysis and comparison of algorithms,a specific modulation recognition method is proposed.Finally,a comprehensive test is carried out.The correct recognition rate meets the design requirements.The method has the advantages of full blind identification,unaffected by the initial phase of the carrier,good realizability and simple expansibility.It proves the feasibility of the method.The design provided an improved symbol rate estimation algorithm which based on the spectrum of envelope square,it used bimodal to instead of single peak,also used mean and variance,which can reduce the symbol rate estimation error and the estimated value fluctuations.Also provided a new membership function,joined the SNR factor for reducing identification miscarriage of justice.Finally,radius fluctuant average and variance of the constellation points are used as recognition features to identify the ? / 4DQPSK,8PSK,16 APSK and 32 APSK modulations,which broadens the modulation recognition range and shows good scalability.