Research And Implementation Of Graph Domain Modulation Recognition Method For MQAM Signals

Automatic Modulation Classification(AMC)can automatically identify the modulation type of the signals that are to be detected.Modulation type is a kind of essential information of communication signals.Hence,AMC is a significant part in the field of communication signal processing.In the practical applications,AMC classifier is an intelligent receiver that plays a key role,In these applications,such as electronic warfare,cognitive radio,spectrum surveillance,airborne or vehicular communication,etc,higher spectrum efficiency and fault tolerance are generally pursued,and MQAM modulation communication signals satisfy these demands.However,the use of MQAM modulation communication signals also faces technical challenges.With the order of MQAM increasing,the cycle spectrum of MQAM signals is consistent and the differences between high-order cumulant of MQAM signals become smaller.So the orders of MQAM signals is difficult to distinguishing,which makes that the receiver identify the orders of MQAM signals using general approach to realize intra-class AMC become a challenging problem.Based on the Digital Signals Processing on Graphs(DSP_G)theory proposed in recent years,this paper conducts a research on intra-class recognition of MQAM communication signals and algorithm computer implementation.This paper also formulates a technical route from simulation verification to computer implementation of the novel recognition algorithm.The main research contents and technical route are composed of three parts.1.According to theDSP_G theory and the mathematical model of MQAM communication system,the constellation diagrams of MQAM signals are studied on the processing method of Constellation Transformation(CT).In order to preserve the characteristics of the signals and simplify the recognition computation,the information and features of the constellation are converted into the graph domain mapping and then mapped to the adjacency matrix.2.That studying the feature extraction and uniform feature selection method of the adjacency matrix generated by MQAM signals is the second step of the research contents.The training feature set,whose elements are multi-dimensional vectors,under ideal channel condition is established,and the angle of feature vector between test and training signal is used as the judgment criterion to identify the order of test MQAM signal.Experimental simulation was carried out to analyze the successful recognition rate of each order MQAM signals under different SNR.At the same time,the robustness of the recognition method based on CT is considered by the successful recognition rate under the terrible receiving conditions that are frequency deviation and phase deviation.3.On the third step,this paper set up the communication system in the actual environment,and then obtain the constellation map of MQAM signal at the receiving machine.Finally we use CT recognition method as the theoretical algorithm to realize the real-time recognition of base band data from the communication system by the computer,and calculate the successful recognition rate.In the aspect of recognition performance,the proposed recognition method based on CT has better performance than the existing modulation recognition method in low SNR environment.At the same time,in the condition of frequency deviation and phase offset,CT recognition method has strong robustness for the most types of MQAM signals.In the actual communication system,the experiment demonstrates that the novel recognition algorithm can be used for real-time recognition,and successful recognition performance in actual communication system is very close to the simulation result.