Research On Modulation Recognition Algorithm Of Communication Signal Based On Random Forest

Recognition of modulation mode is an indispensable key technology in non cooperative communication.It is the basis of research in cognitive radio and other fields.Most of the existing modulation recognition algorithms need to know more signal modulation parameters and require high quality of communication environment.In recent years,many scholars at home and abroad have been studying in this field,and great results have been achieved.A large number of new modulation recognition algorithms are constantly emerging.However,because of the rapid development of modern communication technology,a large number of new modulation methods have been used continuously,making the transmission environment of non cooperative communication more and more complex.How to realize the effective recognition of signal modulation in bad communication environment has become a new test of modulation recognition technology.Based on this,this paper proposes a modulation recognition scheme with good recognition effect,small amount of computation,easy implementation and high robustness under low SNR condition,and validates the scheme through Matlab.The main work of this article is as follows:(1)introduce the research work in this paper,analyze the existing algorithms of signal modulation recognition,summarize it,and illustrate the significance of the algorithm research in this paper.(2)starting from several typical modulation modes of communication signals(SSB,FM,MASK,MFSK and MPSK),the basic principles and characteristics are introduced.Through Matlab modeling and simulation,the characteristics of time and frequency domain are analyzed,and its characteristics are briefly summarized.(3)the analysis of signal characteristics,focusing on the features of the time domain and high-order cumulants,were extracted from the two groups can be distinguished feature of different modulation,and then through the Matlab modeling and simulation,to verify the recognition performance statistics,determine the decision threshold.At last,using the traditional two-fork tree classification algorithm,the recognition of the six modulation schemes selected in this paper is completed.The recognition rate of 2ASK,BPSKand 4ASK is more than 98% when the signal to noise ratio is not less than 2dB,and the time domain parameters are used as the classification characteristics.When the signal to noise ratio is not less than 4dB,the recognition accuracy of BPSK,QPSK,2ASK and 4ASK is up to 100%.Using the higher order cumulants as the feature,the recognition accuracy is over 96% when SNR is not less than 0dB,and the recognition accuracy reaches 100% when signal to noise ratio is not less than 3dB.(4)based on the study of random forest algorithm and combining the characteristics of modulation recognition,a modulation recognition scheme based on random forest algorithm is proposed.Based on the time-domain statistics and higher-order cumulants,the training samples and test samples of random forests are selected to train and test the random forest models.Modeling and Simulation in the Matlab environment,six modulation schemes are selected for modulation recognition.It is verified by experiments that the recognition accuracy of 2FSK,BPSK,4FSK and QPSK can reach more than 78% when the signal to noise ratio is no less than-5dB,and the correct rate of recognition can reach 100% when signal to noise ratio is no less than 3dB.The random forest with higher order cumulant characteristic parameters is used as training sample.When the signal to noise ratio is no less than-10 dB,the correct rate of the six modulation schemes selected is 50%.When the SNR is not less than-5dB,the correct rate is 100%.