Key Technology Research And Application Of Digital Modulated Signal Monitoring And Classification System

In recent years,with the rapid development of the information industry,modern communication has gradually become an indispensable part of people's daily lives.To ensure legal communications and prevent illegal users from illegally occupying and interfering with the spectrum,it is especially important to monitor and classify the signals.At present,the detection and classification of signals mainly rely on manual labor,which is time-consuming and labor-intensive,having a lot of limitations.Regarding this situation,this paper develops a digital modulation signal detection and classification system.This system can complete the estimation of the signal's center frequency,bandwidth,signal-to-noise ratio,symbol rate,and modulation mode classification,and can give warnings to illegal signals as well.First of all,this paper focuses on the parameter estimation algorithms in the key technologies of the system,including symbol rate estimation,frequency offset estimation,and signal-to-noise ratio estimation.In terms of symbol rate estimation,a symbol rate estimation algorithm based on the envelope square spectrum is selected.Aiming at the problem of failure under ASK-type signals and low roll-off factor shaping,an optimized algorithm is proposed.The result of simulation shows that the algorithm proposed in this paper overcomes several drawbacks of the original method,having wider applicability and the advantages of low complexity and accurate estimation,which are suitable for engineering implementation.In terms of frequency offset estimation,for ASK and QAM signals,an estimation algorithm based on the fourth power spectrum is selected,and an improved algorithm combining multiple autocorrelations is proposed to make it have a better estimation performance at the same signal-to-noise ratio.In terms of SNR estimation,an estimation method based on the power spectrum is selected to obtain the bandwidth estimate while estimating the SNR.Finally,this algorithm is simulated and analyzed.Secondly,the modulation classification algorithm as the key technology of the system is deeply studied separately from three different aspects(feature extraction based on spectrum,high-order cumulant,and signal amplitude),analyzing 14 signal types such as Noise,2ASK,4ASK,8ASK,BPSK,QPSK,8PSK,16PSK,OQPSK,?/4DQPSK,8QAM,16QAM,32QAM,64QAM.For these 14 types of signals,by extracting 6 new features and 6 existing features,a joint classification algorithm that is robust to frequency offset is proposed based on the decision tree classifier.Experiments show that this algorithm can reach above 90%classification rate when the SNR is greater than 7dB.Except for 4ASK,the classification rate of all modulation types is above 90%when the SNR is greater than 5dB.Finally,the laboratory instrument Ceyear AV4051 was selected to complete the whole system development,and the vector signal generator Ceyear 1465D-V was used to design and write test cases and methods to complete the verification of system functions and algorithms.