Time-frequency Detection And Modulation Identification Of Communication Signals With Low Signal-noise Ratio

In the post-epidemic period,modern information technology has achieved vigorous development,which has promoted great changes in various fields of economy and society.Especially in the field of communication,Human daily life,home study and work are becoming more and more simple,because of the further improvement of communication network topology and communication methods.But it also makes the communication environment more complicated,and the influence of background noise cannot be ignored.Therefore,reliable signal detection and accurate modulation identification have important research significance under low signal-to-noise ratio.This thesis studies three aspects of time-frequency detection,noise reduction and modulation recognition based on timefrequency images of communication signals.The specific work is as follows:In order to make full use of the statistical properties of white Gaussian noise in the TPST(Three Parameters S Transform)spectral domain and eliminate the influence of frequency variation on the detection threshold,a constant false alarm rate detector based on the TPST is proposed.Detection model.When the false alarm probability is 0.001,0.005 and 0.01 respectively,the detection accuracy of the detection algorithm can reach 100%when the SNR>-5dB.The constant false alarm rate is set to 0.01.Compared with ST(Stockwell Transform)and GST(Generalized Stockwell Transform),TPST has the highest detection accuracy.Then,the same parameters are used to detect 12 different types of modulated signals.When SNR>-4dB,the detection accuracy is 100%.The above experiments show that TPST is more suitable for communication signal detection under low signal-noise ratio.In addition,the method can also be used as a preprocessing method to estimate the parameters of the signal.Taking the LFM(Linear Frequency Modulation)signal as an example,it is first used to estimate appearance and disappearance locations.The appearance and disappearance times are detected under the conditions of TPST and GST respectively.The detection accuracy based on the TPST algorithm is higher,and the accurate detection rate can reach more than 90%when the signal-to-noise ratio SNR>-2dB.Then,the instantaneous frequency of the LFM signal is estimated,which greatly reduces the normalized mean square error compared with the traditional estimation method by ridge extraction directly from the timefrequency transformed image,which is more suitable for a signal that occurs discontinuously.Aiming at the problem of noise reduction of communication signals,a noise reduction method based on vSSMD(Symplectic Singular Mode Decomposition based on Lagrange multiplier)and SSA(Singular Spectrum Analysis)is proposed.The choice of the embedding dimension is very important for the noise reduction effect of vSSMD and SSA methods,and the optimal embedding dimension can be effectively selected by combining the Monte Carlo experimental idea.The vSSMD method can separate the useful signal and the noise signal.Under the low signal-to-noise ratio,the use of SSA can effectively remove a small amount of noise information contained in the useful signal.Taking the simulated communication signal as an example,the denoising effects of the vSSMD-SSA method and the vSSMD and SSA methods are compared.The noise reduction effect is measured by mean square error and output signal-noise ratio.In addition,the denoising effect of vSSMD-SSA method on signals of different modulation types is compared,and the denoising effect is the best for signals with less variation levels,such as 2ASK(2 Amplitude Shift Keying),2FSK(2 Frequency Shift Keying)and BPSK(Binary Phase Shift Keying).Finally,the vSSMD-SSA method is used to denoise the communication signal.The experimental results show that the joint denoising algorithm of vSSMD-SSA can effectively improve the signal-to-noise ratio of the denoised communication signal on the basis of preserving the integrity and details of the useful signal,which is better than other methods,and it has good practical application value in low signal-to-noise ratio environment.Finally,for the modulation identification problem,Time-frequency transform one of the signal characterization methods is studied.The theoretical basis of the quadratic timefrequency analysis method is expounded.Because the time-frequency aggregation degree of the SPWVD(Smooth Pseudo Wigner-Ville Distribution)is not high,introducing the principle of linear time-frequency transform,and a SET(Synchronous Extraction Transform)algorithm based on TPST is proposed,the algorithm has high time-frequency aggregation and anti-noise performance.In order to make full use of the advantages of quadratic timefrequency transform and linear time-frequency transform,a communication signal identification algorithm based on SPWVD and SETPST(Synchroextracting Three Parameters S Transform)is proposed.Finally,combined with the vSSMD-SSA noise reduction method,the modulation identification of the communication signal under low SNR is realized.