Signal Modulation Recognition And Parameter Estimation Based On DCT

Automatic modulation recognition of communication signals is widely used in civilianareas and military areas of signal interception, signal monitoring, software radio and satellitecommunications. Recognize the modulation type correctly, we can carry on interception ordisturbance to the enemy better, constitute defenses and attacks strategy more pertinently. Thetechnologies of modulation recognition and parameters estimation based on Power SpectrumDiscrete Cosine Transform are investigated in this dissertation. The impact of modulationrecognition and parameter estimation caused by noise can be reduced effectively. And it hasbeen verified to be effective in theory and simulation experiment. The details are described asfollows:(1) An intra-pulse modulation recognition algorithm based on Discrete Cosine Transform(DCT) is proposed to improve the recognition accuracy under negative signal-to-noise ratio(SNR) environment. According to DCT’s superior capability and energy compaction, thesignal power spectrum DCT and the characteristic of coefficients of low frequency ofdifferent modulation types have a large difference, so the result of recognition can be obtained.The envelope features and coefficients of low frequency of power spectrums DCT have strongstability, and DCT have a fast algorithm, so these features can be extracted easily.Experimental works have demonstrated that the correct recognition rate (CRR) can increaseby17.4%than ZAM-GTFR method when SNR varies from10dBto10dB.(2) A parameter estimation algorithm of CW signal and BPSK signal based on signalpower spectrum DCT is proposed in this dissertation. Because of the superior energycompaction of DCT, after power spectrum DCT, threshold processing, and making an InverseDiscrete Cosine Transform (IDCT), this method can reduce the impact of noise in parametersestimation. So this method can realize accurate parameters estimation under low SNRenvironment. Our experimental works have demonstrated that when SNR=5dB, theRMSE of carrier frequency and pulse width estimation of CW signal are less than0.42MHzand0.01μs, respectively. The RMSE of carrier frequency, pulse width estimation of BPSKsignal are less than0.44MHz and0.03μs, respectively, and the estimation accuracy canincrease by about22.1%and28.3%with compared algorithms at SNR=5dB, respectively.