Wavelet Transform Applied To Digitally Modulated Signals Identification And Parameter Estimation

This thesis discusses the application of wavelet in signal modulation identification and parameter estimation. In this thesis, continuous Morlet wavelet and discrete Daubechies2 wavelet are used. The main contributions is as follows:The instantaneous frequencies of PSK and FSK signals are estimated by using wavelet transform. Then, the instantaneous frequencies are used for Inter-class classification and parameter estimation. Also, the Intra-class classification of M-ary FSK and M-ary PSK have been made. Based on the number of different frequencies it is easy to identify M-ary FSK signals. As for the M-ary PSK signals, the differential phase peaks at the transients are used. The PDF(Probability Density Function ) of the differential phase peaks of M-ary PSK are the mean of M-1 PDFs of M-1 phase peaks. So the likelihood function of the differential phase peaks can be formed as means to identify M-ary PSK. Furthermore, the influencies of wavelet parameters to signal identification and signal parameter estimation have been systematically studied ,and the results have been adopted to outperform the above procedures.Another method is based on the multi-resolution property of the wavelet. Because different signal modulation has different characteristics at certain resolution, the specific signal information at different resolution are used a vector of signal features to identify signal modulation by RBF neural network. Computer simulations show the methods proposed has good performance even in low SNR ratio.