Modulation Recognition And Parameter Estimation Based On Cyclic Spectral Density

Different modulated signals may have the same spectrum, but their cyclic spectrums are always different. Because the noise doesn’t have the cyclostationary, this makes cyclic spectrums have a very strong anti-noise performance.These advantages make it ideally suite for processing communication signals.In view of this, this thesis will study the applications of the cyclic spectrums in modulation recognition and parameter estimation. The main studies are as follows:Firstly, this thesis points out that if a signal is a cyclostationary signal, its mean function and autocorrelation function must be periodic.Then it analyzes in detail the nature of the cyclic spectrums, from the view of deep significance, and the cyclic spectrums imply the internal period of signals.Quite simply, it reflects the correlation of two spectrum lines separated by a particular frequency.Secondly, this thesis researches carefully on the principle of several typical cyclic spectrum algorithms and their complexity as well as their implementation steps.Then, it deduces several common communication signals’ cyclic spectrum, such as AM, PAM,BPSK, QPSK, MSK, and analyzes the characteristics of their cyclic spectrum and gives their simulation figures by Matlab and C++.Thirdly, this thesis extracts several common cyclic spectral features of communication signals, and designes several signal classifiers based on these features,such as tree structure classifier, neural network classifier, and fuzzy neural network classifier etc.Finally, it researches carefully on several DOA estimation algorithms based on cyclostationary, such as Cyclic MUSIC algorithm, Extend Cyclic MUSIC algorithm,Cyclic ESPRIT algorithm and so on.When the receive array is ULA and the signals are unrelated, this thesis firstly studys the applications of these algorithms in one-dimensional DOA estimation and gives their simulation results. According to these results, these algorithms exhibit selectivity in DOA estimation.When the signals are related, the spatial smoothing technique can make these algorithms also valid.When signals are wide-band, the signal subspace fitting DOA estimation algorithm can accurately estimate signals’ DOA.Then this thesis applies these algorithms to the field of two-dimensional DOA estimation, the receive array is uniform rectangular array or circular array.At last,it researches deeply on the performance of the DOA estimation ofthese algorithms and analyses deeply about the principle and the performance of the cyclic spectrum when using it to estimate signals’ carrier frequency and code rate.