Research On Estimation Methods Of Cyclic Spectrumc And Carrier Frequency Of Cyclostationary Signals In Alpha Stable Distribution Noise

Some man-made signals whose some statistics parameters do vary with time periodically that encountered in communication systems is called cyclostationary signals. In the Gaussian noise environment, cyclic spectrum estimation methods of cyclostationary signals mainly include discrete frequency smoothing algorithm and discrete-time smoothing algorithm, etc. And carrier frequency estimation methods of cyclostationary signals mainly include periodogram method, center frequency method, zero-crossing method, correlation function method and the method based on cyclic spectrum, etc. However, non-Gaussian distribution signals and noises exist widely in practical applications, such as underwater acoustic signal processing, financial data analysis, biomedical signal processing and seismic exploration, the traditional signal processing theories and methods which based on Gaussian distribution model degenerate significantly in these applications, and can not be well applied.The paper used a-stable distribution as noise model, studied the cyclic spectrum estimation methods of cyclostationary signals, and analyzed the fractional lower order cyclic spectrum estimation methods of AM, BPSK and QPSK signals.These methods includes fractional lower order discrete frequency smoothing algorithm, fractional lower order discrete-time smoothing algorithm, fractional lower order FFT accumulation algorithm and fractional lower order strip spectral correlation algorithm.The computer simulation results of these methods had been given in this paper. On this base, we also further studied carrier frequency estimation methods which based on fractional lower order statistics of AM, BPSK and QPSK signals. We had got two conclusions by computer simulation. One is the theories and methods which based on second order statistics of signals processing degenerate significantly under a-stable distribution noise, and can not meet the requirements of the signal processing. The other is the estimation methods which based on the fractional lower order statistics of carrier frequency has strong anti-jamming capability.