Research On Signal Demodulation Technology Under Alpha Stable Distribution Noise

In the traditional field of linear signal processing,the Gaussian noise hypothesis has a solid theoretical basis.However,it is not suitable for modeling in an impulse noise environment,that is,it cannot describe a distribution that may generate a large number of data mutations.Instead,a-stable distribution noise model with longer tails is used.As a promotion of Gaussian distribution,it has solid theoretical foundation,extensive and extremely important application scenarios.Furthermore,it is supported by data which is collected in the nature.Because the theoretical research of the demodulation scheme under the background of a-stable distribution noise is still in its infancy,it is forward-looking and practical to study the demodulation technology under the background of ?-stable distributed noise.Therefore,this paper first generalizes the estimation method of noise parameters to meet the application requirements of the memory modulation method,and then studies the architecture of the overall demodulation system to improve the error performance.Specific results and innovations are as follows:(1)For the generalization of noise parameter estimation method,this paper proposes an estimation method of ?-stable distribution noise parameters with MSK modulation.The method is based on the kernel of the empirical characteristic function algorithm.The state transition matrix is built with Markov chain to calculate the characteristic function of the memory signal.Besides,the characteristic function of the complex ?-stable distribution noise is deduced based on the complex domain.As a result,the estimated values of the characteristic parameter a and the scale parameter? are obtained with better performance.The simulation results show that the proposed scheme has better estimation performance than the existing method with memoryless modulation scheme under the same condition.(2)For the research of the overall demodulation system architecture,this paper proposes a scheme for signal demodulation under additive ?-stable distribution noise environment.The key to this solution lies in the suppression of noise.Subsequently,the signal is demodulated according to the traditional method under the background of Gaussian white noise.In the suppression part,the design of the whitening filter is implemented by establishing an inverse system of an autoregressive model.Moreover,on the basis of analyzing the effects of Gaussianization and whitening on the signal and noise,a preprocessing module is introduced into the transmitter.The preprocessing module is utilized to counteract the influence of the whitening filter on the signal.The simulation results show that compared with the system that only performs Gaussian processing,the proposed system has better noise suppression performance,in other words,the system has lower bit error rates.In addition,this system is universal,which means that,it is also applicable to 2DPSK and GMSK modulation methods.