Analysis On Noise Characteristics And Signal Processing In Low Frequency Communication

Low-frequency communication plays an irreplaceable role in special working environments,such as underwater communication and geological exploration.Low frequency noise is an important factor which will affect communication quality,and it must be effectively suppressed.Low frequency noise is different from the Gaussian white noise in the general communication situation,which has the characteristics of non-Gaussian color noise.The existing research of low-frequency noise has made a preliminary discussion on the statistical characteristics and signal processing,but there are still problems in practical applications.This thesis combines the real data to analyze the low-frequency noise characteristics and discusses the robust processing method of low-frequency noise.The main work can be summarized as follows:1.Analysis of real data processingBased on the real noise data,this thesis uses traditional processing methods to analysis the noise spectrum characteristics and amplitude distribution.The result shows that the real noise data contains strong RF interference.After suppressing the interference,the noise data appears as an amplitude distribution of impulsive noise.Among the various nonlinear transformation methods,the performance of the local optimal detection is best,the clipper and the blanker are effective,and the Gaussian processing performance is poor.2.The Gaussianization and generalized matching(GGM)method for impulsive noiseThis thesis studies the suppression of pulse-type noise,and proposes the GGM method to robustly suppress noise,which can be used for unknown noise distribution.The GGM method is derived from Gaussian processing and generalized matching filtering,and it is a new nonlinear transformation function.Based on the probability density function that obtained by the kernel density estimation(KDE)method,a robust GGM design can be achieved.The simulation shows that the GGM design can achieve the robust processing of unknown distributed noise at the less cost of performance loss.3.Optimization design method for non-linear transformationThis thesis also studies the design of non-linear transformation method,and proposes a new design idea of non-linear function which optimizes the limiter and the Gaussian-tail zero memory non-linearity(GZMNL).The traditional non-linear functions are often based on noise distribution or simulation performance which is not theoretically optimal design criteria.This thesis proposes the nonlinear transformation function design method that maximizes the efficiency and optimizes the function parameters.After that,the adaptive numerical optimization algorithm is used to solve the optimization problem.The idea is applied to the optimization design of the limiter and GZMNL.The results show that the design method is obviously better than the traditional processing method and basically reaches the maximum performance value.