Study On Equalization For QAM Signals Under Complex Noise Environment

As the rapid development of wireless digital communication and the significant demand of communication users,the future communication system is supposed to own the faster data transmission rate,the more accurate communication reliability and the higher spectrum efficiency.However,the inter-symbol interference at the receiver due to the multipath propagation and the limited transmission bandwidth not only harms the communication reliability,but also places a restrict on the transmission rate of the communication system.Therefore,the channel equalization,as an effective way to decrease the inter-symbol interference,has always played a decisive role in the wireless digital communication receiving field.The data-aided channel equalization method acquires the channel response by utilizing a series of training symbols,and then adjusts the tap coefficients of the equalizer based on it to compensate the channel imperfect characteristic,but the transmission of training sequence occupies some of the precious spectrum resource.Moreover,the training sequence is unavailable in the non-cooperative communication systems.Hence,the blind equalization technology emerges.It relies on the prior information of signals rather than the training sequence to recovery the transmitted symbols.Usually,most existing data-aided channel equalization methods and blind equalization methods are based on the assumption of Gaussian noise,however,the ambient noise in practice is normally the impulsive noise that follows the alpha-stable distribution,which then makes the study on the data-aided channel equalization and the blind equalization in the impulsive noise environment meaningful.This dissertation mainly focuses on the data-aided equalization method and the blind equalization method under the complex noise environment(including Gaussian noise and impulsive noise)in the quadrature amplitude modulation(QAM)system,and the specific work is as follows:1.In view of the performance degradation of the traditional least square equalization method under the impulsive noise,a novel non-linear weighted adaptive equalization method based on the least square property is proposed.This method performs robustly because it can effectively combat the impulsiveness of noise.To be specific,we first analysis the reason contributing to the performance degradation of the least square equalization algorithm is the error criterion becomes incorrect under the non-Gaussian noise.Then,we make a novel error criterion by taking a negative exponential of the equalizer output error,which equals a non-linear weighted to the error.Based on this,the proposed method can adaptively control the iteration process of equalizer according to the noise,and then effectively suppress the negative influence on the equalization performance brought by the impulsive noise.Both theoretical analysis and simulation results indicate the robustness of the proposed method in the single input single output system and the multi-input multi-output system respectively.2.In view of the performance degradation and even a failure of the traditional multi-modulus blind equalization method under the impulsive noise,a robust adaptive blind equalization algorithm is proposed for QAM systems.First,the use of higher-order statistics in the cost function amplifies the influence of equalizer output error,which leads to the maladjustment of the multi-modulus algorithm under the alpha-stable noise.Hence,an adaptive weight coefficient is introduced in the proposed method to suppress the negative influence on the iteration process brought by the non-Gaussian noise.On one hand,the proposed blind equalization scheme has the ability to combat noise adaptively via the regulation of the weight coefficient,that is,the blind equalizer only updates its tap coefficients according to the samples polluted by small noise,while it stays unchanged when the samples are distorted by large noise.On the other hand,the step size of the proposed algorithm can be enlarged relatively to resolve the slow convergence speed of least mean square algorithms.This is due to the proposed scheme can avoid the maladjustment caused by large equalizer output error.Both theoretical analysis and simulation results verify the robustness and the fast convergence of the proposed blind equalization algorithm.3.In view of the performance degradation of the classic constant modulus algorithm(CMA)when it is applied to the equalization for high-order QAM signals under the Gaussian noise and the failure of existing blind equalization methods under the impulsive noise,a binary constant modulus algorithm based on the constant modulus property is proposed for high-order QAM systems.The proposed algorithm not only reveals the design defect in the cost function of the CMA,but also resolves the equalization for high-order QAM signals in the Gaussian and non-Gaussian environments.First,we find the excess error and the steady-state error are the reasons which give birth to the performance degradation of CMA for high-order QAM system under the Gaussian noise.It is well known that the gradient of the update formula not being zero at the steady state caused the steady-state error.Besides,the error introduced in the calculation of dispersion constant by the ideal assumption in the constant modulus algorithm is named as the excess error in this dissertation.Then,on the basis of the constant modulus property of high-order QAM signals,a novel cost function and its corresponding update formula are developed with the help of the inherent characteristic of the constellation,which then eliminates the negative influence of the excess error and the steady-state error.Both theoretical analysis and simulation results indicate that the equalization performance is significantly improved by this proposed algorithm at the cost of the sample utilization rate.Finally,since the proposed algorithm filters the samples distorted by large impulsive noise by the decision threshold,thus it can effectively suppress the impulsive noise and avoid the maladjustment under the alpha-stable noise.