| Improving the bandwidth of wireless communication systems is an effective way to meet people's growing diversified demands.However,as the increasing of bandwidth,the existing problems of wireless communications will become more serious and there will also be some new challenges.Firstly,the increasing band-width will lead to the consequence that symbol period is far less than channel delay spread,yielding the severe inter-symbol interference(ISI).Secondly,high bandwidth will also cause the sparse behavior of wireless channels,which brings difficulties in channel estimation.Thirdly,adopting low-resolution analog to dig-ital converter(ADC)at revceiver will remarkably reduce the power consumption under high bandwidth,which,however,could resut in the problem of nolinear distortion.How to cope with the above three difficulties is the key to ensure the quality of wireless communication.To solving the above three problems,the thesis designs iterative receivers based on message passing algorithms and approximate message passing algorithms to realize joint symbol detection,channel estimation and decoding.The main contents of the thesis are as follows:1.Considering ISI issue,we organically combined belief propagation(BP)and expectation propagation(EP)message passing algorithms and partial Gaus-sian approximation(PGA)to design an new iterative receiving algorithm.By analysing the intensity for interference,PGA is exploited to keep the discrete form of strong ones while BP-EP is adopted to project the weak ones onto Gaussian.By adjusting the range of PGA,the proposed reveiver could flexibly generalize the following two algorithms:1)all the interference symbols are approximated into Gaussian,yielding the BP-EP algorithm.Different with the heuristic Gaussian approximation,the proposed receiver projects the beliefs of symbols onto Gaussian through combined BP-EP rules,which decreasing the complexity and increasing the precision of approximation.The simulation results show that the proposed BP-EP receiving algorithm attains the same order of complexity with Gaussian approximate receiver and has significant gains.2)only part of the interference symbols are approximated,yielding BP-EP-PGA algorithm,BP-EP-PGA has the same complexity with PGA and can achieve significant gain in terms of bit error performance and convergence,and approaches the optimal bound,reaching the best tradeoff between complexity and performance.2.Considering single carrier frequency domain equalization(SC-FDE)sys?tem with unknown,sparse channels,we propose an iterative receiving algorithm based on parametric bilinear generalized approximate message passing(PBiGAMP)algorithm.The proposed algorithm can suppress ISI efficiently.We couple the proposed receiver with Gaussian mixture model(GMM)and hidden Markov model(HMM)to tracking the clustered-sparse behavior closely.The proposed algorithm has low complexity,which grows logarithmically in the channel delay spread.Simulation results show that comparing with decision-feedback-equalization-based(DFE)receivers and recently-proposed deconvolution algorithm,the proposed re-ceiver has significant advantages in tems of channel estimation,bit error,conver-gence and spectrum efficiency.3.Considering millimeter-wave systems with low-resolution ADCs,we design an iterative receiver with PBiGAMP.Dealing with ISI,we keep using GMM to estimate 60 GHz sparse channels.By taking the first and second order derivative of normalization function in posteriori probability density function,we obtain the closed-form resolution of posteriori mean and variance of unquantized noiseless randam variable,significantly relieving the nolinear distortion caused by low-resolution ADCs.In addition,the complexity only grows logarithmically in the channel delay spread.Simulation results demonstrate that the proposed receiver outperforms the linearization-based reveivers in terms of bit error and channel esimtation,especially under 1-2 bit quantization. |
PDF Full Text Request