Research On Channel Estimation Algorithms Using The Sparsity Property In Millimeter-wave Systems

Communication over millimeter wave(mm Wave)bands from 30 to 300 GHz is a promising technology for the fifth generation mobile networks to provide multi-gigabit communication services.However,the mm Wave band suffers from severe path loss and rain attenuation.It's thus often used in conjunction with large-scale antenna arrays to provide large beamforming gains.To exploit beamforming gains,the system must have accurate channel state information.With the large-scale antenna array,it is very challenging due to the high dimensionality of the channel matrices,which introduce large training overhead,severe pilot contamination and high complexity.Unlike the conventional rich-scattering channel,the mm Wave channel exhibits sparsity in a scattering environment,especially the non-line-of-sight paths have large attenuation.So the number of dominant paths is limited.Exploiting the sparsity nature,this work considers the channel estimation problem in mm Wave systems with large-scale antenna arrays,and design algorithms to estimate the number of paths,direction of arrivals(Do As)and path gains.The work includes two parts.In the first part,we design channel estimation algorithms using the sparsity property.We consider the uplink transmission where a single-antenna user communicates with a base station employing a large-scale uniform linear array.The spatial channel is transformed into the beamspace channel by the discrete Fourier transform(DFT).Based on the sparsity property of the beamspace channel,we propose two algorithms.The first one is Spectrum Weighted Identification of Signal Sources(SWISS)for the case when the channel statistics are unknown.We introduce a weight vector to amplify the desired signal and suppress the noise.Then we design an optimization problem to obtain the optimum weights from which we can estimate channel parameters.The second one is Neyman-Pearson criterion based-Detector(NPD)assuming the Rician channel.We adopt the NeymanPearson criterion to decide whether there exists a path on each DFT point.The probability of detection is maximized under a constant probability of false alarm(PF).We thus obtain the threshold and estimate channel parameters.Based on NPD,we analyse the relationship between the probability of correctly detecting all paths(PC)and the PF,and propose the iterative NPD to maximize PC.Simulation results show proposed algorithms perform better than the spatial smoothing in terms of the accuracy of the number and location of paths,and the normalized mean square error(NMSE)of the channel.In the second part,we design channel estimation algorithms considering the power leakage problem.In the first part we assume Do As coincide with DFT points.But in practice they are continuous and don't fall on discrete points,leading to the power leakage problem.We propose the Combined Algorithm with Leakage(CAL)that combines the ideas of SWISS and NPD.We detect the strongest path using SWISS and adopt the zero-padding and secant method to estimate the Do A accurately.Then we remove this path from the signal and repeat the step above.We stop until there is only noise in the signal and we can use the threshold of NPD.Simulation results show CAL can work well for solving the power leakage problem and get better performance than SWISS and the spatial smoothing in terms of the accuracy of the number of paths,the MSE of the Do As and the NMSE of the channel.