Research On Angle Estimation Method For Large-scale Antenna Array With Phase Calibration Error In Millimeter-wave Communications

Due to the abundant spectrum resource in millimeter wave(mmW)frequency,mmW communication becomes one of the key technologies for next generation mobile communications.The short wavelength of mmW facilitates the deployment of large-scale antenna array at both the base stations and mobile stations.Great array gains can be achieved via beamforming to compensate the severe path loss in the mmW frequency.Accurate angle information is required by beamforming,thus the angle estimation becomes the base of mmW communications.In practice,the antenna elements may have nonignorable calibration errors on both gain and phase,which severely degrade the performance of angle estimation and consequently the performance of beamforming.In view of the negative effect by calibration error on antenna elements,this thesis studies the angle estimation method to enhance the robustness performance of ESPRIT/MUSIC methods for mmW communications with massive antenna arrays.The main works include the following two aspects.i)We analyze the factors causing antenna array errors and their negative effects,especially the calibration error of the antenna elements.After formulating the signal model with antenna array error,the closed-form expression of the angle estimation error of the ESPRIT/MUSIC method is presented and verified under different settings.ii)Two improved algorithms are designed to improve the angle estimation performance for mmW communications with large random phase calibration errors.Observing that the adjacent elements on the diagonal line of the covariance matrix of the received signal contain the same angle of arrival(AoA)information but independent random phase errors,an algorithm termed Covariance Self-Correction algorithm is developed to estimate the phase error firstly,and then iteratively update the estimation of AoAs and phase error.Since the phase error of the antenna element is unchanged,we can further improve the estimation by choosing the better initial phase error for the iteration under the principle of minimizing the channel recovery error.Simulation results verify the effectiveness of the proposed two algorithms.