Research On Millimeter Wave Low Complexity DOA Estimation And Beamforming Technology

Nowadays the amount of data that needs to be processed has increased dramatically because of the development of wireless communication,the emergence of emerging applications and the increasing number of wireless devices,which leads to the spectrum of low frequency band being unable to meet the demand.The millimeter-wave band has a large amount of spectrum resources,which can complete high-speed and high-capacity wireless network communication.However,since the path loss of millimeter wave is severe,a large-scale antenna array can be used to provide gain at both the transmitter and the receiver.In addition,because the number of multipath of millimeter-wave are less in the outdoor environment,array signal processing techniques can be applied in millimeter wave large-scale antenna system.The traditional array signal processing method has a very high complexity in the millimeter wave large-scale antenna system.This thesis focuses on low-complexity direction of arrival(DOA)estimation and beamforming algorithms for millimeter wave large-scale antenna systems.The concrete contents are as follows:In order to reduce the complexity,power consumption and cost of the system hardware,the performance differences between MUSIC,ESPRIT,ISS and CSS algorithms under low-precision quantization and full-precision are analyzed and compared.Through simulation,it is found that these algorithms still have a high precision with low-resolution quantization.In addition,in view of the high complexity of the traditional algorithm when the number of antennas increases,the feature decomposition of the covariance matrix is avoided by using a method based on an extra vector basis,which lead to reduce the complexity of the algorithm.The adaptive beamforming algorithms which are suitable for low frequency bands can not be used in millimeter wave large-scale antenna systems because the complexity of the adaptive beamforming algorithm is too high to meet the requirements of low latency.This thesis presents a least mean square(LMS)algorithm based on beam scanning.First,the DOA is roughly estimated by beam scanning,and then the initial beamforming weight vector is constructed using this result.Finally,the LMS iteration is performed to obtain the best beamforming weight vector.This algorithm can not only obtain the same performance as the original LMS algorithm,but also effectively shorten the convergence time.Hybrid beamforming architecture can not only reduce RF links,but also meet the needs of multi-stream/multi-user.This paper proposes a simple subarray-based hybrid beamforming technique that can be used with low-precision quantization.The technique first uses the frequency-domain correlation between the received signals of adjacent subarrays at the digital end to estimate the DOA of the source,and then change the value of the phase shifter on the analog side to complete beamforming.This method not only significantly reduces the complexity of the adaptive broadband beamforming technology,but also has excellent performance.