Research On Hybrid Beamforming In Millimeter Wave Communication Systems

As a key technology for the upcoming fifth generation(5G)systems,millimeter-wave(mm Wave)communication has gained sudden attractions.Due to the unfavorable channel characteristic of mm Wave frequency like large path-loss and low penetration,massive multiple-input multiple-output(MIMO)is utilized to combat the large propagation loss with significant beamforming gains.The implement of full digital beamforming requires a dedicated radio frequency(RF)chain for each antenna.Due to the high cost and power consumption,it is impractical to perform full digital beamforming at baseband in mm Wave systems.To overcome these limitations,the hybrid beamforming(HBF)is introduced,which combines the analog beamforming and digital beamforming to reduce the number of RF chains and the implemention complexity of mm Wave systems.Although there are many studies for hybrid beamforming,the optimal design of the HBF has not been fully understood.For single-user mm Wave systems,spatially sparse orthogonal matching pursuit(S-OMP)performs very close to the full digital solution,but it has a high computational complexity.For multiuser mm Wave systems,the hybrid beamforming without codebook exhibits constraints on the number of radio frequency(RF)chains.In order to solve the above problems,a low-complexity HBF algorithm is proposed for single-user mm Wave systems and a greedy HBF algorithm is proposed for multiuser mm Wave systems.For single-user mm Wave systems,the low-complexity HBF algorithm imposes the orthogonal property of the candidate vectors to avoid matrix inversion and eliminate iteration process.An asymptotic sparse orthogonal matching pursuit(AS-OMP)algorithm is proposed by studying the asymptotic orthogonality of the antenna array response vector in the millimeter-wave channel.As the size of the antenna array increases,the AS-OMP algorithm performs approximately to the the performance of the S-OMP.Because the AS-OMP algorithm is affected by the antenna array size,another hybrid beamforming algorithm based on Gram-Schmidt orthogonalization method(GS-HBF)is proposed for general single-user scenes in mm Wave systems.The novel algorithm utilizes the low-complexity Gram-Schmidt orthogonalization method to orthogonalize the candidate vectors to reduce the complexity.Simulation results and complexity analysis demonstrate that the proposed HBF algorithms perform very close to the S-OMP algorithm and enjoy low complexity.For multiuser mm Wave systems,the hybrid beamforming is designed as a matrix factorization problem and an effective greedy algorithm is proposed to solve the problem.In particular,the greedy algorithm based on the singular value decomposition does not depend on the array geometry and requires no extensive accurate information of the angle of the channel which might not be available in practice.The analog matrix is initialized by phase extraction to reduce the iterations.Based on the effective channel,the digital matrix is derived by low-dimensional digital beamforming.Furthermore,in order to avoid the matrix inversion in zero forcing(ZF)algorithm,a low-complexity initialization scheme is then proposed by handling the inter-user interference.Simulation results show that the greedy hybrid beamforming can maintain good performance.