Waveform Optimization And Low-complexity Beam Training For Millimeter Wave Beamforming

The codebook based mm-wave analog beamforming has been the focus of future mm-wave communication development as it's sufficient to use analog beam codebooks for beam training to estimate the dominating channel components due to the sparse mm-wave channel model.However,for the single-user scenario,there is still a lack of research on beam pattern optimization based on data transmission requirements.In addition,the specific impact of beam synthesis on system capactiy is still left to be found.At the same time,for multi-user scenario,the increasing number of users also causes a great amount of complexity of multi-user beam training using exhausive search.According to above problems,for single user scenario,the optimization problem about the optimal top-flat main gain is established and the corresponding algorithms are proposed to generate the desired beam pattern in this dissertation.After that,a closed form relationship is approximately derived in this dissertation in terms of the beam gains,the antenna number and the beam width of the main lobe.Following the characteristics of codebook based analog beam communication,a new communication mechanism with longer coherence time which can improve the system capacity during data transmission is proposed in this dissertation.The closed solution of the capacity is then derived under the new mechanism using optimized beam codebook.At last,the optimal setting range of the main lobe which can reach the maximum system capacity is obtained by simulation results.For multi-user scenario,the codebook based multi-user beam training process is constructed into an optimal assignment problem.To reduce the huge amount of complexity caused by traditional exhaustive method,based on the “greedy” idea,conventional greedy method is improved by proposing the mean as well as variance as the new enhanced preference of greedy algorithm,making a satisfying compromise between capacity and complexity performance.However,with the number of users becomes greatly large,the complexity of the proposed enhanced greedy algorithm prohibitively rises.This dissertation further improves the enhanced greedy algorithm and a single enhanced greedy algorithm is proposed,which not only significantly reduces the complexity,but also restrain the proportional relationship between the complexity and the number of users.