Research On Fast Search Algorithm For Beamforming Technology In Millimeter Wave System

In recent years,with the continuous development of multimedia application,the data traffic of wireless communication applications continues to grow.In order to deal with the growth of wireless link capacity,adapt to the requirements of increasing transmission rate in wireless communication,and avoid channel crowdness and mutual interference between channels,indoors and outdoor millimeter wave communication technology has become a new frontier for deploying high-speed data links.Although there are many advantages of millimeter wave communication,such as the large communication capacity,good confidentiality,and high transmission quality,it also exists limitations.The most basic challenge of millimeter-wave communication is that communication in such a high frequency band causes extremely high path loss.To compensate for this important link budget gap,joint transceiver beamforming technology is required to generate a very directional narrow beam so as to provide a large antenna array gain.The large-scale antenna array makes it expensive and difficult to obtain complete channel state information(CSI)and channel estimation will be more difficult especially in multi-user MIMO system.Compared with the adaptive beamforming technology which needs to adjust the antenna dynamically according to the feedback about CSI from the receiver,the analog beam training based on fixed codebook is much easier to implement.For analog beam training,the most important thing is to find the best beam pair in the pre-designed codebook quickly and efficiently and the best beam pair is found according to the current channel conditions.In order to improve the performance of beam training,the fast search algorithm of beamforming technology based on fixed codebook in millimeter-wave system are studied in this paper and the main work is organized as follows:(1)The exhaustive search is time-consuming in the codebook-based beamforming technology,a hierarchical search method based on hierarchical codebook is adopted to reduce the complexity of beam searching in this paper.In the case where the total transmit power of the transmitting antenna is limited or the transmit power of the single antenna is limited,the performance of two different hierarchical codebooks is different in both line of sight(LOS)channle and non-LOS(NLOS)channel.Three different hierarchical search algorithms,unilateral search,crossed search and improved unilateral search are proposed and the best one is obtained in this paper.Further,under LOS channel,a single-ended search mode that further reduces the search complexity and improves the search success rate is proposed and the best hierarchical approach for hierarchical codebook is obtained with the fixed search complexity.(2)For the hierarchical codebook with aligned structure,the beam gain is low in the low layer and the beam gain at the beam boundary in the low layer is much lower.Besides,the boundary angles of the wide beams in the low layer are still the boundary angles of the narrow beams in the high layer.Those above make large probability of the search error concentrated on the boundary angle of the wide beams in the low layer and aligned structure makes there is no chance to correct the search error brought by the the wide beam searching.As a result,the accuracy of the hierarchical search algorithm has a decrease.Therefore,a hierarchical codebook with non-aligned structure is proposed to improve this situation,and the number of beams which should be checked in each layer is adjusted according to the non-aligned codebook structure.Compared with the aligned hierarchical codebook,the hierarchical codebook with non-aligned structure significantly improves the success rate of the hierarchical search algorithm under both the LOS channel and the NLOS channel.In order to realize the multiplexing gain in the millimeter-wave system,hierarchical codebook with non-aligned structure is further used in this paper to implement multi-beam search,and the accuracy of multi-beam search is greatly improved by increasing the oversampling layer search.