Research On Channel Estimation Algorithm Based On Beam Search In Massive MIMO System

Millimeter wave(mmWave)massive multiple-input multiple-output(MIMO)has been regarded as a popular technology for the fifth generation(5G)cellular networks,due to its significant superiorities in energy efficiency and frequency efficiency.Accurate channel state information(CSI)is crucial for improving beamforming gain in the mmWave Massive MIMO systems,which can compensate the transmission loss of the mm Wave signal.However,the complexity of channel estimation schemes increases with the number of antennas.Considering the sparse property of the mmWave in the angle domain,scholars has proposed sparse mmWave channel estimation methods.However,the conflict between high estimation precision and heavy training overhead is a bottleneck to apply channel estimation schemes.Therefore,the study of fast yet accurate channel estimation methods has significant research value.This dissertation introduces the characteristics and superiorities of mmWave Massive MIMO,and the research status of beam searching techniques and sparse channel estimation schemes.Then we study and analyze the beam training-based sparse channel estimation method of mmWave massive MIMO system,which mainly include the following two aspects:Considering the limitations of beam searching techniques,such as the limited estimation precision and the poor performance in low signal-to-noise ratio(SNR),a novel adaptive high-resolution channel estimation(AHRCE)approach is proposed.Hierarchical search based on normal-resolution codebooks and maximum likelihood(ML)detection criteria is exploited to acquire the coarse estimated multipath components.The number of measurements and the set of candidate propagation paths can be dynamically adjusted among training levels.In the second stage,a sparse signal recovery method based on compressed sensing is designed to achieve high-resolution angles of arrival(AoAs)and angles of departure(AoDs)is introduced.The accurately estimated AoAs and AoDs are estimated by calculating a set of ratio metrics in virtual measurements process,without quantization error and additional pilot overhead.Numerical results indicate that the proposed scheme achieves a more efficient tradeoff between the pilot overhead and detection probability performance in broader SNR region than the other existing method.For the time-varying channel estimation in a mobile network,a two-stage fast beam tracking approach is designed to reduce the polit overhead of the beam search approach.In the first stage,using the temporal correlation existed between practical MIMO channels,the updated beam pairs are sought in the adjacent angular range based on the Neyman-Person criterion.In the second stage,the attained angular coarse range is refined based on hierarchical search approach.A discrete-state Markov process is applied to model the variation of AoA/AoD,which is used to analyze the detection probability,polit overhead and data rate of the proposed approach.The optimization function is constructed considering the polit overhead and data rate,and then a near-optimal probing beam width of the first stage in high SNR is obtained.Numerical results show that with the optimized beam width,the proposed approach can significantly achieve an efficient tradeoff between the data rate and pilot overhead under the guarantee of high successful tracking probability.