Study On Double Codebook Beamforming Technique

Multi-antenna technology can make use of spatial degrees of freedom to achievespatial diversity and multiplexing, effectively improve the spectrum utilization. Amongthem, beamforming is a main technology to implement space division multiple access inMIMO system, mainly divided into two categories. One is a long-term beamformingtechnology, applied to a closely spaced antenna array. The main principle is to use thestrong correlation of spatial channel and interference principle of waves producingdirectional radiation pattern, so that the main lobe of the radiation pattern adaptivelydirects to users. The other is a short-term beamforming technology, based on real-timechannel impulse response, using a weak correlation between antenna array elements tobuild signal processing algorithms. With complete channel information in the senderand receiver, beamforming technology can easily and effectively achieve maximumdiversity gain. When the distance of antenna is between half and10times of thewavelength, and the azimuth spread is small, the long-term and short-term beamformingtechnology can work together. Therefore, double codebook technique is introduced intothe LTE-Advanced (LTE-A) system, which indicates the long-term and short-termbeamforming characteristics by two codebooks respectively. Further, since the changespeeds of long-term and short-term characteristics are not the same, two estimation andfeedback periods of them are different. Hence the feedback period of long-termcodebook can be extended to reduce the average amount of feedback and improve theresource efficiency.Although the double codebook is defined in LTE-A, the estimationand feedback of double codebook are the key problems to be solved.This thesis studies double codebook beamforming technology, and based on therelevant theoretical research proposes a series of solutions. The main contributions areorganized as follows:Firstly, it analyzes the design characteristics and estimation algorithm of doublecodebook. In the OFDM system of wideband signal, frequency offset exists betweendifferent subcarrier, so there is codebook estimation deviation in the independent use ofthe channel information carried on the different carriers. The thesis proposes a coherentestimation algorithm based on the correlation between multi-carriers of wideband signal,focusing the different channel information to a reference frequency for the equivalentchannel information. What is more, weighted average of the channel information atdifferent sample times is proposed, which take the SNR (Signal Noise Ratio) as the weighting factor. As a result, a large deviation due to some poor samples can beovercome. Through theoretical analysis and numerical simulation, coherent estimationalgorithm can improve performance, especially for wider band signals.Secondly, it studies the effect of IQ (in-phase/quadrature-phase) imbalance on thedouble codebook beamforming system, and derives the codebook estimate deviationcaused by imbalance. According to signal processing theory, the character of codebookestimation with IQ imbalance is deduced, and a complete joint Tx_Rx IQ imbalanceestimation algorithm is proposed, which can estimate and compensate Tx_Rx IQimbalance parameters step by step so as to improve system capacity.Thirdly, in order to reduce the uplink feedback resource, the feedback period offirst codebook for long-term codebook is usually longer. However, due to thetime-varying channel change speed, a constant feedback cycle will affect the accuracyof the codebook1. Thus it proposes an adaptive feedback strategy of that codebookbased on improved non-equidistant gray prediction algorithm. Further, the short-termcodebook estimation with higher real-time requirement may be severely degraded bythe outdated channel information. Kalman prediction algorithm is introduced to predictthe channel estimate matrix, and the state transition matrix is corrected by grayprediction algorithm, so as to improve the estimation accuracy.Finally, in Massive MIMO system model, with the increasing number of antennas,the channel estimation information is multiplied with complexity of the feedbackinformation. In addition, antenna array of massive MIMO is always a two-dimensionalstructure, and channel direction estimation also needs to be expanded intotwo-dimensional direction estimation. Thus, a double codebook design and estimationscheme is proposed based on the two-dimensional channel estimation principle, andeffectiveness of the scheme is verified by simulation. With a variety of codebookestimation algorithm simulation comparison, the thesis also compares the performanceand complexity of three estimation algorithms. Among them, the complexity ofestimation algorithm based on ESPRIT principle is the lowest, and the performance isalso close to the improved two-dimensional double codebook estimation algorithm.