针对FDD模式下大规模MIMO系统中下行信道信息获取及有效克服信道时变性问题，本项目拟基于长时码本和短时码本对下行信道信息进行量化反馈，由网络端基于各用户波束方向动态估计长时码本，终端基于压缩感知及卡尔曼滤波理论估计短时码本。重点研究内容是：利用上下行波束方向的对偶性，直接由网络端基于各用户的波束方向估计获取长时码本，无需进行长时码本反馈，难点在于研究Massive MIMO系统中在不同天线阵列下的多用户波束方向估计算法；为了适应长时码本的时变性，搭建用户移动与波束方向之间的变化关系特征模型，并据此提出动态长时码本估计策略；根据压缩感知理论获得压缩的信道估计信息，终端基于压缩信道进行卡尔曼信道预测克服短时码本时变性，并估计短时码本，无需信道重构。整个策略通过分步估计以及压缩估计算法降低估计及预测复杂度，降低反馈资源占用量，同时通过动态估计长时码本和卡尔曼滤波预测信道克服时变信道问题。

With respect to acquiring the downlink channel information(CSI) and solving its time-varying problem, we attend to feed back by quantifying CSI with a long-term codebook and a short-term codebook in this project. The eNodeB estimates the long-term codebook dynamically based on the DOA estimation of each UE. The UE estimates the short-term codebook based on compression perception and Kalman filter theory . The key research contents are as follows: Firstly, based on the uplink-downlink duality of the beam direction, long-term codebook is estimated by eNodeB based on the DOA estimation, which can avoid the feedback. The difficult point focuses on the multi-user DOA estimation under different antenna arrays in the Massive MIMO system. Secondly, in order to adapt to time-variability of long-term codebook, the relationship of user mobility and DOA variation is studied. Accordingly, a dynamic long-term codebook estimation strategy is put forward. Finally, Based on the compressed sensed theory, Kalman filter is applied to predict the compressed CSI to overcome the time-variability of short-time codebook, which is estimated based on compressed CSI without reconstruction by UEs. In a whole, with this project research, classification estimation and compression estimation algorithm can decrease the complexity of estimation and prediction, and reduce the feedback resource. What’s more, the problem of channel time-variability is overcome by dynamical long-term codebook estimation and channel prediction with Kalman filter.