Research On Limited Feedback Precoding Schemes Using Statistic CSI For Three Dimensional MIMO

With the ability of improving the spectrum efficiency and system capacity significantly by fully exploiting the spatial resource, MIMO has become one of the widely-adopted key technologies in the advanced mobile communication systems. The system performance would be further enhanced by precoding at the BS with the knowledge of CSIT.In systems without reciprocity between the uplink and the downlink(e.g., FDD systems and uncalibrated TDD systems), the CSIT is obtained with the aid of limited-feedback technology. Different from the 2D MIMO system considering the horizontal CSI,3D MIMO systems in which AAS is deployed at the BS take both horizontal and vertical CSI into consideration. With the freedom degrees provided by the AAS, BS steer beams to the desired UEs in both azimuth and elevation that increase the received SNR and suppress the interference from other UEs and adjacent cells. However, the channel dimension expansion caused by the deployment of AAS makes the codebooks designed for 2D MIMO are no longer suitable for 3D MIMO. It is necessary to redesign strategies in accordance with the characteristics of 3D MIMO channel. Fortunately, because of the character of changing slowly and the ability to indicate the channel statistic properties, statistic CSI is used for the design of codebook and MU-MIMO transmission with a favorable tradeoff between performance and feedback overhead. With the aforementioned background, the research focuses on limited feedback precoding scheme using statistic CSI for 3D MIMO system.Firstly, some basic principles of MIMO is given, followed by an introduction to the AAS and its antenna model. After that, a summary of MIMO channel models is demonstrated. Besides, two kinds of 3D MIMO channel models are investigated deeply, including the channel model proposed by 3GPP TR 36.873 and a simplified ray-based one.Then, a survey of precoding schemes based on both instantaneous CSI and Statistic CSI feedback is presented. Simulations in both SCM and 3GPP TR36.873 channels demonstrates that double codebooks(such as AC, WDFT and DC) using statistic CSI obtain certain performance gain in comparison with single codebooks(e.g., GLP, DFT, RVQ etc.).In the precoding strategies using statistic CSI, statistic eigenmode-spatial division multiple access(SE-SDMA) is mainly introduced and analyzed. With the aid of Statistic CSI, SE-SDMA can obtain considerable system capacity and spectral efficiency with low feedback overhead in aid of statistic CSI.Subsequently, a double codebook based on weighted KPC is put forward. Considering that double codebooks outperform single codebooks in 2D MIMO channel, it is found that 3D channel can be approximated by a linear combination of paths directing to multiple scattering clusters and the direction information of the paths can be drawn from the channel correlation matrix. Therefore, KPC is used as the long-term codebook to quantify the direction information, and a short-term codebook(e.g., GLP or RVQ) can be used to combine the long-term codewords.BS can reconstruction the CSI with the feedback. Simulation results shows that weighted KPC codebook approaches the spectrum efficiency achieved by ideal feedback with moderate feedback overload.Finally, a SDMA scheme using statistic CSI for 3D MIMO is proposed. It is verified in the 3GPP TR36.873 channel model that 3D MIMO channel can support MU-MIMO excellently. Considering the spatial correlation characteristics of 3D MIMO channel, the user-clustering codebook is designed as the kronecker product of rotated DFT matrices, and SE-SDMA can fully exploit the spatial degrees of freedom provided by the differences between UEs. It is demonstrated that the achievable system average through-output is considerable.