Research On Channel State Information Quantization In Full Dimention Multi-input Multi-output Systems For 5G

Discussions on the fifth-generation(5G)mobile communications system is already in full swing to meet new and unprecedented demands,which include explosive growth of data traffic,massive increase in the number of devices and continuous emergence of new services and application scenarios.As a key can-didate for 5G,full dimension multi-input multi-output(FD-MIMO)utilizes a large number of antennas placed in a two dimension(2D)antenna array panel for realizing spatially separated transmission links to a large number of mo-bile stations.Meanwhile,it can realize three dimension(3D)beamforming by making use of the vertical freedom of the channels.Channel state information(CSI)quantization is one of the main constraints on the improvement of FD-MIMO.On one hand,FD-MIMO support up to 64 antenna ports placed in a 2D array,largely increasing the complexity of CSI quantization.On the other hand,to completely explore the freedom of vertical dimension,it is critical to take the vertical channel information into considera-tion when quantifying the CSI.To solve the problems above,this thesis first validates the effectiveness of the Kronecker product based codebook for 3D MIMO channels.Further-more,the optimal codebook construction criterion is derived and a near-optimal codebook construction algorithm is proposed.Through designing the codebook which matches the characteristics of 3D MIMO channels,this thesis makes full use of the vertical freedom of 3D channels.Moreover,to reduce the high CSI quantization complexity due to the large scale antenna array,this thesis fur-ther proposes a low complexity channel quantization scheme which reduces the complexity from exponential to linear with the number of antennas by mak-ing use of the duality between codebook searching and non-coherent sequence detection.The main contributions are described as follows:1.This thesis demonstrates that the codewords in the Kronecker product based codebook(KPC)approximately match the distribution of the opti-mal beamforming vectors under spatially correlated 3D channels,showing the effectiveness of such codebook structure.Moreover,this thesis inves-tigates the structure of the codebook and then find a criterion of construct-ing the optimal KPC to achieve the lowest cross-correlation amplitude.Furthermore,considering the optimal codebook only exists under certain conditions,a modified discrete stochastic approximation algorithm is used to construct a near-optimal codebook.Simulation results show that our al-gorithm greatly reducesImaxand respectively provide 1dB and 0.5dB cod-ing gain compared to the traditional DFT codebook and the original KPC.2.In FD-MIMO systems where hundreds of antennas are deployed,the quan-tization complexity will be unacceptable.To reduce the complexity as well as make use of the vertical freedom of channels,this thesis proposes a low complexity channel quantization scheme based on KPC which reduces the complexity from exponential to linear with the number of antennas by making use of the duality between codebook searching and non-coherent sequence detection.The full diversity order is also demonstrated to be achieved,showing that the proposed limited feedback scheme does not affect the diversity gain of the system.Monte Carlo simulation results show that compared with traditional 2D CSI quantization schemes,the proposed scheme provides at least 0.8dB coding gain.Moreover,the pro-posed scheme outperforms other FD/3D CSI quantization schemes with moderate complexity when the channel is highly spatial correlated.3.System-level evaluation for the KPC based 3D beamforming is made by building the platform according to 3GPP standard.The simulation re-sults show that under the current antenna configuration,the KPC based 3D beamforming performs better than the convention codebook based 3D beamforming.However,the gain is not obvious due to the limited antenna number.To further improve the system performance,antenna number should be increased or other enhancement technology need to be adopted.Finally,a brief summary of this thesis is made and possible future research contents are pointed out.