Research And Evaluation On CSI Enhancement For 5G NR Massive Mimo System

Massive MIMO is one of the major Physical Layer technologies of 5th-Generation(5G)New Radio(NR),and with R15 standard frozen,the performance of massive MIMO system can basically meet the requirements of the two major application scenerios in 5G,enhanced Mobile BroadBand(eMBB)and Ultra Reliable and Low Latency Communication(URLLC),which needs further improvement to make 5G better combined with vertical industries.High system performance depends on precisely obtaining Channel State Information(CSI)by accurate channel estimation and efficient CSI feedback.For NR system at the present stage,it can't be ignored that the reference signals related to channel estimation have higher Peak-to-Average Power Ratio(PAPR)than data signal and that CSI feedback has large overhead in case of multi-user transmission,which will affect transmission and coverage of signals and thus affect obtainment of CSI and also makes the corresponding codebook only supports up to two-layer transmission.Therefore,implementing CSI enhancement in R16 has great value to improvement of NR system.The topic of this thesis comes from National Science and Technology Major Project,5G International Standard Candidate Evaluation and Vertification.Based on the topic,combined with the background of 5G standardization research,this paper proposes two schemes on CSI enhancement in the perspective of obtaining and feedback:reducing the PAPR of the DeModulation Reference Signal(DMRS)and compressing the CSI feedback,which are evaluated based on the link-level and system-level simulation platform respectively.The main study content of this paper is divided into the following three parts:(1)Review on enhancement technologies of 5G NR massive MIMO system and simulation platform.The related technical background of 5G NR massive MIMO system enhancement is analyzed,especially the research situation at present of CSI enhancement.Also,the simulation flows of the general simulation platforms are given in the paper,including link-level and system-level simulation platform.(2)On the issue of the high PAPR channel estimation related reference signals affecting signal transmission and coverage,the paper proposes the technical scheme of CSI enhancement based on PAPR reduction of DMRS.Firstly,the reason why there exists PAPR issue in DMRS is explained,and by simulation,it can be seen that in case of Cyclic Prefix-Orthogonal Frequency Division Multiplexing(CP-OFDM)waveform and Discrete Fourier Transformation-spread-Orthogonal Frequency Division Multiplexing(DFT-s-OFDM)waveform,PAPR of DMRS is up to about 5dB and 2.5dB higher than that of data signal respectively.Then for both waveforms,new DMRS sequence generation schemes are proposed and implemented and evaluated based on link-level simulation platform.Simulation results show that new schemes can reduce PAPR of DMRS to the level of data signal,that is the gap is not more than 1dB.(3)On the issue of CSI feedback with large overhead being an obstacle of to extention of the number of layers the codebook supports,the paper proposes the scheme of CSI enhancement based on CSI feedback compression.Firstly,the overhead issue of CSI feedback is introduced by computing the overhead of Type ? and Type ? CSI feedback with the number of layers,quantization precision and etc.fixed.The overheads of them are 21 bits and 687bits respectively.Then,in respect of the two schemes determined by the 3rd Generation Partnership Project(3GPP),DFT beam-based CSI feedback compression and Singular Value Decomposition(SVD)decomposition-based CSI feedback compression,this paper makes specific designs and implement them based on the system-level simulation platform.It is found that in terms of the ratio of overhead reduction and cell average UE throughput,the two schemes can bring gain of 4.13dB and 3.07dB respectively compared with R15,while in terms of the ratio of overhead reduction and cell edge UE throughput,the two schemes can bring gain of 3.21dB and 2.85dB respectively compared with R15.It can be concluded that DFT beam-based CSI feedback compression has higher gain,laying the technical foundation for extention of the number of layers Type ? codebook supports.