Implementation Of 5G-NR Uplink Sounding Channel Based On Vector Processor

With the development of mobile communication technology,the 5th Generation Mobile Communication Technology(5G)with higher speed,lower delay and more users has appeared in people’s lives,promoting the progress of communication technology.Research on the uplink sounding reference signal(Sounding Reference Signal,SRS)is an essential part of 5G.Based on the 3GPP R15 5G-NR physical layer protocol,this thesis designs and implements the uplink SRS transmitter and receiver.The main contents of the thesis are as follows:First,the time-frequency resources,transmission configurations and resource sets of uplink sounding reference signals in the 5G-NR standard are studied.On this basis,the uplink SRS transmitter technology is studied,including the configuration of high-level parameters,the generation of SRS sequences,the mapping of physical time-frequency resources,and the generation of baseband signals.In addition,according to the processing flow of the uplink SRS transmitter,the floating-point simulation platform of the SRS transmitter is built,which proves that the uplink SRS transmitter technology has certain feasibility.Then,the channel estimation technique of the uplink SRS is studied.In order to solve the problem of multi-user separation,the conjugate separation method and the DFT window separation method are used for channel estimation.Without considering the noise in the window,the DFT windowing and separation method under different window lengths is simulated,and the results show that the window length L is M/8 The denoising DFT windowing method in the window is M/8 The separation method has the best performance.In order to reduce the influence of noise on the channel estimation performance,the noise threshold is set to denoise the conjugate separation method and the DFT window separation method.The noise threshold is affected by the δtime domain points with the largest power in the time domain.After simulation,the denoising conjugate separation method with δ being 25 and the DFT windowed separation method with δ being 25 and window length L being M/4 have better performance it is good.At the same time,comparing the algorithms before and after denoising,the results show that the window length L is M/8,the performance of the unwindowed DFT and windowed separation method is the best.Then,the performance of linear interpolation and Wiener filter interpolation is compared.The simulation results show that the performance of Wiener filter interpolation is better than that of linear interpolation.On this basis,the signal-to-noise ratio(SNR)estimation performance of the three noise estimation calculations of the idle sub-carrier method,the minimum mean square error method,and the energy concentration area method is compared.The simulation results show that the signal-to-noise ratio(SNR)estimation performance of the idle subcarrier noise estimation method is the best,and the realization scheme of channel estimation is proposed as DFT window separation method of the window length being M/8 without denoising in the window,Wiener filter interpolation and noise estimation method of idle subcarrier.Finally,the SRS transceiver is designed and implemented on the CEVA XC4500 DSP vector processor based on the uplink SRS transmitter and receiver processing flow.Based on the proposed channel estimation implementation scheme,the parallelism analysis and research of the DSP fixed-point implementation of the SRS channel estimation module are carried out,and the channel estimation module is optimized.At the same time,the performance analysis of the DSP fixed-point implementation of the SRS transceiver is carried out from the aspects of memory occupancy and running time.In addition,the performance of multi-user separation,Wiener filter interpolation,and signal-to-noise ratio(SNR)estimation floating-point simulation and CEVA XC4500 fixed-point simulation is compared to verify the feasibility of channel estimation scheme on the DSP platform.