Implementation Of 5G-NR Random Access Channel Based On Vector Processor

In recent years,the fifth-generation mobile communication system has been applied to commercial systems in various industries due to its superior characteristics.As an indispensable part of 5G,the random access process is the process required for the user to establish a basic connection with the network.Therefore,accurately analyzing the physical random access channel and realizing uplink synchronization is the basis of the subsequent communication process.How to accurately detect the preamble signal and meet the requirements of low latency is a key issue to design an acceleration algorithm according to the characteristics of the hardware computing platform.Therefore,this thesis mainly studies the realization of 5G NR physical random access channel based on vector processor.The main work is as follows:Firstly,the random access process of 5G system is studied in this thesis,and the key technologies of PRACH transmission and reception are deeply analyzed in combination with the relevant provisions of the3 GPP protocol on the physical layer of random access.Then this thesis simulates the PRACH sending and receiving process.Among them,for the generation of the preamble sequence at the transmitting end,this thesis combines the research and uses the generation algorithm of the frequency domain preamble sequence,which reduces the time complexity compared with the traditional time domain generation and the use of DFT operation;in the selection of the correlation algorithm at the receiving end,this thesis Through the algorithm derivation and comparison,the frequency domain correlation operation is finally selected between the time domain correlation algorithm and the frequency domain correlation algorithm.For peak detection,this thesis proposes an improved multi-threshold decision peak detection algorithm.The simulation results show that for AWGN and TDLC300-100 channels and long and short preamble sequences,the algorithm is better than the traditional peak detection algorithm,the performance is improved by 1-2d B,and it meets the requirements of the protocol,and has engineering realization value.Finally,combined with the improved peak detection algorithm,the whole process of PRACH sending and receiving is implemented on the CEVA-XC 4500 DSP platform by vector processor.According to the platform’s vector parallel design idea,the complex multiplication,complex addition,data shift and data comparison operations in the algorithm are implemented in parallel.The implemented algorithms include frequency domain correlation,time domain filtering and peak detection algorithms.This thesis compares the success rate of preamble detection between floating-point simulation and vector processor fixed-point implementation.The results show that the vector processor implementation scheme can meet the parameter requirements specified in the protocol,and the parallel implementation based on the vector processor can effectively shorten the running time of the code.