Research On Channel Estimation Technology Of 5G-NR Demodulation Reference Signal Based On Vector Processor

With the increasing demand for high-capacity,low-latency,high-reliability,high-speed,high-connection,and energy-efficient communications,5G,a new generation of mobile communication systems,emerges as the times require.Channel estimation is an extremely important link in the physical layer of wireless communication systems,and channel estimation has a very critical impact on the performance of wireless communication systems.In order to meet the ultra-low latency requirements of 5G communication systems,optimizing the real-time performance of baseband signal processing in vector processors is a hot trend in current research.From the perspective of 5G communication base station,the thesis conducts a detailed study on the channel estimation module in the physical layer uplink shared channel receiver,and implements the channel estimation module on the CEVA-XC4500 DSP vector processor platform.The first part of this thesis introduces the research background of the thesis,gives a brief overview of the design of the base station chip and introduces the structural characteristics and advantages of the DSP vector processor,then explores the research status of channel estimation in detail,and arrange the structure of the text.The second part of this thesis studies the signal propagation characteristics and wireless communication channel model of wireless mobile communication,introduces the OFDM technology and MIMO technology used in the 5G communication system,builds a single-antenna OFDM system model and studies the 5G frame structure,time-frequency domain resources and transmission time interval in combination with the3 GPP protocol R15 version,which provides a theoretical basis for the following research.The third part of this thesis studies the channel estimation algorithm based on DMRS for PUSCH channel in 5G-NR,introduces the generation of DMRS for PUSCH channel in 5G standard protocol and the timely frequency domain resource mapping,introduces several channel estimation algorithms for DMRS locations,performs interpolation filtering based on the channel frequency response values of DMRS locations to obtain channel frequency response values for non-DMRS locations,studies the OFDM symbol timing deviation and its influence on the interpolation results,builds a 5G PUSCH channel link on the MATLAB simulation platform,and compares the performance and complexity of the aforementioned algorithms through this link.The fourth part of this thesis introduces the CEVA-XC4500 DSP vector processor platform,describes the engineering development process based on the platform,and introduces the fixed-point processing of data in the platform.Then,according to the comparison results of the algorithms in the third part,a set of algorithm combinations with excellent performance and in line with the characteristics of the vector processor are selected.Based on the hardware structure of the vector processor,the Vec-C instruction set is used to design the algorithm of each sub-module in the channel estimation module.By dismantling and subdividing the algorithm,changing the internal instruction execution order of the algorithm,making full use of the parallelism of the vector processor,and optimizing the modules to maximize the hardware resources in the vector processor.Finally,a set of channel estimation solution for 5G-NR PUSCH based on this platform is designed.Under the condition that the performance of floating-point operations is unchanged,the running time is greatly reduced.According to the experimental results,when the system is configured with 25 MHz bandwidth,30 k Hz subcarrier spacing,and 2 × 2 transceiver antennas,the total number of running Cycles is180669,which takes 0.1390 ms in total,which meets the experimental requirements.