Research On Doppler Frequency Shift Estimation Methods In 5G-NR High-speed Mobile Scenarios

High-speed railways have received more and more attention due to their economic,environmental protection,and all-weather operation advantages.Advances in economy,science and technology have promoted the rapid development of high-speed railways.At present,the speed of high-speed rail can reach 300km/h or even faster.As one of the important application scenarios of5G-NR,high-speed railways have increasing communication requirements and have higher requirements on the quality of communication systems.The International Telecommunication Union?ITU?in the formulation of key performance requirements for 5G requires the downlink rate to reach 50Mbit/s and the uplink to 25Mbit/s in high-speed rail communication scenario with a speed of 500km/h.In 5G-NR communications,high-speed rail broadband mobile communication technology faces huge challenges,which mainly include:large car body penetration loss,frequent handoffs,and large Doppler frequency shifts.Therefore,in order to meet the key performance requirements in 5G-NR high-speed rail communication scenarios,more research on related communication technologies is required.Considering the overall performance and requirements of the 5G-NR communication system,after evaluating the advantages and disadvantages of the single carrier and multiple subcarrier waveforms,3^rd Generation Partnership Project?3GPP?finally chose Orthogonal Frequency Division Multiple?OFDM?as the Uplink and Downlink transmission technology.However,in the 5G-NR high-speed mobile communication scenario,the higher speed movement of the vehicles?>=500km/h?and higher carrier frequencies will cause larger Doppler frequency offsets,which will seriously destroy the orthogonality between the subcarriers of the OFDM system,and make the channel time-varying rapidly,thereby reducing the transmission performance of the system.To solve the impact of Doppler frequency offset on the communication system,the anti-Doppler frequency shift technology is needed and the Doppler frequency shift extimation and compensation technology is the basis.This thesis aims to improve the accuracy of Doppler frequency offset estimation and reduce the complexity of Doppler frequency offset estimation.Based on the existing Doppler frequency offset estimation methods,we research more efficient and practical Doppler frequency offset estimation methods in 5G-NR high-speed mobile scenarios.The main contents and innovations are as follows:?1?This thesis uses the technology of wireless environment map to propose a Doppler frequency offset estimation method?radio map and pilot estimator,RMPE?,which combines the environment map and the segmented pilot.This method mainly solves the problem of limited accuracy of Doppler frequency offset estimation of the 5G-NR high-speed railway communication uplink transmission system.In the method,based on the regularity and predictability of the route of the high-speed railway,the environment map method is firstly used to obtain the initial estimation of the Doppler frequency offset.To improve the accuracy of frequency offset estimation and meet the needs of the large frequency offset estimation range in this scenario,a maximum posterior probability estimation method based on pilot segmentation is used.This method takes the initial estimate of the environment map as a priori information and uses the pilot segmentation method to obtain the maximum a posteriori estimate.Theoretical analysis and computer simulation results show that the method's estimation range and accuracy can meet the requirements for large frequency offset estimation in the 5G-NR millimeter wave high-speed railway scnario.Its estimation performance is much better than the existing methods,and it has lower computational complexity.?2?Based on the research of neural network and machine learning in the field of wireless communication,this thesis proposes a back propagation neural network estimator?BPNE?based on BP neural network for 5G-NR high-speed railway communication system.The method is mainly divided into two stages of offline training and online estimation.First,the random Doppler frequency offset and pilot data symbols are used to construct a training sample set and the training sample set are used to train the BP neural network offline to build the mapping relationship between the input and output data.Then based on the mapping relationship,the Doppler frequency offset estimation is performed using the pilot data symbols in the received signal.This method requires only a part of the pilot data to participate in the calculation of the frequency offset estimation,which has a good estimation performance.