Channel Estimation Technology Implementation For 5G-NR Shared Channel Based On Vector Processor

With the rapid development of mobile communication technology,the fifth generation of mobile communication systems with its superior characteristics have been applied to commercial systems in various industries.As one of the key technology carriers of the 5G-NR physical layer,the shared channel carries uplink and downlink business data,which has the characteristics of high bandwidth and large data volume,which put forward higher requirements for the accuracy and speed of channel estimation.Therefore,this thesis mainly studies the channel estimation technology of 5G-NR shared channel based on vector processor.The main tasks are as follows:Firstly,this thesis introduces the research background and current status of the thesis,and analyzes the research status of channel estimation and the development status of hardware computing platform.Then,the shared channel of the 5G-NR physical layer is described in detail,and the frame structure parameters,resource configuration,and DM-RS design scheme are given.According to the system design ideas,the basic process and structure of the sending and receiving ends,and some key technologies of the receiving end are discussed.In addition,the basic characteristics of the wireless fading channel are briefly introduced,and the phenomena such as channel delay expansion and Doppler frequency shift are analyzed.Secondly,in order to improve the channel estimation performance,this thesis studies a channel estimation interpolation coefficient generation scheme based on the channel power delay profile(PDP)and the Doppler spectrum,derives the channel frequency domain correlation coefficient based on channel PDP,and derives the time domain correlation coefficient based on the Doppler spectrum.In addition,according to the relationship between the channel time domain correlation coefficient and the Doppler spectrum,the maximum Doppler frequency coarse estimation algorithm based on the channel autocorrelation function is studied.Simulation results show that the method of using preset channel Wiener filter interpolation coefficients will have partial performance loss,but the algorithm complexity is low,and the algorithm that generates the Wiener filter interpolation coefficients in real time from the current channel PDP and Doppler spectrum can obtain better channel estimation performance,but the algorithm complexity is higher.Furthermore,in order to reduce the complexity of the algorithm,this thesis proposes a discrete cosine transform(DCT)channel estimation algorithm based on channel PDP.Firstly,two classic transformation domain interpolation channel estimation algorithms are studied: discrete Fourier transform(DFT)interpolation algorithm and DCT interpolation algorithm,and how to reduce noise in the transformation domain and improve the performance of transformation domain interpolation under the condition of transformation domain interpolation is analyzed.On this basis,according to the characteristics of DCT,the Wiener filter noise reduction vector is simplified.Simulation results show that compared with the traditional DCT channel estimation scheme,the proposed algorithm can improve the channel estimation performance under low complexity.Finally,according to the parallel characteristics of the vector processor,the channel estimation algorithm of the shared channel is optimized and programmed.The implementation scheme includes the least squares channel estimation algorithm,the frequency domain DCT interpolation algorithm,and the time domain Wiener filter interpolation algorithm.The C language and parallel instructions are used for programming,parallel acceleration is carried out by using the feature of large-bit wide vectors,and the compilation optimization option is used to improve the efficiency of code execution.The implementation performance of the parallel design scheme is then shown,and the test results show that the parallel operation based on vector processor can greatly reduce the runtime of the code.