Research On Channel Estimation And Signal Detection Techniques For Fast Time-varying Channels In 6G

The air interface and related waveforms of future communication systems must support various high-speed application scenarios,but orthogonal frequency division multiplexing(OFDM)technology widely used in fourth and fifth generation mobile communication systems faces serious subcarrier interference in high-speed mobile scenarios.In order to provide stable and reliable communication services in high mobility scenarios,6G systems can consider using Orthogonal Time Frequency Space(OTFS)in the delayed Doppler domain as their modulation scheme.The multipath channel in the OTFS system has orthogonal,sparse,and separable characteristics,and all transmitted symbols can obtain full diversity,which can effectively combat fast time-varying channels.Firstly,this thesis introduces the input-output relationship of the OTFS system,derives the equivalent channel matrix,and analyzes the time-invariant and sparse characteristics of the delayed Doppler domain channel.Based on this analysis,the single pilot channel estimation and its existing problems in OTFS systems were analyzed.Then,the use of CAZAC(Constant Amplitude Zero Auto Correlation)sequences for multi pilot channel estimation was proposed,and pilot insertion along the Doppler direction was proposed while removing some noise at the receiver.Secondly,this thesis introduces the commonly used MP(Message Passing)algorithm for signal detection in OTFS systems and analyzes its existing problems.After deduction,it was found that the symbol by symbol maximum likelihood detection algorithm can be used for OTFS signal detection,but the complexity of directly using this algorithm is too high.To address this issue,a maximum likelihood detection based on continuous parallel interference cancellation is proposed,which is mainly based on the two-dimensional convolution of the transmitted signal and channel information,while fully utilizing the detected signal to remove interference from the detected signal and reduce the complexity of the maximum likelihood detection algorithm.On this basis,zero values are inserted into the data matrix of the sending end,enabling the receiving end to achieve separation of received signals in the delay direction.Finally,both of the above contents are based on integer multiple Doppler frequency shift.However,due to the limitations of OTFS system parameters,the actual channel will exhibit fractional Doppler frequency shift at the receiving end,causing severe Doppler inter frequency interference.At the same time,due to the periodic limitation of Fourier transform,the OTFS system cannot describe a frequency offset greater than that between carriers in the delayed Doppler domain.In order to solve these two problems,this thesis analyzes the input-output relationships of OTFS systems in three domains.When comparing the input-output relationships in the delayed Doppler domain and the input-output relationships in the time-frequency domain,it is found that both have amplitude,phase distortion,and frequency to frequency interference caused by fractional Doppler frequency shift.The interference problem caused by fractional Doppler frequency shift can only be alleviated from the time-frequency domain to the delayed Doppler domain,but cannot be completely solved.In the delayed time domain,frequency offset is a constant without considering resolution,and it exhibits better sparsity compared to the delayed Doppler domain.This thesis believes that the good performance of OTFS systems is mainly due to the orthogonal and separable characteristics of channels with small delay and Doppler resolution,rather than transmitting signals through delayed Doppler domain,which expands the understanding of OTFS and OFDM systems.This thesis proposes a signal transmission scheme using an OTFS system channel model to transmit signals in the time-frequency domain,changing the symbol transmission order in the delay time domain to enable hard decision at the receiving end.At the same time,it proposes using maximum ratio merging for signal detection in the delay time domain of the receiving end and provides the corresponding algorithm flow design.