Research On Reference Signal Design And Channel Estimation In OTFS Systems

With the development of beyond the fifth generation communications(B5G)and the sixth generation communications(6G),researchers have begun to focus on how to ensure the feasibility and stability of communication systems in highmobility scenarios.In high-frequency and high-mobility environments supported by B5G/6G,Doppler shifts lead to serious inter-carrier interference(ICI)and cause a decline in the overall performance of the orthogonal frequency division multiplexing(OFDM)system.A technique dubbed orthogonal time and frequency space(OTFS)was published.In the OTFS system,a time-varying channel can be converted into an invariant channel in the delay-Doppler(DD)domain through a two-dimensional transformation.When there are high Doppler shifts and high time delays,the OTFS system performs better.We mainly study the reference signal design and channel estimation for the OTFS system.We analyze the OTFS system model firstly.Due to the limited frame durations,fractional Doppler shifts usually exist.The fractional Doppler shift results in DD channel dispersion,which increases the difficulty of channel estimation.The existing channel estimation schemes have good performance when there are only integer Doppler shifts.However,they are unable to accurately estimate the fractional Doppler shifts,resulting in the loss of performance when fractional Doppler shifts exist.Therefore,a novel channel estimation scheme is proposed.It firstly estimates the Doppler shift by linear fitting based on the channel dispersion in the DD domain and then calculates the power of dispersion paths as combining coefficients to obtain channel fading.Simulation results show that the proposed scheme can significantly reduce the normalized mean square error(NMSE)compared with conventional schemes.Due to the limitation of the number of reference signals,channel estimation is more difficult in complex environments.Therefore,a new double-impulse reference signal design is proposed.We also propose three estimation schemes.The first channel estimation scheme is based on the least square criterion and has the lowest computational complexity.The second channel estimation firstly compensates for the phase difference between two reference signals and then estimates the fractional Doppler shift,which has higher complexity.When two channels have the same delay but different Doppler shifts,the third channel estimation scheme calculated the channel correlation function in the DD domain,then enhances the channel estimation.It has the highest complexity.Simulation results show that the performance advantages of the dual-impulse reference signal design are significant,especially in complex channel environments.Moreover,the reference signal design and channel estimation in the timefrequency(TF)domain are rarely studied.Considering that the receiver in the TF domain needs to be supported,the reference signal design and channel estimation in the TF domain are also studied in this thesis.Two schemes of reference signal design are proposed:the full transmission block reference signal scheme is designed for scenarios where channels change slowly,whereas the comb reference signal scheme is designed for scenarios where channels change fast.We also propose two TF domain channel estimation schemes:the linear filtering channel estimation and the channel parameter iterative estimation.The former has lower complexity,whereas the latter has better performance.The iterative channel parameter estimation scheme decomposes the channel matrix and then estimates the time delay,the Doppler shift,and the channel gain.It can be designed as an iterative process to improve accuracy.Finally,we compare the reference signal design and channel estimation schemes in the TF domain and in the DD domain.Simulation results show that the reference signal scheme in the TF domain is more flexible.However,because of ICI,the performance of channel estimation schemes in the TF domain are worse than those in the DD domain under the same overhead.