Key Technology Research On Orthogonal Time-frequency Space(OTFS) System

On the time-varying channel with high Doppler spread,the performance of traditional Orthogonal Frequency Division Multiplexing(OFDM)systems deteriorates seriously,and the complexity of signal processing increases sharply.Orthogonal Time Frequency Space(OTFS)technology introduces equivalent channel description in the time-delay Doppler domain and two-dimensional orthogonal extension in the timefrequency domain,each modulation symbol in the transmission block experiences almost the same and slowly changing sparse channel in the time-delay Doppler domain and obtains full channel diversity in the timefrequency domain.This feature makes the OTFS system significantly improve the robustness to Doppler spread and effectively balance the complexity of the receiver.Therefore,OTFS is considered as an important technical means to overcome the communication challenges in high mobile scenarios,which has attracted extensive attention in the industry.This thesis focuses on key technologies of OTFS system,including Out-of-band Emission(OOBE),Peak to Average Power Ratio(PAPR)and channel equalization.The main work is as follows:(1)Aiming at the problem of high OOBE in the current rectangular pulse shaped OTFS system,an OTFS system based on Bi-orthogonal Frequency Division Multiplexing(OTFS-BFDM)is proposed.On this basis,a low complexity transceiver structure compatible with OTFS is designed to implement OTFS-BFDM system,and the equivalent channel matrix of the system is derived to prove its sparsity.The linear complexity of Message Passing(MP)signal detection is adopted to approach the maximum likelihood detection.The simulation results show that the OOBE performance of OTFS-BFDM system under any non-rectangular pulse forming is significantly better than that of OTFS system.The better the frequency domain localization of shaping pulse,the smaller the bit error rate and PAPR.Compared with OTFS system,the bit error rate of OTFSBFDM system is almost no loss,but the OOBE is evidently reduced when the shaping pulse is located as a rectangle in the frequency domain.(2)Aiming at the problem of clipping noise interference in clipped OTFS system used for PAPR suppression,a Message Passing-assisted Iterative Noise Cancellation(MP-AIC)method is proposed.This method first constructs a reference clipping noise at the receiver to extract the residual signal fed to the MP detector,and then uses the residual signal and channel noise as the noise input of the MP detector,and gradually eliminates the influence of clipping noise on signal detection through multiple iterations.In this process,the convergence probability of the probability quality function of the modulated alphabet in the current iteration is taken as the initial probability in the next iteration to speed up the convergence rate of MP signal detection.Simulation results show that the proposed MP-AIC method significantly improves the BER while reducing PAPR.Compared with the bit error rate without clipping noise processing,the bit error rate of the OTFS clipping system after only two iterations of MP-AIC can be reduced by 72%.(3)Aiming at the problem that current OTFS receivers cannot balance the complexity and accuracy at the same time,an LSMR-Successive Interference Cancellation(LSMR-SIC)channel equalization method based on Least Square Minimum Residual(LSMR)is proposed.Time-domain equalization in OTFS system with symbol-level cyclic prefix makes the equivalent channel matrix more sparse in time-domain channel than in time-delay Doppler channel.In this method,LSMR algorithm is used to equalize the received signal in time domain,so that the receiver can obtain stable equalized data under the condition of large-dimensional and illconditioned equivalent channel matrix,and then the interference signal is further eliminated by SIC technology.Simulation results show that the proposed LSMR-SIC method has the characteristics of fast convergence,low complexity and high accuracy.In high-speed moving scenarios at 500 km/h,the LSMR-SIC equalizer can reduce the BER of the system by 59.5%compared to the Linear Minimum Mean Squared Error(LMMSE)equalizer.