Research On Channel Estimation And Precoding For MIMO-OFDM In High-Speed Mobility Scenarios

With the explosive development of highways,high-speed railways and urban roads,problems such as safety,effectiveness and timeliness of traffic control need to be solved urgently.Intelligent transportation system is promising as the future transportation system.With the rapid development of the Internet of Vehicles,the demands for low-latency,large-capacity and high-reliability communication are becoming increasingly urgent.However,the traditional mobile communication is designed for the scenarios with stationary or low-speed terminals.When it is applied to high-speed mobility scenarios,there are many challenges.For example in the 5G communication standards,the bandwidth of 100 MHz can provide a transmission rate greater than1 Gbps when the terminal is moving slowly.However,in high-speed mobility scenarios,for example larger than 350 km/h,only the data rate of 50 Mbps can be achieved,which is only 10 % of the throughput required by the vehicle communication systems.The main reasons for the severe deterioration of system transmission capacity are the fast time-varying channel and Doppler frequency spread effect.Therefore,improving data transmission rate in high-speed mobility scenarios is very important and scientifically significant.In order to overcome the deterioration of data transmission efficiency in high-speed mobility scenarios,this paper fully considers the typical problems of fast time-varying channels,such as Doppler frequency spread effects and challenges in tracking channel variations.Through jointly using theoretical analysis,mathematical modeling,and numerology evaluation,the channel estimation,ICI suppression and precoding for Massive antenna system are analyzed and researched.The innovative work of the thesis includes the following four aspects:First,a novel channel prediction method based on beam-space decomposition is proposed to solve the difficulty of tracking channel variation in high speed mobility scenarios.It is well know that AMC technology based on CSI plays an important role in communication systems.However,when the terminal is fast moving,the wireless channel changes rapidly.The CSI reported by the terminal does not well match the channel at the time when the user is being scheduled.As a result,the system performance deteriorates.In order to relieve this problem,this paper proposes a novel channel prediction method.Based on extended Saleh-Valenzuela model,the wireless channel is divided into multiple clusters in spatial domain.Each cluster is composed of sub-paths with similar spatial angles.By decomposition in the beam domain and frequency domain,the wireless time-varying channel is characterized by weighted summation of a number of spatial and frequency domain basis vectors.An improved FIIB algorithm is proposed to overcome the high complexity of calculating basis vectors and weighting parameters.The variation of these parameters can be approximated by polynomials.The base station can accurately predict the channel variation based on the reported polynomial coefficients.This method greatly improves the accuracy of time-varying channel prediction.Simulation results show that the method can accurately predict channel changes even with three-order coefficients,and the resources to carry these polynomial coefficients are far fewer than those traditional methods.Second,a FS-BEM channel estimation method is proposed to solve the problem of performance deterioration of channel estimation in high speed mobility scenarios.In OFDM communication system,channel estimation is usually conducted using inserted reference signal in an OFDM symbol.In the traditional channel estimation,the channel is assumed to be constant within an OFDM symbol.However,this assumption does not hold when the terminal is fasting moving,which will make the channel estimation methods invalid.Based on the theory of BEM,this paper analyzes the issues in CE-BEM,and proposes a FS-BEM channel estimation method.By oversampling the basis,the MMSE or MIR criterion is adopted to select the subsampled basis.Therefore,accuracy of channel modeling is improved.Simulation results show that this method overcomes the drawbacks of traditional CE-BEM models,and has good performance at different moving speeds.Third,a novel precoding and equalization method is proposed,which can effectively suppress Doppler-induced ICI while keeping the calculation complexity low at the receiver.When a terminal is fast moving,ICI would appear,and seriously affect the performance of OFDM based wireless communication systems.In traditional equalization algorithms,iterative interference cancellation methods are usually adopted to overcome the impact of ICI.These methods have problems such as high complexity and deteriorated performance when the terminal moves fast.This paper proposes a novel precoding and equalization method,where the transmitted signal is segmented in frequency domain,and redundant subcarriers are inserted for each segment.As a result,the ICI is constrained within one segment,and inter segment interference is avoid.At the receiver,the ICI in each segment can be easily removed through simple linear transformation operation and order-1 equalization.This method can greatly reduce the implementation complexity of the receiver,and has good ICI suppression performance.Last,a novel two-stage codebook based on beam domain representation of precoding matrix is proposed to solve the problems of system performance deterioration when traditional codebook is applied in medium or high speed moving scenarios.In MIMO systems,the feedback of the precoding matrix consumes a large amount of uplink radio resources.Especially when the terminal is fast moving,very short feedback period is necessary in order to track the CSI variation,which further increases feedback overheads.A novel method based on beam domain representation of precoding matrix is proposed to solve this problem.Based on the matrix decomposition theory,the precoding matrix is represented as linear combination of several orthogonal vectors.The terminal needs to feedback the indices of the selected orthogonal vectors and the coefficients corresponding to each selected vector.By jointly configuring long and short-period feedback,the method can track channel variation meanwhile maintaining low resource consumption.The simulation results show that the proposed CSI feedback method not only harvests the precoding gain,but also greatly reduces the amount of information to be reported.Some contents of this method have been accepted in the chapter 5.2.2.2.3 of ‘TS38.214: NR Physical layer procedures for data'.