Research On Channel Estimation Based On Superimposed Pilot In High-speed Mobile Scenarios

With the convergence of mobile communications and broadband wireless access technologies,high mobility scenarios,such as high-speed rail and vehicle-to-vehicle(V2V)scenarios,are typical scenarios of the fifth generation(5G)mobile communication system.Therefore,providing ultra-reliable,high-quality,and fast mobile communication services for high-speed mobile scenarios,has become one of the technical goals of 5G.On the other hand,the challenges of orthogonal frequency division multiplexing(OFDM)communication systems is increasingly severe for the scarcity of wireless spectrum resources and high mobility.So,how to make full use of limited spectrum resources is a problem that must be solved by modern wireless communication technology.Among them,the wireless transmission scheme based on the superimposed pilot has been widely concerned because it can reduce the frequency band overhead occupied by the pilot sequence.However,a serious degradation of channel estimation and symbol detection performance because of the mutual interference between pilot and information data in the superimposed pilot scheme.The key task is how to decouple the channel estimation from symbol detection at the receiver.Aiming at higher bandwidth efficiency and channel estimation performance of wireless channels,the dissertation discusses the wireless transmission technologies using superimposed pilot and focuses on interference elimination in channel estimation and superimposed signal separation,researches on OFDM system based on superimposed pilot in high-speed mobile scenarios.The main contents and innovation on this dissertation are listed as follows:Firstly,based on the characteristics of wireless channels in high-speed scenarios,the reasons for the time-frequency domain selective fading and the non-stationary characteristics are specifically analyzed.In order to reduce the estimation complexity of fast time-varying channels,the idea of basis expansion models(BEM)is introduced.It shows that the BEM model can effectively simulate the channel characteristics in highspeed scenarios from simulation analysis.And on this basis,an OFDM system based on complex exponential BEM(CE-BEM)is established.Aiming at the pilot insertion method,a block superimposed pilot pattern is designed,and the traditional channel estimation method to eliminate data interference is analyzed.In addition,the power allocation problem of superimposed pilots is also researched,and a suitable power allocation factor is selected through simulation verification.Based on this,the performance of the traditional channel estimation algorithm to eliminate data interference is simulated and verified,and the iteratively detected channels as the number of iterations increases,the estimation performance improves significantly,which verifies the feasibility of the iterative detection channel estimation method.However,for the channel estimation challenges in high-speed scenarios,it needs to adopt other channel methods that can better track the channel.For the non-stationary time-domain characteristics of the channel,this paper proposes a combination of BEM and extended Kalman filter(EKF)to jointly estimate the channel time-domain autocorrelation coefficient and channel impulse response,and reconstruct the transmitted symbol at the receiver to update the observation matrix,and finally use iterative BEM-EKF channel estimation method to gradually eliminate the estimation of data symbol interference at the superimposed position.It shows that iterative BEM-EKF can effectively improve the accuracy and bit error rate(BER)of channel estimation with the increase of the number of iterations,and it has a significant improvement in throughput compared to the traditional superimposed pilot algorithm from simulation analysis.Further,the prior channel fading statistics is necessary for EKF estimation method,but in actual communication,the receiver is unknown.Therefore,this paper further adopts expectation-maximization(EM)parameter estimation,and uses the maximum likelihood to estimate the parameter matrix of the first-order time-varying auto regression(TVAR),observation noise,and the covariance of the channel parameter shift noise according to the received data.To further enhance the system’s robustness to channel time-varying fading,EKF and Rauch-Tung-Striebel smoother(RTSS)are used to implement timevarying channel tracking.Through simulation analysis,the proposed EM-EKF channel estimation algorithm has better estimation performance than the existing representative algorithms and is more suitable for the actual OFDM communication system in highspeed mobile scenarios.