Research Of OFDM Channel Estimation Based On BEM

Orthogonal frequency division multiplexing(OFDM)technology has strong resistance to multipath fading and high spectrum efficiency.Therefore,it is often used as a transmission scheme for high-speed systems.CP-OFDM is still one of the key physical layer technologies in the 5G communication standard released in 2018.In recent years,due to the large-scale popularization of high-speed railway and the rise of vehicle networking technology,wireless communication in high-speed mobile scene has attracted extensive research and attention.Different from the traditional low-speed scenario,the wireless channel in the high-speed environment presents the dual selectivity of frequency and time.As a result,the Doppler spread seriously destroys the orthogonality between subcarriers.At the same time,it also seriously affects the accuracy of channel estimation,which is one of the key modules of the receiver,and indirectly leads to the overall deterioration of system performance.Therefore,this paper takes the basis expansion model(BEM)as the research object,and makes an indepth analysis and research on the doubly selective channel estimation of OFDM system around the two aspects of inter carrier interference(ICI)and noise interference cancellation.The specific research results are as follows:1)This paper analyzes the generation and elimination principle of inter carrier interference in doubly selective channel in the beginning,and studies the BEM channel estimation method based on pilot cluster and time-domain windowing.Aiming at the problem of high computational complexity in traditional BEM estimation,an improved BEM channel estimation algorithm with simplified windowing at the receiver is proposed.Firstly,the windowing process of the receiving end is simplified by deriving the relationship between the base coefficient and the channel frequency domain matrix and the window function.Then,according to the windowing principle,the final channel frequency domain response is approximated as a banded matrix,and a low complexity equalization method is used for this special matrix.Finally,the system simulation is carried out under the extended vehicle channel model with the normalized maximum Doppler shift of 0.2 and 0.3 respectively.The results show that compared with the existing BEM estimation methods,the proposed algorithm greatly reduces the computational complexity of BEM estimation,and its bit error rate performance is only lost by 0.5d B,which ensures its performance advantage in dual channel.2)Given the fact that limited system performance in BEM channel estimation based on one independent block,considering the wide and stable characteristics of the channel in a certain period of time,the BEM estimation method using multi block OFDM symbols for joint modeling is studied.The characteristics of the base coefficients of adjacent symbols when using least squares estimation(LS)are analyzed.Aiming at the problem that the LS estimation value of the base coefficient is greatly affected by noise in non ideal channel,a multi symbol BEM channel estimation method using preamble sequence to preprocess and reduce the noise of the base coefficient is proposed.In the proposed method,a small preamble sequence is added to the head of the frame structure composed of multiple OFDM symbols.When estimating the base coefficient,the sequence is first used to estimate the channel tap position and filter the noise,and then the noise of the base coefficient is eliminated according to the obtained delay position information of the effective tap,Finally,the final channel frequency domain response matrix is calculated by using the denoised basis coefficient.The simulation results show that under the extended vehicle channel model with normalized maximum Doppler shift of 0.2 and 0.3,the proposed method can effectively eliminate the influence of noise in the solution of base coefficients,and significantly improve the MSE performance of channel estimation and the bit error rate performance of the system.