The OFDM System Channel Estimation Based On Wavelet Transform Weighted Projection TSVR

During the transmission of mobile signal communication,the change of channel characteristics will have a great impact on the communication quality.The change of channel parameters is caused by many factors,such as the multi-path effect of transmission,delay spread and relative motion,etc.In addition,the multipath effect in the propagation makes the wireless channel present frequency selective characteristics,which is the main reason for fading in wireless communication system,so it is more important to understand the channel characteristics.Because the traditional support vector regression algorithm gives the same weight to the training samples with different degrees of noise pollution,the performance is reduced.In order to improve the performance of OFDM channel estimation,this paper proposes a channel estimation algorithm for orthogonal frequency division multiplexing(OFDM)system based on projection wavelet transform weighted twin support vector regression(TSVR)using the advantages of wavelet transform and support vector machine.Most channel estimation algorithms in OFDM systems are based on the linear assumption of channel model.In PWWTSVR algorithm,the channel of OFDM system presents nonlinear and fading characteristics in time domain and frequency domain.PWWTSVR uses the pilot signal to estimate the response of the nonlinear wireless channel,finds a projection axis in each objective function,and uses the prior information of the training data to find the projection axis,so as to minimize the variance of the projection point and improve the performance.PWWTSVR algorithm is an improvement of the traditional TSVR algorithm.Different from the traditional support vector regression algorithm,the PWWTSVR model proposed in this paper uses wavelet transform to get different penalty weights of training samples in different positions.Intuitively,for the samples with larger noise pollution,we give smaller weights,and for the samples with smaller noise pollution,we give larger weights.In order to reduce the influence of outliers,the quadratic term and the first term of channel regression function are inserted.The final regressor can avoid the overfitting problem to a certain extent,and yields great generalization ability.Numerical experiments on artificial and benchmark datasets demonstrate the feasibility and validity of the proposed algorithm.Bit error rate and the mean square error criterion for the evaluation of the simulation results show that under the Jakes model fast fading channel conditions,PWWTSVR channel estimation method takes better performance compared with several other methods.