Channel Estimation Technology Research Based On Message Passing Algorithm For OFDM System

Orthogonal frequency division multiplexing(OFDM)is a multi-carrier transmission technology whose sub-carriers are orthogonal to each other.It has advantages of high spectrum efficiency and robustness against to the inter-symbol interference(ISI)and is one of the key technologies of the physical layer for the current mainstream mobile communication standards such as long term evolution(LTE).On the receiver side of the OFDM system,the acquisition of accurate channel state information(CSI)is a necessary condition for realizing coherent demodulation.This has led to a lot of research on channel estimation technology.Traditional pilot-based channel estimation algorithm needs a large amount of system overhead to obtain excellent estimation performance.Blind channel estimation exists the problem of slow convergence and high computational complexity.Iterative channel estimation is an effective way to make up for the shortcomings of the above channel estimation algorithms and has attracted widespread attention in recent years.Factor graph(FG)provides a visual selection for the design of iterative algorithm.It has many advantages,such as intuition,flexibility,modest computational complexity and easy combination with other algorithms.Therefore,FG has a great development potential in the design of iterative channel estimation algorithm.Based on the channel statistical properties and message passing theory,several iterative channel estimation algorithms based on factor graph are proposed.The main research contributions of this thesis are summarized as follows:In the first part,an iterative channel estimation algorithm based on one-dimensional factor graph is proposed.The first-order auto-regressive(AR)model is used to characterize the time-domian correlation of the time-varying channel.Time-domain transfer nodes are introduced and one dimensional factor graph model is constructed.Then the iterative updating of the message on the channel coefficient nodes and the time-domain transfer nodes is carried out on the above factor graph according to the sum-product algorithm(SPA),which improves the accuracy of the channel frequency responses(CFR)estimation.Simulation results show that the proposed algorithm can achieve convergence by no more than three iterations and obtain better performance of mean square error(MSE)compared with the traditional Winner-filtering channel estimation algorithm,avoiding complex matrix inversion at the same time.In addition,under different time-varying channel conditions,the estimation performance of the proposed algorithm is superior to the Kalman filter algorithm based on comb-like pilots,showing better tracking capability of time-varying channel.To further improve channel estimation performance,an iterative channel estimation algorithm based on two-dimensional factor graph is proposed in the second part.For the OFDM system under time-varying multipath channel conditions,firstly we introduce the time-domain transfer nodes and the frequency-domain transfer nodes and construct the two-dimensional factor graph model.The messages on the channel coefficient nodes,the frequency domain transfer nodes and the time domain transfer nodes are iteratively updated according to the parallel scheduling mechanism.The estimation values of all channel frequency responses are obtained after several iterations.We adopt COST207 channel model and Rayleigh time-varying multipath channel model to verify the performance of the proposed algorithm.Simulation results show that under both fast time-varying and slow time-varying channel conditions,the proposed algorithm can achieve better estimation performance compared with the channel estimation algorithm based on one-dimensional factor graph and improve the performance of the traditional interpolation algorithm more significantly.A joint channel estimation and data detection algorithm is proposed in the third part.Firstly,a two-dimensional factor graph model is constructed for the coding transmission OFDM system.Then soft information is exchanged iteratively between the channel estimator and data detector and the channel estimation is further refined by the channel statistical properties.Simulation results show that the proposed algorithm can greatly improve the performance of the interpolation algorithm at different pilot interval and bit error rate(BER)of the system is significantly dropped.It is worth mentioning that at the same signal to noise ratio(SNR)level,the difference in BER between the two algorithms can be up to three orders of magnitude.