中文摘要：

依据时变信道的时域相关性以及和积算法（SPA）规则，提出了一种正交频分复用（OFDM）系统中基于图模型的信道估计算法。首先，将相邻OFDM符号之间的信道频率响应建模为一阶自回归（AR）过程；然后，在因子图上实现相邻信道系数节点之间的信息交换，通过多次迭代提高了导频处信道估计的精度。仿真结果表明，本算法可以有效改善最小二乘（LS）算法对于时变多径信道的估计性能，性能优于传统的最小均方（MMSE）信道估计。

英文摘要：

In accordance with the time-domain correlation of time-varying channels and the SPA （Sum and Product Algorithm） rules, a channel estimation algorithm based on factor graph is proposed. Firstly, the channel frequency responses between adjacent the OFDM （Orthogonal Frequency Division Multiple- xing） symbols are modeled as first-order AR （Auto-Regression） process, and then, information exchange between adjacent channel coefficient nodes is realized on the factor gragh, and via multiple iterations, the accuracy of least-square （LS） channel estimation on pilot positions is improved. Simulation results show that the proposed algorithm could effectively improve the estimation performance of LS algorithm for time- varying multipath channels, and performs better than the traditional MMSE （Minimum Mean Square Error） channel estimation algorithm.