In this paper

a novel joint channel estimation and symbol detection algorithm is proposed.Based on the idea of generalized distribution law

the likelihood function which is used to achieve the optimal sequence detection is factored into products of several terms of local probability function.An intuitive factor graph is adopted to describe the relationship of these terms.Then a distributive joint channel estimation and symbol detection method is deduced by applying sum-product algorithm on the factor graph directly

which approximates the performance of optimal sequence detection with low computational complexity.Meanwhile

particle filter is introduced to solve integrals of continuous variables in the algorithm

by which a real-time channel tracking is presented.Thus the demand for pilot symbols can be reduced and the spectrum utilization efficiency is improved.Simulation results show that the new algorithm can achieve good performance with low computational complexity.