Abstract: Particle filtering is a Monte Carlo method which is based on the sequential importance sampling, its re-sampling step results in the particle impoverishment. The traditional OFDM time-varying channel estimation algorithm based on particle filtering suffered several problems, such as low accuracy and high computation complexity. In view of eliminating particle impoverishment, the method of particle flow is used to replace re-sampling. Through constructing the differential equation to achieve Bayesian estimation and using particle flow to smoothly migrate the particles to the posterior distribution in state space, prior particles are updated to posterior particles, and then an OFDM time-varying channel algorithm based particle flow filtering is proposed. Compared with the channel estimation algorithm based particle filtering, the algorithm has lower computation complexity, higher estimation accuracy and more robust to the environmental noise.