Abstract: Particle Filter (PF) is an efficient parameter estimation method. By analyzing the mathematical model of single carrier frequency domain equalization (SC-FDE) system and the principles of particle filter, the time-varying channel is modeled as a first-order AR process. An attempt is made to combine the PF method with the channel estimation of SC-FDE system based on UW, and the algorithm steps are given in detail. Then simulations and bit error rate performance comparison with LS estimation are done considering three kinds of channels with different time-varying degree. The results demonstrate that in the time-varying conditions, PF based channel estimation can improve the bit error rate (BER) performance comparing with linear LS method, and the slower the channel varies, the more obvious the performance gain is.