Abstract: The results of time-frequency transfer will be affected by the non-modeling error and observation noise, which is often high-frequency signals, and the construction of low-pass filter can eliminate the high-frequency noise signals in the observation sequence to a certain extent. In this paper, we analyze the nature of the Vondrak filter function, weight the clock difference sequence by the IGG3 algorithm and select the suitable filtering factor by the frequency response method; we conduct the satellite two-way time and frequency transfer experiment (TWSTFT), the satellite two-way time and frequency transfer experiment based on the software receiver (SDR-TWSTFT), and the short-baseline common-view time and frequency transfer experiment on the different links, and use the short-baseline common-sight time and frequency transfer method on the results. experiments, and the clock difference results are smoothed and denoised by anti-differential Vondrak filtering. The results show that: the filtered clock sequence can well reflect the trend of the original clock sequence; the daily fluctuation effect is effectively suppressed and the accuracy of the smoothed TWSTFT clock result is significantly improved; the accuracy of the common-view clock result is significantly improved, and the difference between the result and the precise point positiening (PPP) timing result is kept in the range of −1.0 to 1.0 ns.