Uncombined precise point positioning (PPP) could be used to extract ionospheric delay with high accuracy. However, parameters estimation in PPP requires long converge time due to the high correlation between ionosphere and ambiguity parameters. Further more, the multipath effects at tracking station degrade the precision of code and phase measurements, thus impact the performance of PPP ionospheric delay estimation. For static observation stations, sidereal filtering could be used to eliminate multipath errors by taking advantage of the ground track repeat period of GPS satellites. After extracting the code and carrier phase residuals of the past few days in post-processing, multipath error correction model could be established with historical residual series by sidereal filtering, so as to improve the performance of real-time ionospheric delay estimation. Experiments with IGS observation data showed that with the application of sidereal filtering, real-time ionospheric delay extraction error decreased from 0.185 m to 0.028 m and convergence time of ionospheric parameters for newly rising satellites reduced from 80 minutes to 35 minutes. Improvements in single station ionosphere delay estimation could refine ionosphere model of local network. On the other hand, precise satellite slant ionosphere delay can be obtained at a lower elevation, which can reduce the layout density need of reference station network.