In this paper, shortcoming of traditional wavelet denoising in real-time signal processing is discussed, requirements of online denoising are considered, and a moving window is introduced into traditional wavelet transform. Using the moving window,an online wavelet denoising approach is proposed. Some problems of online denoising, such as border distortion and pseudo-Gibbs phenomena, are discussed. To solve these problems, window extension and window cycle spinning are also proposed. Different approaches are tested by the signal widely used in denoising domain. Both the visual results and the quantitative measures are presented to highlight the availability of the new approach.