Untitled Document
Untitled Document
> Archives > To Be Published
Recently Accepted Papers
Transform Domain Wiener Filtering Algorithm: Time Varying Non-stationary Case
Transform Domain Wiener Filtering Algorithm: Time Varying Non-stationary Case
Abstract The spectral distribution of signal and noise are assumed stationary but in many applications stationary characteristics of signal and noise are not satisfied completely. The time dependent Power Spectral Density (PSD) of the input AR process allows the Wiener filter as timevarying nature. The analysis of transform-domain LMS adaptive filter for fixed data AR process has been carried out in recent work but the time-variance has not been explored fully. Most vibration data are in time-varying nature and the analytical solution is very difficult using some useful transform. The time varying application of Wiener filter allows data processing in block by block, and Wavelet transform is the unique transform scheme. In this work, SWT domain Adaptive-Wiener filter is implemented for time-varying input process that has been not implemented in recent work. Wavelet domain LMS and RLS adaptive algorithm is presented and analyzed in this work. Performance of the proposed algorithm is evaluated through Matlab to validate the analysis, and the simulation shows that the proposed algorithm provides better signal denoising and convergence performance than other transform domain adaptive algorithm. Parameters of the simulation selected according to the derived condition, and their effect on system performance are also investigated.
Keyword Denoising, Wavelet transform, AR model, Adaptive filtering, Mean square error
PDF
Status Before proofreading
Untitled Document