Variable Step Size Pre-Whitening Transform Domain LMS Adaptive Noise Canceller

In this paper, we propose a novel variable step size pre-whitening transform domain LMS algorithm for adaptive noise cancellation by introducing a new expression for updating the step size using a smoother gradient vector estimated by weighted averaging. The resulting algorithm presents a good tradeoff between performance and computational complexity and significantly outperforms the existing transform domain LMS algorithms in terms of the convergence speed, level of the steady state reached by the excess mean square error (EMSE), the steady state of the EMSE, misadjustment and output SNR.