Laguerre escalator lattice and feedforward/feedback active noise control

Abstract

In this paper, a new Laguerre Escalator Lattice structure for an adaptive filter is proposed. This new structure orthogonalizes a nonstationary signal with computational efficiency. In this, the escalator lattice orthogonalizes a nonstationary signal and the Laguerre structure provides the computational efficiency. Further, its nonstationary orthogonalization has been exploited for active noise control (ANC). In the feedforward ANC (FFANC), the new orthogonalization of the input is used both for the main path identification and for the noise canceller for the secondary path identification. In the feedback ANC (FBANC), it is used to predict the nonstationary primary noise field component from the residual noise, for deriving the desired error for the secondary path identification. For the system identification with a nonstationary input, the proposed structure has a significantly improved performance both in terms of convergence speed and error, over that of the Laguerre lattice. Its application to FFANC for a nonstationary noise filed results in a very good performance both in terms of convergence speed and the error magnitude over that of Laguerre lattice. In FBANC, the ability of the new adaptive filter to accurately predict the nonstationary primary noise component from the error for the noise canceller improves the secondary path identification. This results in a significant improvement in ANC performance of about 5 dB over that one uses Laguerre lattice predictor. Further, the escalator realization by Laguerre structure reduces the computations significantly (by about 50%).