Low-Delay and Low-Cost Sigma-Delta Adaptive Controller for Active Noise Control.

Most digital active noise controllers have an intrinsic delay due to reconstruction and anti-aliasing filters. These delays may make them less attractive when compared with analog implementations. This delay is critical for broadband active noise control performance, even though the same is not valid in the narrowband case. The delay can be reduced by using custom design filters for the application with large transition bands but cannot be made close to zero. Also, the sampling rate can be increased, but this causes a significant increase in the computational cost since the number of control filter output calculations per second increases and the size of the control filter increases. This work proposes a way to reduce these delays almost to zero without a substantial increase in the computational cost by using the quantization and noise shaping techniques used in sigma-delta analog-to-digital (AD) and digital-to-analog (DA) converters. Sigma-delta AD and DA converters are used but without the typical filter. The controller filter impulse response is also oversampled, and the resulting filter is implemented at a high frequency but using just a few bits, resulting in low computational complexity. The AD downsampling filter and the DA reconstruction are not needed because noise shaping moves the quantization error to high frequencies that are not audible and that are filtered naturally by the acoustic transducers. However, standard adaptive filtering algorithms working at a low sampling frequency are still used to choose the control filter. [ABSTRACT FROM AUTHOR]