Reduced complexity diffusion filtered x least mean square algorithm for distributed active noise cancellation

Abstract

A computationally efficient diffusion cooperation scheme-based distributed active noise control (DANC) system is proposed in this paper. It is observed that the conventional centralized multi-channel ANC (MANC) systems employed for noise reduction in a wide region are computationally complex and lack scalability. Additionally, the noise reduction for practically encountered noises is a challenging task, especially for multi-point environments. To overcome these drawbacks, in this paper, a diffusion filtered x least mean square (DFxLMS) algorithm is developed for DANC systems. The proposed DFxLMS-DANC scheme is modified using proximal secondary path bounds to reduce computational overhead. Also, the practical application of air-conditioner noise control is addressed in the presence of real primary and secondary path scenarios. It is shown that the total computational improvement in proposed DFxLMS-DANC and modified DFxLMS-DANC systems is 23.13% and 49.87%, respectively, over multiple error FxLMS-based MANC system. It is also demonstrated that the proposed method helps to achieve \(\sim \)  18 dB reduction in the air-conditioner noise levels in practical environments.

http://bit.ly/2ss945k