Diffusion-Based Distributed Wave-Domain Active Noise Control With Convex Sum of Non-Convex Quadratic Costs
Résumé
Active noise control (ANC) in the wave-domain has been widely used to perform noise cancellation in large spatial areas with time-varying acoustic characteristics. Most of existing works on ANC focus on centralized strategies where the residual signals from all microphones are required for performing each estimate update. Nonetheless, as the number of speakers and microphones increases, the centralized strategy might encounter challenges related to flexibility and scalability. The aim of this work is to introduce a distributed wave-domain ANC method based on a convex sum of non-convex quadratic costs to spread the computational load amongst nodes to overcome these issues. In order to minimize the ANC cost, we consider a group diffusion adaptation strategy, where the errors measured at each node, and the driving signal estimated at each node, involve those nodes individually and their neighbors. As a result, additional nodes with microphones or speakers can be easily integrated into the system, and each node can process with low computational complexity. Comparing the proposed algorithm to its centralized and distributed counterparts, we evaluate the algorithm’s efficacy in various environments.
Origine | Fichiers produits par l'(les) auteur(s) |
---|