Affine-projection-like algorithm for active noise control over distributed networks

This paper considers the active noise control over a network of distributed acoustic nodes. In active noise control practical scenarios, it is necessary to improve the performance of the well-known least mean squares algorithms in order to improve the convergence behavior in transient state. Affine projection algorithms have shown to speed up the convergence speed of the algorithms. However, affine projection algorithms require matrix inversion, whose calculation is not straightforwardly distributed among the nodes. Therefore, we present a distributed version of the affine-projection-like algorithm (which avoids matrix inversion) together with an incremental collaborative strategy in the network. Results demonstrate that the proposed scalable and versatile distributed active noise control algorithm exhibits faster convergence than the conventional least-mean-squares-based algorithms as well as the same performance than its centralized version.