Active noise cancellation in space containing scattering objects based on kernel interpolation

A spatial active noise control (ANC) method for canceling noise in a region containing scattering objects is proposed. Spatial ANC aims to suppress noise over a target control region using multiple microphones and loudspeakers. The driving signals of the loudspeakers are obtained adaptively based on the interpolation of the sound field from the microphone measurements. However, most spatial ANC methods assume that the target region is free space, i.e., contains no scattering objects, because they are based on the sound field interpolation methods that do not take into account the presence of scatterers within the target estimation region. On the contrary, in spatial ANC, ANC users will be present within the target region, which may deteriorate their regional noise reduction performance, especially at high frequencies. We propose a spatial ANC method based on the sound field estimation for the region containing scattering objects. The incident sound field in the target region is estimated based on the kernel ridge regression under the condition that the interpolated function satisfies the homogeneous Helmholtz equation, with the scattering effects eliminated. The numerical experiments indicated that the proposed method outperforms the existing methods at high frequencies.