A wave-domain model for acoustic MIMO systems with reduced complexity

The identification of loudspeaker enclosure microphone (LEM) systems with large numbers of microphones and loudspeakers is a challenging task for both, algorithmic and computational reasons. Wavedomain adaptive filtering offers an alternative to point-to-point channel identification where the complexity of the LEM system model can be reduced by using a physically motivated approximation. In this paper we present for the first time an explicit discrete-time and discrete-frequency formulation of the necessary transforms to the wave domain for circular loudspeaker and microphone arrays in a practically relevant setup. We further propose a generalization of the LEM system model in the wave domain allowing for a scalable complexity and a flexible trade-off between low complexity and high accuracy. An experimental evaluation of an acoustic echo cancellation scenario supports the applicability of the presented model.