Multichannel adaptive filtering in compressive domains

In this paper, we give a study on reducing the coefficients to be estimated in an adaptive sparse multichannel system identification problem. We present an approach to perform the adaptation in a compressed representation of the sparse system without requiring prior knowledge about the dimensions in which the system has significant components. The presented technique exploits the ability of sparse systems to be compressed offering a reduction of the adaptive filter coefficients in addition to high convergence rates.