A Multichannel Recursive Least-Squares Algorithm Based on a Kronecker Product Decomposition

Abstract

In this paper, a multichannel recursive least-squares (RLS) algorithm is proposed for the identification of multiple-input single-output (MISO) systems. As compared to the conventional (multichannel) RLS algorithm, the proposed solution exploits a Kronecker product decomposition of the global impulse response, together with low-rank approximations. The gain is twofold, in terms of both performance and complexity. Simulations performed in the context of stereophonic acoustic echo cancellation indicate the appealing features of this algorithm.