A Decomposition-Based RLS Algorithm for Stereophonic Acoustic Echo Cancellation

In hands-free teleconferencing systems, stereo transmission provides telepresence thanks to our binaural hearing system and gives a realistic presence that actual single-channel systems cannot offer. In this context, stereophonic acoustic echo cancellation (SAEC) is necessary for full-duplex quality communication. Nevertheless, the long length of the acoustic impulse responses represents one of the main challenges in SAEC, which significantly impacts the overall performance of the adaptive filters used to model the acoustic echo paths. In this paper, we propose a decomposition-based approach using the nearest Kronecker product, which results in a combination of shorter filters to solve the original system identification problem. The solution is formulated in terms of the recursive least-squares (RLS) algorithm, which is very appealing for its convergence features. Besides the lower computational complexity, the proposed decomposition-based RLS algorithm achieves a better tracking behavior as compared to its conventional counterpart. Simulation results support these practical features.