An efficient, low-complexity, normalized LMS algorithm for echo cancellation

Modem teleconferencing systems contain multiple audio channels. Algorithms for acoustic echo cancellation form an integral part of these systems. They need to be computationally efficient and rapidly converging. Normalized least mean square (NLMS) algorithms, because of their simple architecture and robust performance, form the backbone of the echo cancellers used in the industry. They become computationally expensive if used for echo cancellation in multi-channel systems. The algorithm proposed in this paper addresses this problem by incorporating the principle of partial updating of the filter coefficients in the NLMS algorithm. The performance of the proposed algorithm is compared with other adaptive algorithms for acoustic echo cancellation. It is shown that the proposed algorithm has a reduced complexity, while providing a good overall performance.