IMPROVEMENT OF NLMS ADAPTIVE ALGORITHM PERFORMANCE IN DOUBLE TALK MODE

A. N. Tarakanov, A. E. Nazarovsky, A. L. Moseev, Yaroslavl State University, Yaroslavl, Russia

SYNCHROINFO JOURNAL. Volume 5, Number 1 (2019). P. 9-12.

Abstract

Line echo cancellers are widely used in telephony. Most commonly used adaptive algorithm is Normalized Least Mean Squares (NLMS). Unfortunately it has poor performance in double-talk mode. In this paper the performance of NLMS algorithm with Double-Talk Detector (DTD) based on Geigel algorithm is researched. A new robust adaptive algorithm based on NLMS is proposed, and its properties are studied. The DSP based comparison of echo canceller models using NLMS and proposed algorithm is considered.

Keywords: line echo canceller, normalized least mean square algorithm, double-talk detector, misalignment, robust algorithm, digital signal processor.

References

[1] Haykin S. Adaptive filter theory. Englewood Cliffs, NJ: Prentice-Hall, 1996. 996 p.
[2] ITU-T Recommendation G.165. Echo cancellers. 1993.
[3] ITU-T Recommendation G.168. Digital network echo cancellers. 1997.
[4] C.F.N. Cowan, P.M. Grant. Adaptive filters. Prentice-Hall, Inc., Englewood Cliffs, 1985.
[5] Huber P.J. Robust Statistics. N.-Y.: Wiley. 308 p.
[6] Tarakanov A.N., Moseev A.L., Nazarovsky A.E. Sensitivity decreasing in adaptive algorithm for echo cancellation. Proceedings of 58-th scientific session devoted to radio day. Moscow, 2003. Vol.1. P.90-92.