Correlation LMS Algorithm and Its Application to Double-Talk Echo Canceling
Conventional algorithms for echo canceling systems for hands-free mobile radiotelephone or tele-conferencing sets pose problems in double-talk scenarios. This paper describes a new algorithm for tap adaptations in echo cancelers that is based on the processing of the correlation function (CLMS) of the input signal where the near-end signal is located. In this approach, the gradient search algorithm is obtained from the correlation function of the input signal instead of from the input signal itself. Since near-end and far-end signals are supposed to be uncorrelated, the new gradient is not disturbed by the near-end signal. The tap coefficient data are copied in a conventional digital filter replica path for echo canceling.
Please disable any pop-up blockers for proper viewing of this Whitepaper.