Tuning non-linear adaptive compensators in non-ideal noise environment

Adaptive compensation has been popular in a variety of biomedical signal pr ocessing applications. One of the major limitations of this method is the r equirement for a primary signal + noise channel and a secondary noise chann el, with the noises in the two channels being strongly con elated. In the m ajority of real-life biomedical applications this requirement can rarely be fulfilled since the secondary channel inevitably contains certain represen tation of the signal from the primary channel thus jeopardizing the adaptiv e compensation process. In the present work we describe a method for tuning an adaptive compensator containing a non-linear real-time multiplier in th e reference channel in such non-ideal noise environment. The method is base d on iterative adjustments of an exponential non-linear function which modi fies in real-time the signal-to-noise balance in the reference channel in f avor of the noise, preserving the correlation of the latter with the noise in the primary (signal + noise) channel but diminishing the signal content in the reference channel. Sets of model signals and noises are used to illu strate this methodology with a particular emphasis on biomedical signal app lications. A quantitative description of the iterative tuning procedure is provided and the capabilities and limitations of this novel method for adap tive compensation are outlined. (C) 1999 IPEM. Published by Elsevier Scienc e Ltd. All rights reserved.