Communications with chaotic time series: probabilistic methods for noise reduction

Abstract

The paper deals with noise decontamination of chaotic time series under the assumption that some a priori information about the system which produced the time series is known in advance. We show that this a priori information can be quite naturally used in standard maximum likelihood approaches. Focusing on low complexity implementations we derive different quasioptimal maximum-likelihood solutions aimed at off-line and on-line noise cleaning. The obtained results show attractive capabilities for on-line and low-cost implementation. Copyright © 1999 John Wiley & Sons, Ltd.