Affiliation:
1. Signal Processing Laboratory, Swiss Federal Institute of Technology, EPFL, 1015-Lausanne, Switzerland
Abstract
Reliable chaos detection in real-world time series is attracting increasing attention in the scientific community. This work shows that it is possible to use chaos analysis methods such as attractor dimension estimation, Lyapunov exponents estimation and nonlinear prediction, under the condition that the limitations and drawbacks of the algorithms used are kept in mind. Three existing algorithms for chaos characterization are analyzed in terms of classification performances and robustness with respect to noise and data length. It is shown that all three help detect chaos and even classify different types of signals, but that their results are not devoid of ambiguity. An illustrative example is given, in which the algorithms presented are applied to heart rate variability signals, and directions of research are proposed for the design of a straightforward and simple chaos detection methodology.
Publisher
World Scientific Pub Co Pte Lt
Subject
Applied Mathematics,Modelling and Simulation,Engineering (miscellaneous)
Cited by
16 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献