Discriminating Chaos from Non-Gaussian Noise on Analog Circuits

Abstract

This paper presents an approach for studying non-linear circuits dynamics. In particular, we study deterministic vs. stochastic dynamics by distinguishing chaotic behavior from noise. The proposed approach is general enough to detect any type of noise with no restriction on the noise distribution. It is based on a statistical proof by contradiction using hypothesis testing with Lempel-Ziv Complexity (LZC) method as discriminating statistics. The LZC method is adopted as a nonparametric test that has the advantage of detecting any kind of noise. Experimental results on a $\Sigma-\triangle$ Modulator and a Lorentz circuit showed that the proposed approach provides a reliable distinction between chaotic/hyperchaotic behavior and non-Gaussian noise.