Use of recurrence plots in analysis of time-series data

Abstract

Recovering information about the periodic behavior of a non-stationary signal typically requires a large data set and extensive off-line analysis. Alternatively, we've modified Eckmann, Kamphorst and Ruelle's idea of a recurrence plot, which they defined to appreciate the dynamics of a time series without either of these constraints. Geometrical properties of our plots provide dimensional complexity and local rate of divergence estimates.