https://doi.org/10.1140/epjs/s11734-021-00169-y
Regular Article
Detection of data corruption in stationary time series using recurrence microstates probabilities
1
Departamento de Física, Universidade Federal do Paraná, 81531-980, Curitiba, PR, Brazil
2
Universidade Federal de São Paulo, Instituto de Ciência e Tecnologia—ICT, 12231-280, São José dos Campos, SP, Brazil
3
Laboratório Associado de Computação e Matemática Aplicada, Instituto Nacional de Pesquisas Espaciais, 12227-010, São José dos Campos, Brazil
Received:
30
November
2020
Accepted:
23
April
2021
Published online:
25
June
2021
Recurrence microstates can be used to analyze many properties of stationary states of stochastic and deterministic time series, including the level of correlation of stochastic signals. Here, we show how artificially inserted data (data that does not belong to a original stationary signal) may be detected using recurrence microstates statistics. We show that the method is sensitive enough to detect the breaking of the stationary signal even when the corrupted inserted data span into the same domain of the original data. Examples of our analyses are applied to two numerically generated time series of dynamical systems, namely the logistic map, and the Lorenz equations. Finally to show results applied to experimental time series, we analyze a digital audio signal of a human speech.
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of Springer Nature 2021
