Multifractal Security Analysis of Cyberphysical Systems

Abstract

In this paper authors propose to use multifractal analysis and Kalman filter for security analysis of cyberphysical systems (CPS). First stage of the proposed approach is data reduction caused by time series processing. The second stage is anomaly detection and the third stage is attack prediction. For anomaly detection it has been proposed to use multifractal analysis and time series prediction. Most real processes, as scientific researches are shown, have multifractal nature. Multifractal analysis and Legendre spectrum characteristics calculation allow to detect violation of multifractal properties caused by cyberattack. Kalman filter is used for cyberattack detection at early stage, because it performs system state prediction with high accuracy. The effectiveness of the proposed approach is confirmed by experimental results.