Bayesian Estimation of Oscillator Parameters: Toward Anomaly Detection and Cyber-Physical System Security

Area of Science:

• Cyber-physical Systems
• Measurement Science and Technology
• Statistical Inference

Background:

• Cyber-physical systems (CPS) security faces challenges due to limited parameter spaces in anomaly detection.
• Software-assisted physical systems, like those in additive manufacturing or DNA synthesis, often present ill-posed problems for anomaly detection.
• Conventional methods struggle with the inherent complexities of CPS anomaly detection.

Purpose of the Study:

• To present a novel Bayesian inference method for detecting anomalies in cyber-physical systems.
• To address the challenge of limited data and parameter spaces in CPS anomaly detection.
• To quantify the uncertainty associated with anomaly detection in CPS.

Main Methods:

• Utilized Bayesian inference to estimate unknown parameters of a generic cyber-physical system actuator.
• Obtained a numerical transfer function model from experimental input-output measurements.
• Simulated a code-based malicious signal to test the detection efficacy of the Bayesian approach.

Main Results:

• Demonstrated the efficacy of Bayesian inference for anomaly detection using a minimal number of data points.
• Achieved effective anomaly detection with uncertainty quantification.
• The developed model successfully identified malicious signals within the cyber-physical system.

Conclusions:

• Bayesian inference offers a robust solution for real-time anomaly detection in cyber-physical systems.
• The proposed method is adaptable to various CPS anomaly detection scenarios.
• This approach enhances the security of cyber-physical systems by providing reliable detection with uncertainty quantification.