A differentially private square root unscented Kalman filter for protecting process parameters in ICPSs

Abstract

Industrial cyber–physical systems (ICPSs) have received increasing attention from both academia and industry. However, the privacy-utility trade-off of process parameters is still a challenge in the ICPSs. To address this challenge, a Kalman filter-based differential privacy and an unscented Kalman filter-based differential privacy algorithms are derived. In order to increase the utility of process parameters while protecting the privacy of process parameters, a differentially private square root unscented Kalman filter algorithm is proposed by employing the square root unscented Kalman filter and differential privacy. The experiments based on a numerical control lathe are presented to prove the privacy and evaluate the utility of process parameters.