Variance Value Stream Data Publishing Based on Differential Privacy Under Wearable Devices
Open Access
- 22 September 2023
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 1470-1478
Abstract
The publishing of real-time data streams with differential privacy is one of the current research hotspots in the field of privacy protection. The existing data stream publishing mainly revolves around count publishing, while user vital signs data collected under wearable devices often need to be released using mean and variance values. For the release of variance value, this paper proposes a differentially private variance value release method based on Kalman filtering and adaptive sampling under wearable devices. This method combines Kalman filtering and adaptive sampling technology to dynamically allocate a privacy budget according to the fluctuation characteristics of the data stream. By introducing the global sensitivity of the variance value that adapts to the data fluctuation characteristics of wearable devices, it reduces the amount of added noise and improves the accuracy of published data. Experimental results show that compared with the filtering and adaptive sampling (FAST) algorithm for differential privacy time series monitoring, the algorithm proposed in this paper has higher availability of published data under the condition of meeting the same privacy budget.Keywords
This publication has 10 references indexed in Scilit:
- PrivApprox: Privacy-Preserving Stream AnalyticsInformatik Spektrum, 2019
- Calibrating Noise to Sensitivity in Private Data AnalysisJournal of Privacy and Confidentiality, 2017
- Differentially private event sequences over infinite streamsProceedings of the VLDB Endowment, 2014
- Efficient and accurate strategies for differentially-private sliding window queries2013
- Differential privacy under continual observation2010
- Privacy integrated queries2009
- Robust De-anonymization of Large Sparse Datasets2008 IEEE Symposium on Security and Privacy (SP 2008), 2008
- ACHIEVING k-ANONYMITY PRIVACY PROTECTION USING GENERALIZATION AND SUPPRESSIONInternational Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2002
- k-ANONYMITY: A MODEL FOR PROTECTING PRIVACYInternational Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2002
- Protecting respondents identities in microdata releaseIEEE Transactions on Knowledge and Data Engineering, 2001