Security vs Privacy in Cyber-Physical Systems
This CPS proposal focuses on developing scientific foundations for the research area of CPS Security and Privacy. Our novelty is that while security and privacy are two properties we want these systems to provide, they have not been addressed jointly before.
Privacy concerns have increased in the last couple of years due to the unprecedented scale of data collected related to human activity. These privacy challenges are exacerbated by the widespread deployment of embedded sensors and Internet of Things (IoT) devices, where users are generally unaware of their exposure. In addition to privacy, there is also a growing concern about data integrity in a variety of applications. In particular, the data collected by sensors modernizing our infrastructures (e.g., intelligent transportation systems or smart grids) needs to be trustworthy to achieve their intended objectives, but widespread vulnerabilities in these systems, force us to take into account that part of the data may be corrupted and needs to be filtered out.
Privacy and data integrity in CPS have been studied independently before, but as far as we know, they have not been addressed jointly.
The proposed research will study how privacy-protection mechanism such as differential privacy affects the state-of-art statistical data integrity attack-detection mechanisms for cyber-physical control systems, and analyze how an attacker can take advantage of the noise added to protect privacy to design stealthy attacks that increase the dissemination of false information while remaining hidden in the differential privacy noise. Furthermore, we propose novel defenses against those attacks.
Keywords: privacy, security.
- PDF document
- 806.27 KB
- 123 downloads
- Download
- PDF version
- Printer-friendly version