BES - Differentially Private and Distributed Event Aggregation in Advanced Metering Infrastructures
Paper i proceeding, 2016

Significant challenges for online event aggregation in the context of Cyber-Physical Systems stem from the computational requirements of their distributed nature, as well as from their privacy concerns. In the context of the latter, differential privacy has gained popularity because of its strong privacy protection guarantees, holding against very powerful adversaries. Despite such strong guarantees, though, its adoption in real-world applications is limited by the privacy-preserving noise it introduces to the analysis, which might compromise its usefulness. We investigate the above problem from a system-perspective in the context of Advanced Metering Infrastructures, providing strong privacy guarantees together with useful results for event aggregation taking into account the distributed nature of such systems. We present a streaming-based framework, Bes, and propose methods to limit the noise introduced by differential privacy in real-world scenarios, thus reducing the resulting utility degradation, while still holding against the adversary model adhering with the original definition of differential privacy. We provide a thorough evaluation based on a fully implemented Bes prototype and conducted with real energy consumption data. We show how a large number of events can be aggregated in a private fashion with low processing latency by a single-board device, similar in performance to the devices deployed in Advanced Metering Infrastructures.

Författare

Vincenzo Massimiliano Gulisano

Chalmers, Data- och informationsteknik, Nätverk och system

Valentin Tudor

Chalmers, Data- och informationsteknik, Nätverk och system

Magnus Almgren

Chalmers, Data- och informationsteknik, Nätverk och system

Marina Papatriantafilou

Chalmers, Data- och informationsteknik, Nätverk och system

2nd ACM Cyber-Physical System Security Workshop (CPSS 2016)

59-69

Styrkeområden

Energi

Ämneskategorier

Datavetenskap (datalogi)

DOI

10.1145/2899015.2899021

ISBN

978-145034288-9