On the Impact of Side Information on Smart Meter Privacy-Preserving Methods

TL;DR

This paper investigates how side information impacts the effectiveness of real-time privacy-preserving algorithms for smart meters, comparing two deep adversarial learning approaches and highlighting the importance of side information in privacy performance.

Contribution

It introduces and compares CAL and DI-based adversarial learning methods for smart meter privacy, analyzing their robustness against side information in real-world data.

Findings

CAL is less sensitive to side information than DI.

Privacy is significantly affected by multiple sources of side information.

Both methods perform similarly without side information.

Abstract

Smart meters (SMs) can pose privacy threats for consumers, an issue that has received significant attention in recent years. This paper studies the impact of Side Information (SI) on the performance of distortion-based real-time privacy-preserving algorithms for SMs. In particular, we consider a deep adversarial learning framework, in which the desired releaser (a recurrent neural network) is trained by fighting against an adversary network until convergence. To define the loss functions, two different approaches are considered: the Causal Adversarial Learning (CAL) and the Directed Information (DI)-based learning. The main difference between these approaches is in how the privacy term is measured during the training process. On the one hand, the releaser in the CAL method, by getting supervision from the actual values of the private variables and feedback from the adversary performance, tries to minimize the adversary log-likelihood. On the other hand, the releaser in the DI approach completely relies on the feedback received from the adversary and is optimized to maximize its uncertainty. The performance of these two algorithms is evaluated empirically using real-world SMs data, considering an attacker with access to SI (e.g., the day of the week) that tries to infer the occupancy status from the released SMs data. The results show that, although they perform similarly when the attacker does not exploit the SI, in general, the CAL method is less sensitive to the inclusion of SI. However, in both cases, privacy levels are significantly affected, particularly when multiple sources of SI are included.