Biblio

The interaction design community increasingly addresses how digital technologies may contribute to societal transformations. This paper aims at understanding transformation ignited by a particular constructive design research project. This transformation will be discussed and analysed using resilience thinking, an established approach within sustainability science. By creating a common language between these two disciplines, we start to identify what kind of transformation took place, what factors played a role in the transformation, and which transformative qualities played a role in creating these factors. Our intention is to set out how the notion of resilience might provide a new perspective to understand how constructive design research may produce results that have a sustainable social impact. The findings point towards ways in which these two different perspectives on transformation the analytical perspective of resilience thinking and the generative perspective of constructive design research - may become complementary in both igniting and understanding transformations.

This paper proposes a novel privacy-preserving smart metering system for aggregating distributed smart meter data. It addresses two important challenges: (i) individual users wish to publish sensitive smart metering data for specific purposes, and (ii) an untrusted aggregator aims to make queries on the aggregate data. We handle these challenges using two main techniques. First, we propose Fourier Perturbation Algorithm (FPA) and Wavelet Perturbation Algorithm (WPA) which utilize Fourier/Wavelet transformation and distributed differential privacy (DDP) to provide privacy for the released statistic with provable sensitivity and error bounds. Second, we leverage an exponential ElGamal encryption mechanism to enable secure communications between the users and the untrusted aggregator. Standard differential privacy techniques perform poorly for time-series data as it results in a Θ(n) noise to answer n queries, rendering the answers practically useless if n is large. Our proposed distributed differential privacy mechanism relies on Gaussian principles to generate distributed noise, which guarantees differential privacy for each user with O(1) error, and provides computational simplicity and scalability. Compared with Gaussian Perturbation Algorithm (GPA) which adds distributed Gaussian noise to the original data, the experimental results demonstrate the superiority of the proposed FPA and WPA by adding noise to the transformed coefficients.