ProTECting: garantindo a privacidade de dados gerados em casas inteligentes localmente na borda da rede

Abstract

With the growth of the Internet of Things (IoT) and Smart Homes, there is an ever-growing amount of data coming from within people’s houses. These data are valuable for analysis and to discover patterns in order to improve services and produce resources more efficiently, e.g., using smart meter data to generate energy with less waste. Despite their high value for analysis, these data are intrinsically private and should be treated carefully. IoT data are fundamentally infinite, and this property makes it even more challenging to apply conventional models to achieve privacy. In this work, we propose a locally differentially private solution to estimate frequencies of values in the context of Smart Home data. Our solution is divided into two different strategies, the first one being based on the concept of sliding window, and guaranteeing differential privacy for a defined number of reports. The second strategy uses two rounds of randomization and works even for an infinite number of reports while focusing on getting better utility than the baseline. Concerning specific results of comparison with baseline, the double randomization strategy was able to achieve a reduction of at least 35% in Erro Quadrático Médio (EQM). As for Jensen-Shannon Distance / Distância de Jensen-Shannon (JSD), the other metric used to quantify the utility in this work, the minimum average reduction was approximately 17%.