Opportunistic multi-party shuffling for data reporting privacy
This addresses privacy concerns for voluntary participants in data collection, but it is incremental as it builds on existing mixing strategies.
The paper tackled the problem of ensuring privacy in data collection frameworks by proposing an opportunistic multi-party shuffling scheme to mix data and obfuscate its source, simulating cases with artificial and real-world data to show effectiveness depending on time constraints.
An important feature of data collection frameworks, in which voluntary participants are involved, is that of privacy. Besides data encryption, which protects the data from third parties in case the communication channel is compromised, there are schemes to obfuscate the data and thus provide some anonymity in the data itself, as well as schemes that 'mix' the data to prevent tracing the data back to the source by using network identifiers. This mixing is usually implemented by utilizing special mix networks in the data collection framework. In this paper we focus on schemes for mixing the data where the participants do not need to trust the mix network or the data collector with hiding the source of the data so that we can evaluate the efficacy of peer to peer mixing strategies in the real world. To achieve this, we present a simple opportunistic multi-party shuffling scheme to mix the data and effectively obfuscate the source of the data. We successfully simulate 3 cases with artificial parameters and then use the real-world Mobile Data Challenge (MDC) data to simulate an additional 2 scenarios with realistic parameters. Our results show that such approaches can be effective depending on the time constraints of the data collection and we conclude with design implications for the implementation of the proposed data collection scheme in real life deployments.