With the convergence of mobile communications, sensors and online social networks technologies, we are witnessing an exponential increase in the acquisition of personal data. All these data end up in corporate servers or in the Cloud and introduce an unprecedented threat on privacy. The PerSoCloud project aims at designing and implementing a robust proof-of-concept of a full-fledged Privacy-by-Design Personal Cloud Sharing Platform, and validating it with jurists and sociologists.
The Personal Cloud is a paradigm giving users the ability to store their complete digital environment, synchronize it among various devices and share it with other users and applications under their control. Several academic works have already tackled different dimensions of the personal cloud problem and some pioneer systems are arising in the market place, usually with limited functionalities in terms of data sharing and privacy protection. The PerSoCloud project precisely tackles the sharing and privacy issue from the architectural angle, following a fully decentralized approach. We identify three major dimensions of the problem: (1) device sharing (sharing owner's data among the various devices of a same user), (2) peer-sharing (sharing personal data among identified collaborating users) and (3) community-sharing (sharing personal data among a large community of users), all this with strong privacy guarantees.
Scientific challenges are associated to each of these dimensions. Device-sharing requires the design of a data synchronization protocol combined with security primitives guaranteeing that the data is never at risk whatever device using or hosting it. Peer-sharing means being able to exchange personal data among users based on rules and policies to control their access and usage. PerSoCloud will focus on the enforcement of these policies, letting the model to express these policies as open as possible. Finally, Community-sharing means being able to perform privacy-preserving global/big data computations crossing personal data without reintroducing the need to centralize these data. In other word, this means designing large-scale privacy-preserving peer-to-peer architectures and protocols in a decentralized infrastructure mixing storage and computing nodes of various powers and trustworthiness levels (from secure personal devices to high-end servers).
This platform will be validated through three scenarios illustrating each of the targeted challenge. The first two scenarios correspond to usual actions performed today by Cozy users willing to access their documents from multiple devices and willing to share documents with partners. The third scenario is a privacy-preserving community-based application where users can get advices from their community by distributing queries among social graphs, or can be safely involved in global/big data computations over personal data. In addition, a legal analysis will guarantee compliance of this platform with the security and privacy French and EU regulation, which firmly promotes the Privacy by Design principle, including the current reforms of personal data regulation.
The solution proposed in PerSoCloud will capitalize on complementary technologies and expertise from the partners, namely on personal cloud platform, security of personal data, decentralized cloud architectures, as well as on legal and sociological issues.
Besides expected scientific contributions, the PerSoCloud platform should have a significant societal and industrial impact, at the time of an explosion of the management of personal data and of an increasing distrust of individuals towards central servers.
ORANGE (Entreprise autre que TPE ou PME)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Inria - Centre de recherche Saclay - Ile-de-France - Equipe projet SMIS
Université de Versailles Saint-Quentin-en-Yvelines
Help of the ANR 677,493 euros
Beginning and duration of the scientific project: January 2017 - 36 Months