Project DetailsOptimizing Privacy in the Personal Cloud
|Laboratory : LSIR||Semester / Master||Completed|
Traditional personal clouds (e.g. Dropbox, Google Drive…) give the cloud provider complete access to uploaded users' data. Other solutions (e.g. SpiderOak, Mega, …) allow the user to encrypt his files before sending them to the cloud. In the first case, the users' privacy is at risk. In the second case, the provider cannot deliver services (e.g. document editing, file viewing…) based on users' files. Hence, there is a tradeoff between privacy and utility in the personal cloud.
In this project, we aim to handle this issue by optimizing the choice of privacy policies, in order to automatically control the privacy-utility tradeoff.
Given this aim, it becomes apparent that the Personal Cloud needs to be considered as a techno social system, in which the people collaborate to achieve their goals. As the name indicates, such systems are in essence social systems driven by technology. To put it in another way, such systems reside on the overlap of traditional computing systems and social systems. For pursuing our goals, we need to take into account factors such as: user behavior; their privacy needs; the utility they hope to extract from using the system, etc. Such research will be interdisciplinary in nature and would benefit from ideas in economics, psychology and the social sciences in general.
We aim to generate synthetic workloads and user-behavior models to get some insight into the nature of the problems and potential solutions. Towards this end we would use agent-based simulations that analyze a considerable space of behavioral traits or strategies.
On the privacy front, we will be aiming to design and deploy solutions, which can, in an automated fashion, manage privacy risks. We focus on two of the problem's elements: risk estimation and risk mitigation. The former involves quantifying the risk of data sharing, in order to first inform the users about it and to also weigh it against the utility of data sharing in the management step. We aim to tackle privacy risk mitigation issue by recommending optimized privacy policies, thus relieving the user from the burden of deciding on the policy to match the privacy-utility tradeoff the users envision.
In order to accomplish the above, we will be carrying out a variety of activities including but not limited to:
- analysis of real-world cloud traces for specifying privacy metrics;
- agent based simulations for modeling user behavior and testing proposed solutions;
- implementation of proposed solutions and integration with deployed cloud service(s).
- Having an inclination for interdisciplinary approach
- Also helpful, but not essential, is the passion for building deployable systems.
- Knowledge of Java development
ContactsIn case of any questions, please drop us an email or come to our offices: