One of the main impediment to a wider adoption of cloud storage and computing services is the need to keep data private. Recently, several research communities have begun exploring new techniques that allow a client to outsource storage and computation over private data to a potentially untrusted cloud service, while maintaining the privacy of the data. These techniques are highly varied depending on a large number of factors, like the ownership of the input and output data, the type of processing required, the desired level of protection, and the level of trust that the client is willing to put in different components of the outsourcing service. This highly collaborative project combines many different areas of expertise with the aim of substantially changing the way outsourced data and computation are perceived and used in the real world.
This project focuses on finding a modular approach to the specification, design and analysis of privacy preserving mechanisms in the context of outsourced computation. The investigators explore a broad solution space that combines algorithmic techniques, limited use of trusted hardware, and distributed computation and trust, and create a common framework that allows one to compare and analyze vastly different solutions in a modular and composable manner. The project will also explore the theoretical and practical limits of what each of these solutions can achieve and hone in on a practical solution that possibly combines many different techniques to achieve the right balance of efficiency and security.
