Security and privacy research

I'm interested in problems arising in computer security surrounding outsourced computation and privacy-preserving protocols. With Mike Walfish and his students (notably Srinath Setty), I've been working on using the theory of probabilistically-checkable proofs to design systems which allow a client to outsource a computation to an untrusted server (e.g., "the cloud"). The server returns a result and a proof of correctness; ideally, it takes the client less time to verify the proof then to check by just recomputing the result. (This setup is sometimes referred to as "verifiable computation" or "verifiable computing".)

Our project has a webpage, and we've written a few papers on our work.

A hybrid architecture for interactive verified computation
joint with Victor Vu, Srinath Setty, and Michael Walfish
To appear in IEEE Symposium on Security and Privacy (Oakland), 2013.
Resolving the conflict between generality and plausibility in verified computation
joint with Srinath Setty, Benjamin Braun, Victor Vu, Bryan Parno, and Michael Walfish
To appear in ACM European Conference on Computer Systems (EuroSys), 2013. 		preprint.
Taking proof-based verified computation a few steps closer to actual practicality
joint with Srinath Setty, Victor Vu, Nikhil Panpalia, Benjamin Braun, and Michael Walfish
21st Usenix Security Symposium (Usenix Security), 2012. 		published version.
Making argument systems for outsourced computation practical (sometimes)
joint with Srinath Setty, Richard McPherson, and Michael Walfish
19th Network and Distributed System Security Symposium (NDSS), 2012. 		published version.
Toward practical and unconditional verification of remote computations
joint with Srinath Setty and Michael Walfish
13th Workshop on Hot Topics in Operating Systems (HotOS), 2011.		 published version
With various people, I've done work on designing location-based services that preserve locational privacy. The idea is to demonstrate it is possible to provide useful services (e.g., automatic mileage-based tolling) without requiring pervasive tracking of individuals.