Share This
Center for Algorithms and Interactive Scientific Software (CAISS)

Cloud Computing Security

Several trends are contributing to a growing desire to ``outsource'' computing from a (relatively) weak computational device to a more powerful computation service. Consider cloud computing, where businesses buy computing time from a service, rather than purchase, provision, and maintain their own computing resources. Sometimes the applications outsourced to the cloud are so critical that it is imperative to rule out accidental errors during the computation, let alone malicious behavior from the cloud service provider.

The proliferation of mobile devices, such as smart phones and netbooks, provides yet another venue in which a computationally weak device would like to be able to outsource a computation, e.g., a cryptographic operation or a photo manipulation, to a third-party and yet obtain a strong assurance that the result returned is correct.

In all of these scenarios, a key requirement is that the amount of work performed by the client to generate and verify work instances must be substantially cheaper than performing the computation on its own. It is also desirable to keep the work performed by the workers as close as possible to the amount of work needed to compute the original function. Otherwise, the worker may be unable to complete the task in a reasonable amount of time, or the cost to the client may become prohibitive.

Group members are actively researching methods for efficient verifiable computation, i.e. schemes that allow a client to efficiently verify the result of an outsourced computation, by investing much less work than required by the computation itself.

Affiliated Publications

D. Fiore and R.Gennaro Improved Publicly Verifiable Delegation of Large Polynomials and Matrix Computations. ACM CCS 2012

S.Benabbas, R.Gennaro and Y.Vahlis. Verifiable Delegation of Computation over Large Datasets, CRYPTO 2011.

R.Gennaro, C.Gentry, B.Parno. Non-Interactive Verifiable Computing: Outsourcing Computation to Untrusted Workers, CRYPTO 2010.

This research is supported in part by the National Science Foundation under Grant SaTC #1017660 and by the U.S. Army Research Laboratory and the U.K. Ministry of Defence and was accomplished under Agreement Number W911NF-06-3-0001.