CAREER: Practical verification of outsourced computations

University Of Texas At Austin, Austin TX

Investigators

Abstract

Many computing services are moving to the "cloud" because cloud computing offers economic and operational benefits. However, cloud services not only are complex black boxes operated by third parties but also refuse to give service guarantees. Thus, instead of trying to make cloud services more trustworthy (that is, giving customers reasons to place faith in the cloud), this project explicitly removes trust from the cloud: the project is building systems that give cloud customers guarantees, even under arbitrary service malfunctions. Two such systems concern storage and function computation. Under the storage system, the challenge is to use client-side verification to resist arbitrary or malicious cloud malfunctions (losing data, returning stale data, corrupting data, compromising data, etc.). Under the function computation system, a client specifies a computation to a server, the server executes it and returns the output and auxiliary information, and the client uses the auxiliary information to efficiently check whether the output is correct. The approach is to adapt, and translate into practice, a fascinating body of theory (on probabilistically checkable and interactive proofs) that is currently thought to be impractical. With qualitative advances in reliable yet practical computing, this project will make cloud computing safer for existing customers and spur the adoption of cloud services. This will mean more people paying less for computing, producing beneficial effects throughout the computational ecosystem. The educational thrusts include teaching graduate students to teach (applying the K-12 student teaching model), graduate and undergraduate mentoring, curriculum development, and outreach to underrepresented groups.