Faster Bootstrapping with Polynomial Error

Bootstrapping is a technique, originally due to Gentry (STOC 2009), for “refreshing” ciphertexts of a somewhat homomorphic encryption scheme so that they can support further homomorphic operations. To date, bootstrapping remains the only known way of obtaining fully homomorphic encryption for arbitrary unbounded computations. Over the past few years, several works have dramatically improved the efficiency of bootstrapping and the hardness assumptions needed to implement it. Recently, Brakerski and Vaikuntanathan (ITCS 2014) reached the major milestone of a bootstrapping algorithm based on Learning With Errors for polynomial approximation factors. Their method uses the Gentry-Sahai-Waters (GSW) cryptosystem (CRYPTO 2013) in conjunction with Barrington’s “circuit sequentialization” theorem (STOC 1986). This approach, however, results in very large polynomial runtimes and approximation factors. (The approximation factors can be improved, but at even greater costs in runtime and space.) In this work we give a new bootstrapping algorithm whose runtime and associated approximation factor are both small polynomials. Unlike most previous methods, ours implements an elementary and efficient arithmetic procedure, thereby avoiding the inefficiencies inherent to the use of boolean circuits and Barrington’s Theorem. For 2 security under conventional lattice assumptions, our method requires only a quasi-linear Õ(λ) number of homomorphic operations on GSW ciphertexts, which is optimal (up to polylogarithmic factors) for schemes that encrypt just one bit per ciphertext. As a contribution of independent interest, we also give a technically simpler variant of the GSW system and a tighter error analysis for its homomorphic operations. ∗School of Computer Science, College of Computing, Georgia Institute of Technology. Email: [email protected] †School of Computer Science, Georgia Institute of Technology. Email: [email protected]. This material is based upon work supported by the National Science Foundation under CAREER Award CCF-1054495, by the Alfred P. Sloan Foundation, and by the Defense Advanced Research Projects Agency (DARPA) and the Air Force Research Laboratory (AFRL) under Contract No. FA8750-11-C-0098. The views expressed are those of the authors and do not necessarily reflect the official policy or position of the National Science Foundation, the Sloan Foundation, DARPA or the U.S. Government.