Toward Basing Fully Homomorphic Encryption on Worst-Case Hardness

Gentry proposed a fully homomorphic public key encryption scheme that uses ideal lattices. He based the security of his scheme on the hardness of two problems: an average-case decision problem over ideal lattices, and the sparse (or “low-weight”) subset sum problem (SSSP). We provide a key generation algorithm for Gentry’s scheme that generates ideal lattices according to a “nice” average-case distribution. Then, we prove a worst-case / average-case connection that bases Gentry’s scheme (in part) on the quantum hardness of the shortest independent vector problem (SIVP) over ideal lattices in the worst-case. (We cannot remove the need to assume that the SSSP is hard.) Our worst-case / average-case connection is the first where the average-case lattice is an ideal lattice, which seems to be necessary to support the security of