We proposc multi-bit versions of several single-bit cryptosystems based on lattioe problems, the error-ffee version of the Ajtai-Dwork cryptosystem by Goldoeich, Goldwasser, and Halevi
[CRYPΓO′97] , the Regev cryptosystems [STOC
2α13 and STOC 205], and the Ajtai
cιyp− tosystem
[ST∝2W5] .Based on a common
stIucture amongst them, we devclop a generic oechnique for constructin
g their multi-bit versions without increase in the size of ciphertexts.By analyzing the trade-off between the decryption enor and the hardness of underlying lattice problems.it is shown that our multi-bit versions encrypt
O(logn) -bit plaintexts into ciphertexts of the same length as the original ones with the reasonable sacrifices of the hxdness of the underlying lattice problems.Our technique also provides a ncw algebraic property pseudo-homomorphism of the lattice-based crypto- systems.
We construct public-key cryptosystems that are secure assuming the *worst-case* hardness of approximating the shortest vector problem on lattices. Prior cryptosystems with worst-case connections (e.g., the Ajtai-Dwork system) were based either on a *special case* of the shortest vector problem, or on the conjectured hardness of lattice problems for *quantum* algorithms. Our main technical innovation is a reduction from certain variants of the shortest vector problem to corresponding versions of the "learning with errors" (LWE) problem; previously, only a quantum reduction of this kind was known. In addition, we construct new cryptosystems based on LWE, including a very natural chosen ciphertext-secure system that has a much simpler description and tighter underlying worst-case approximation factor than prior constructions. (Duration: 30 minutes, on or before Wednesday.)