International Association
for Cryptologic Research

Most of the current fully homomorphic encryption (FHE) schemes are based on either the learning-with-errors (LWE) problem or on its ring variant (RLWE) for storing plaintexts. During the homomorphic computation of FHE schemes, RLWE formats provide high throughput when considering several messages, and LWE formats provide a low latency when there are only a few messages. Efficient conversion can bridge the advantages of each format. However, converting LWE formats into RLWE format, which is called \textit{ring packing}, has been a challenging problem. We propose an efficient solution for ring packing for FHE. The main improvement of this work is twofold. First, we accelerate the existing ring packing methods by using bootstrapping and ring switching techniques, achieving practical runtimes. Second, we propose a new method for efficient ring packing, \textsc{HERMES}, by using ciphertexts in Module-LWE (MLWE) formats, to also reduce the memory. To this end, we generalize the tools of LWE and RLWE formats for MLWE formats. On a single-thread implementation, \textsc{HERMES} consumes $10.2$s for the ring packing of $2^{15}$ LWE-format ciphertexts into an RLWE-format ciphertext. This gives $41$x higher throughput compared to the state-of-the-art ring packing for FHE, \textsc{PEGASUS} [S\&P'21], which takes $51.7$s for packing $2^{12}$ LWE ciphertexts with similar homomorphic capacity. We also illustrate the efficiency of \textsc{HERMES} by using it for transciphering from LWE symmetric encryption to CKKS fully homomorphic encryption, significantly outperforming the recent proposals \textsc{HERA} [Asiacrypt'21] and \textsc{Rubato} [Eurocrypt'22].

Homomorphic encryption (HE) has open an entirely new world up in the privacy-preserving use of sensitive data by conducting computations on encrypted data. Amongst many HE schemes targeting on computation in various contexts, Cheon--Kim--Kim--Song (CKKS) scheme is distinguished since it allows computations for encrypted real number data, which have greater impact in real-world applications. CKKS scheme is a levelled homomorphic encryption scheme, consuming one level for each homomorphic multiplication. When the level runs out, a special computational circuit called bootstrapping is required in order to conduct further multiplications. The algorithm proposed by Cheon et al. has been regarded as a standard way to do bootstrapping in the CKKS scheme, and it consists of the following four steps: ModRaise, CoeffToSlot, EvalMod and SlotToCoeff. However, the steps consume a number of levels themselves, and thus optimizing this extra consumption has been a major focus of the series of recent research. Among the total levels consumed in the bootstrapping steps, about a half of them is spent in CoeffToSlot and SlotToCoeff steps to scale up the real number components of DFT matrices and round them to the nearest integers. Each scale-up factor is very large so that it takes up one level to rescale it down. Scale-up factors can be taken smaller to save levels, but the error of rounding would be transmitted to EvalMod and eventually corrupt the accuracy of bootstrapping. EvalMod aims to get rid of the superfluous $qI$ term from a plaintext $\pt + qI$ resulting from ModRaise, where $q$ is the bottom modulus and $I$ is a polynomial with small integer coefficients. EvalRound is referred to as its opposite, obtaining $qI$. We introduce a novel bootstrapping algorithm consisting of ModRaise, CoeffToSlot, EvalRound and SlotToCoeff, which yields taking smaller scale-up factors without the damage of rounding errors.