Fully homomorphic encryption over the integers for non-binary plaintexts without the sparse subset sum problem

In this work, we solve the open problem of designing a fully homomorphic encryption scheme over the integers for non-binary plaintexts in Z Q for prime Q (Q-FHE-OI) without the hardness of the sparse subset sum problem (SSSP). Furthermore, we show that our Q-FHE-OI scheme is a useful optimization fo...

Full description

Saved in:
Bibliographic Details
Main Authors: Aung, Khin Mi Mi, Lee, Hyung Tae, Tan, Benjamin Hong Meng, Wang, Huaxiong
Other Authors: School of Physical and Mathematical Sciences
Format: Article
Language:English
Published: 2020
Subjects:
Online Access:https://hdl.handle.net/10356/140697
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:In this work, we solve the open problem of designing a fully homomorphic encryption scheme over the integers for non-binary plaintexts in Z Q for prime Q (Q-FHE-OI) without the hardness of the sparse subset sum problem (SSSP). Furthermore, we show that our Q-FHE-OI scheme is a useful optimization for evaluating arithmetic circuits on encrypted data for some primes. To that end, we provide a natural extension of the somewhat homomorphic encryption (SHE) scheme over the integers proposed by Cheon and Stehlé (Eurocrypt 2015) to support non-binary plaintexts. Then, a novel bootstrapping algorithm is proposed for this extended SHE scheme by introducing generalizations of several functions in binary arithmetic. As a result, we obtain a Q-FHE-OI scheme for any constant-sized prime Q≥3 without the hardness of the SSSP, whose bootstrapping algorithm is asymptotically as efficient as previous best results. Beyond that, we compare the efficiency of our scheme against a Q-FHE-OI scheme obtained by emulating mod-Q gates with boolean circuits as proposed by Kim and Tibouchi (CANS 2016). Our analysis indicates our proposed scheme performs better for prime Q up to 11287, which improves on the result of Kim and Tibouchi, who showed there is at most one prime, Q=3 where the Q-FHE-OI scheme by Nuida and Kurosawa (Eurocrypt 2015) is a better approach. This overturns our previous understanding that Q-FHE-OI schemes do not provide significant benefits.