Paper 2017/105

A Secure and Fast Dispersal Storage Scheme Based on the Learning with Errors Problem

Ling Yang, Fuyang Fang, Xianhui Lu, Wen-Tao Zhu, Qiongxiao Wang, Shen Yan, and Shiran Pan

Abstract

Data confidentiality and availability are of primary concern in data storage. Dispersal storage schemes achieve these two security properties by transforming the data into multiple codewords and dispersing them across multiple storage servers. Existing schemes achieve confidentiality and availability by various cryptographic and coding algorithms, but only under the assumption that an adversary cannot obtain more than a certain number of codewords. Meanwhile existing schemes are designed for storing archives. In this paper, we propose a novel dispersal storage scheme based on the learning with errors problem, known as storage with errors (SWE). SWE can resist even more powerful adversaries. Besides, SWE favorably supports dynamic data operations that are both efficient and secure, which is more practical for cloud storage. Furthermore, SWE achieves security at relatively low computational overhead, but the same storage cost compared with the state of the art. We also develop a prototype to validate and evaluate SWE. Analysis and experiments show that with proper configurations, SWE outperforms existing schemes in encoding/decoding speed.