基于LT码的损坏数据探测与动态更新方案
LT Codes-based Damaged Data Detection and Dynamic Update Scheme
作者：徐明(国防科技大学 计算机学院, 湖南 长沙 410073)；黄宇锋(国防科技大学 计算机学院, 湖南 长沙 410073)；付绍静(国防科技大学 计算机学院, 湖南 长沙 410073;密码科学技术国家重点实验室, 北京 100878;信息保障技术国防科技重点实验室, 北京 100072)；罗玉川(国防科技大学 计算机学院, 湖南 长沙 410073)
Author：XU Ming(College of Computer, National Univ. of Defense Technol., Changsha 410073, China)；HUANG Yufeng(College of Computer, National Univ. of Defense Technol., Changsha 410073, China)；FU Shaojing(College of Computer, National Univ. of Defense Technol., Changsha 410073, China;State Key Lab. of Cryptology, Beijing 100878, China;Sci. and Technol. on Info. Assurance Lab., Beijing 100072, China)；LUO Yuchuan(College of Computer, National Univ. of Defense Technol., Changsha 410073, China)
收稿日期：2017-08-06 年卷（期）页码：2018,50(5):145-151
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：云存储;LT码;BLS签名;动态更新
Key words：cloud storage;LT code;BLS signature;dynamic update
基金项目：国家自然科学基金面上项目资助（61572026；61672195；61379144）；密码科学技术国家重点实验室开放课题资助项目（MMKFKT201617）；信息保障技术国防科技重点实验室开放课题资助项目（KJ-15-001）
中文摘要
随着云存储的日益普及，确保数据的安全性与可用性也越来越重要，然而数据损坏的情况不可避免，因此如何能够快速探测出损坏数据，是一个亟待解决的问题。已有基于LT编码的云存储方案与基于认证跳跃表的云存储方案对数据用户来说具有较高的通信开销与计算开销，负担较重。作者基于LT编码设计了两级分布式二叉树的云存储方案。该方案通过对探测数据建立两级分布式二叉树结构，利用BLS签名的聚合性质对数据块进行计算并生成聚合标签，验证聚合标签是否正确以定位出损坏数据所处的位置，提升了探测损坏数据块的效率；同时，基于LT码的编码性质，对更新数据块进行异或运算，实现了数据的动态更新并且能对云服务器是否进行数据更新进行验证，从而解决快速准确探测损坏数据与数据动态更新的问题，且该方案具有较好的整体性能。实验结果表明：所设计的方案是基于LT编码的两级分布式二叉树的方案，在损坏数据探测方面，与基于LT编码的逐一探测损坏数据方案相比，存储开销相当，但是计算开销与通信开销大大减少；在数据动态更新方面，与基于RS编码的动态更新方案相比，存储开销与通信开销相当，但是计算开销大大减少。
英文摘要
With the growing popularity of cloud storage, it become more and more important to ensure the safety and availability of data. However, data corruption is inevitable. Therefore, to efficiently detect the damaged data is an urgent issue. The LT codes-based cloud storage scheme and the rank-based skip list cloud storage scheme bring high communication and computation cost to data user. To address these problems, a LT code-based two level distributed binary tree cloud storage scheme was proposed in this paper. The scheme improved the efficiency of detecting damaged data blocks by constructing two level distributed binary tree structure of detection data and generating aggregation labels via BLS signature. At the same time, based on the coding property of LT code, the updated data block was used in exclusive or operation. The dynamic update of the data was realized and the data update of the cloud server can be verified, so as to solve the problem of fast and accurate detection of the damaged data and the data dynamic update. It also has a better overall performance. Compared with the original LT code-based detecting damaged data scheme, our experimental results showed that our approach has the same storage cost but lower communication and computational cost. Compared with the RS code-based updating dynamic data scheme, our experimental results showed that our approach has the same storage and communication cost but lower computational cost.
【关闭】
