通信学报 ›› 2022, Vol. 43 ›› Issue (1): 127-137.doi: 10.11959/j.issn.1000-436x.2022004

• 学术论文 • 上一篇    

边缘协同的轻量级隐私保护分类框架

熊金波1,2,3, 周永洁1,4, 毕仁万2, 万良1, 田有亮1,3,4   

  1. 1 贵州大学计算机科学与技术学院,贵州 贵阳 550025
    2 福建师范大学计算机与网络空间安全学院,福建 福州 350117
    3 贵州省公共大数据重点实验室,贵州 贵阳 550025
    4 贵州大学密码学与数据安全研究所,贵州 贵阳 550025
  • 修回日期:2021-12-15 出版日期:2022-01-01 发布日期:2022-01-01
  • 作者简介:熊金波(1981- ),男,湖南益阳人,博士,福建师范大学教授、博士生导师,主要研究方向为安全深度学习、移动群智感知、隐私保护技术等
    周永洁(1996- ),女,贵州镇远人,贵州大学硕士生,主要研究方向为安全深度学习、隐私保护技术等
    毕仁万(1996- ),男,湖南常德人,福建师范大学博士生,主要研究方向为安全深度学习、安全多方计算等
    万良(1974- ),男,贵州铜仁人,博士,贵州大学教授、硕士生导师,主要研究方向为网络空间安全等
    田有亮(1982- ),男,贵州六盘水人,博士,贵州大学教授、博士生导师,主要研究方向为算法博弈论、密码学与安全协议、大数据安全与隐私保护、区块链与电子货币等
  • 基金资助:
    国家自然科学基金资助项目(61872090);国家自然科学基金资助项目(61872088);国家自然科学基金资助项目(U1836205);国家自然科学基金资助项目(61772008);贵州省科技重大专项计划基金资助项目(20183001);福建省自然科学基金资助项目(2019J01276);贵州省科技计划基金资助项目([2019]1098);贵州省高层次创新型人才基金资助项目([2020]6008);贵阳市科技计划基金资助项目([2021]1-5)

Towards edge-collaborative, lightweight and privacy-preserving classification framework

Jinbo XIONG1,2,3, Yongjie ZHOU1,4, Renwan BI2, Liang WAN1, Youliang TIAN1,3,4   

  1. 1 College of Computer Science and Technology, Guizhou University, Guiyang 550025, China
    2 College of Computer and Cyber Security, Fujian Normal University, Fuzhou 350117, China
    3 State Key Laboratory of Public Big Date, Guiyang 550025, China
    4 Institute of Cryptography &Data Security, Guizhou University, Guiyang 550025, China
  • Revised:2021-12-15 Online:2022-01-01 Published:2022-01-01
  • Supported by:
    The National Natural Science Foundation of China(61872090);The National Natural Science Foundation of China(61872088);The National Natural Science Foundation of China(U1836205);The National Natural Science Foundation of China(61772008);Science and Technology Major Support Program of Guizhou Province(20183001);The Natural Science Foundation of Fujian Prov-ince(2019J01276);Science and Tech-nology Program of Guizhou Province([2019]1098);Project of High-level Innovative Talents of Guizhou Province([2020]6008);Science and Technology Program of Guiyang Province([2021]1-5)

摘要:

针对边端计算环境下存在感知图像数据泄露与隐私保护分类框架计算低效的问题,提出一种边缘协同的轻量级隐私保护分类框架(PPCF),该框架支持加密特征提取和分类,在边缘节点协同分类过程中实现对数据传输和计算过程的隐私保护。首先,基于加性秘密共享技术设计一系列安全计算协议;在此基础上,两台非共谋的边缘服务器协同执行安全卷积、安全批量归一化、安全激活、安全池化等深度神经网络计算层以实现 PPCF。理论与安全性分析证明了PPCF的正确性和安全性,性能评估结果显示,PPCF可达到与明文环境等同的分类精度;与同态加密和多轮迭代计算方案相比,PPCF在计算开销和通信开销方面具有明显优势。

关键词: 边缘协同, 隐私保护目标分类, 加性秘密共享, 深度神经网络, 安全计算协议

Abstract:

Aiming at the problems of data leakage of perceptual image and computational inefficiency of privacy-preserving classification framework in edge-side computing environment, a lightweight and privacy-preserving classification framework (PPCF) was proposed to supports encryption feature extraction and classification, and achieve the goal of data transmission and computing security under the collaborative classification process of edge nodes.Firstly, a series of secure computing protocols were designed based on additive secret sharing.Furthermore, two non-collusive edge servers were used to perform secure convolution, secure batch normalization, secure activation, secure pooling and other deep neural network computing layers to realize PPCF.Theoretical and security analysis indicate that PPCF has excellent accuracy and proved to be security.Actual performance evaluation show that PPCF can achieve the same classification accuracy as plaintext environment.At the same time, compared with homomorphic encryption and multi-round iterative calculation schemes, PPCF has obvious advantages in terms of computational cost and communication overhead.

Key words: edge-collaborative, privacy-preserving object classification, additive secret sharing, deep neural network, se-cure computing protocol

中图分类号: 

No Suggested Reading articles found!