物联网感知数据隐私保护关键技术研究

Owing to the dramatic growth of acquiring information from the physical world, various types of Internet-of-Thing systems have been pervasively adopted in our daily life and work. Furthermore, their scale and functionality are still under developing. Currently, the IoT systems may facilitate the improvement of urban life, accelerate the economic growth, and enhance the integration of the whole nation, while the privacy preservation of sensory data inside is a critical, essential, and unavoidable issue for implementing all these purposes. However, existing results cannot be applied for sophisticated IoT systems as they mainly focus on single user, single data dimension, or single service. Therefore, it is an essential problem and great challenge to design the privacy preservation techniques for IoT systems. In this project, we exploit the correlations among sensory data, the correlations among users, and the diversity among sensing devices, to design a novel privacy preservation framework from multiple perspectives. The objective of the privacy preservation techniques is to comprehensively derive and preserve the private information beneath the sensory data, while retaining the efficiency of the IoT systems. More specifically, our research topics include the analysis and interaction of private information, the algorithms for privacy preservation, and the techniques for high-efficient privacy-preserved data acquisition in IoT systems.

随着人类对获取物理信息的诉求不断增长，各类物联网系统被广泛的应用于生活生产，并且其规模和功能性仍在不断发展。物联网系统具有能够改善民生、促进经济增长、增强国力的巨大价值，而针对其中的感知数据隐私保护问题是促成物联网系统被广泛应用中不可回避的关键问题。然而现有的基于单个用户、单维数据、单种服务的隐私保护技术无法胜任物联网系统的复杂环境。如何有效的支持物联网感知数据隐私保护是目前人们面临的重点挑战。本项目针对物联网中感知数据和用户关联性强、设备多样性的特点，从隐私保护的多个环节入手，构建适用于物联网感知数据的隐私保护框架。本项目以全面的发现并保护隐私信息、保证物联网系统的执行效率为目标，研究隐私信息的保护技术，包括隐私信息分析和交互技术研究；隐私信息保护机制研究；隐私保护的高效数据获取技术研究；通过构建的原型系统验证所提出的技术的正确性和高效性。

物联网数据内容丰富，蕴含着大量与用户关联的信息，在提供大量价值信息的同时，也为隐私保护提出了极大的要求和挑战。然而，现有隐私保护方法主要讲感知数据视为孤立的数据条目进行保护，忽视了数据间的关联性、数据模态的多样性、数据用途的多样性、参与者的复杂性以及网络特性等，无法有效应对挑战。为此，本课题针对物联网数据隐私保护的基础理论和方法开展研究，形成包括隐私信息分析及交互技术、隐私数据保护技术、数据隐私保护高效实现技术等方面的一系列成果，涵盖工业互联网、智慧城市、群智感知、数据交易等多个场景，解决了数据收集、数据交易、数据共享、数据发布等一系列关键途径的隐私保护问题，并构建感知数据隐私保护原型系统用于验证算法有效性。本课题执行期间取得了良好的进展，形成了一系列研究成果，并发表于知名国际学术期刊及会议。