【作者】 刘恒；

【导师】 周明天；

【作者基本信息】 电子科技大学 ， 计算机应用技术， 2010， 博士

【摘要】 近年来,随着无线设备和移动计算技术的不断发展,信息空间和物理空间已经相互交融重叠,形成了一个新的研究领域:普适计算。在普适计算空间中,用户可以随时、随地获得数字化的服务。但是这种无处不在的移动计算环境促使了一种新的基于位置的应用服务(Location Based Services,LBS)的产生和发展。然而,LBS服务在给用户提供方便服务的同时也带来了新的安全问题,由于LBS服务的产生,用户的个人隐私信息由原来的虚拟信息(如用户名、密码等)扩展到物理信息(如用户的地理位置信息),这种扩展信息的物理性特征使得其保护变得更加困难。同时普适计算环境中计算主体的移动,使得相互合作的主体具有事先不可预知性,传统的针对静态网络或者封闭系统的隐私保护机制将不再适用于普适计算环境。因此普适计算环境下位置服务的隐私安全问题将是普适计算实施所面临的最大挑战。本文的工作就是要针对现有普适计算环境下的位置服务中用户隐私保护技术存在的薄弱环节,进行新型的位置隐私保护技术及其应用的研究。主要研究内容涉及用户位置坐标的隐私保护、用户身份隐私保护、用户查询请求隐私保护等多种隐私对象,考虑了密集用户环境和稀疏用户环境等不同空间特征对隐私的影响,针对定制类位置服务和非定制类位置服务两种不同应用下的用户隐私内涵和保护技术进行分析研究,取得了如下创新性成果:(1)提出了一种基于密度的匿名空间生成算法,通过对普适空间中的网格化转换,巧妙地将最优模糊空间问题转化为最大密度空间问题,通过用户邻近分布矩阵和用户相对密度实现了用户最优模糊空间的寻找。该算法在利用匿名空间实现对用户位置隐私的保护的同时,利用匿名空间中的多个不可见用户实现对目标用户身份隐私的保护。(2)针对匿名空间保护技术在稀疏用户环境下的匿名失效和服务质量衰减问题进行分析,提出了一种基于干扰信息的隐私保护增强技术,通过干扰用户或干扰坐标对稀疏环境下的用户匿名空间进行调整,降低匿名空间大小提高服务质量,该技术包括了两个关键算法:干扰用户触发算法平衡了用户隐私需求和系统服务性能;干扰信息拟真算法通过周期分布密度矩阵调整干扰用户的分布,增强了干扰信息的拟真性。(3)提出了一种以变化假名切断用户运动轨迹的隐私保护技术,以及与该技术相关的利用服务信息进行空间分割的方法。该技术利用空间分割将用户轨迹划分为多个不连续的轨迹段,通过不同轨迹段之间的假名变换实现对用户轨迹隐私的保护,有效的避免了习惯路径预测带来的匿名失效问题。(4)针对用户查询请求中的查询位置隐私和查询对象隐私泄漏问题,提出了一种双阈值隐私保护技术,该技术通过用户查询敏感模型和服务质量模型在查询生命周期内对用户匿名空间进行动态变换,实现了用户隐私需求和LBS服务质量需求之间的平衡。仿真结果证实该算法具有较好的保护能力和较高的运行效率。(5)基于用户的隐私需求的多样性和环境变化的随机性特征,研究了用户的多角度隐私保护需求,在此基础上实现了基于安全中间件的用户隐私保护平台,该平台通过对用户需求和环境因子的预判断,自动为用户选择合理的隐私保护技术组合,使得用户在不同信任域中移动时可以实现自动的隐私技术切换和隐私策略匹配,从而屏蔽了底层隐私保护平台的复杂性、异构性,使应用服务开发者和用户可以更加关注应用服务的逻辑,不需要考虑隐私域下的隐私保护方式的转换。更多还原

【Abstract】 In recent years, with the rapid development of wireless and mobile computing, Ubiquitous/Pervasive computing is the integration of cyberspace and physical space. In pervasive computing, users can obtain digital services non-intrusively anytime and anywhere. This mobile computing environment promotes the birth and development of Location Based Services(LBS).However, LBS not only make the services more convenient but also introduces a new security challenge. Users private information extend to physical information(such as location) from visual information ( such as ID, password etc.) because of the generation of LBSs. The physical characteristics of the extend privacy information makes it more difficult to protect users’privacy. At the same time, due to the unpredictability of the cooperated principals, it is unsuitable to use traditional security mechanisms suited for static network or closed systems to solve problems occur in pervasive computing environment. Therefore privacy problem is the biggest challenge the pervasive computing will meet.This dissertation aims to research the key issue in both theories and applications of the new privacy protection technology according to the weak link of users’privacy in location based services of pervasive computing. The main research contains several kinds of privacy, such as user location privacy, user identity privacy, user query privacy etc. According to the difference of Customization Service and anonymous service, densely environment and sparse environment are considered in the analysis of privacy connotation and protection technology. The major contributions of this dissertation are as below:(1) A Density-Based Spatial Cloaking Algorithm (DBSCA) is presented. The DBSCA subtly transforms the optimal obfuscating region problem into the maximum density region problem by grid processing of pervasive space .In DBSCA, the Optimized Cloaking Region (OCR) is generated with distribution matrix and relative density matrix of neighboring users. User location privacy is protected with cloaking region, meanwhile, the identity privacy is confused with other users in cloaking. (2) According to analysis of anonymity failure of cloaking and decrease of QoS, a privacy enhance technology based on interference information is proposed. The user cloaking in sparse environment has been mixed with a set of dummy user or interference location to minimize the size of each cloaking area and improve QoS. This technology include two algorithms: The first one called dummy user trigger algorithm is able to achieve the balance of users Location privacy and system performance. The second algorithm enhanced the truthfulness of dummy user through dummy user generation algorithm based on period density matrix.(3) A new method of cutting off users trace by mutil-pseudonym transform is described, as well as associated method of space partition using services information. This technology disconnects users’trajectory into several discontinuous segments. User path privacy has been protected by change pseudonyms between different segments. Adopting this method can effectively avoid the failure of pseudonyms bring by usual route prediction attack.(4) A double threshold privacy technology based on user distribution matrix is proposed for protecting user query privacy, include query location privacy and query content privacy. Users cloaking region had been dynamic modification in query lifecycle by user query privacy model and QoS model to balance user privacy and LBS quality. Experimental results validate the efficiency and effective of proposed technique.(5) Combined with the various demand of protecting privacy and randomness of environmental changes, The users privacy needs of multi-perspective was studied in detail and Platform for Privacy Based on Security Middleware(PPBSM) realized on this basis. Through pre-judgment of users’privacy and environment, the reasonable protection technology been automatic chose by PPBSM. User can realize automatic switching of protection methods and privacy policy. These make the APPs developers and users pay more attention to service logic but not switching of protection methods in different privacy domain by shielding of system complexity and system heterogeneity.更多还原