近邻移动社交网络中邻居发现和好友匹配研究

发布时间：2018-01-15 01:05

  本文关键词：近邻移动社交网络中邻居发现和好友匹配研究　出处：《燕山大学》2015年博士论文　论文类型：学位论文

  更多相关文章： 近邻移动社交网络 近邻发现 好友匹配 占空比 帕耶加密

【摘要】：随着移动网络技术和移动终端设备的迅速发展,越来越多的人们开始使用各种移动社交网络服务,位置邻近的移动用户无需接入互联网络就可以直接通过安装在智能终端上的蓝牙或者Wi Fi接口进行用户交互,从而构建近邻移动社交网络,这种新型的社交网络模式便于人们结交新朋友并可以进行面对面地交流,受到越来越广泛的关注。近邻发现和好友匹配是近邻移动社交网络工作的基础,研究高效的解决方法对于推动这种社交网络的理论研究和实际应用都具有重要意义。首先,提出基于Bi-direction的近邻发现方法。该方法使用一个静态活跃时间槽和两个动态活跃时间槽,在每个周期内动态活跃时间槽分别从左右两个方向相对移动。针对非对齐模式,Bi-direction中采用了条纹探测原理,只需要增加一个信标,即可减少一半的活跃时间槽,极大地降低了近邻发现所需占空比,同时为了使条纹探测也能服务于对齐模式,设计了溢出条纹探测方法,提出随机选择动态活跃时间槽的首次开启位置可以进一步提高发现效率。其次,提出基于邻居加速的近邻发现方法。该方法利用间接邻居发现的思想,将已经发现的邻居节点的邻居加入到自身邻居列表,通过活跃时间槽调度算法和节点自身的电能存量选择额外开启活跃时间槽的位置和个数,进一步提高间接邻居发现效率。该方法可以与任何基础方法结合使用,实现了性能的提升,更有利于应用在移动环境中。再次,提出基于分布式计算的动态好友匹配方法。定义了两个隐私保护级别,用户属性优先级可以按照用户需求动态调整,采用帕耶加密方法对用户属性优先级信息进行加密,利用其同态性和自我屏蔽性实现用户信息不被泄露,设计提前过滤协议首先淘汰掉不满足阈值条件的候选用户,通过相关系数法好友匹配协议实现隐私保护级别Ⅰ下的好友匹配,设计广义Jaccard系数法匹配协议实现隐私保护级别Ⅱ下的匹配过程。最后,提出基于双服务器的第三方好友匹配方法。该方法中同时使用匿名服务器和计算服务器进行好友匹配。用户信息被分为用户ID信息和用户属性优先级信息两部分,分别用匿名服务器公钥和计算服务器公钥对其加密传输,匿名服务器实现用户ID信息的置换和反置换处理,实现了用户ID信息和用户属性优先级信息对应关系的破坏和还原的目标,计算服务器使用改进的广义Jaccard系数法计算用户相似度,其不能获取用户属性优先级对应的真实用户ID信息,弥补了传统第三方计算方法的缺陷,实现了隐私保护级别Ⅲ下的好友匹配过程。实验结果表明,本文提出的近邻发现方法提高了近邻发现的效率,可以更好地满足移动环境下的需求,分布式和第三方计算两种好友匹配方法不仅大幅降低了移动终端的计算和通信开销,而且可以应用于不同的应用场景,有利于近邻移动社交网络的应用推广。
[Abstract]:With the rapid development of mobile network technology and mobile terminal devices, more and more people begin to use various mobile social network services. Adjacent mobile users can interact with each other directly through Bluetooth or Wi Fi interfaces installed on intelligent terminals without access to the Internet, so as to build a neighboring mobile social network. This new model of social network makes it easy for people to make new friends and can communicate face to face, which has attracted more and more attention. The discovery of close neighbors and the matching of friends are the basis of the work of mobile social networks. The study of efficient solutions is of great significance to promote the theoretical research and practical application of this kind of social networks. First of all. A method of neighbor discovery based on Bi-direction is proposed, which uses a static active time slot and two dynamic active time slots. The dynamic active time slot moves relatively from the left and right directions in each cycle. The fringes detection principle is used in the non-aligned mode Bi-direction, and only one beacon is added. The active time slot can be reduced by half, which greatly reduces the duty cycle required by the nearest neighbor detection. In order to make the fringe detection also serve the alignment mode, a method for detecting overflow fringes is designed. It is proposed that random selection of the first opening position of the dynamic active time slot can further improve the discovery efficiency. Secondly, a neighbor accelerated neighbor discovery method is proposed, which utilizes the idea of indirect neighbor discovery. The neighbor of the neighbor node is added to the neighbor list, and the location and number of the additional active time slot are selected by the active time slot scheduling algorithm and the power storage of the node itself. Further improve the efficiency of indirect neighbor discovery. This method can be combined with any basic method to achieve improved performance, more conducive to the application in mobile environment. Again. A dynamic friend matching method based on distributed computing is proposed. Two levels of privacy protection are defined, and the priority of user attributes can be dynamically adjusted according to the needs of users. The user attribute priority information is encrypted by using Paya encryption method, and the user information is not leaked by its homomorphism and self-shielding. First, the filter protocol is designed to eliminate the candidate users who do not meet the threshold condition, and then realize the friend matching under the privacy protection level I through the correlation coefficient method friend matching protocol. A generalized Jaccard coefficient matching protocol is designed to implement the matching process under the privacy protection level 鈪，

本文编号：1426079