基于k-匿名的轨迹隐私保护技术研究

发布时间：2018-04-02 00:40

本文选题：轨迹隐私保护　切入点：DTW距离函数　出处：《江西理工大学》2017年硕士论文

【摘要】：随着移动网络和定位设备的飞速发展,各种移动应用中的数据成井喷式增长,致使大数据普遍存在。同时,由于各种公司和研究机构对数据的分析与挖掘。因此,只要存在数据的地方,就存在数据安全隐患。保证发布数据的轨迹隐私不泄露是本文的研究目的。使用经典轨迹k-匿名算法对轨迹信息进行发布的过程中,构造轨迹等价类造成较大信息损失,导致数据可用性较低。同时,经典轨迹匿名算法使用欧几里德函数度量轨迹距离时,无法降低噪声采样点给距离度量带来的较大误差。为此,本文着重对降低轨迹等价类构造信息损失和避免噪声干扰的轨迹匿名算法进行研究。本文的研究内容主要分为以下几部分:1.经典算法基于欧几里德距离度量函数的轨迹相似性计算过程中,要求时间点一一对应,无法度量有局部时间偏移的轨迹间的相似性。在构造轨迹等价类的过程中删除较多位置信息,对轨迹数据的可用性影响较大。为此,本文提出了一种可以度量不等长距离度量函数——DTW距离函数,并结合一种新的轨迹匿名模型——(k,?,p)-匿名模型,构造了DTW-TA轨迹匿名算法。在满足轨迹隐私保护要求的前提下,有效地控制了信息损失。2.在传统算法计算轨迹距离时,由于噪声采样点的存在,会导致轨迹距离出现较大误差,降低轨迹相似性,降低轨迹数据的精度。针对这一情况,本文结合LCSS距离函数和(k,?)-匿名模型提出了LCSS-TA轨迹匿名算法。该算法通过将两个轨迹采样点之间的距离映射成0或1,降低了噪声采样点导致的距离误差,有效地提高轨迹信息的实用性。综上所述,本文提出的基于DTW距离函数的DTW-TA轨迹匿名算法使发布轨迹数据的可用性得到有效提高,k-匿名改进模型下的LCSS-TA算法较大地降低了噪声点对轨迹距离度量的影响,提高了轨迹数据的精度。
[Abstract]:With the rapid development of mobile network and positioning equipment, the data in various mobile applications are blowout growth, which makes big data widely exist. At the same time, because of the analysis and mining of data by various companies and research institutions, As long as there are places where data exist, there are hidden dangers of data security. The purpose of this paper is to ensure that the privacy of the trajectory of the published data is not leaked. In the process of publishing the trace information, we use the classical trajectory k- anonymous algorithm to publish the trace information. The construction of locus equivalents causes a great loss of information, which results in low availability of data. Meanwhile, when the classical trajectory anonymous algorithm uses Euclidean function to measure the distance of trajectory, It is impossible to reduce the large error caused by noise sampling points to distance measurement. For this reason, In this paper, we focus on the research of trajectory anonymity algorithm, which can reduce the loss of information and avoid noise interference. The main contents of this paper are as follows: 1. The classical algorithm is based on Euclidean distance metric function. In the course of calculating the trajectory similarity of, It is impossible to measure the similarity of locus with local time offset because of the one-to-one correspondence of time points. In the process of constructing locus equivalence class, more position information is deleted, which has a great influence on the usability of locus data. In this paper, we present a new metric function, DTW distance function, which can be used to measure unequal distances, and combine it with a new locus anonymous model. In this paper, a DTW-TA path anonymous algorithm is constructed, which can effectively control the loss of information under the premise of satisfying the requirement of trajectory privacy protection. 2. When the traditional algorithm is used to calculate the trajectory distance, because of the presence of noise sampling points, In this paper, LCSS distance function and LCSS distance function are combined to reduce the accuracy of trajectory data. By mapping the distance between two locus sampling points to 0 or 1, the algorithm reduces the distance error caused by noise sampling points and effectively improves the practicability of trajectory information. An anonymous DTW-TA locus algorithm based on DTW distance function is proposed in this paper, which can effectively improve the availability of published trajectory data and reduce the influence of noise points on trajectory distance measurement. The accuracy of track data is improved.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP309

【参考文献】