匿名性能衡量方法研究

Research on Anonymity Measure

对现有的多种匿名性衡量方法进行分析和比较,对熵值和θ方法的局限性作了详细分析,指出这两种方法都不能很好地反映出攻击者作局部判定的难易.在此基础上,提出一种新的基于部分熵值的匿名性能衡量方法,该方式在假设匿名攻击有效的前提下,给出相近匿名集的定义,并且基于相近匿名集的元素个数、概率总和以及概率分布重新定义了匿名性衡量的指标.文中详细给出了该方法设计的思想,分析和计算表明,新的衡量方法能较好地体现攻击中常用的排除法,反映出匿名概率集中概率突出且接近者影响攻击判定的事实,能更好地衡量出系统的匿名性能.

Some typical anonymity measures are introduced and compared in this paper, which points out that neither the entropy measure nor the θ-method can accurately the difficulty for anonymity attacker. Based on the analysis, a new anonymity measure based on the entropy value of partial members is proposed. In the new measure, on the premise of the attack being effective, the definition of close anonymity set is given and a new metric for anonymity evaluation is proposed considering the number of members, sum of and distribution of probability in the close anonymity set. The idea of the new measure is discussed detailedly in this paper. The results of analysis and calculation show that the new measure can well explain the fact that elimination is always used in practical attacks and only the objects with relative large adjacent probability could confuse the attaeker＇s decisions, which is a preferable measure for anonymity evaluation.