基于鼠标及窗口行为的持续身份认证研究

Research on Continuous Identity Authentication Based on Mouse and Window Behavior

针对当前基于鼠标动力学的身份认证方法在开放环境中认证性能低的问题,提出了一种联合鼠标及窗口行为进行跨域分析的方案.首先从采集的数据中提取鼠标动力学以及点击窗口名双域特征;然后使用投票制特征选择算法进行特征筛选,以降低鼠标行为的变异性;最后通过支持向量机构建认证模型,在完全自由的内网中进行持续身份认证实验.实验结果表明,该方法在2min的检测时间中表现出较低的误报率(5.38%)和漏报率(3.16%),且准确率可达到94.2%,优于当前在开放环境下基于鼠标动力学的分类识别研究.

Aiming at the problem of the low authentication performance of the current mouse dynamicsbased authentication method in an open environment,a cross-domain analysis scheme combining mouse and window behavior was proposed.First,the mouse dynamics and the dual-domain features of the click window name were extracted from the collected data,and then feature selection algorithms were used to perform feature selection to reduce the variability of mouse behavior.Finally,the support vector machine was used to build an authentication model,and continuous identity authentication experiments were performed in a completely free intranet.The experimental results show that the method shows a low false alarm rate of 5.38%and a false negative rate of 3.16%in a 2-minute detection time,and the accuracy rate can reach 94.2%,which is better than the current mouse-based power in open environments classification and identification studies.