【摘要】 针对计算机系统安全中的用户身份识别和监控等基本问题,提出了一种利用鼠标动力学行为特征进行身份识别的新方法.通过采集各种应用环境下的鼠标行为数据,从交互和生理2个层面上对人机交互过程中计算机用户的鼠标行为特征进行建模、分析,以达到实时监测用户身份、检测非法用户的目的.所提方法可为身份监控原型系统实时采集用户的行为数据,并将当前的行为与用户的历史行为模型进行比较,以判断和检测用户身份,再依据判断检测结果产生实时的响应,从而有效防止非法用户侵入.实验采集并分析了10个用户的鼠标行为数据,通过特征降维与神经网络分类相结合的算法,得到了0.48%的误识率和2.86%的拒识率,充分展示了基于鼠标动力学行为模型进行身份认证和监控的可行性.更多还原

【Abstract】 User identification and monitoring is one of the most important issues in computer system security.A new method for user identification is presented based on the dynamics of computer mouse behavior.Data of mouse behavior in various applications are collected,and users’ mouse behavior in human computer interaction is analyzed and modeled,specifically from both the interaction layer and the physiological layer.Based on the dynamic model,a real-time identity authentication and monitoring prototype system is developed,which can intercept users’ mouse behavior data,and compare user’s current behavior with his history behavior model in order to detect and authenticate current user’s identity.According to the result of authentication,system responds real-time and defends against the intrusion of illegal user.An algorithm that uses feature dimension reduction and neural network for classification is applied in experiments for ten users.The experimental results show that mouse dynamics is effective for authenticating and monitoring user identities with a false accept rate(FAR) of 0.48% and a false rejection rate(FRR) of 2.86%.更多还原