A Real-Time Authentication Method Based on Cursor-hidden Scene

摘要

Identity theft, as a common insider attack method, is difficult to detect because it is hard to distinguish the legal user from the masquerader with a stolen legal identity. Compared with traditional identity authentication method such as password or fingerprint, behavior biometrics based on HCI (Human-Computer Interaction) has been more useful and effective in real-time authentication. However, existed approaches either require longer authentication time or are designed for special scenario. In this paper, we propose a real-time authentication approach based on mouse-hidden scene to detect identify theft attack. When operations of the suspicious masquerader are detected, the cursor is hidden deliberately. Under the scene, we assume that mouse operators become anxious and show some unique and instinct mouse movement operations. Based on the movement traces, behavioral model is generated and used for detecting the suspicious masquerader. The experiments show the approach can achieve 2.6 percent of FAR and 3.3 percent of FRR while authentication time consumed is acceptable in practice.