Database Intrusion Detection Using Role Profiling with Role Hierarchy

摘要

Insider threats cause the majority of computer system security problems. An anomaly-based intrusion detection system (IDS), which can profile normal behaviors for all users and detect anomalies when a user's behaviors deviate from his/her profiles, can be effective to protect computer systems against insider threats. Although many ID-Ses have been developed at the network or host level, there are still very few IDSes specifically tailored to database systems. We build our anomaly-based database IDS using two different profiling methods: one is to build profiles for each individual user (user profiling) and the other is to mine profiles for roles (role profiling) when role-based access control (RBAC) is supported by the database management system (DBMS). Detailed comparative evaluations between role profiling and user profiling are conducted, and we also analyze the reasons why role profiling is more effective and efficient than user profiling. Another contribution of our work is that we introduce role hierarchies into database IDS and remarkably reduce the false positive rate without increasing the false negative rate.