An Immunity-Based Intrusion Detection Solution for Database Systems

Abstract

Database intrusion detection has been an important research area in database security. It focuses on malicious transaction attacks, which cannot be handled by traditional database security mechanisms, such as authorization, access control, integrity control, and so on. Although there have appeared some intrusion detection systems, current researches on malicious transaction detection are limited in accuracy and efficiency. Inspired by natural immune system, we propose a novel immunity-based intrusion detection solution for database system in this paper. It provides an additional layer of defense against DBMS misuse, especially malicious transactions. The ability to learn and to adapt to the environment dynamically entitles the system to detect both known and unknown malicious transaction intrusions efficiently. Simulations show that the database intrusion detection system based on data immunity can accelerate detection of malicious transaction attacks and improve its accuracy without causing any other performance penalty.