Detection and prevention of malicious of relational database management systems (RDBMS)

Abstract

Insider attack is formsthe biggest threat against database management systems. Although many mechanisms have been developed to detect and prevent the misuse activities on the database systems, such as authorized modification on the approved records by the authorized users in malicious intent. However, thesemechanisms still have some limitations in detecting insider attacks. This study proposes a mechanism called dependency mechanism (DM) by utilizing the dependency relationship among database attributes to detect and prevent the authorized modification on approved database records, which have been already used, approved and closed by the system. The proposed mechanism is based on high and low dependency among attributes(columns). These dependencies based on the high and low repetition and usage of the attribute in the database. When the DM recognizes and detects any authorized modification on the approved records, it considers it as a malicious. The results of proposed mechanism DM showed high ability to detect and prevent the malicious modification on the approved records. Flag Based Mechanism (FBM) is considered as a baseline for this study. The evaluation parameter is a detection rate, by which the accuracy of the proposed mechanism is evaluated and compared to the FBM technique.