Research And Implementation On The Multilevel Secure Relational Database Management System

In multilevel secure database systems, the "Write-Up" protocol of classical BLP model violates integrity constraints and produces covert channel and inference channel problem. Polyinstance can solve it, but it takes semantic ambiguity problem at the same time. This paper presents a multiple structure tables secure model to solve the question of semantic ambiguity in DBMS. The new model avoids the convert channel in BLP model. It is a preferable method to solve semantic ambiguity. The new model not only ensures the security but also strengthen the integrity and availability by large margin. Covert channel analysis is an inevitable problem of high security information system reseach. Covert channels can leak confidential information under the supervision of security mechanism, so its threat is very serious.When developing a secure system at or above B2, it needs covert channel analysis by TCSEC. This paper introduces the definition and classification of covert channel, compares with the advantages and drawbacks of present covert channel identification, and proposes the CFT method in detail,and analyses the security model of multiple structure table. Database inference control problem is a well-known problem in database security research. Users with relatively low clearence may retrieve the lower data and infer higher classified data. This paper investigates the inference problems due to functional dependencies(FD) in a multilevel relational database with element classification schemes, analyzes the faults of CLA algorithm presented by Su and Ozsoyolu, correlative better algorithms are proved. According to the theory mentioned above, a multilevel secure DBMS is implemented and the theory mentioned in this paper is correct by the result of our experiment.