Abstract:
In the midst of the technology advancements that we are witnessing nowadays, information
security is of an essence. Information warfare is actively introducing new
offensive strategies to exploit databases that include sensitive data, or even classified
and secret data in the case of the multilevel ones. The attacks vary from stealing sensitive
data, to modifying crucial records, or simply corrupting the database. Many
approaches to detect and prevent from such attacks are available, from firewalls to intrusion
detection systems. But in many cases, prevention can fail, and detection could
happen long after the attack. That implies the necessity of damage assessment and recovery
to restore the database integrity. We introduce in this work a hybrid lightweight
approach based on clustering and graphs to track transactional dependencies in multilevel
databases. To the best of our knowledge, targeting multilevel databases has not been tackled before, so we will compare our approach with single level databases
approaches, taking into consideration the additional security and integrity constraints
imposed by the multilevel nature.
