Statically checking confidentiality via dynamic labels
Bart Jacobs, Wolter Pieters, Martijn Warnier · 2005
This paper presents a new approach for verifying confidentiality for programs, based on abstract interpretation. The framework is formally developed and proved correct in the theorem prover PVS. We use dynamic labeling functions to abstractly interpret a simple programming language via modification of security levels of variables. Our approach is sound and compositional and results in an algorithm for statically checking confidentiality.