Tools for model-based security engineering: models vs. code

We present tools to support model-based security engineering at both the model and the code level. In the approach supported by these tools, one firstly specifies the security-critical part of the system (e.g. a crypto protocol) using the UML security extension UMLsec. The models are automatically verified for security properties using automated theorem provers. These are implemented within a framework that supports implementing verification routines, based on XMI output of the diagrams from UML CASE tools. Advanced users can use this open-source framework to implement verification routines for the constraints of self-defined security requirement

In a second step, one verifies that security-critical parts of the model are correctly implemented in the code (which might be a legacy implementation), and applies security hardening transformations where that is not the case. This is supported by tools that (1) establish traceability through refactoring scripts and (2) modularize security hardening advices through aspect-oriented programming. The proposed method has been applied to an open-source implementation of a cryptographic protocol implementation (Jessie) in Java to build up traceability mappings and security aspects. In that application, we found a security weakness which could be fixed using our approach. The resulting refactoring scripts and security aspects have found reusability in the Java Secure Socket Extension (JSSE) library