Enhancing Model Driven Security through Pattern Refinement Techniques

Security requirements are typically defined at a business abstract level by non-technical security officers. However, in order to fulfill the security requirements, technical security controls or mechanisms have to be considered and deployed on the target system. Based on these security controls security patterns have to be selected. The MDS (Model Driven Security) approach uses security requirement models at a high level of abstraction to automatically generate security artefacts that configure security services. The main drawback of the current MDS solutions is that they consider just one security pattern for each security requirement. Current SOA and cloud services are scattered across multiple heterogeneous security domains. Partners and clients with different security infrastructures are changing continuously, which requires the support of multiple patterns for the same security service. The challenge is to provide configurable security services that can support different patterns. In order to overcome this shortcoming we propose a framework that integrates pattern refinement to the MDS approach. In this approach a security pattern refinement layer is added to the traditional MDS layers. The pattern refinement layer supports the configuration of one security service with different patterns, which are stored in a pattern catalog. For example, our approach enables the generation of security artefacts that configure a non-repudiation service to support both fair non-repudiation and naive non-repudiation patterns.