Model driven design of secure properties for vision-based applications: A case study

In this paper we discuss an approach to overcome difficulties and gaps which are typically encountered when dealing with security-oriented model-driven approaches. In particular, we state that state-of-the-art MDS approaches are not suitable for modern companies and industry in general, and address security only at a late stage of development, often causing big delays and reengineering costs due to extensive reworks. Instead, we propose to adopt in the SEcube platform an OTA-based XMDD approach to integrate security ab-initio. In addition, since our approach is based on a set of reusable SIBs organized within dedicated palettes in DIME, we decouple the issue of guaranteeing that the SIBs are correct and secure from the issue of analyzing the applications, which can be greatly simplified by knowing the characterization of each SIB in advance. We apply our approach to the concrete realm of computer vision steering robotics, present the safety and security properties elicited on the specific case study, and discuss the ways they can be enforced.