nForge: A Workbench for System Designers

ABSTRACT: Early design analysis of embedded real-time systems is vital to prevent late stage architecture changes, which can be costly from program risk and cost perspectives. System designers typically use various standards for modeling, including UML and SysML with MARTE augmentations, with supporting tools. These do not integrate well with analysis tools - e.g., for schedulability, performance - due to proprietary and in-house nature of analysis tools, non-standard data exchange mechanisms, and broad general-purpose standards that can be difficult to interpret and apply for individual projects"design analysis needs. We propose to investigate and design nForge, a workbench that will include (a) mechanisms for project-specific customization/enforcement of modeling standards overcoming specific limitations, (b) open, scalable and standards-based data interchange framework, and (c) automated change propagation and alert generation upon design violations. nForge will support UML and SysML with MARTE augmentations, and integrate with widely used design modeling tools. Demonstrations of feasibility and scalability will be conducted in the context of performance analysis for a DoD avionics challenge problem, in collaboration with an industry partner with access to rich program-scale sanitized datasets. The approach will also be applicable to other design analysis - e.g., power consumption and reliability. BENEFIT: Defense and aerospace programs will benefit from an open, scalable and automated means of integrating system design models with domain-specific analysis tools and methods. Facilitating early stage design analysis through such standard means will save significant amount of late stage rework and will eliminate program risks. Software-intensive cyber-physical systems in telecom, automotive, industrial controls, consumer products, wearable electronics and medial devices markets will also benefit from a scalable solution that can work across the range - from small, embedded systems to large system-of-systems.