Trade-off Analysis for SysML Models Using Decision Points and CSPs

Abstract

The expected benefits of Model-Based System Engineering (MBSE) include assistance to the system designer in finding the set of optimal architectures and making trade-off analysis. Design objectives such as cost, performance and reliability are often conflicting. The SysML-based method OOSEM and the ARCADIA method focus on the design and analysis of one alternative of the system. They freeze the topology and the execution platform before optimization starts. Further, their limitation quickly appears when a large number of alternatives must be evaluated. The paper avoids these problems and improves trade-off analysis in a MBSE approach by combining the SysML modeling language and so-called “decision points”. An enhanced SysML model with decision points shows up alternatives for component redundancy, and instance selection and allocation. The same SysML model is extended with constraints and objective functions using an optimization context and parametric diagrams. Then a representation of a constraint satisfaction multi-criteria objective problem (CSMOP) is generated and solved with a combination of solvers. A demonstrator implements the proposed approach into an Eclipse plug-in; it uses the Papyrus and CSP solvers, both are open-source tools. A case study illustrates the methodology: a mission controller for an Unmanned Aerial Vehicle (UAV) that includes a stereoscopic camera sensor module. Keywords: MBSE, Optimization, SysML, CSP, Papyrus, System engineering, Optimal architecture design, Decision points.