Model-based Systems Engineering Efficiencies

NASA, JPL, and other space agencies rely on ever larger, more complex software systems to do challenging science. This trend raises the stakes and drives the need for breakthrough solutions which reduce mission risk. At the laboratory, the emergence of model-based design reformulates the design process, and therefore the nature of software systems engineering must change accordingly. For example, Mars Polar Lander is thought to have failed when the software turned off its engines too soon – the flaw in the software-system design was introduced when the software requirements were derived from the system requirements; had there been a software-system model, this flaw would have been easily detected with the approach we advocate. Model based engineering is used on other missions, but we're specifically talking about 3 Mars missions (Mars Science Laboratory, Mars 2020, and Mars Sample Retrieval Lander) in this paper. Model-based systems engineering promises to shift the way space mission designs are captured, reviewed and elaborated from being centered on drawings and text-based documents to computer processable models. In this context, model-based software systems engineering aims to validate mission-specific, model-based components of spacecraft flight software designs. Our goal is to identify the places where applications of model-based design allows for 3-fold or better improvements in defect detection rates. For example, if current requirements inspections are able to detect 70% of a certain class of defects – that leaves 30% undetected. We are seeking techniques that are able to detect at least 90% of that class of defects – thus reducing the undetected rate to 10% or less (a 3-fold reduction or better). The results of this research are necessary to ensure the laboratory is ready for model-based design and the increasingly challenging missions developed with this paradigm. This paper gives a synopsis of how these methods and techniques evolved and makes recommendations as to how they should be infused into systems engineering at the laboratory.