Model-Based Design Driven by Embedded Systems

With software and control systems becoming standard fare even in what used to be considered straight mechanical-based products, model-based design -- an engineering approach rooted in industries like aerospace and defense -- is gaining traction as companies seek ways to reduce the growing complexity of development while reducing time-to-market.

In traditional development processes, engineers gather requirements from multiple sources, create a paper specification, and then work off that paper spec to produce a detailed design concept. The concept is then prototyped using simulation or physical hardware, and it is continually checked against requirements until a suitable design is achieved. Testing typically occurs toward the end of this multistage process -- a design workflow that can be problematic if errors or flaws are detected late in the game, when they are far more costly to remedy.

In contrast, with the model-based design approach, a system model serves as the centerpiece of the development process, starting as early as the requirements phase and evolving through design, implementation, and testing. This single model is an executable specification that can be linked to the original requirements. The approach allows for two-way traceability between the design and the requirements while enabling a multidomain engineering team to provide continual input and refine the model throughout the development process. Simulation is employed every step of the way to determine whether the model is behaving as desired.

The MathWorks, which markets one of the leading tools for model-based design, says the practice was originally tuned for the development of dynamic systems such as control systems, signal processing, and communications systems but is systematically being embraced by other industry segments grappling with complexity issues.

"Software complexity is the big driver," says Ken Karnofsky, senior strategist for signal processing applications at The MathWorks.

Model-based design first took hold in the automotive, aerospace and other industries dealing with extremely complex, large-scale embedded applications that needed a lot of process rigor beyond some of the ad hoc engineering happening in other segments, he said. Other fields, like industrial machinery and medical communications, are now facing similar requirements, which are prompting a recasting of traditional design approaches in the hopes of finding a more efficient way.