Abstract

In this presentation, I will consider the questions

• What is Model-Based Design?

• Who is using it?

• What are the challenges?

I will use MATLAB and Simulink examples to illustrate important techniques such as algorithm modelling, automatic code generation and verification techniques including formal methods. I will describe how these techniques are being used today – for the development of complex and high reliability software in the aerospace, automotive and other industries. An underlying principle of Model-Based Design is the concept of a “golden-model” that pulls together the combined effort of an entire design team including algorithm developers, software engineers, verification engineers and documentation writers. From this model, engineers can automatically generate production quality code as well as documentation, test results and additional artefacts required to meet the requirements of standards such as DO-178B for development of avionics software.

Despite the widespread adoption of Model-Based Design, new technologies are becoming available and challenges remain that make this a very exciting area to be involved with. Some of the challenges relate to existing workflows that cannot easily be changed in the conservative world of avionics software development; however, change is being pushed through by the need to reduce costs and shorten development cycles. With the efficiency of automatically generated code now matching manually-written software it becoming difficult to justify the time and cost of hand-coding. I will describe a traditional workflow and how this can be evolved to embrace Model-Based Design.

Increasing demands for computational power and availability of multi-core and multi-processor systems are opening up new opportunities. I will discuss how Model-Based Design can take advantage of such technologies in order to provide a seamless workflow from modelling through to verification and production deployment of an embedded system.