Get Smart with Embedded Software

By Mike Evans

Unlike a lot of media campaigns, IBM’s “Smarter Planet” ads feature a basic truth behind them: A high proportion of the products and devices we interact with are getting smarter. The proposition is that these will enable a decent standard of living for the soon-to-be 7 billion people on Earth to live sustainably within the resources of the planet.

Of course, technology only enables. The real drivers are people’s needs, cultures and organizational capabilities. The Internet hosts new types of organizations, accelerating the trend to inter-connect us. So far, it links people to content and other people. In the future, though, it will link people to things—and even link things to things.

Smart devices are products and equipment that are a combination of mechanical, electrical, semiconductor and software elements. It is the embedded software that controls these devices and enables the increasingly complex interactions between people and devices. But a consequence of this is that the design process to develop smart devices cannot simply break the smart product into independent parallel designs by engineers in each discipline for each element. The elements interact.

A function to meet a customer requirement, such as triggering an actuator, might be realized in mechanical, electrical or software controls. The design trade-offs are complex, and extend beyond the traditional engineering function. The unit volume, the unit price, the power consumption, the time to market and the flexibility to change the product capability all interact.

There is also a time dimension to the design trade-offs. As volume increases, implementing a function in a different discipline might reduce cost. Downloading new software to an existing product in the field, creating new functionality, might generate post-sales revenue and maintain customer loyalty.

One group that will be affected first by these changes are design engineers. Smart products change the design process and the role of the engineers. Fortunately, engineers are highly skilled and adaptable people! But they will need smarter tools.

Looking Back to Look Ahead
When I was an engineering student, more than 40 years ago, undergraduate courses were for specific disciplines: electrical, mechanical, electronics. We did not expect to need sales or commercial skills. Our success would be measured in terms of product innovation—introducing new products to the market. Selling them profitably would be somebody else’s problem.

But when we were unleashed onto the industry, we quickly found out that making sure our innovative product met a market demand turned out to be our problem, too. Many of us acquired commercial skills on the job.

Since I studied, the narrow course I took has transformed to a comprehensive grounding in engineering, management and economic principles. Company-based projects are a significant feature of the course, enabling students to apply what they have learned to real industrial problems. The head of the Institute of Manufacturing at Cambridge University, Dr. Mike Gregory, sets out the objective as “combining intellectual rigour with real-world relevance with an integrated community of academics, students and industrialists with a shared passion for modern manufacturing.”

Real-world problems turn out to be much more complex than excellence in design in one’s own discipline. Engineers have developed techniques to handle complexity and interactions, such as multi-disciplinary design teams. A systems engineering approach has become common. The International Council for Systems Engineering (INCOSE) defines systems engineering as “an interdisciplinary approach and means to enable the realization of successful systems. It focuses on defining customer needs and required functionality early in the development cycle, documenting requirements, then proceeding with design synthesis and system validation while considering the complete problem.”

Software Development Becomes Key
Smart products mean that all product development engineers will have to engage more effectively with one particular discipline: software development. Traditionally, engineers have always been able to program computers a bit. But they considered this as supporting their discipline, not an integral part of the product they are designing. Embedded software as an element of a product must be developed with the same professional engineering techniques. Engineers from all disciplines will have to understand these constraints, and how product functionality realized in software interplays with their own discipline.

Smart products add flexibility to products, but complexity to the design process. It can be very difficult to track which requirements are satisfied where. For smart products, a successful design process will need further emphasis on requirements definition; modeling the design at successive stages of refinement; simulation and analysis of a virtual product; and test and verification that the design requirements are met as we replace the virtual product design with physical realization.

Many engineering software tools still only work within a single discipline. What is needed are tools that comprehensively support all the different disciplines, and allow designs to be verified both at the detail of the problem and as a complete solution to the requirements.  

Mike Evans is research director of Cambashi Ltd, based in Cambridge, UK. Contact him via [email protected].