Formal Methods in Embedded Design

F or more than two decades, applied formal methods have remained the unexplored frontier of embedded system design— just beyond the reach of practice. There have been inroads, certainly, but no sign of a revolution—even a quiet one—in industry. Considering the rapid progress of technology over this time, the dynamic expansion of applications, and the meager investment in formal methods, just keeping pace with the advancing frontier speaks pretty well for the research. Having watched these trends over the years, I would not forecast any dramatic changes in practice. We should be looking for a sea change, not a revolution. Formal methods have a cumulative impact, reflected in languages and “informal” methods as much as in automated reasoning tools. Nevertheless, I see their use in embedded systems design accelerating in the coming five years, especially if we consider the full spectrum of approaches described in the “Formal Methods Terminology” sidebar. System developers must watch this trend closely and foster appropriate longterm strategies. The most important strategic investment is in expertise. The big winners will be the successful risk takers, but the risks are still significant enough to make expertise more important than technology.