According to Edward A. Lee, in “Computing Needs Time”, Communications of the ACM, May 2009, pp. 70-79,
Embedded systems will no longer be black boxes, designed once and immutable in the field; they will be pieces of larger systems, a dance of electronics, networking, and physical processes. An emerging buzzword for such systems is cyber-physical systems, or CPS.
Realizing the potential of CPS requires first rethinking the core abstractions of computing. Incremental improvements will continue to help, but effective orchestration of software and physical processes requires semantic models that reflect properties of interest in both.
I’ve focused on making temporal dynamics explicit in computing abstractions so timing properties become correctness criteria rather than a QoS measure. The timing of programs and networks should be as repeatable and predictable as is technologically feasible at reasonable cost. Repeatability and predictability will not eliminate timing variability and hence not eliminate the need for adaptive techniques and validation methods that work with bounds on timing. But they do eliminate spurious sources of timing variability, enabling precise and repeatable timing when needed. The result will be computing and networking technologies that enable vastly more sophisticated CPS applications.
More blog entries on Computing.