Activity-Based Implementations of Systems of Systems
A System of Systems (SoS) is a composition of component systems and is naturally modeled as a DEVS coupled model which is the computerized representation to support “virtual build and test.” However, given that SoS refers to something existing in the real world, there is also an implementation model, a model that reflects more about the characteristics of the environment in which the SoS will live. This chapter presented activity concepts as a means to bridge the gap between information-level requirements (behavior and timing) and energy consumption in the implementation. This bridge enables implementations of SoS that minimize energy while meeting information-level requirements. In particular, we show how hardware synthesis from DEVS coupled models can exploit disparity in the activity level of its components, giving rise to design for low-power optimization methods. The basic approach is a globally asynchronous, locally synchronous design pattern that enables efficient clock management and clock gating of individual design elements. Explicitly capturing timing requirements within the system model enables optimization to create a design that differentially assigns clock frequencies. This allows component clocks to run only when needed and at frequencies that may be much less then would be needed in the standard single clock design. A DEVS-based hardware implementation of an SoS can exploit the timing requirements and achieve significantly lower power consumption than conventional approaches.
- Hu, X., & Zeigler, B. P. (2011). Linking information and energy—activity-based energy-aware information processing. Simulation.Google Scholar
- Pifer, T. (2012). DEVS-based hardware design, synthesis, and power optimization using explicit time specifications and deterministic path-based latency. Masters Thesis, University of Arizona.Google Scholar
- Zeigler, B. P., & Hammonds, P. E. (2007). Modeling and simulation-based data engineering: introducing pragmatics into ontologies for net-centric information exchange. New York: Academic Press.Google Scholar