Disciplined Design of System of Systems
The Disciplined Design of Systems of Systems (ddosos) is sponsored by
the Army Research Laboratory (ARL) as contract W911NF-11-2-0038 from June 12, 2011 to June 11, 2012. The project covers these areas:
- 1. Multiform Models of Time
- The Ptolemy Project, in collaboration with ARL, will adapt Ptolemy II to provide a simulation framework that supports a
multiform notion of time, where components distributed across a network can disagree about both absolute time and
the passage of time. The Recipient will use the design drivers to evaluate this simulation capability.
- 2. Temporal Isolation
- The Ptolemy Project will develop a notation and formalism for expressing unambiguously in a model the temporal
properties that a system must have to work correctly. These temporal properties become correctness criteria for any
implementation of the model.
- 3. Hybrid Models
- The Ptolemy Project will use state-of-art hybrid system modeling techniques to build and collect examples from the driver
problems that illustrate the pitfalls of hybrid system modeling. These examples can serve as test cases for
improvements in modeling.
- 4. Correct Composition
- The Ptolemy Project, in collaboration with ARL, will use their previously developed ontology system to develop domain-specific
ontologies for the challenge problems. This exercise will highlight weaknesses and missing capabilities.
- 5. Linking Behaviors to Implementation
- The Ptolemy Project will prototype in Ptolemy II a quantity manager mechanism that enables accurate modeling of the
effects of media access control in networks.
- 6. Design Drivers
- Distributed fuel control and power generation systems include the continuous dynamics of physical plants, discrete
behaviors of event sensors, the behavior of communication networks, and the behavior of software. The Recipient
will construct a series of small models that illustrate the challenges in each of the problem domains, and then
compose these small models pairwise to assess the ability of existing technology to handle heterogeneous
composition and to scale to reasonably large systems. The individual component models and their pairwise
combination will be recorded as a set of regression tests.
This group has the following subpages: