Statement of Work
for NEPHEST
The objective of the NEPHEST project,
jointly funded between DARPA/ITO and NSF/CISE, is the development of
a National Experimental Platform for Hybrid and Embedded Systems
Technology. The
University of California at Berkeley is developing theories,
software, and computational tools for the hierarchical modeling of
distributed hybrid and embedded systems by providing technologies
for their composable specification, analysis, simulation, and
synthesis. In our
subcontract to the NEPHEST effort, we will provide assistance and
support to the statement of work outlined in the NEPHEST
proposal.
1.
We shall assist in surveying the state-of-the-art in
hybrid and embedded system technology. The Berkeley contribution
to the report (Deliverable 1) will focus on established research
projects and major industrial R&D and standardization
efforts. Specifically included in this survey will be the SystemC
initiative (http://systemc.org)
and other component-based methods from the hardware design
technology community, real-time Java and related language-based
efforts to provide a design framework for embedded systems,
real-time CORBA and related middleware aimed at embedded real-time
systems, and synchronous languages and related computational
paradigms aimed at embedded systems. These will be evaluated with
respect to their emphasis on effective composition of components for
hybrid and embedded systems.
2.
We will develop an architecture design and
demonstration of an initial experimental prototype of an open
framework for integrating hardware and software components of
large-scale experiments for hybrid and embedded system
research. Because of the very short delivery schedule, this design
and prototype will be based on the Ptolemy II framework from
Berkeley (http://ptolemy.eecs.berkeley.edu), adapted to include
recently developed concepts of interface definition encompassing
dynamic properties of components. These interface concepts bring well-established
notions of information hiding, polymorphism, and inheritance from
object-oriented architectures into actor-oriented component
architectures, which are better suited to hybrid and embedded
system design. The
design will be given as Vanderbilt-style meta models, which are
static structure object models (in UML) annotated as necessary
with semantic information and constraints. The initial
experimental prototype will be Java implementations of this static
structure. This design will be the Berkeley contribution to
Deliverable no. 2.
3.
We will develop a suite of re-usable components
that demonstrate the concepts of deliverable 2 by showing how
polymorphic interface-based component design can lead to effective
integratable component libraries. We will propose component and interface
specification formats based on established syntaxes like XML and
IDLs and will identify a suite a tools (such as graphical editors,
visualization tools, and engineering process support tools) that form
the essential framework for a national experimental
platform.
4.
We will help clarify the role of challenge problems
in future NEPHEST efforts by formulating the experiment that a
challenge problem development performs. That is, we will help define the metrics by which
positive and negative experimental outcomes will be recognized. It
will not be sufficient in challenge problems for the outcome to be
“it works” or “it flies.” Instead, there must be some
demonstrable improvement in modularity, robustness, performance,
design process, or cost.
5.
We will adapt the GSRC web-based collaboration
framework (http://gigascale.org) to form a
prototype for a web accessible national repository for components,
tools, verification results and documents. We will integrate with
this framework executable components and tools based on the
Ptolemy II framework, as adapted in deliverable 2, and in
collaboration with Vanderbilt, will design an architecture for
versioning and executable component assembly over the net.
6.
We will evaluate the feasibility of NEPHEST in
collaboration with other team members.