Project Title:  Object-Oriented Analysis, Design, Modeling and Composition
(OOADMC) Tools for HPC Modeling and Simulation

Background

In the last year project, we completed our core infrastructure environment 
for Metacomputing FMS using our HLA-compliant, Web/Commodity based WebHLA
architecture. The whole environment was first fine-tuned and tested one
one specific large scale FMS application - Parallel CMS (Comprehensive
Minefield Simulator for Ft.Belvoir - and then packaged in terms of coarse
grain modules / federates that are now ready for reuse and adaptation
for NAVO applications.

Based on lessons learned in our work on parallelizing CMS given by
complex object-orineted C++ simulation code, and discussions with various
DoD contractors such as Metron, Teknowledge and Mak,  we realized that
there is an urgent need for parallel tools in the area of object-oriented
analysis, design modeling and composition. We propose here to build such
high level visual tools on top of our Metacomputing FMS infrastructure 
by integrating our previous WebFlow system with the new industry standards
for visual modeling such as UML (Uniform Modeling Lanaguage).

Project Description

Several HPC programming tools for data parallel programming intense CTAs
such as CFD or SIP were collected and transferred to the DoD users
by the PET program. However, Modeling and Simulation CTAs such as FMS or
IMT that are based on dynamic object-oriented programming with large
number of objects and complex associations require dedicated tools for OO
HPC software engineering. In our previous projects with DoE, NSF and
DoD, we  were experimenting with one such Web based visual composition
environment called WebFlow that offered Java applet based visual graph
authoring tools for distributed dataflow simulations. Meanwhile, new
software industry standards such as UML (Uniform Modeling Language) and
the associated authoring tools such as Rational Rose appeared recently in
the areas of object-oriented software analysis and design for the desktop
computing. We propose here to construct suitable extensions and
adaptations of Web based authoring environments such as WebFlow and
OOAD standards such as UML and to develop new generation tools
that will offer user-friendly standards based visual authoring support for
building HPC object-oriented applications.

More specifically, we will analyze and adapt for the HPC purposes the
ARGO/UML package available from collab.net or tigris.org
that offers open source Java based implementation of the UML standard.
We will combine this tool for fine grain OA&D with our WebFlow model for
visual authoring of coarse grain module based distributed Web
computing applications and with our HLA-compliant WebHLA infrastructure
for high performance / metacomputing federations. The resulting
environment that we call WebFlow/UML will offer consistent, standards
based, all grain visual authoring support ranging from fine grain event
and objects to medium grain components and modules to coarse grain
federates. Hence, WebFlow/UML will support and integrate various standard
middleware encapsulation frameworks ranging from Enterprise Java Beans to
CORBA components to HLA federates. WebFlow/UML middleware to support
such multi-framework integration will be implemented by extending our
JWROB based WebHLA architecture and by adding suitable new CORBA facilties
to JWORB such as the core UML infrastructure (including Meta-Object
Facility (MOF) for general purpose metamodeling, UML Metamodel for
representing UML diagrammatic topologoes and XML based Metadata Interface
(XMI) for Web scripting and exchange of UML models) and the associated
extensions (such as Common Warehouse Metamodel and Software Process
Engineering Metamodels). 

We will layer WebFlow/UML on top of selected FMS programming environments
such as HPC RTI, Tempo/Thema, Parallel NSS or SPEEDES, and DMSO M&S
environments such as the HLA FEDEP (Federation Development Process).
We will interact with the corresponding DoD user communities and with
NAVO researchers to solicite their specific requirements and then to
fine-tune our design and prototypes based on the user feedback.                

Project Deliverables


* WebFlow/UML environment for all grain visual authoring of
  object-oriented component based HPC modeling and simulation software;

* WebFlow/UML adaptations and bindings for selected FMS, DMSO or
  NAVO modeling and simulation environments, based on requirements
  gathered from the DoD users of the current systems.                  

Project Benefits

WebFlow/UML will offer a user-friendly and standards based environment
for object-oriented analysis, design, modeling and composition of high
performance simulation codes. It will improve the quality of DoD
HPC codes, shorten the development time, reduce development costs,
enforce the robustness of deployed systems, enable interoperability
between various encapsulation frameworks and facilitate reuse
in terms of standardized components of various granularities. 

Collaboration

We will interact both with NAVO DoD users to solicitate their requirements
and to align WebFlow/UML best to their needs, and with a broader DoD
community of HPC Modeling and Simulation, engaged in various aspects
of high performance object-orineted software engineering and code
development. For example, we will adapt WebFlow/UML to the needs of
Parallel CMS and we were also encouraged by Jeff Steinman of Metron,
Inc. who developes general purpose SPEEDES simulation engine for PDES
(Parallel/Distributed Event Driven Simulations) to explore UML
extensions towards Visual SPEEDES. The requirement of UML compliance
was presented to Metron by their DoD customer, Wargame2000 enterprise
at the Joint National Test Facility in Colorado Springs, CO.
DMSO is also encouraging the whole modeling and simulation community to
experiment with, propose, prototype and standardize via SISO various
high level user friendly authoring technologies within the HLA
Federation Development Process (FEDEP). We also received positive feedback
for our initial WebFlow/UML concepts from Mak and SAIC after our recent
presentation at the HLA Users Group meeting in Boston.    

Requested Budget: $107,763