Return-Path: furm@npac.syr.edu Delivery-Date: Tue May 2 03:06 EDT 200 Received: from vega.npac.syr.edu (vega.npac.syr.edu [128.230.8.146]) by postoffice.npac.syr.edu (8.9.3/8.9.3) with ESMTP id DAA08746; Tue, 2 May 2000 03:06:12 -0400 (EDT) Received: from localhost (furm@localhost) by vega.npac.syr.edu (8.9.3/8.9.3) with ESMTP id DAA27306; Tue, 2 May 2000 03:06:29 -0400 (EDT) X-Authentication-Warning: vega.npac.syr.edu: furm owned process doing -bs Date: Tue, 2 May 2000 03:06:29 -0400 (EDT) From: Wojtek Furmanski To: "David E. Bernholdt" cc: Wojtek Furmanski Subject: ARL Year 5 Proposals Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Content-Length: 24443 David, ARL proposals follow. The best split is 65% of my time and 1 GRA since I lose SU benefits if go below 60-something %. Regarding ERDS final report - I cannot deliver it before my trip, I'm sorry. These ARL proposals took much time and effort then I expected. I hope you can excuse me at ERDC and I will submit my reports right after I'm back (May 10). thanks Wojtek ------------------------------------------------------ INSTITUTION NAME: Syracuse University, Department of Electrical Engineering and Computer Science WORK PACKAGE TITLE: FMS Support for SBA PROJECT TITLE: FMS Core Support POC NAME: Wojtek Furmanski EMAIL: furm@npac.syr.edu PHONE: 315-443-1799 FAX: 315-443-1973 CTA or PEI: FMS PROJECT DESCRIPTION Modeling and Simulation plays a strategic, mission-critical role in several domains of DoD, including wargaming, training, weapon engineering and acquisition. As a computationally intense domain, it also calls naturally for HPC support. Our FMS work so far was focused on the wargaming domain and included: a) successful Origin2000 implementation and metacomputing deployment of Scalable Parallel CMS (a "million mines battlefield"); b) support for FMS CHSSI projects and infrastructure such as the SPEEDES simulation engine for parallel object-oriented event driven "optimistic" time-warp simulations. Meanwhile, ARL MSRC established ties with the T&E community and advanced the IMT CTA domain that is focused on single weapon/vehicle/device engineering level Modeling and Simulation. Within the new DoD vision of Simulation Based Acquisition (SBA), FMS and IMT efforts and technologies need to be merged to support collaborative, full life cycle simulations of weapons systems, so that they can be continuously refined while being tested and evaluated in synthetic battlefield environments. Starting in Year 5, we therefore propose to establish closer ties with the IMT CTA, both PET and CHSSI, and to start addressing tangible and practical tasks that would facilitate collaboration and lead us gradually towards SBA environments. The FMS Core Effort will provide the overall support to enable such collaboration between FMS and IMT and the associated Focused Efforts will address specific Year 5 development and training activities. PROJECT OBJECTIVE(S): More specifically, we propose in Year 5: a) to transition our successfully completed Parallel/Metacomputing CMS as a testbed application to the Metacomputing Initiative (lead by ERDC); b) to continue support for CHSSI training, including: 1) preparation for SPEEDES training while Metron, Inc. is working towards the SPEEDES v1.0 software release and documentation, and 2) an initial training activity in support of data intense simulations of relevance for T&E community; c) to initiate two new focused efforts that will offer some immediate advantages to current MSRC users and will facilitate longer term FMS and IMT integration towards SBA: 1) Oracle 8i Support for DICE and HPC T&E; and 2) WebFlow/UML Infrastructure to Enable SBA via Cross-CTA PSEe. The core FMS effort will support the activities listed above at the level of: 1) base interaction between FMS and IMT users interested in collaboratory work towards SBA, 2) identifying needs and opportunities for training in the SBA domain; and 3) tracking relevant technologies such as SPEEDES, Oracle, UML, HLA and Web/Commodity suite including CORBA, Java, COM, XML. DELIVERABLES: 1) Contacts established between FMS and IMT CTAs, both at the PET and CHSSI levels; 2) Long range collaboration plan towards SBA outlined; 3) Near term Year 5 focused activities supported at the level of overall concept development, coordination and technology tracking (Oracle, UML, Web, HLA); 4) Status of SPEEDES clarified and recommendation provided or action taken regarding training development and delivery; 5) Participation in broader Technical Collaboration and Data Mining activities as outlined during the Year 4 PET Review Meeting in ARL, April 2000. CUSTOMERS/END USERS: 1) ARL MSRC and their T&E customers that need data intense simulation support; 2) FMS and IMT CHSSI communities. BENEFIT TO THE WARFIGHTER: Modeling and Simulation in Support of SBA has been identified to be of critical importance for the whole DoD acquisition process. We believe that HPC will play an important role in enabling and shaping the SBA landscape. The end result of successful SBA deployment are higher quality weapon systems produced faster and in a more cost-effective way. PROJECT DEPENDENCIES AND SCOPE: This core project is intended to support all other FMS Year 5 projects, including one Training and two Focused Efforts described in dedicated proposals. There is no bottleneck dependencies between Core and Focused efforts as all focused activities are intended to proceed autonomously and to deliver near term results, with Core playing the role of facilitator, communicator and non-intrusive integrator effort. RISK ELEMENT: SBA is a new DoD initiative which is bound to face several fiscal, technological and cultural barriers. We are fully aware of the associated risks and we build our approach, outlined in the present set of year 5 proposals, around a set of tangible focused efforts that both fit and accelerate the long term vision and will result in near term deliverables for specific DoD users who were already consulted and expressed interest in our proposed deliverables. REQUIRED FUNDING LEVEL: Year X: Year X+1: Year X+2: INSTITUTION NAME: Syracuse University, Department of Electrical Engineering and Computer Science WORK PACKAGE TITLE: FMS Support for SBA PROJECT TITLE: Training Support for Data Intense Simulation Infrastructure POC NAME: Wojtek Furmanski EMAIL: furm@npac.syr.edu PHONE: 315-443-1799 FAX: 315-443-1973 CTA or PEI: FMS PROJECT DESCRIPTION In the FMS Training sector, we have a pending task in area of SPEEDES training - we are monitoring the SPEEDES system (currently at the level of undocumented v0.81 release) and we are waiting for the documented v1.0 release to address the training task. Meanwhile, we intend to prepare high level visual authoring tools that will facilitate the use of SPEEDES for beginners (such as WebFlow/UML environment discussed in another Year 5 FMS proposal). Also, while waiting for completion of the SPEEDES system by Metron, we propose a smaller training task in the area of Oracle middleware that is associated with our other proposed focused effort (Oracle 8i support for DICE). PROJECT OBJECTIVE(S): By fall 2000, we propose to deliver a tutorial at ARL on Oracle 8i support for CORBA and Data Mining. We view it as a good way to establish working relations with ARL MSRC and their T&E users, interested in use of Oracle and in Data Mining. We also expect the topic of Oracle 8i support for CORBA to be of interest for other CTAs interested in ORB based PSEs such as CCM (where the corresponding system is represented by Gateway) and IMT + FMS (where we propose initial effort towards a PSE in Year 5 based on WebFlow/UML). After SPEEDES 1.0 is released and documented, we intend to extend this (initially IMT oriented) training effort towards the FMS domain and to work jointly with Metron to deliver SPEEDES seminars for beginners, hands-on tutorials for early adopters (using our WebFlow/UML tools for Visual SPEEDES discussed in the the associated proposal), and advanced SPEEDES courses for professional FMS programmers. DELIVERABLES: A single tutorial (or a few more elaborate tutorials if needed) on Oracle 8i support for CORBA and Data Mining, delivered in Fall 2000 at ARL. Possible formation of a working group towards SBA during the tutorial sessions. CUSTOMERS/END USERS: We expect CORBA support in Oracle to be of interest to PSE oriented S&T users, and Data Mining support in Oracle to attract the T&E users, thereby creating a creative mix that could stirr some interesting discussions and perhaps form a critical mass towards an SBA oriented working group. BENEFIT TO THE WARFIGHTER: Oracle 8i offers a powerful infrastructure for enterprise computing. Several data-intense DoD enteprises can significantly benefit from the latest capabilities of Oracle 8i in the areas of distributed objects (CORBA) and data mining. PROJECT DEPENDENCIES AND SCOPE: This training effort is linked with the focused project on Oracle-DICE integration. One of the goals of the training task is to establish early contacts with DICE users and developers, hence the training needs to be delivered in the first part of Year 5 (fall 2000). RISK ELEMENT: Risks of training include poor attendance and poor response of the audience. We intend to mitigate the poor attendance risk by addressing both computation (CORBA) and information (Data Mining) communities. We intend to mitigate the poor response risk by preparing an excellent set of lecutures with carefully selected hands-on examples and preceeded by our thorough study of the relevant sectors of the Oracle 8i system and documentation. REQUIRED FUNDING LEVEL: Year X: Year X+1: Year X+2: INSTITUTION NAME: Syracuse University, Department of Electrical Engineering and Computer Science WORK PACKAGE TITLE: FMS Support for SBA PROJECT TITLE: Oracle 8i Support for DICE (Distributed Interactive Computing Environment) and HPC T&E POC NAME: Wojtek Furmanski EMAIL: furm@npac.syr.edu PHONE: 315-443-1799 FAX: 315-443-1973 CTA or PEI: FMS PROJECT DESCRIPTION SBA vision calls for data-intense large scale high performance simulations in support of test and evaluation. A robust interoperatbility framework is needed to transfer, maintain, process and visualize large datasets, associated with typical T&E applications. Oracle 8i seems to be a natural base platform to build upon, due to growing momentum of Oracle in the Internet software industry (as exemplified by the May 8 Business Week Cover Story "Despite the Tech Stock Slide - Oracle Is Cool Again"), and the growing installation base of this technology in the DoD labs (including the recently completed secure installation at ARL MSRC). Syracuse Unviersity developed over the last few years a significant expertise in the use of Oracle database as a backend for Web based collaboration and training environments. Examples include CareWeb - our telemedicine collaboratory system for school nurses, deployed by a recent pilot project in Syracuse City School District, and Language Connect University - a Web based system for foreign language training with database management support for students, teachers, courses, multimedia materials, interactive courses etc., developed in collaboration with a local multimedia training corporation. These products were constructed using Oracle 7, whereas the most recent Oracle 8i offers even much broader range of data storage, management, retrieval and processing capabilities that can significantly accelerate T&E data analysis. The problem, however, is that Oracle 8i is a gigantic software system and the professional comprehensive DBA level know-how in this area comes with a high price tag. We propose to transfer our in-depth expertise in Oracle to DoD users and to further extend it by the latest Oracle 8i capabilities for the benefit of ARL MSRC and their T&E customers. This includes support both for the core database server and the more recent Oracle Application Server (OAS) that offers a flexible and powerful middleware platform. OAS can handle both HTTP (Web) and IIOP (CORBA) protocols and it offers a plug-and-play integration and interoperability architecture in terms of Oracle 'cartridges' that can encapsulate multiple programming languages and paradigms. We view this flexibility of OAS to be of critical importance for SBA system engineering that calls for plug-and- play integration of diverse multi-disciplinary components. We propose to initiate the process of integrating Oracle with HPC systems by a focused Year 5 project that will add Oracle capability to the DICE (Distributed Interactive Computing Environment) system at ARL MSRC. DICE currently uses HDF5 (Hierarchical Data Format v5) based middleware layer that offers interim storage for heavy data that are fetched either from files or from HPC system memories to become target of interactive visualization at the client workstations. In this project we will address the issue of cooperation between HDF5 transient storage and Oracle 8i based persistent storage layers, and we will augment DICE functionality by adding the Oracle storage support for the benefit of DICE users at ARL. PROJECT OBJECTIVE(S): Specific design and implementation strategy requires deeper insight into DICE and will be addressed in the course of this project. DICE offers now a new object based XML scripted abstraction layer on top of HDF5 called DICE Object Directory. Oracle 8i also offers now a new object-relational mapping framework based on CORBA and Java. A new activity at OMG (Object Management Group), led by Oracle and including IBM, Unisys and other vendors, develops Common Warehouse Metamodel that offers CORBA interoperability across several popular data storage formats, inclduing XML-based, object-based, record-based, multi-dimensional and relational models. In this project, we will analyze DICE, Oracle 8i and CWM object / data models, we will enumerate possible interoperability modes between Oracle and DICE, and we will select and implement the most promising integration framework. DICE offers various data access, scripting and visualization capabilities - we will ensure that our inclusion of Oracle backend will fully sustain the current DICE functionality and will augment it by Oracle specific tools such as WebDB page authoring, Enterprise JavaBeans based Java component scripting or SQL*Plus, PL/SQL and SQLJ based database (SQL) scripting. DELIVERABLES: We will develop CORBA/Java based middleware object layer on top Oracle 8i database that will cooperate smoothly with the DICE Object Directory model on top of HDF5. We will also explore the relevance of CWM for DICE-Oracle integration and possibly implement a proof-of-the-concept prototype. As a net result of this project, the DICE users will be able to use Oracle storage model, interchangeably with currently supported (flat file, dynamic memory) storage models, transparently underneath the HDF5 data buffer, and consistently with all current DICE visualization and scripting capabilities. CUSTOMERS/END USERS: ARL MSRC, their T&E customers, all other CTAs that already use or want to use Oracle (for example PET SIP CTA). BENEFIT TO THE WARFIGHTER: Visualization of HPC simulations, offered by DICE, accelerates the HPCMP mission of transfering the latest S&T achievements to the warfighter. Added robust support for persistence storage of large data sets in terms of industry leading Oracle database, proposed in this project, will further the relevance and usefullness of DICE for the analysis and visualization of the mission-criticial HPC applications at the ARL MSRC. PROJECT DEPENDENCIES AND SCOPE: Oracle-DICE integration is a new project that does not depend on other activities. We could mention here the link between the Oracle Application Server to be used in this project and our JWORB (Java Web Object Request Broker) server which follows a similar Web+CORBA integration concept. JWORB was developed in our previous FMS projects to support WebHLA simulations. As a consequence, our HLA-compliant simulation backends can naturally cooperate via JWORB<=>OAS middleware bus with Oracle based data backends and, after completion of the proposed project, can be also dynamically visualized using DICE. RISK ELEMENT: Oracle 8i is a big and complex system. We have working experience with modest scale NT installation at Syracuse, but no experience with the high end SMP Solaris installation at ARL MSRC. Some platform dependent problems might therefore appear in the course of this project with the potential risk of slowing down the transition of our work to ARL. We will mitigate this risk by periodic visits or remote code transfers to ARL MSRC aimed at on-site testing of our Oracle software while it is being developed at Syracuse. REQUIRED FUNDING LEVEL: Year X: Year X+1: Year X+2: INSTITUTION NAME: Syracuse University, Department of Electrical Engineering and Computer Science WORK PACKAGE TITLE: FMS Support for SBA PROJECT TITLE: WebFlow/UML Infrastructure to Enable SBA via Cross-CTA PSEs POC NAME: Wojtek Furmanski EMAIL: furm@npac.syr.edu PHONE: 315-443-1799 FAX: 315-443-1973 CTA or PEI: FMS PROJECT DESCRIPTION Our WebFlow environment for visual graph based authoring of distributed Web applications was successfully adapted to CCM CTA within the Gateway project. We also notice that there is growing industry interest in visual software engineering standards for various computational domains. We therefore believe that WebFlow authoring has significant potential for other CTAs as well. For domains that make essential use of object-oriented programming such as Modeling and Simulation, represented by FMS and IMT, we need a visual object authoring framework - the corresponding standard is being provided now by the industry in terms of the rapidly adopted Uniform Modeling Language (UML). UML offers a set of standardized diagrammatic topologies to visually address several complementary aspects of object-oriented modeling, analysis, design and composition, including Class Diagrams, Object Diagrams, Collaboration Diagrams, Sequence Diagrams, Activity Diagrams, Statechart Diagrams, Component Diagrams, Deployment Diagrams and Use Case Diagrams. UML can be also further extended and customized for various computational domains in terms of CORBA based UML metamodel, standardized by the Object Management Group. We propose here a focused visual modeling and simulation effort, aimed at adapting and customizing UML for selected tasks within the FMS and IMT CTAs, and setting the stage for building FMS and IMT PSEs in support of SBA. Since our WebFlow environment already offers custom visual graph authoring model, a natural next step is to replace its front-end by the UML based extensible standard and to link it with suitable CORBA middleware. Some commercial UML tools already appeared on the market such as Rational Rose by Rational Corporation, but we intend to use instead a promising open source Java based UML editor called ARGO/UML, developed by DARPA and now maintained by the Tigris program within Collab.net ( www.tigris.org ). For a CORBA middleware, we will use our JWORB (Java Web Object Request Broker) developed in our previous FMS projects and used as plug-and-play software bus for WebHLA federations. For comparison, Gateway system was based on commercial ORBacus product. Regardless of a particular ORB vendor, all CORBA middleware based PSEs can naturally collaborate via the common IIOP (Internet Inter-ORB Protocol). By layering UML standard based visual authoring model on top of any such CORBA middleware, we will offer a uniform user-friendly interface to multidisciplinay Problem Solving Environments and we will demonstrate it by prototyping and integrating some initial IMT and FMS tools towards SBA. PROJECT OBJECTIVE(S): In this project, we will construct WebFlow/UML by merging ARGO/UML visual front-end with our JWORB middleware and we will ensure smooth interoperability with other CORBA based models of relevance for HPCMP such as Gateway or Oracle Application Server. In Year 5, we will develop the core system and will illustrate its capability to act as a skeleton of a multi-disciplinary PSE in terms of the following selected trial IMT and FMS application demos: a) Visual Database Schema Editor for Oracle 8i; b) UML Profile for DICE Visualization Dataflow; c) Initial support for Visual SPEEDES. Application a) will allow T&E users to quickly design and implement a set of Oracle tables and to specify visually their relations, constraints, triggers etc. in terms of suitably extended UML diagrams. In Application b), DICE users will be able to use suitably extended UML editor to specify visually the requested dataflow between the HPC backends, interim storage module and visualization filters and viewers. In Appplication c) (subject to availability of documented SPEEDES v1.0), the FMS users will be able to visually assembly a set of coarse grain simulation components (such as ships, tanks, radars etc.) and their interactions, and then to generate automatically the corresponding parallel event-driven SPEEDES based simulation code. By linking a), b) and c) modules (using CORBA), we will be able to select requested simulation components from Oracle database, visualize them on the fly during the composition process using DICE, and to construct SPEEDES based runtime executable - with all these operations performed in a consistent and user-friendly visual framework based on standard compliant UML notation. DELIVERABLES: 1) Core WebFlow/UML as a skeleton of multi-disciplinary PSE, constructed by merging ARGO/UML visual front-end with a Java/CORBA middleware server - here we will support JWORB which is of relevance for WebHLA (FMS), ORBacus which is of relevance for Gateway (CCM) and Oracle Application Server which is of relevance for DICE (IMT & SciVis). 2)Initial IMT application demos of WebFlow/UML such as: a) ER-style visual database schema editor for Oracle; and b) AVS-style dataflow editor for DICE; 3) Initial FMS application - Visual SPEEDES - at the level of Parallel NSS demo, and subject to availability of SPEEDES 1.0 with documentation in the first part of Year 5. 4) Proof of the concept integration of IMT and FMS application demos listed above: a WebFlow/UML based Visual Simulation Editor that selects simulation components from Oracle database, visualizes them using DICE, and simulates them using SPEEDES. CUSTOMERS/END USERS: Initial users of Year 5 deliverables include IMT/T&E customers who use Oracle or DICE and FMS users who use SPEEDES or take SPEEDES training and want to quickly build HPC simulations without expert C++ programming. In the following years, after the WebFlow/UML will be more broadly adopted as a skeleton framework for a multi-disciplinary PSE, users could come from all CTA areas relevant for the HPCMP mission. BENEFIT TO THE WARFIGHTER: Problem Solving Environments facilitate addressing challenges that result in the leading edge S&T transfer to the warfighter. Building usable and efficient PSEs is a challenge by itself which is addressed by the proposed WebFlow/UML framework. Success of this project would accelerate engineering and adoption of PSEs in various CTA areas, starting from IMT and FMS domains. Integration of IMT and FMS PSEs in terms of WebFlow/UML would provide a natural starting framework towards SBA, to be then joined by other dedicated CTAs such as CCM, CSM, SIP etc. PROJECT DEPENDENCIES AND SCOPE: This is a new project and its core part does not depend on other projects: both ARGO/UML and JWORB are available and ready for integration towards WebFlow/UML. DICE application demo proposed here is linked and synergistic with the Oracle-DICE integration effort, proposed in another Year 5 focused task. SPEEDES application depends on the availability of SPEEDES 1.0 in Year 5. RISK ELEMENT: Marketing innovative technologies such as WebFlow/UML to the DoD community might take more time and effort than we could anticipate at the current stage. We try to mitigate this risk by addressing initially realatively simple application problems such as Oracle schema design and DICE visualization (which is backed up by another focused effort). On the other hand, PSEs are relevant only when capable to attack truly complex and challenging problems - therefore we also include the Visual SPEEDES application which carries considerable risk due to current uncertainties regarding SPEEDES availability and manageability. REQUIRED FUNDING LEVEL: Year X: Year X+1: Year X+2: