3.5.1 Warfighter Needs
Computing and Software Technologies (as is the case with Modeling and Simulation) are significantly important to all Joint Warfighting Needs and Capabilities. In the information age, warfighting is significantly influenced by the speed, accuracy, and quality of information provided for applications such as C4I, precision weapons, logistics support, and readiness support. In such applications, a few well-trained humans, augmented or assisted by high performance, automated systems, can outperform dozens, hundreds, and sometimes, thousands of unautomated or poorly automated well-trained personnel. The computing foundations developed by this subarea provide the advantage in any conflict or operation permitting early, decisive victory or rapid non-combat response in operations other than war with minimal cost in assets and human life. Advancements in software and software development productivity support both the capability and the affordability of new and upgraded defense systems. Computing and software technologies such as software re-engineering, software reuse, and software acquisition strategy have significant impact on weapon systems' upgrades, product improvements, and system performance advancements. Pervasive throughout all technology areas, and as important, are embedded high performance computing, embedded software and system engineering, knowledge-base design, advanced hardware/software system architecture technology, information presentation and interaction, and intelligent information technology.
3.5.2 Computing And Software Technology Overview
3.5.2.1 Goals and Timeframes. For ease of portrayal and discussion, the four related but distinct technical aspects that make up this subarea are considered separately. These four aspects are:
Specific computing and software goals and time frames are shown
in the Figure III.10. A roadmap that focuses on the linkages and
key relationships associated with the corresponding DTOs is provided
in Section 4.
| Information
Presentation and Interaction | Heavily
tethered, helmet-mounted 3-D displays | Single user immersible VR with pre-calculated displays, tether-free 3-D displays | Multiple user immersible VR, real-time displays |
| Limited (10 k word), speaker-dependent, vocabularies | Medium(50 k word), speaker-independent vocabularies | Large, speaker-independent vocabularies | |
| Single user, single discipline | Multiple user, single discipline collaboration | Multiple user, multiple discipline collaboration | |
| Machine Intelligence | Real-time planning for intelligent devices | Real-time adaptation of intelligent devices to changing situations | Plan creation and execution among cooperating intelligent robots |
| Autonomous devices operating independently on single tasks | Team tactics demonstrated for autonomous multi-agent behavior | Self-initiated plan creation and execution among cooperating intelligent robots | |
| Unintegrated and unfiltered data searches in Cyberspace | Filtered information searches by intelligent agents in Cyberspace | Filtered and integrated responses to information queries in Cyberspace | |
| Computing Systems | 100 GFLOP/cubic foot for militarized HPC | 500 GFLOP/cubic foot. | T-FLOP/cubic foot |
| 10-20 Percent efficiency on MPP | 50 Percent overall efficiency on MPP | ||
| Max 300 GFLOP sustained | Scalable Designs to 10 T-FLOPS | Scalable Designs to 100 T-FLOPS | |
| Baseline Performance on ATR | 10X Baseline on ATR | ||
| System Design and Evolution | Cost/quality/performance certification process for COTS or other reusable components | Commercial vendor's adhere to cost/quality/certification process | Certification of self-learning/adaptive systems demonstrated |
| No criteria for applications architecture evaluation/comparison | Criteria for non-real-time algorithmic architectures in place | Criteria for real-time and knowledge-based architectures in place | |
| Ad-hoc architectures approach | Standards for architectures based on object orientation demonstration | Standard architectures for real-time, secure and knowledge-based software in place | |
| Developed component by component with extensive use of COTS/GOTS | Domain specific development through specification process only | Warfighter modifiable systems in place |
3.5.2.2.1 Information Presentation and Interaction. The overall goal is to allow the humans in the loop to exploit all information relevant to their individual tasks, without reaching "information overload" and respond in a timely manner. The approach is to make maximum effective use of all the human senses and intellect to perform this task in as natural appearing an environment as possible. One technical challenge is to build and incorporate affordable, high-resolution large 3-D displays into systems to depict an accurate picture of the situation to the visual senses and enhance those system interfaces with natural language and gesture I/O. A second technical challenge is to provide a truly interactive, virtual reality depiction of the situation with human "immersion." The depiction will then be further enhanced by removing artificial tethers such as helmet mounted displays and data gloves, providing real-time updates to the depiction, and allowing the immersion of multiple humans at the same time. Some aids to visualization rely on sensing the location and orientation of the participant. Participant location and orientation are used to adjust the presented visual scene. Inaccurate or delayed tracking can induce virtual reality sickness (akin to motion sickness). Therefore, a third challenge is improvements in measurement resolution, accuracy, and responsiveness needed to promote improvements in aids to visualization.
3.5.2.2.2 Machine Intelligence. The overall goal is to provide techniques for automated reasoning to advance the capabilities of single and multiple autonomous robotic systems and intelligent aids to human decision making. Deep challenges remain in integrating reactive and reflective planning, achieving control in unstructured, real-time environments; automating cooperation among multiple intelligent systems; using simulated devices and environments to debug and optimize complex control software; and exploiting varieties of machine learning methods in conjunction with those simulations. Meeting these challenges contributes to the ability for autonomous and semi-autonomous vehicles and weapons platforms to operate undersea, on the ground, or in the air, as well as allowing intelligent software agents to better aid in decision making and efficient information searches in Cyberspace.
3.5.2.2.3 Computing Systems. For the technologies involved with computing systems, the overall goals are to overcome the inevitable (and already present in some applications) limitations in computational throughput, as well as the cost, power, size, and weight requirements for single or small numbers of networked computers. These requirements and limitations impact weapons systems, simulators, and engineering support systems. The technical challenges are to cut the costs and size of giga-and tera-floating point operation (FLOP) computers so they might fit into a weapons package, to make efficient use of parallel and massively parallel computing assets, and to design systems that will allow easier, and hopefully, very low cost transition to the future commercial advancements in both hardware and software to achieve truly architecture-independent, high-performance computing.
3.5.2.2.4 System Design and Evolution. In the area of System Design and Evolution, the overall goal is to reduce the time and cost of building, deploying, and modifying software-intensive systems while increasing the overall quality in terms such as residual errors, reusability, and integrity. A subgoal is to allow the warfighter to program the system he needs for his specific battle environment. The technical challenges are to develop and/or automate techniques for forward system engineering approaches and provide tools which have a systematic, holistic approach to design and information management across the system life-cycle, as well as for the activities associated with the reengineering and reverse engineering of legacy systems for modernization or reuse.
3.5.2.3 Related Federal and Private Sector Efforts
3.5.2.3.1 For Information Presentation and Interaction, major related efforts include NASA work on human-computer interfaces for the Space Shuttle, the proposed NASA Space Station and ground-control workstations. and private sector applications in the business domain. Private sector participants include Apple Corporation, Microsoft, Sun Microsystems, IBM, Xerox PARC, and AT&T. Universities having substantial programs in this area, funded through domestic and international customers include Carnegie Mellon, Stanford, the University of Southern California, Georgia Institute of Technology, Massachusetts Institute of Technology, Virginia Polytechnic Institute, and the University of Arizona. The university focus is almost solely on commercial applications of novel interface technologies, virtual reality, and intelligent user interfaces.
3.5.2.3.2 Machine Intelligence Research and Development, while getting the majority of its funding from DoD, is augmented with programs from NASA, dealing with expert systems and intelligent controls; DOT, dealing with applications on land and air traffic control; and the DOE, concentrating on control of industrial processes. NSF supports a broad program of basic research in robotics and intelligent systems. In addition, there are a significant number of industrial firms which have internal research and development (IR&D) projects studying related machine intelligence technology.
3.5.2.3.3 Computing Systems represents a significant part of the U. S. program in High Performance Computing and Communications (HPCC), which involves the Department of Education, NSA, DOE, the EPA, NASA, NSF, NIST, and the National Oceanic and Atmospheric Administration (NOAA). Most of the major vendors of high performance computing systems, including the workstation manufacturers Hewlett-Packard, Sun Microsystems, and Silicon Graphics are building on the scaleable computing technology developed by the programs that are part of this subarea. There are major technology efforts underway at industry/Government sponsored consortia such as MCC (Microelectronics and Computer Technology Consortium), often involving DoD sponsorship matched with significant industrial cost sharing. The TRP has also targeted some of this technology for dual-use development and defense conversion.
3.5.2.3.4 System Design and Evolution is an area of widespread national concern. The DoD R&D organizations work very closely with the industrial sector to address the challenges in this area. NSF has Government-Industry-University Cooperative Research Centers, two examples of which are the Software Engineering Research Center collocated at Purdue University and the University of Florida; and the Center for Information Management Research, collocated at Georgia Institute of Technology and the University of Arizona. The Software Productivity Consortium (SPC) in Herndon, VA gets its basic funding from about 15 major companies with a strong interest in increasing their software productivity and from the DoD. The DoD labs and the SPC have worked out or are currently working out agreements to share and expand on their successful software engineering products. By Congressional action, the National Applied Software Engineering Center has been established, due in no small part to the efforts of the Service Labs and ARPA.
3.5.3 S&T Investment Strategy
3.5.3.1 Technology Demonstrations
3.5.3.1.1 Information Presentation and Interaction is a supporting technology, and as such has no current formal technology demonstrations. However, a Technology Demonstration for Developing Speech Recognition for Future Digital Signal Processors (DSPs) in Handheld Computers was funded by the Technology Reinvestment Program (TRP) and demonstrated a family of continuous speech recognition capabilities ranging from small vocabulary for command and control to a large vocabulary for dictation.
3.5.3.1.2 Machine Intelligence has technology demonstrations in Autonomous Vehicles and Image Understanding Architecture (IUA) Vision and covers a technology demonstration that is key to the Decision Making IS&T subarea, namely USTRANSCOM Planning Tools. The USTRANSCOM Planning Tools demonstration is aimed at developing the next generation of generic AI planning, resource allocation, and scheduling technology. It is responsible for the capture of new AI planning capabilities in robust, application ready software tools, and the demonstration of the feasibility of their application against employment and deployment crisis action planning tasks within the context of USTRANSCOM exercises.
Autonomous Vehicles focuses on ground vehicles for phase IV of a four-phased program. This phase will integrate a reconnaissance, surveillance, and target acquisition subsystem, with a multiple vehicle mission subsystem, resulting in the robust navigation of a team of four vehicles as a screening force in support of manned vehicles. The IUA Vision Demonstration is a TRP to develop and demonstrate important image understanding (IU) products by using and enhancing existing IU software technology and COTS hardware technology in a common architecture. The result will be an architecture that will allow IU capabilities in deployed systems to improve as rapidly as the technology is delivered.
3.5.3.1.3 Computing Systems has a large technology demonstration that provides many enabling technologies for the Information Management and Distribution IS&T subarea as well as the business and combat missions of DoD. Information Infrastructure Services focuses on the Cyberspace areas of electronic transactions, information management, and transaction support services, including common authentication, authorization, and accounting services; resource registration and discovery, real-time multimedia interoperability; and adaptive computing services.
3.5.3.1.4 System Design and Evolution has two technology demonstrations. The first demonstration, Software Life Cycle Support, is intended to develop, demonstrate, and transition state-of-the-art computer system and software engineering technology that supports the complete system life cycle and improves the productivity of the process and quality of the products. Aimed primarily at the embedded world, this program has already demonstrated to Joint Surveillance Target Acquisition Radar System (JOINT STARS), the Next Generation Attack Submarine (NSSN), F-16, and other military programs as well as civil applications in air traffic control. Additional demonstrations to these organizations are planned.
The second demonstration, called the Evolutionary Design of Complex Systems (EDCS) will demonstrate the next generation of technologies, processes and development environments beyond computer-aided software engineering (CASE) and knowledge-based engineering to address the unique requirements of large-scale, complex systems with long life cycles where missions and performance requirements tend to evolve over the life of the system. The objective is to scale-up the incremental development and prototyping paradigms as a means to increase effectiveness of systems and systematically reduce risks over the entire system life cycle. Technology demonstrations will address applications software system architecture technologies to support: (1) the evolution of a system implementation, (2) a knowledge-based environment in which all aspects of a system life cycle are formalized, (3) language support for evolutionary development of software components, (4) a specification techniques for complex software architectures, (5) software systems with components written in multiple programming languages, and (6) experimental tools to support the refinement of software prototypes into production quality systems.
3.5.3.2 Technology Development. The Information Presentation and Interaction aspect of this subarea requires advancements in many technologies to achieve the goals for optimizing human performance in the information rich combat environments of the present and future. Some of the advancements needed will be funded by the commercial sector, but in many cases, the DoD must scale up the "game oriented developments" to real-world military applications. DoD must make continue investments in: (1) real-time adaptable user interfaces, (2) crew-aiding systems, (3) intuitive multi-user interfaces, (4) real-time processing and display of solid objects, (5) locomotion control, (6) robust real-time speech recognition and understanding, (7) text processing, understanding, and multi-lingual translation, and (8) interface and interaction development tools to facilitate design and development of interfaces, human-computer dialogs, integration of human and computer control, system composition and integration, and virtual reality fidelity.
Machine Intelligence, which is most heavily influenced by DoD investment, investments in the following areas to achieve its objectives for automated reasoning and autonomous vehicles: (1) extensions and alternatives to rule-based systems, particularly CASE-based reasoning, (2) machine learning and machine vision, (3) tools for the off-line design and on-line adaptation of fielded systems, (4) hybrid (incorporating both human and computer) control strategies including behavior-based architectures and for semi-autonomous control, incorporation of the technologies covered in Information Presentation and Interaction into robotic systems, (5) real-time, autonomous planning and scheduling of tasks that can deal with temporal aspects and uncertainty, (6) integration of intelligent systems across domains, and (7) the general tools and technology to build and validate intelligent systems.
Computing Systems needs advancements in both hardware and software technologies to achieve its objectives for overcoming the DoD mission shortfalls resulting from high cost and/or computationally-bound computing systems. These listed advancements recognize the progress being made for commercial applications, and are geared to capitalize on that progress as well as the projected progress for commercial applications. The needed investment areas are: (1) real-time operating systems for high performance computing, (2) common building blocks architectures for military applications that are portable between underlying components, (3) tera-FLOP scalable embedded systems, (4) computing networks that keep up with processor performance, (5) caching models, hardware-accelerated communications and multithreaded systems to overcome latency problems, (6) architectures and commercial leveraging and packaging to overcome the still prohibitive costs per MFLOP for embedded and non-embedded systems, (7) tools, strategies, and architectures for reducing the huge gap between sustained and peak performance, (8) tools to overcome the verification gap for complex computing hardware (e.g., the Pentium bug), (9) scalable design algorithms to overcome design complexity, (10) computational prototyping and low-cost evaluation to overcome the high cost of physical prototyping of complex hardware, and (11) software engineering tools to assist in making most efficient use of parallel and massively parallel computation architectures.
The System Design and Evolution aspect needs technology advances in a number of areas to meet its goals for rapid development and delivery of low-cost, high quality software-intensive systems. There is a broad reliance on tools and techniques for improving the processes associated with the software-intensive system from conception through system retirement. These include: (1) software reengineering and reuse technologies including rationale recapture and understanding, domain analysis, respecification, architecture transformation; (2) fault-tolerance methods for critical system applications; (3) system and software engineering frameworks and components including requirements analysis, real-time analysis, common design records (with tailored views for the various stakeholders in system development), constraint management, truth-maintenance, group collaboration, analysis and mining of design information, multi-media, simulation/modeling, and interoperability; (4) high-assurance techniques to support software for distributed, real-time, and heterogeneous computing architectures; (5) integrated automation tools, in some cases embodying knowledge-based technology, to assist system builders in exploring the total system design space and in synthesizing alternative system configurations including hardware, software and human processes; (6) evaluation and assessment technologies that enable uniform and consistent measurements of critical attributes throughout the system development process and support revalidation of outcomes at each level of system abstraction and step of refinement; and (7) production technology for building complex computing systems including automated capabilities to precisely specify software architectures, to analyze architectural influences on systems performance, and to facilitate composition of software systems through software module reuse.
3.5.3.3 Basic Research. In support of the above technology development program, significant investments in basic research should continue to overcome gaps in both theory and the scale of the existing theories for computing and software technologies including:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|