ISI researchers design, model, and implement systems. Their goals include:

• New core technologies, along with supporting architectures, toolkits, and testbeds for deployment in real-world applications, leading to fully integrated practical prototypes that can be employed as is, or adapted for national defense, commercial, or academic use
• Intensive design development of systems and technologies to improve robustness, scalability, and versatility
• Integration of heterogeneous systems and technologies for synergistic performance

Complementing ISI's systems approach is a culture of collaborative research. Over the years, ISI has built extensive ties with research and development groups around the world through joint projects, provision of core services to the research community, and coordination and leadership of national and regional coalitions. From this unique vantage point, ISI frequently serves as an important liaison between federal agencies, industry, academic institutions, and independent research centers. Appropriately, ISI has received strong funding support throughout its history from the Defense Department Advanced Research Projects Agency. Other government funders include NSF, NSA, DOE, NASA, and NIH. Additional support is provided by such companies as Intel, HP, IBM, Microsoft, Cisco Systems, Sun Microsystems, Nortel and Sprint.

Themes include:

• Autonomous Agents
• Geographical Information Systems
• Digital Government
• Networking
• Grid Technology
• Privacy and Security
• Natural Language
• Knowledge Systems
• Sensor Networks
• Robotics
• Soft Hardware & FPGA
• Compiling and Flexibility
• Chips, PIM & Embedded
   

Additional ISI Projects
 

Autonomous Agents

USC is an internationally recognized center for research on agents, systems that exhibit autonomy, purposefulness, behavioral flexibility, and even social capability. Researchers work on the construction of individual agents, including "virtual humans," on web-based information agents, and on the coordination of multiple agent systems.

ISI researchers played a key role in the development of Soar, the leading cognitive agent architecture in use today. Soar applications modeled behavior among teams of autonomous helicopter pilots under fire in battlefield simulations, and among agents modeling soccer players in virtual matches. Researchers are also developing softbot agents and intelligent proxies with variable levels of autonomy and multiagent team coordination for applications. In one project, Electric Elves, a committee of agents has, over a period of many months, negotiated with each other to plan and schedule the entire suite of daily worktime activities of the (human) members of the research group that is developing them. The agents use GPS systems to track group members and PDAs and desktop computers to communicate with them. In addition to applications, ISI researchers are deepening understanding of the theory of agent group behavior.

Another standout project is CAMERA, developed for the U.S. Marine Corps, which schedules operations for squadrons of Harrier attack jets. Top USMC air commanders have positively reviewed the system, which is now deployed on carriers and under consideration for wider adoption by Marine aviation.

ISI's Center for Advanced Research in Technology for Education (CARTE) is a world center in the creation of interactive technologies for education. It is developing pedagogical agents--computer-animated teachers and guides that interact with students in real time. Such teacher agents--early versions were named "Steve" and "Adele"--pose questions to students, correct and comment on the students' responses, and can understand and respond to students' follow-up queries. Pedagogical agents have been deployed in classes at USC and at other universities, and are being tested in elementary school classrooms. Recent CARTE efforts focus on giving such animated robotic agents "social intelligence" to engage in natural tutorial interactions with learners, and monitor and influence their confidence and motivation. As a foundation for this effort, CARTE also conducts basic research in speech and gesture synthesis. In addition to teaching academic or vocational courses, CARTE is exploring the use of agents in such delicate situations as giving parents of children with cancer background information.

Guidebots of the type developed by CARTE researchers are also in use at the USC Institute for Creative Technology to develop effective training material for the U.S. Army notably the Mission Rehearsal Exercise (MRE).
 

Digital Government

The Digital Government Research Center (DGRC), a joint effort of ISI and Columbia University funded by the National Science Foundation, applies advanced information technologies to help government become more effective and responsive to citizens, while strengthening security and protecting privacy. The DRGC brings together researchers from the fields of computer science, engineering, and social sciences with experts from local, state, and federal government agencies.

Some DGRC projects involving ISI and other USC researchers are:

Information Integration: The Energy Data Collection project integrates statistical data about energy products from databases and web sources from diverse government agencies and makes the integrated data accessible through a variety of modes designed to assist users at all levels of comfort with computers.

Transportation Research: The Argos project provides a flexible data query and analysis system based on the web services paradigm. As an application domain we examine several goods movement planning problems and their effects on spatial urban structure. In particular, we are modeling the flow of freight in the Los Angeles metropolitan area by integrating statistical and geospatial data.

Security: The FedStats Secure Collaborative Environment provides coordinated teleconferencing and remote data access and manipulation, allowing experts in different locations to discuss and analyze data in real time through secure channels.

Social Sciences: The Neighborhood Participation Project studies ways information technology can facilitate meaningful and constructive participation by neighborhood groups in city governance. The DGRC hosts the National Conferences on Digital Government Research, operates the NSF's Digital Government Consortium web site www.digitalgovernment.org, and publishes the monthly dg.Online e-newsletter.
 

Geographical Information Systems

GeoWorlds is a sophisticated information management system that generates highly detailed, continuously updated dossiers about developing situations. The system integrates digital libraries, geographic data systems, and other sources including satellite and/or overflight surveillance. In experimental use by the U.S. military's Pacific Command (PACOM), GeoWorlds assembles custom repositories of geographic information about a region, bi-directionally links this data to document-based information from the World Wide Web and a variety of data bases, and monitors real time sensors and telecommunications for developments that might change conclusions or decisions. One application is GeoTopics, scanning news sources around the world for headlines. The system facilitates collaboration by decision makers and experts. It has been tested in emergency management, intelligence analysis, economic and business information gathering, and scientific research applications.
 

Networking

One of the birthplaces of the Internet, ISI continues to research new ways to link computational resources--and safeguard --> --existing ones.

ISI researchers developed the Domain Name System, as well as many of the basic protocols underlying email, telnet, and ftp. ISI researchers also developed and licensed many of the core multicast technologies that support teleconferencing and distance education.

ISI has developed key elements of the Next Generation Internet, such as ultrahigh capacity local networks, and ISI scientists are among the leaders in the development of Internet Protocol version 6. Other key research interests include the X-Bone system for automating deployment and management of overlay networks, routing policy and management systems, mapping and fault detection for large networks, and active networking for dynamic network control and signaling. Applying architectural design principles learned in the development of the Internet, researchers are exploring ways to scale networks that are pervasive, unattended, and widely embedded throughout the physical environment. ISI also manages or participates in several national and regional operational networks such as Los Nettos, CAIRN, TRAIL, CalRen-2, and the Active Nets Backbone as collaborative research test beds. It coordinates routing, interconnects, and multiple test bed experiments for the Next Generation Internet program's SuperNet, a coast-to-coast network providing individual researchers with multigigabit-per-second bandwidth.

Applications for this bandwidth include the "Digital Amphitheater," which will allow new dimensions in meetings, education and even artistic performance. ISI also develops simulation, design, and testing tools such as the Virtual Internet Testbed and the Multicast Internet Testbed for projects that go beyond the scale and testing capabilities of today's networks. ISI and partners have demonstrated real-time transmission of full bandwidth high definition TV over the Internet, along with a new way of transmitting high definition TV using native Internet protocols, and also GRIP, a technique to enable PCs to communicate securely at gigabit/sec speeds.

The Postel Center for Experimental Networking (PCEN) was established at ISI in 1999 to commemorate Jon Postel's contributions to the development of the Internet. PCEN provides facilities, funding, and support for distinguished visiting scholars and graduate fellows to perform applied research and develop tools of general utility in the spirit of its namesake's work. Its endowment has been generously funded by Cisco Systems, Centergate, and Sun Microsystems, as well as by a number of private individuals.

Work at PCEN is driven by the interests of its visiting scholars and graduate research fellows. Our first scholar made progress developing a proximity networks architecture, in which a user's physical location is integrated with automated configuration of nearby devices to provide a coordinated service environment. Our first graduate research fellow is currently applying proximity networking principles to the development of alternate wide-scale Internet architectures.
 

Grid Technology

ISI, through its Center for Grid Technologies, collaborates with Argonne National Laboratory and the University of Chicago on the Globus Project, which distributes the Globus Toolkit, open-source middleware that makes it possible for geographically distributed researchers to share a wide variety of resources, including computational power, storage systems, data collections, and unique scientific instruments.

In 2001 the Grid approach pioneered at ISI and Argonne/Chicago became the de-facto standard for Grid Computing. The National Science Foundation began a major project, the National Middleware Initiative, to build on Globus beginnings and expand Grid capabilities, that is now using a 320-node Linux cluster at USC as a test bed. Late in 2001 in a simultaneous announcement, IBM, Microsoft and ten other major industrial companies adopted Globus tools. R&D magazine named the Toolkit as one of its R&D100 innovations for 2002.
 

Privacy and Security

ISI's Center for Computer Systems Security is a significant contributor to research in privacy and computer security. The center was established to bring together academic researchers from USC and other schools, along with experts from government and industry, in order to develop computer security and privacy technologies, and promote their use throughout the Internet. The center also offers an opportunity for students to work on such systems and is developing curriculum options in security and privacy.

Center researchers include the principal designer of the Kerberos authentication system, who with colleagues created and standardized, through the Internet Engineering Task Force, Kerberos extensions for public key cryptography. These extensions are now part of a growing number of commercial products including MS Windows 2000 and Windows XP.

ISI continues to pioneer technologies and toolkits for group security, intrusion detection, information survivability, detection of forged or altered electronic documents, and managing security policy. ISI researchers are also working proactively to protect Web sites from the threat posed by denial of service attacks.
 

Natural Language

Natural language processing research at ISI combines new statistical techniques and traditional symbolic methods to produce practical tools for translation, summarization, and text and speech generation.

The natural language processing group is one of the largest in the world. ISI researchers have developed ONTOSAURUS, a large-scale ontology (concept thesaurus) for machine interpretation of texts, and other ontological tools. Recent projects include a Japanese-English translator, statistical methods to decipher ancient texts, and a toolkit for rapid construction of a translator between any two languages from a set of bilingual texts. Current work also is developing advanced techniques to extract comprehensive linguistic knowledge automatically from billion-word text databases in diverse languages.
 

Knowledge Systems

ISI research combines Web- and user-oriented knowledge acquisition and information gathering techniques with knowledge representation and reasoning architectures. ISI researchers are developing editors and annotation tools such as WebScripter and TRELLIS to make Web content more accessible to software agents and search engines. Another effort, Ariadne/SIMS, facilitates use of agents to integrate and reason about information from heterogeneous collections of databases and on-line resources.

The Expect Project helps users with no formal training in knowledge representation, such as biologists and military planners, to structure data into knowledge bases. Novel collaborative dialogue and learning techniques help broaden the range of knowledge that can be so structured.

Loom, a knowledge representation and reasoning architecture that has been licensed to hundreds of users, was pioneered at ISI and continues to evolve as PowerLoom, with a wide range of applications from expert systems design to data mining. Creation of large ontologies and knowledge bases from distributed and semi-structured sources is an ongoing area of research.

Sensor Networks

A new way to continuously monitor the environment has been taking shape at ISI (DSN, I-LENSE), one that uses walnut-sized modular sensor devices that link themselves wirelessly together to form autonomous networks. Ultimately, these will draw enough energy from their environment to remain active indefinitely, continuously reporting on their surroundings. ISI leads in advanced multihop networking techniques for organizing and managing such systems (PADS, SCADDS), and at application techniques to minimize power consumption. Using new techniques of adaptive computingsuch a network can be reconfigured while deployed to serve as a intrusion alarm or as an imaging system that tracks moving objects. It can also serve as a longterm environmental listening post reporting on temperature, the presence or absence of chemicals, or a wide variety of other parameters in remote locations, such as the ocean or deep space. Sensor network technology was successfully tested in the California desert in 2001. The demonstration featured innovative ISI technological contributions such as power-aware routing, directed diffusion, adaptive fidelity algorithms, data aggregation, and localization of sensors, while minimizing use of energy in communication through GPS synchronization, and power-aware sensor management. A follow-on effort is designing low-power media-access control protocols for longterm deployment in the James Reserve Wildlife Refuge in the San Jacinto Mountains in a joint project with UCLA/CENS. New initiative, STRETCH, is deploying these in fabric matrices.
 

Robotics

Award-winning robotics researchers at ISI develop modular hardware systems and teams of independently-controlled autonomous agents that work in synergy. The ISI's Polymorphic Robotics Laboratory is one of the national-leading places for self-reconfigurable robots and collaborative robot teams. CONRO (Configurable Robots) is a prototype miniature reconfigurable modular robotic system built at ISI. Its potential applications include reconnaissance and search-and-rescue tasks in urban, seashore and other field environments. Each robotic module is programmable to function independently, but communicate with the others attached to it for joint tasks without a central controller. A group of such mini-robots might adopt one shape to crawl through rubble for search and rescue or reconnaissance, and then change shape when they arrived at their destination to better aid victims or observe activity. Software "hormones"--modular programs that function like the chemical messengers in biological systems--are a key element in making such --> --systems work. A single standard hormone message will produce different behavior, according to a given module's location in an assembly.
 

Soft Hardware

Researchers at ISI are exploring applications for Field Programmable Gate Arrays (FPGAs), a new tool that combines the speed of dedicated, optimized hardware with the flexibility of software. The SLAAC (Systems Level Applications of Adaptive Computing) program has developed a series of new computing platforms based on FPGAs that are proving successful in a variety of uses where development of new hardware chips would be too expensive, too slow, or both.

SLAAC applications include high-speed secure network communications and image processing of many different varieties, including infrared, sonar, radar, and high-bandwidth sensor information. In one military application, ISI has developed a SLAAC-based unit that can realize a 95 percent reduction in size when replacing last-generation circuit boards in the Navy's Aegis defense system. Other ISI researchers are working with developers of new chips to integrate them quickly and effectively into new platform architectures.
 

Compiling and Flexibility

New compilation tools can be a major resource to enable the faster creation of more flexible platforms. The DEFACTO (Design Environment for Adaptive Computing Technology) program has successfully automated design space exploration of FPGA designs expressed in high-level programming languages such as C through collaboration between parallelizing compiler technology and high-level hardware synthesis. The compiler engages synthesis in an iterative process, where an implementation automatically derived by the compiler is partially synthesized to provide feedback to the compiler's optimization search algorithm. The compiler research group at ISI is currently applying this basic approach, called "empirically-based optimization," [ps file] to other domains, including embedded and high- performance computing.
 

Chips, PIM & Embedded

For years, a large group of ISI researchers representing numerous disciplines of information technology has been working on a longstanding goal of chip design: a unit in which the processing function is integrated with memory to avoid the performance bottleneck caused by lagging memory access speed. A new "processor-in-memory" (PIM) chip created by the DIVA (DataIntensiVe Architecture) project has been successfully prototyped and is now undergoing testing. Two other projects are now leveraging the DIVA research results to explore other aspects of PIM technology. Godiva, a collaboration with Hewlett-Packard and Rice University, is exploring how PIM technology can be used to develop high-productivity computing systems for the 2010 timeframe. MONARCH, a collaboration with Raytheon and Mercury, is aiming to develop a chip that can change its architecture to perform different functions more efficiently.

ISI scientists are working to expand the usefulness of embedded computer architectures. Simple chips are now part of numerous devices, and the methods for tailoring them to their tasks and integrating them into their surroundings have become more and more sophisticated. ISI researchers believe that similar techniques can be used to create more powerful multiprocessor supercomputers with reduced power and space requirements.

Additional Current Projects at ISI

• (ACC) Active Congestion Control - The Active Congestion Control project is applying Active Networking techniques to feedback congestion control. Feedback congestion control is a very effective system for sharing network bandwidth when the bandwidth delay product of the network is low, but loses its effectiveness in high bandwidth-delay networks. Using Active Networking techniques ACC seeks to increase the range over which feedback is effective. ACC is a task under the ARP project.
   
• (ACTF on HPC) - The Computational Sciences Division is working to expand their customers' capabilities to do large-scale simulations on distributed Linux Clusters and is based on the ISI team's two-decade experience in parallelizing this type of simulation program in a way that makes it better able to scale, (i.e., more effectively use additional processors). ISI is supporting the migration of the Army Constructive Training Federation into High Performance Computing to enhance the ACTF's capabilities to simulate urban environments and civilian inhabitants.
   
• (ADGEN) Advanced Generation for Question Answering - In this project, we are studying and formalizing the properties of texts that make them coherent (or not). We make use of extensive statistical analysis over a vast corpus of coherent documents. This work is applied to machine document creation, as in multidocument summarization and question answering.
   
• AMP - The AMP project, a collaboration with Lockheed Martin, BAE Systems, and the University of Maryland, is developing programming models, libraries, system prototypes, and application demonstrations for polymorphous computer architectures. Please see http://www.east.isi.edu/AMP/index.htm for more on AMP.
   
• AMPS - The AMPS project is developing power-aware multiprocessing technology for space. USC/ISI is working with BAE Systems to show how technology developed in the PAMA project can be applied to space applications and rad-hard components.
   
• AnimalWatch: An intelligent tutoring system for Grade 6 mathematics - The transition from arithmetic to algebra can be difficult for many students, and is associated with a decline in mathematics motivation. The AnimalWatch project provides individualized web-based tutoring in early algebra concepts by integrating mathematics instruction with environmental science materials. Students solve mathematics problems about endangered species, receive multimedia help as needed, and take part in ôvirtual adventuresö led by animated characters based on real scientists who use mathematics in their research. Evaluation studies will be conducted in Grade 6 classrooms in the Pasadena Unified School District.
   
• (AQ2-Time) Temporal Awareness Algorythms for Natural Language Text - This project is a collaboration with Brandeis University and Georgetown University to improve our ability to extract temporal information from text and then to reason about it. The primary foci of ISI's part of the project are to determine the temporal information implicit in our knowledge of typical durations of events and the temporal information implicit in the rhetorical structure of the text. For example, in the text
   
  A bomb exploded on a bus yesterday. Police are investigating.
   
  we know that the duration of a bomb explosion is less than the duration of a bus ride, and that the investigation happened after the explosion. The principal product of this project will be an annotation scheme for text that makes this information explicit.
   
• (AQ2-Wordnet) Extending WordNet for Question-Answering - WordNet is an widely used, extensive on-line dictionary of English, developed over the years in an effort led by George Miller of Princeton University. For every word there is a corresponding list of word senses and for each word sense there are lists of synonyms, hypernyms, and antonyms, and there is a "gloss" or definition, which is in the form of an English sentence. The aim of the current project, a joint project between Princeton and ISI, is to extend this resource in several ways that will make it more useful for question-answering.
   
  The focus of Princeton's effort is to augment the glosses by indicating for all the lexically ambiguous words which word sense is intended. The focus of ISI's effort is to translate the English glosses into axioms in formal logic. The effect of this will be to create a huge linguistically-oriented knowledge base that can be used by question-answering systems that employ a theorem prover. Previous research suggests that this can very significantly enhance the performance of question-answering systems.
   
• Ariadne - Ariadne is a system for extracting and integrating data from semi-structured web sources. Ariadne enables users to rapidly create information agents for the Web. The system provides tools for dynamically extracting data from Web sources, resolving inconsistencies in naming across different sources, and constructing efficient plans for integrating the data. Please see http://www.isi.edu/info-agents/ariadne for more on Ariadne.
   
• (ARP) Active Reservation Protocol - The ARP (Active Reservation Protocol) project is developing a Java-based active network Execution Environment (EE) within which network control and signaling code can be dynamically deployed and executed. The result will be portable control and signaling software, dynamic installation of new versions, and customization of features for different user groups. Please see http://www.isi.edu/div7/arp/ for more on ARP.
   
• (ATTEND) Automated Tools To Evaluate Negotiation Difficulty - ATTEND seeks to develop a theoretical understanding of systems in which negotiation between multiple agents is needed to determine how a set of resources will be utilized. The goal is to be able to partition negotiation problems into sub-problems that make the negotiations simpler and, ultimately, to be able to provide warnings about the difficulty of each sub-problem. ATTEND seeks not just to do so on a theoretical basis, but to interface efficient implementations of its analytical tools to existing negotiation systems, and to demonstrate that the information supplied by ATTEND enables those systems to change their behavior in order to solve problems more effectively.
   
• (BAE) Awareness and Management of Power for Space - BAE-AMPS will develop a power-aware multiprocessor based on the RAD750, which is a radiation-hardened PowerPC microprocessor for space platforms. The project will investigate power management techniques for multiprocessors, including voltage and frequency scaling and node state management, for satellite-based DoD multiprocessor applications. The project will investigate power management techniques in hardware, run-time software, and design synthesis software.
   
• (BCBM) Believable Communicative Behavior Middleware - The BCBM (Believable Communicative Behavior Middleware) project is developing a model of nonverbal communication in animated agents, based upon research in psychology, sociology, and linguistics. This project is particularly interested in modeling how contextual factors can influence the manner in which nonverbal behavior expresses communicative intent. This will be realized in an extension to the Unreal Tournament game for controlling the behavior of animated characters based on their communicative intent and relevant contextual factors.
   
• CACE-UI-IPT - This effort coordinates multiple projects at ISI and Vanderbilt University to provide comprehensive decision support tools for military planning. See CARTE.
   
• (CAMERA) Coordination and Management Environments for Responsive Agents - CAMERA is an aggressive, systems-oriented attack on the complex problem of ensuring effective, purposeful action in an environment where behavior is dynamically determined and control relies upon cooperation between autonomous agents. Our vision is an agent environment in which collections of agents form and organize to accomplish focused, purposeful results. In such an environment, it is essential to have configurable negotiation mechanisms between agents, complemented by mechanisms enabling the agent collection to detect problems and negotiate self-correcting behaviors. The CAMERA approach focuses on developing capabilities to: (1) proactively adapt to problems and opportunities; (2) systematically reevaluate priorities in the face of conflicts or time constraints; and (3) robustly handle communication breakdowns inherent in a distributed environment.
   
• (CARTE) Comprehensive Analytic Real-Time Execution - CARTE produces a comprehensive decision support toolset for mission planning and control, allowing extremely rapid evaluation of alternatives and total control of the situation. The technical starting points for CARTE are negotiation technology (a new approach to resource allocation) and context-aware information management technology. Negotiation technology performs scheduling responsive to the time available for doing that scheduling and does it in a way that helps balance considerations, explore tradeoffs, and re-evaluate priorities. Context-aware information management proactively selects, filters, and routes critical information to both human and automated participants in the planning process. This concept has been tested in the ONR-managed CACE ACTD on coordinated flight scheduling and operations scheduling for Harrier squadrons. CARTE extends the CACE initial product both horizontally and vertically. Horizontally, it will extend to deployments going from squadrons of a single type/model/series to composite squadrons of multiple types/model/series, eventually covering all of TACAIR. Vertically, the level of support moves up from the squadron level toward the group level. This will dramatically reduce the time to required to handle the dynamics of supporting an Air Tasking Order, and allow us to provide forecasts of squadrons and groups capabilities to support upcoming operations.
   
• Chatterbots: Speech Dialogue - We are building speech dialogue systems to pass the Turing test. These systems operate in immersive, graphical training simulators where realism is critical.
   
• (CHIME-II) Component-based Heterogeneous Information Mosaicking Environment - CHIME addresses a key challenge in today's intelligence community: to provide intelligence analysts with a data visualization tool that supports temporal navigation, multi-media linking, and flexible call-out of multi-dimensional data, including integrity and context data. Today's analysts can process only a portion of the information relevant to their area of interest, and the amount of that information continually increases. They needs tools that help to rapidly extract information from large, disparate data sets, to focus and hone the datsets, and then to develop dynamic presentations that allow viewing of the data from multiple perspectives, along multiple dimensions. CHIME is a collaborative analyst environment that presents a common geo-spatial framework for rapid assembly, organization and correlation, temporal and spatial analysis, and flexible visualization of multi-media intelligence information.
   
• (CMMD)Coordinated Multisource Maintenance on Demand - CMMD explores how to help NASA Space Exploration units balance between multiple sources of demands on available resources. Demands arise from the competing interests of science, exploration, maintenance and safety upon teams of humans and robots which must be self-supporting for extended operations on the Moon and Mars.

  The research has applications to mission scenarios for Crew Exploration Vehicle and other systems including fleets of unmanned and manned vehicles and robotic probes. Research issues include distributed architectures for information exchange, mixed-initiative planning, open-world distributed planning, minimally-disruptive plan and schedule repair, interleaved planning and scheduling processes, and collaborative context-sensitive interfaces to decision support tools.

  CMMD is a collaboration of USC ISI, Vanderbilt University, and NASA Ames and Johnson Space Centers, scheduled to extend over four years. The current award represents the university portion of the first year of effort; the NASA Centers are funded separately.

• CogGrids - This research project will combine Artificial Intelligence and Distributed Computing techniques to create knowledge-rich workflow services that can support the execution of large-scale scientific workflows. The main foundation will be provided by expressive formal representations of the application workflow and of the execution environment. Please see http://www.isi.edu/cognitive-grids for more on Cognitive Grids.
   
• (CRCNS) Assembling Visible Neurons for Simulations: Merging of High-Throughput 3D Microscopies with Advanced Computational Tools - This NIH work, performed in collaboration with SDSC and the University of Utah, will use Globus technologies to enhance the process of very large-field 3D laser-scanning light microscopy and electron tomography of neurological structures. Improving the resolution of light microscopy and electron tomography will create an unprecedented opportunity to probe the effects of neurological structures in the behavior of the nervous system.
   
  Algorithmic resolution enhancements to 3D light microscopy can provide detailed information at the limit of the light microscope of large cellular structures and whole cells but involve the creation of enormous amounts of data, computationally intensive image processing requirements for assembling large volumes, and labor and time intensive data management and manipulation overhead. Similarly, tomography also involves the acquisition and generation of huge amounts of data. Globus technologies will enable efficient data and resource management for these processes.
   
• (CONSER) Collaborative Simulation for Education and Research - CONSER is developing network simulation and visualization tools (ns and nam) to support networking research for protocol development and evaluation, and network education, illustrating concepts about existing and new network protocols.
   
• (CONRO) Configurable Robots - The CONRO Project has a goal of providing the Warfighter with a miniature reconfigurable robot that can be tasked to perform reconnaissance and search-and-identification tasks in urban, seashore and other field environments. CONRO is made from identical miniature modules, each about 2.5 cm long. It can be programmed to alter its topology in order to respond to environmental challenges, such as obstacles. The base topology is simply connected, as in a snake, but the system can reconfigure itself in order to grow a set of legs or other specialized appendages. Each module consists of a CPU, some memory, a battery, and a micro-motor, plus a variety of other sensors and functionality, including vision and wireless connection and docking sensors. Major challenges include packaging, power and cooling, as well as the major issue of programming and program control. The system will be fully distributed, with no central controller, with the top-level structure being JAVA-based agents.
   
• (Constable) Helping users to examine and modify constraints - Constable is a constraint editing tool that have been developed under the Temple program. Constable can be used to view the results of checking a plan with a number of constraints, and can also be used to modify those constraints to suit the particular circumstances under which the plan is being developed. It is a powerful tool to help developing plans and also illustrates our work with constraints, which can be used in a number of related tools and shared between them. Please see http://www.isi.edu/ikcap/constable for more information on Constable.
   
• (COSSACK) Coordinated Suppression of Simultaneous Attacks - Distributed denial-of-service attacks are becoming increasingly sophisticated, both in the geographical scope of their origins, and in the complexity of their software. One class of existing responses is attack signature specific, and requires manual intervention. Another class is focused on traceback based attribution. The Cossack project will investigate a different approach: one that uses coordination between local intrusion detection systems to detect the onset of an attack, and automatically mitigates or suppresses the attack. Please see http://www.isi.edu/cossack/ for more on COSSACK.
   
• Content Representation Canonical Ontology - Large ontologies (networks defining entities, events, relations, etc. in domains) are recognized by many as providing information useful to systems that attempt intelligent behavior. In this project we merge the WordNet taxonomy, built at Princeton University, the Mikrokosmos ontology, built at New Mexico State University, and the Upper Model of the SENSUS ontology, built at ISI, to form an ontology of over 100,000 nodes. We employ several new representational innovations and add some tens of thousands of instances of various kinds. Please see http://blombos.isi.edu:8001/dino.
   
• (CSC) Criticality-Sensitive Coordination - Participants will work on multi-agent systems to develop decision support assistants that enable fielded human units to dynamically adapt their mission plans in response to change. Research problems include distributed coordination over large interconnected mission structures that change dynamically, supporting coordination of large-scale operations where units may have roles in multiple missions, learning to support the units better by automating decision making when data is potentially sparse, responding in (fast enough) "real time" to change, and reasoning about decision-making policies and procedures during coordination.
   
• (DEFACTO) Design Environment for Adaptive Computing Systems - The project has successfully automated design space exploration of FPGA designs expressed in high-level programming languages such as C through collaboration between parallelizing compiler technology and high-level hardware synthesis.
   
• (DEFCN) Dynamic Policy Evaluation for Containing Network - Developing an access control framework and reference implementation supporting access control policies that are sensitive to network threat conditions. They will integrate the mechanisms for evaluation and execution of such policies with intrusion detection and response systems, and through continued development and extension of a Generic Authorization and Access-control Application Programming Interface (GAA-API) they will ensure that the the framework is usable and integrated with a wide range of applications. Please see http://www.isi.edu/gost/gost-group/projects/defcn/ for more on DEFCN.
   
• DEN Lord Foundation - Develop advanced educational technologies in support of USC School of Engineering's Distance Education Network, including web-casting technology, course management, interactive simulations, learning objects, and student assessment. Please see http://den.usc.edu/ for more on DEN.
   
• (DGO 2000) dg.o Workshop and Publicity - To support the National Science Foundation's Digital Government (DG) Program, this funding allows ISI personnel to host annual conferences for DG project researchers, Government employees, commercial enterprises, and interested parties. The dg.o conference series has grown from a 30-person workshop in 1999 to a conference of about 200 people in 2002. Please see http://www.dgrc.org/dgrc/dgo2002.
   
• (DOCKER) Helping People Connect Software - DOCKER is an intelligent interface to enable unsophisticated users to configure and compose components as Web services with semantic markup. Web services is a useful paradigm to encapsulate components including information sites on the Web, complex software, and simple conversion routines. Please see http://www.isi.edu/ikcap/docker/ for more on DOCKER.
   
• (DRAGON) Dynamic Resource Allocation via GMPLS Optical Networks - The DRAGON project is developing dynamic, deterministic, and manageable end-to-end network transport services for high-end e-Science applications. This service capability builds on the success and reach of the IP infrastructure, and complements it with deterministic end-to-end multi-protocol services spanning multiple administrative domains and a variety of conventional campus network technologies. Initial focus is on the dynamic control and provisioning of lambda paths.
   
  DRAGON is developing the necessary software components to address unresolved IP control plane issues as well as a host of practical end system issues in order to provide rapid provisioning of inter-domain services with associated authentication, authorization, accounting, scheduling, and end system instantiation. The project is instantiating a GMPLS capable optical core network in the Washington D.C. metropolitan area.
   
• (DynaBone) Fault Tolerant and Security Adaptive Dynamic Overlay Netwokrs - DynaBone is a system for the rapid configuration, deployment, and management of protective layered overlays that both proactively and reactively resist distributed denial-of-service (DDOS) attacks. DynaBone uses X-Bone's unique ability to layer and compose overlays (VPNs), merging a parallel set of VPNs, each with different properties that resist different types of attacks, as a single virtual VPN with the combined strength of its components. Please see http://www.isi.edu/dynabone.
   
• (DYNAMITE) Dynamic Negotiating Adaptive Multi-Agent Teams - Focuses on new algorithms for distributed constraint optimization and their realization on physical hardware, such as distributed sensors to track mobile targets.
   
• (DSN) Dynamic Sensor Networks - The DSN project funded under DARPA SensIT is investigating human interfaces for sensor networks, sensor network modeling tools, and mission planning and coverage analysis tools.
   
• (EDC) Multilinguality in the Energy Data Collection System - In a collaboration between ISI and Columbia University, we have built EDC: a database query system for information about gasoline that is provided by several Government agencies in over 50,000 data tables in a variety of formats. This project extends the natural language question interface to the EDC system by allowing users to pose queries and receive answers in Spanish and Chinese.
   
• EPA-Air - The EPA's Air Quality Management Office of California every year has to integrate the air quality readings from its 36 regional offices and send them to US-EPA in North Carolina (which in turn has to integrate the data from all 50 states and from neighboring countries). To investigate the problem of cross-database mapping, this project will employ new statistical algorithms developed for Machine Translation to discover correspondences.
   
• (EQUAKE) SCEC Community Modeling Environment - Information Infrastructrue for System Level Earthquake Research - In this large-scale collaboration funded by the NSF we are contributing ontologies and specialized domain and translation models to support semantic interoperability between heterogeneous data sets and simulation codes.
   
• eSpeech - Develop high quality, expressive synthetic speech for interactive training applications.
   
• (eRule) Language Processing Technology for Electronic - Many people today - including news analysts, opinion pollsters, advertisers, and government regulation writers - need to interpret, structure, and rapidly master large quantities of opinion-based text.
   
  We focus on the federal government's several thousand regulation writers, employed in some 200 agencies, who formulate, in a tightly scripted procedure, the rules and regulations that define the details of our laws. In this procedure they invite, and then process in detail, comments from the public on their proposed regulations. They may receive several hundred thousand form letters by email, academic studies and manuscripts of a few thousand pages, and anything in between.
   
  We will develop new text processing tools that can perform advanced analysis of such collections, including text clustering, text searching using information retrieval, opinion identification, stakeholder characterization, and extractive summarization.
   
  Our Rule-Writer's Workbench will be built by Computer Science researchers at ISI and CMU, deployed at our government partners at DOT and EPA, and evaluated by our partners from Social and Political Science departments at the Universities of Pittsburgh and San Francisco. This is a three-year project funded under the National Science Foundation's Digital Government program.
   
• (ETIQUETTE) Evaluating Social Skills in Conversational Avatars - This project will develop a model for assessing the believability of animated characters based on the extent to which they adhere to social and cultural norms of behavior, including etiquette. It will then test this model in an experimental study in which subjects judge the degree of believability of animated character behavior in various contexts.
   
• Evaluation Standards for Machine Translation and Speech Processing - The evaluation of Machine Translation (MT) systems is as old as MT itself, and to date over 50 different evaluation measures have been proposed. This project develops a systematic taxonomization of the measures, with definitions, pointers to their origins, uses, etc. Please see http://www.isi.edu/natural-language/mteval. This work is in collaboration with partners at the University of Geneva, Switzerland.
   
• (EXPECT) A Reflective Architecture for Knowledge Acquisition - The research focus of EXPECT is the development of acquisition interfaces that help users to extend the knowledge of intelligent systems and their knowledge bases. Much of our work has been on planning tasks and applications, including process models. A recent area is grounding knowledge bases in the Semantic Web to enable problem solving in heterogeneous distributed environments, including projects such as DOCKER, IKRAFT, TRELLIS, and Pegasus. Please see http://www.isi.edu/ikcap/expect/ for more on EXPECT.
   
• (FAZD) System Design & Development for an Integrated Risk Analysis Tool - Under funding from the USC DHS CREATE Center, Information Sciences Institute is developing a Risk Analyst Workbench (RAW). This is a software modeling system that can be used by risk analysts to assess terrorism threats and to evaluate strategies for countering these terrorist threats. RAW is currently being designed to support risk analysis in aviation and ports and will be extended during year 2 under CREATE funds to support risk analysis in two new areas related to economic and logistical risks.

  Under FAZD, a companion project to RAW, the ISI team will work with members of the DHS Center on Foreign Animal and Zoonotic Disease (FAZD) at Texas A&M, and risk analysts in CREATE to further extend RAW to provide a risk analysis platform for the risk analysis of intentional introduction of foreign animal and zoonotic diseases. In addition to extending the design of RAW to this knowledge domain, ISI will work with CREATE risk analysts to integrate advances in risk methodology, and improved protocols for conducting risk analysis that integrates the new data and analysis technologies.
   

• (FMESHD) Fault Tolerant Mesh of Trust Applied to DNSSEC - Two fundmamental changes in RFC 2535 have been developed as a result of DNSSEC testbed operations. First, the RFC 2535 mechanism for exchanging KEY data between parent and child must be replaced by the Delegation Signer (DS) solution. FMESHD is currently testing a DS patch to BIND 9 and will be make this patch publically available. Second, the RFC 2535 mechanims for handling application keys (such as SSH and IPSEC keys) must be seperated from essential DNSSEC zone management tasks. Please see http://fmeshd.nge.isi.edu/ for more on FMESHD.
   
• (FNIISC) Fault-Tolerant Networking Through Intrusion Identification and Secure Compartments - FNIISC is a collaboration between UCLA, UC Davis, USC/ISI, and NSCU. The objective is to understand the fundamental principles in developing a scalable and robust fault-tolerant network that can continue to function despite the presence of component failures. Much of the FNIISC work is applied to protecting the BGP routing infrastructure from faults and attacks. Please see http://fniisc.nge.isi.edu/ for more on FNIISC.
   
• (FSCE) FedStats Secure Collaborative Environment - The FSCE project is investigating the effects on firewalls and other middleboxes on teleconferencing and data access protocols. The focus is primarily on the needs of the Digital Government and FedStats communities. Please see http://www.east.isi.edu/projects/FSCE/.
   
• (FACS) First Aid for Computer Systems - The FACS project is designed to provide a level of quick protection for computer systems in the event of an attack. Traditionally, response systems provide either immediate but coarse-grained blocking of the network, or specific but time-consuming analysis of the attack, followed by patch distribution. The former uses little information gained about the attack, while the latter allows considerable damage to mount before the fix is installed.
   
  In contrast, FACS uses what information is quickly available about the *symptoms* of the attack, as opposed to its root cause. Files may be sequestered, specific services may be interrupted, or processes and user accounts suspended. The attack may persist, but we expect that its impact will be reduced and that the system may be able to sustain ordinary operations throughout the duration.
   
• (FREIGHT) Better Freight Flow Data for Analysis and Planning - The Freight Analysis and Planning project models the flow of freight in the Los Angeles metropolitan area by integrating statistical and geospatial data, using the integrated data to estimate freight flow throughout the transportation network and perform what-if analyses.
   
• GALE: AGILE at ISI - DARPA has recently funded three macro-projects in Human Language Technology, in the largest language- and speech-related program ever to date. All three projects are teams led by companies, and each team includes research on speech recognition, machine translation, text summarization, information extraction, and question answering. Several people at ISI have joined BBN's team AGILE.

  Part 1: ISI is leading a project aimed at significant improvements in automated language translation from both text and speech inputs.

  Part 2: ISI is participating with BBN and LCC Inc. to jointly construct a 'Distillation Engine' that performs information extraction, text summarization, and question answering in an integrated way. This engine produces output to user questions from the information that has been input as either speech or text in several languages, and translated.

  Part 3: ISI is collaborating with BBN, Colorado, and UPenn to construct the OntoBank, a corpus of eventually 1 million words that is being constructed by manual insertion of very simple semantic information into Wall Street Journal and other text in English, Chinese, and Arabic. This corpus has the potential to enable a whole new generation of research into shallow semantic language processing.
  

• (GeoWorlds) GeoSpatial Information Management - GeoWorlds extends, integrates, tests, and evaluates a unique combination of Digital Library and Geographical Information System technology. The integration of these technologies is being overseen by USC/ISI in collaboration with multiple sources (including the USC ISI DASHER Project, the University of Arizona, the University of Illinois at Urbana-Champaign, the University of California at Berkeley, and the University of California at Santa Barbara). The goal is to demonstrate a vision of a system that helps a user understand facts and events in relation to space and time by presenting and exploring those relationships in a visual environment that integrates information search and analysis tools with geospatial displays and multimedia documents.
   
• (GRIP) GigaBit Rate IPsec - GRIP is a research project to develop the technologies necessary for secure host-to-host Internet Protocol Security (IPsec) communications at gigabit rates. This project will define an architecture, develop a design, and implement this capability. Please see http://www.east.isi.edu/projects/GRIP/grip_overview.htm for more on GRIP.
   
• (GrilPhyn) Towards Petascale Data Grids - The GriPhyN Project is developing Grid technologies for scientific and engineering projects that must collect and analyze distributed, petabyte-scale datasets. GriPhyN research will enable the development of Petascale Virtual Data Grids (PVDGs) through its Virtual Data Toolkit (VDT) Please see http://www.griphyn.org/index.php for more on GrilPhyn.
   
• Heracles - Heracles is a system for building web-based information assistants. This system provides the infrastructure to rapidly construct new applications that extract information from multiple Web sources and interactively integrating the data using a dynamic, hierarchical constraint network. Please see http://www.isi.edu/info-agents/heracles for more on Hercales.
   
• (iVDGL) International Virtual Data Grid Laboratory - The iVDGL is a global Data Grid that will serve forefront experiments in physics and astronomy. Its computing, storage and networking resources in the U.S., Europe, Asia and South America provide a unique laboratory that will test and validate Grid technologies at international and global scales. Sites in Europe and the U.S. will be linked by a multi-gigabit per second transatlantic link funded by the European DataTAG project. Please see http://www.ivdgl.org/index.php for more on iVDGL.
   
• (JESPP-II) Joint Experimentation on Scalable Parallel Processors, Phase II - JESPP-II is the second phase of a long-term multi-phase effort to support very large scale 'wargame' like experiments, involving millions of entities which represent a rich mixture of current-day and hypothetical capabilities and behaviors. These experiments allow military organizations such as Joint Forces Command to explore and evaluate the interactions of alternative equipment, doctrine, training, personnel, environmental, and other factors in settings with a very high degree of fidelity and realism. The long-term goal is to (a) build the infrastructure for the enormous computational demands of very large scale experiments; (b) develop human-n-the-loop capabilities for initiating, monitoring, controlling, and analyzing the experiments; and, (c) design and demonstrate techniques for programming much richer and more sophisticated simulated entities for participation in the experiments. This current phase will develop new tools to enhance communications between the processors of the SPPs, ensure reliable communication to the operator's consoles, enable better reporting and analysis of both exercise and computer performance, and encourage new research into simulation of human behavior. To that end, personnel from Divisions I, II and VIII are collaborating with subcontractors the Jet Propulsion Laboratory and OTCI, as well as with JFCOM contractors.
   
• (KANAL) Knowledge Analysis on Process Models - KANAL is a tool for checking process models entered by users. By relating different pieces of information in process models among themselves and to the existing KB, it performs a variety of verification and validation checks and propose useful fixes. Please see http://www.isi.edu/ikcap/kanal.
   
• (KOJAK) Scalable Semantic Link Discovery via Integrated Knowledge-Based and Statistical Reasoning - This project supported by DARPA's EELD program is applying a combination of knowledge-based and statistical techniques to detect patterns of interest in large-scale evidence warehouses.
   
• (LATTICED) An Algebra for Intrusion Correlation - LATTICE-D is an effort to formulate a basis for integrating multiple intrusion detection (ID) systems. Because different ID systems use unique methods of detecting intrusions, and the attacks they detect are constantly and often subtly changing, there is a clear benefit to using several systems in parallel. LATTICE-D attempts to provide a sound theory for combining the hypotheses of several ID systems in a meaningful way.
   
• (L-GIST) Learning generalized image schemas for transfer - The L-GIST project uses theories of human cognitive development as a foundation for more robust machine learning systems. In the project, Carole Beal and Paul Cohen will design an agent to acquire tactical expertise in ground warfare through active experience in a rich simulated environment, using representations based on image schemas: representations that capture spatial relations, trajectories, and force dynamics and that some theories suggest form the foundation of human semantic knowledge. Evaluation will focus on the agent's ability to transfer its learning to new battlefield scenarios, and will be conducted in collaboration with military experts.
   
• (MACC) Media Aware Congestion Control - Recently the use of media rich networked applications, such as video conferencing, telepresence, distance learning and voice-over-IP has greatly increased. While these tools enhance our ability to collaborate and communicate, their increased use poses a potential threat to the stability of the Internet, as they typically utilize non-congestion controlled transport protocols in a tradeoff for timely delivery. MACC addresses the lack of a readily available and standardized way of deploying congestion control for interactive networked multimedia applications.. Please see http://www.east.isi.edu/projects/MACC/.
   
• (MACSS) (MAC Protocols Specific for Sensor Networks, started in 2002) - MACSS is investigating Media Access Control (MAC) protocol design for sensor networks: energy efficiency via coordinated sleep, and interactions between the MAC and the physical layer and sensors. We also plan to study how sensor net applications differ from Internet-style applications. MACSS is supported by NSF and Intel Corporation. Please see http://www.isi.edu/ilense/macss/.
   
• MARBLES - Marbles schemes are a family of cooperative and adaptive algorithms for distributed resource allocation problems. Long-term goals for these schemes emphasize fault tolerance and real-time performance in which a good timely solution is preferable to an optimal but too late solution. This project develops base technology for MARBLES algorithms, as well as exploring issues of application to Unmanned Combat Autonomous Vehicles.
   
• (Marine Iraqi) Marine Corps Support for Tactical Iraqi- In collaboration with Tactical Language Training LLC, this project will adapt the Tactical Iraqi language and culture trainer to meet Marine Corps training needs. Changes include giving game characters Marine Corps uniforms, adding translations for Marine Corps ranks, providing training for Marine Corps missions such as "meet and greet" and "cordon and knock".
   
• Maternal Problem Solving Training in Childhood Cancer - This project developed a model for instruction via interactive computer-based stories, called interactive pedagogical drama, and applied it to develop a problem solving skills training program for mothers of children with cancer. This program was tested at cancer centers around the US.
   
• (MERL) Task-Oriented Dialogues between Students and Pedagogical Agents - This project is exploring the relationships between work on intelligent tutoring systems and work on collaborative discourse theory, with the goal of using the latter as a foundation for the former. The project is a collaboration with researchers at the Mitsubishi Electric Research Laboratories.
   
• (MONARCH) A MOrphable Networked microARCHtecture - Monarch, is a research project under the PCA (Polymorphous Computing Architectures) Program at DARPA. MONARCH will define a novel computing model, chip architecture and support environment while creating a software environment that permits affordable, rapid, regular system upgrades for performance growth and parts obsolescense mitigation. The MONARCH architecture will support reactive multi-mission, multi-sensor, and in-flight retargetable missions. Please see http://www.isi.edu/asd/monarch/ for more on MONARCH.
   
• (MOVER) Model-based Object and Video Event Recognition - This project is funded under ARDA's Video Analysis and Content Extraction (VACE) program. Its aim is to develop enabling technologies that - provide more robust object detection and tracking, including camera parameter estimation - provide means for analyzing object motion and deformation, object recognition, multi-model fusion, mensuration, and scene modeling - provide an event ontology, supported by an event representation language to attain and document event understanding.
   
  ISI will work primarily on the third task. Specifically, the event representation language will enable the recognition of and communication about the hierarchical structure of events, so complex events can be viewed as compositions of more simple primitive events. Among the potential applications is the analysis of surveillance videos to determine automatically, for example, whether a sequence of observable events constitutes a holdup, shoplifting, or a legitimate purchase. In general, the focus will be on interpreting observed events in surveillance of indoor and outdoor scenes, and in interpreting the actions of participants in decision-making meetings.
   
• MS ASSESS - Develop a framework for collecting and analyzing student performance data for assessment purposes, build on the Microsoft .NET framework.
   
• (Net-FS) Protected Virtual Networking API Using a File System Interface - NetFS is a file system interface to the networking components of an operating system. Similar to procfs and kernfs file systems, NetFS represents the system's various networking components as a directory of files, providing fine-grained access control, a uniform API, and the opportunity for virtual per-process views.. Please see http://www.isi.edu/netfs.
   
• (NMAA) NGI Multimedia Applications and Architecture - The NMAA project is exploring the limits of IP-based streaming media applications. It focuses on two tasks: scaling to very high quality using HDTV-over-IP, and scaling to very large numbers of participants: the Digital Amphitheatre. Please see http://www.east.isi.edu/projects/NMAA.
   
• ONTOSAUR - A large-scale ontology (concept thesaurus) for machine interpretation of texts.
   
• (OTDR) Operational Test bed for New DNS features in the Internet Root - This project will acquire, configure, and deploy systems in the Internet that support IPv6 native and dual-stack (IPv4 and IPv6) protocols that are configured as DNS root servers. We will work with the existing root server operators and ICANN to ensure that these systems accurately represent the ICANN maintained root zone, to the extent that they are supporting legacy Internet services. Areas where changes to the ICANN maintained data include changes to the "cache" or hints file. We will be using our servers, with their IPv6 addresses. Following this work will be the addition of DNS integrity checks to both the hints file and the projects version of the root zone file. Please see http://www.isi.edu/otdr/ for more on OTDR.
   
• Optimizing Signal and Image Processing in a Dynamic, Data-Driven Application System - This project will develop a dynamic, data-driven application system for signal and image processing under severe resource constraints. We propose a multidisciplinary approach that optimizes from algorithm specification, to mathematical representation, to software and hardware (FPGA) implementation, based on properties of data and unique requirements of the environment and the target hardware device. The proposed approach will perform joint optimization across mathematical, software and hardware (system-on-a-chip FPGA) domains in a dynamic and data driven fashion in that signal-processing transforms are tailored to algorithm requirements and input signals, for reduced distortion and increased compression. System implementations will be based on the best mathematical formulation of the problem coupled with automated selection of the best implementation among a space of alternatives, through the integration of models relating mathematical properties to implementation behavior. Both hardware and software optimization are treated in a unified way. Collaboration with Intel will facilitate industrial adoption of the technology.
   
• PAMA - The PAMA project is developing a power-aware multiprocessor architecture and application demonstrations in collaboration with Los Alamos National Laboratory. The project is investigating hardware architectures, power management libraries, and scheduling techniques. Please see http://www.east.isi.edu/PAMA/index.htm.
   
• (PADS) Power Aware Distributed Systems - The PADS project funded under DARPA PAC/C studied architectures, protocols, and algorithms of unattended ground sensor systems to identify appropriate power management knobs in this domain. This project has transitioned to the PASTA project in Phase 2 of the PAC/C program.
   
• (PASTA) Power Aware Sensing Tracking Analysis - The PASTA project funded under DARPA PAC/C is investigating advanced power management techniques for unattended ground sensor systems including a power aware microsensor architecture, link and network protocols, sensor specific processing devices, power aware signal processing algorithms, and field-level energy management.
   
• PedAgents - PedAgents (Supporting science learning with pedagogical agents). Modeling how students' solve scientific problems is important for understanding how scientific reasoning develops, and for using this understanding to improve all students=92 learning. Previous work in the domain of chemistry indicates that students frequently settle early on a strategic approach and continue to use the same approach on future problems, even if the approach is not effective or efficient. The PedAgents project focuses on the deployment of individualized interventions for students in high school science classes using web-based multimedia problem solving exercises collectively called IMMEX (Interactive Multi-Media Exercises), already in use in urban California schools.
   
• (PCA) Polymorphous Computing Architecture - The DARPA PCA (Polymorphous Computing Architecture) program is developing streaming languages and compilers. These are intended for future, polymorphic computing systems. However, in the near-term, they may also enable Defense applications to exploit modern graphics co-processors (GPUs). DARPA is exploring the possibility of using PCA compiler technology to accelerate the performance of the U.S. Army's OneSAF code by porting computationally intensive functions such as inter-visibility calculations to GPUs.
   
• (PHOSPHORUS) Rapidly Extending and Building Agents to Form, Robust, Adaptive Teams- This work addresses the issue of how to match requests and agent capabilities through ontologies. The hypothesis is that a broad and shallow ontology (such as the HPKB upper ontology or SENSUS) is sufficient for locating agents. Please see http://www.isi.edu/expect/projects/agents/phosphorus.html for more on PHOSPHORUS.
   
• (POWER99) Mathematical Modeling of Multi-Agent Systems - The goal of the project is to create a science of multi-agent systems. We are developing approaches that will allow us to mathematically model and analyze collective behavior of systems of interacting artificial agents, including intelligent software agents and robots.
   
• (POWERLOOM) Tools for Assembling and Managing Scalable Knowledge Bases - PowerLoom is the successor to the Loom knowledge representation system. It provides a language and environment for constructing intelligent applications. PowerLoom uses a fully expressive, logic-based representation language (a variant of KIF), and it uses a natural-deduction-style backward and forward chainer as its inference engine. Please see http://www.isi.edu/isd/LOOM/PowerLoom/ for more on POWERLOOM.
   
• Proteus - We are investigating approaches to turn existing sources into web servcies and then to rapidly compose these web services into new services. The resulting system will support the automatic construction of new web services from existing components. Please see http://dblab.usc.edu/WebServices for more on Proteus.
   
• (PSST- PDA) Problem Solving Skills Training in Pediatric Cancer - This project will create and deploy an animated pedagogical agent on a personal digital assistant (PDA). The system is intended to help mothers of pediatric cancer patients to learn to solve their problems better. The agent will guide the mothers through the process of identifying and classifying their problems, developing possible solutions for them, evaluating the solutions, and seeing how well they worked. The system will be tested in clinical trials at several cancer centers around the United States.
   
• (RABBIT) Iraqi-English speech-to-speech system deployment - The Rabbit project focuses on adapting training and decoding algorithms developed in the context of text-to-text statistical machine translation to speech-to-speech translation environments. Research challenges include dealing with limited amounts of training data, domain-specific jargons and dialects, and speech recognition errors. The research findings specific to this project are incorporated into deployable portable speech-to-speech translation devices built by SRI International and Language Weaver Inc.
   
• (RADHARD) Foundry and Test Coordination/Characterization and Technology Development/Evaluation - The radiation-hardened-by-design (RHBD) effort addresses design techniques for integrated circuits destined for use in satellite system payload electronics. Historically integrated circuits for space electronics have been manufactured at foundries that develop and maintain costly, highly specialized semiconductor processes. These processes are created to ensure that integrated circuits can survive space radiation environments. The RHBD program will develop and validate novel integrated circuit design techniques that can be manufactured on standard commercial fabrication processes. The success of this program will validate a methodology to deliver more advanced capabilities to space quicker and for far less cost.
   
• (RAISE) Rational ATM Internet Suite - The purpose of RAISE is to create a combined hardware/software platform suitable for high-performance internet routing and switching research. A further objective is to include the RAISE platform in DARTnet2 and to operate high link-speed and high fan-out topology in DARTnet2. Please see http://www.isi.edu/div7/raise/ for more on RAISE.
   
• (RapidLang) Rapid Development of Mission-Oriented Communication Skills - The Tactical Language Training Project is developing tools for creating training systems that help people develop a quick understanding of foreign language and cultures. Learners interact with animated agents representing people in a foreign culture, as well as with a virtual tutor agent. This work is part of the DARPA Training Superiority program.
   
• RAP Teams - Develop highly effective, large-scale, heterogeneous, robot-agent-person (RAP) teams.
   
• RATS - The RATS project is studying VCSEL-based free-space optical networks for multiprocessors in collaboration with Applied Photonics. Please see http://www.east.isi.edu/RATS/index.htm.
   
• The Refinery: Theory Refinement for RKF - In this project we are developing methods and tools to support domain experts who are not knowledge engineers in the development of sophisticated knowledge bases. In particular, we have developed a query debugging tool called WhyNot that can pinpoint missing knowledge that caused a particular query to fail (supported by DARPA's Rapid Knowledge Formation program).
   
• Responding to the Unexpected - Exploring new developments in information technology, engineering, and social science that make possible the dynamic construction of highly effective virtual organizations for responding to unexpected events, whether natural or man-made..
   
• (ROLE-ITS) Achieving Motivational and Cognitive Outcomes in Mathematics Using Enhanced Intelligent Tutoring Technology - This Research on Learning and Education project will use inexpensive web cameras to track high school students' attention and motivation in real time while working on math problems at computers in their classroom. The ITS software will make teaching decisions based on a model of expert human tutoring expertise which balances motivational and cognitive goals to optimize learning outcomes. Animated virtual characters will be used to direct the learner's attention to important information on the screen, and to increase student interest and motivation. The software will be deployed and evaluated in Los Angeles public high schools serving diverse student populations.
   
• (ReWrite) A Problistic approach to rewriting for Machine Translation, Lang Generation & Abstracting - A statistical approach to machine translation and summarization. This project is investigating statistical methods for automatic language translation and summarization, exploiting very large collections of on-line linguistic data.
   
• (SAMAN) Simulation Augmented by Measurement and Analysis for Networks - SAMAN looks at rapidly generating representative traffic models for network simulation, applying analytic techniques to speed simulation, and how to make current networks more robust to failure.
   
• (SCADDS) Scalable Coordination Architectures for Deeply Distributed Systems - SCADDS is developing algorithms and software to communicate and control sensor networks. Research includes directed diffusion,adaptive fidelity, localization, and MAC protocols suitable for long-lived, self-configuring networks.
   
• (SCEC) Community Modeling Environment-Information Infrastructure for System Level Earthquake Research - This is a multi-site collaboration to provide information technology infrastructure for earthquake research, including knowledge representation and reasoning, Grid technologies, digital libraries, and interactive knowledge acquisition. Please see http://www.isi.edu/ikcap/scec-it/ for more on SCEC.
   
• (SIM-TBASSCO) Semantic Interoperability Measures - Template-Based Assurance of Semantic interoperability in Software Composition - SIM-TBASSCO addresses a critical problem in the longstanding software-engineering goal of assembling software from components: adaptive composition that is sensitive to quality concerns. Conventional approaches support composition up to a point, but they cannot handle qualitative considerations in composition, such as implementation effort, performance, resource requirements, or reliability. SIM-TBASSCO helps software developers engage in guided, efficient searches and evaluations of the set of alternative system implementations that can be built with the components available to them. It will let developers evaluate components' functional and data equivalence compatibility, find pertinent data conversion mappings, and predict performance (time, space, network) of a component architecture under specific usage situations and hardware/networking environments.
   
• (SI Agents) Social Intelligence in Interfaces for Educational Software - Develop a capability for guidebots to interact with learners in socially intelligent ways, to promote learner motivation and avoid and recover from communication breakdowns.
   
• (SLAAC) System Level Applications of Adaptive Computing - The SLAAC project funded under DARPA ACS has produced three generations of FPGA-based computing system architectures, developed programming models and design tools, and has validated the technology in a number of Defense applications. This project is currently investigating low power optimization techniques for signal processing modules.
   
• (SLATE) Compiler-driven Design Space Exploration for Heterogeneous System-on-a-Chip - This project focuses on system-level mapping of applications written in imperative programming languages such as C to System-On-A-Chip systems implemented using multiple Field-Programmable Gate Arrays (FPGAs) architectures. The proposed research addresses system-level partitioning and scheduling of the execution of tasks among computing cores based on high-level program analysis as well as managing the storage and movement of data between both internal and external memories and between tasks. The proposed research addresses these issues by the synergistic collaboration of program analysis, parallelizing compiler technology and behavioral synthesis tools. Please see http://www.isi.edu/~pedro/projects/SLATE/slate.html.
   
• (SLICK) Skills for Learning to Interactively Capture Knowledge - Developing acquisition interfaces that are proactive learners, able to reason about learning activities and with initiative in participating in the process accordingly. Please see http://www.isi.edu/ikcap/slick for more on SLICK.
   
• SLIIC - The SLIIC project is mapping Navy radar applications developed by Lockheed Martin and developing hardware prototype for data-intensive systems, including PIM and streaming architectures. Please see http://www.east.isi.edu/SLIIC/index.htm.
   
• STEVECO - The STEVECO project is a collaboration between USC's CARTE and Intelligent Systems Technology, Inc. Its purpose is to develop productizable pedagogical agent technology. There is close collaboration with the VET project, which has been conducting research in pedagogical agents, and the ADE project, which is also developing commercial grade agent technology for distance learning.
   
• (TEAMCORE) Rapidly Extending and Building Agents to Form Robust, Adaptive Teams - Focuses on team-oriented programming (TOP), and team coordination algorithms to interpret such TOPs. The team coordination algorithm (in this case STEAM) is packaged in proxies that enable different heterogeneous entities to work together as a team. The key idea is to provide a new level of abstraction in programming teams that takes a step beyond agent-oriented programming (AOP).
   
• (Temple) Template Enhancement through Knowledge Acquisition - Under the Temple project, we are developing tools to guide users in adding planning constraints and preferences, including wizards and smart editors. Please see http://www.isi.edu/ikcap/temple for more on Temple.
   
• TetherNet - A system renting and relocating Internet subnets. Provides true Internet access behind NATs and short-lease dynamic DHCP environments, providing temporary infrastructure for experiments, demos, and remote access (travel, remote offices, etc.). Please see http://www.isi.edu/tethernet.
   
• TerraWorld - The focus of this project is to develop the technology and infrastructure to integrate the huge amount of data available on the web with the widely available geospatial data sources. The challenges in performing this type of integration are first, how do we automatically perform the mapping between the structured data and the geospatial data, second, how do we identify and resolve differences in accuracy of the information being integrated. Please see http://www.isi.edu/info-agents/terraworld for more on TerraWorld.
   
• (TextMap) An Intelligent Question Answering Assistant - In this project, we develop algorithms capable of mining vast amounts of text in order to find short answers to questions posed in natural language. We are primarily focusing on algorithms for answering - factoid questions ("What's the population of Los Angeles?"); - causal questions ("Why do people snore?"); - biography questions ("What do you know about Condoleezza Rice?"); - opinion questions ("What's Rice's opinion about the Middle-East conflict?").
   
• Theseus - Theseus is an execution platform for information agents. The system provides an expressive language for defining information agents and is based on a streaming dataflow architecture that permits a high degree of parallelism and pipelining of data. Please see http://www.isi.edu/info-agents/theseus for more on Theseus.
   
• (TRANSONIC) Enabling Human-Human Communication via Automated Two-way Speech-to-Speech Language Translation - In collaboration with USC/EE and HRL, we are developing a prototype speech translation system (English/Farsi and English/Dari) running on a handheld platform. The focus domains for the system are medical triage, force protection, and refugee processing.
   
• TRELLIS - There are two key research objectives in