| |
 |
|
Visualization of a directory tree, created with Walrus software
tool (see inside back cover for details). Image courtesy of CAIDA.
|
Representative FY 2003
agency activities
NSF: Research in middleware to optimize the performance
of networked applications; high-performance
connections for colleges and universities; strategic Internet
technologies such as network monitoring, problem detection and
resolution, automated advanced tools for active and intelligent
networks, collaborative applications, and innovative access methods
NIH: Demonstrate application of scalable, network-aware,
wireless, geographic information system (GIS), and security technologies
for networked health-related environments
DARPA: Scalable network modeling and simulation
tools capable of predicting behavior at scales ranging from milliseconds
to hours on networks of hundreds of nodes;demonstrate hybrid optical/RF
self-healing networks
NASA: Implement a high-speed testbed network to
develop and demonstrate advanced computing, networking, and collaborative
technologies; integrate network services (QoS, passive monitoring,
resource reservation) for grid environments; demonstrate hybrid
satellite/mobile wireless/ad hoc
network applications and Office of the Future work environment
DOE Office of Science: Research in high-performance
transport protocols enabling reliable TCP
delivery of terabits/sec throughput to distributed high-end science
applications; development of end-to-
end performance monitoring, network diagnosis, and scalable cybersecurity
services for large-scale
scientific collaborations
NSA: Research on advanced network topologies and
protocols, network convergence, all-optical
networking, and network management, including burst switch technology,
provisioning, message passing, low-power wireless nets, firewalls
in high-speed systems, and security and interoperability issues
NIST: Standards for networked communication of
pervasive computing devices, including models of industry protocols,
validation of emerging specifications, evaluation of adaptive control
mechanisms;
metrics and protocols for ad hoc wireless networks; approaches to
agile switching infrastructure;
protocols and standards for Internet infrastructure security
NOAA: Early adoption of scalable network capabilities
and applications in support of severe
weather forecasting and warning, and hazardous materials response
ODDR&E: University-based research in real-time
fault-tolerant network protocols
|
Three decades ago, a handful of Federally
funded researchers invented a way to send messages from one computer
to another over telephone lines. Their world-changing
breakthrough, elaborated in subsequent Federal and private sector
R&D, has evolved into the Internet - the basic, if awesomely
diverse and powerful, infrastructure for human enterprise in the
new century.
Federal research in large-scale and broadband networking continues
to support U.S. leadership in advanced communications by developing
and prototyping next-generation technologies to dramatically increase
the speed, reliability, security, and versatility of networks. In
the wake of 9/11, the NITRD agencies' ongoing research emphasis
on network reliability, security, and privacy has also become a
shared national concern.
Although the Internet still has the aura of novelty about it, the
reality is that the Net is the product of older technologies that
limit the speed and size of data transfers, open networked devices
to cyber attacks, and are not scalable enough to extend reliable
connectivity to fast-growing numbers of wireless, mobile, and embedded
devices. NITRD research focuses on achieving the fundamental technical
advances needed to make end-to-end high speeds, reliability, security,
and flexible access the standard features of the Nation's digital
communications systems.
One key research area is optical technologies, which offer exponentially
higher bandwidths than today's Internet and thus make possible advanced
applications and future network expansion. The NITRD agencies have
demonstrated the world's first end-to-end optical network that is
thousands
of times faster than the Internet and are working to further develop
optical systems, which also offer far greater network security and
reliability than are currently possible. For example, the NSF-supported
STAR TAP, a Chicago-based cross-connect of U.S. and international
high-speed networks, is evolving into StarLight, an experimental
optical testbed and infrastructure for network services optimized
for bandwidth-intensive applications. StarLight will support experimental
networks at 1, 2.5, and 10 gigabits per second and research networks
at speeds up to 10 gigabits. The connections will allow production
(ordinary uses) networks to operate at 1 gigabit levels.
NITRD research at the nexus of high-speed
optical and wireless technologies provides the basis for the significant
networking advances the Nation needs to keep pace with accelerating
demands, including:
- High levels of network trust (the system
is highly reliable and the information it carries is secure and
private)
- Anytime, anywhere network connectivity
- End-to-end high bandwidths for high-performance applications
- Grids to connect computing systems, storage, and instrumentation
- Collaboratory security and quality of service (QoS), meaning uninterrupted,
uniform high network speeds with low latency
- Sensor nets - billions of networked, embedded sensors
In March 2001, the agencies held a major "Workshop on New
Visions for Large-Scale Networks: Research and Applications"
in Vienna, Virginia. More than 160 participants from government,
academia, and industry analyzed six scenarios for the networking
future (including intelligent warfare, disaster response, and air
transport) and identified the long-range research needed to realize
these visions (http://www.nitrd.gov/iwg/pca/lsn.html).
Their recommendations are helping guide NITRD research planning.
For FY 2003, NITRD focus areas include
technologies and services to enable wireless, optical, mobile, and
hybrid communications; networking software to enable information
to be disseminated to individuals, multicast to select groups, or
broadcast to an entire network; research on scalability and on modeling
and simulation of the Internet; improved end-to-end performance
and performance measurement; software for efficient development
and execution of scalable distributed applications; software components
for distributed applications, such as electronic commerce, digital
libraries, and health care; and infrastructure support and testbeds.
NIH's National Library of Medicine, for example, will extend its
leading-edge research in telemedicine with a new program to demonstrate
the application of scalable, network-aware, wireless, geographic
information systems (GIS), and security technologies to networked
health-care environments. Project proposals will focus on applications
for health-care delivery systems, medical decision making, public
health networks, largescale emergencies, health education, and medical
research.
Major Research Challenges
- Trust: security, privacy, and reliability
- Adaptive, dynamic, and smart networking
- Measurement and modeling of network performance
- Scalable technologies for massive increases in heterogeneous
network traffic, including billions of wireless devices and sensors
- Networking applications, including vertical integration and
supporting tools and services such as middleware (see next section
"Middleware MAGIC To Outfit Networks for Grids")
- Revolutionary research: theories of complexity, generalized
control theory, other models to address evolution of network functionality
amid exponential growth in connectivity
Middleware MAGIC To Outfit Networks for Grids
Middleware is vertical integration software
that enables networked resources and multiple applications to work
smoothly together to provide end user services. As the name suggests,
middleware operates in between top-level software that end users
interact with and core networking and operating system software
at the lower end of the software stack. Middleware is a facilitator
and middle manager. It provides transparency among network service
providers, for example, to enable information to flow seamlessly
and securely in a trustworthy framework, and assures software functionality
across heterogeneous computing and storage systems to meet user
requirements, including the ability to develop new applications.
Although state-of-the-art middleware is crucial to improving the
networked performance of most
applications, middleware development frequently confronts two competing
and contradictory
demands: optimizing the entire network for a single application
and sharing limited resources for the common good of all applications.
Middleware R&D has begun to sort out
these seemingly incompatible requirements, most notably through
experimentation with implementations of the Globus middleware suite
(details on page New Technologies
To Explore the Frontier of Complexity). But the area still requires
significant attention from the research community. To close the
gap, a new NSF middleware initiative aims to assemble components
already available and pinpoint those areas that require new scientific
knowledge and insight. The initiative will focus on:
- Applied infrastructure research specifically directed at middleware
services, with the goal of producing working prototypes
- A software distribution based on research prototypes and operational
middleware services
- Deployment of this middleware infrastrucure to experiment with
new distributed applications
In addition to the NITRD agencies’ middleware research,
in January 2002 the Large Scale Networking Coordinating Group added
a new Middleware And Grid Infrastructure Coordination
(MAGIC) team to its research program.The new
team joins two existing teams, the Joint Engineering
Team and the Network Research Team. MAGIC is
chartered to:
- Coordinate interagency middleware and grid efforts
- Enhance and encourage interoperable grid technologies and deployments
- Promote usable, widely deployable middleware tools and services
- Provide a forum for effective international coordination of these
technologies
MAGIC membership includes representatives of Federal agencies with
responsibility for middleware and grid projects and researchers,
implementers, operators, and users of middleware and grid technologies
from academia, the commercial sector, and other institutions. Participants
view the group as a mechanism for exchanging information about major
agency efforts, such as NSF's TeraGrid, DOE's Science Grid, and
NASA's Information Power Grid, and addressing common technical issues
and concerns. At its initial meeting, the MAGIC team set its sights
on completing the following three near-term tasks:
- Document the significant number of domestic and international
grid projects
- Develop a Summer 2002 workshop focusing on middleware and grids
to identify research, development, implementation, and maintenance needs to provide guidance
to Federal research and funding agencies
- Increase participation in MAGIC by application developers, the
commercial sector, and middleware and grid users
|
