XOM Project Overview
The ability to perform many-to-many multicast over an open network is very important to the RT-DVS
community and key to implementing the Extensible Modeling & Simulation Framework (XMSF). This framework,
recently recognized by the Simulation Interoperability Standards Organization (SISO), has the objective
of using XML and Web based technologies for expanding the user base of RT-DVS. Implementing end system
or overlay multicast for real-time distributed simulations allows the continued use of open protocols as
implemented across the Internet. RT-DVS is then no longer dependent on consistency of network policy
implementation across an open network such as the Internet and supports the RT-DVS communityís effort
to move to Web based technologies.
The XOM project's objectives are first to develop a proposed architecture for an overlay multicast protocol
and then to validate its feasibility in a working prototype that can be made available to interested users for
further experimentation. This project explores the concept of end system multicast with QoS and its suitability
for RT-DVS. The research project will develop the basic requirements of an end system based multicast protocol
to support RT-DVS. After a review of these basic requirements, a high-level architecture and an overlay protocol
will be proposed for use in RT-DVS. A prototype of the proposed protocol will also be demonstrated and used to
validate functionality and performance.
The project was originally conceived following the development of the first requirements
for XMSF, which specified network requirements that can best be provided using multicast networking. The work
group that helped develop the original XMSF requirement also recognized that many open issues with IP multicast
over the Internet are likely to continue to be insurmountable and would prevent deployment in a manner that will
meet the Quality of Service (QoS) and the many-to-many multicast needs of RT-DVS.
Distributed virtual simulations operating across a network in human time generate large amounts
of message traffic among the computers hosting the simulation applications. This requires many-to-many
communications in a dynamic group environment where N computers in the group scales as O(N) message transmissions
from each member or O(N2) total message transmissions in the group.
(See publication). In addition, this
simulation environment may not necessarily be homogenous, e.g. each simulator
is likely to be different but they dynamically share common simulation objects over time. The result is that
simulation objects may have membership in multiple groups with each groupís membership changing at different rates.
Distributed virtual simulations also require specific delay bounds to support the delivery of real-time,
interactive visual and audio information at human response times. This environment can be described as a
multiparty collaborative environment running multimedia applications. The underlying networking environment
needs to support a large number of participants dynamically joining and leaving the applications across the
myriad of public and private networks that make up the Internet. Because each of these networks is
independently managed, the RT-DVS applications cannot solely rely on the Internet to deliver the necessary
QoS even where QoS mechanisms are deployed. As a result, networking real-time simulators together generally
has had deployment only in specialized local area networks or on private networks dedicated to the simulation
- Characterize modeling and simulation network traffic load using
- Using the traffic characteristics and XMSF documented requirements to develop high level architecture for
the XOM protocol.
- Demonstrate key concepts of the architecture in a XOM prototype using live modeling and simulation network traffic.