James H. Hammond Electronics Engineer Information Technology Group Applied Research Laboratories University of Texas at Austin PO Box 8029 Austin, TX 78713-8029 email: hammond@arlut.utexas.edu voice: 512/835-3761 FAX: 512/835-3100 Michael Myjak Senior Research Scientist Institute for Simulation and Training University of Central Florida 3280 Progress Drive Orlando, Florida 32826 email: mmyjak@ist.ucf.edu voice: 407/658-5043 FAX: 407/658-5059 Brian E. Barnes Research Engineer ERB Building Rm 128 Georgia Tech Research Institute Georgia Institute of Technology Atlanta, GA 30332-0832 email: brian.barnes@gtri.gatech.edu voice: 404/894-1676 FAX: 404/894-9081 *************** Advanced Distributed Simulation: A Requirements Generator for the Next Generation Internet Michael Myjak University of Central Florida mmyjak@ist.ucf.edu Brian Barnes Georgia Institute of Technology brian.barnes@gtri.gatech.edu James Hammond University of Texas at Austin hammond@arlut.utexas.edu Introduction The emergent requirements of Distributed Simulation Systems are an important example of the requirements for the Next Generation Internet. The capabilities of current networks and protocols are not sufficient for potentially demanding simulation paradigms such as the DoD High Level Architecture. In this white paper, the above institutions suggest research to explore requirements posed by advanced distributed simulation systems and potential solutions to be incorporated in the Next Generation Internet. Rationale Distributed simulation is a significant area of research and development in the U.S. and abroad. The current international standard for Distributed Interactive Simulation (DIS) is IEEE Standard 1278. It is utilized worldwide for defense-related interactive simulations, and has been adapted to many other uses such as air traffic control simulations, virtual manufacturing, and entertainment. Recently, the Under Secretary of Defense for Acquisition and Technology has mandated a new High Level Architecture (HLA) for defense applications which will establish a common technical framework to facilitate the interoperability of all types of models and simulations. While IEEE 1278 was designed perform acceptably in a broadcast environment, the HLA has incorporated new concepts of data distribution, faster-than-real-time progress, and dynamic publication and subscription to data. The HLA has placed new requirements upon the network infrastructure which cannot be satisfied using existing technology. Characterization of Distributed Simulation Requirements In general terms, Distributed Simulation systems are heterogeneous systems collaborating to perpetuate synthetic environments. Frequently they are designed to be interactive, incorporating legacy simulators, and analytical models. The integration of these complex systems into a virtual environment depends on the ability of the systems to reliably share data in real-time. Most significant simulations comprise several integrated networks. This introduces complications of bandwidth allocation, and multi-point communications. Finally, many of the simulations must be designed to provide predictable and repeatable simulations. This implies predicable and reliable quality of service, and intolerance to latency irregularity. Given such demanding system requirements, current IP based solutions have proven kludgy and unreliable. A new paradigm is needed. Requirements not Addressed by Current Internet Capabilities In addressing the requirements in the context of the NGI, one must first accept responsibility to work in a heterogeneous network environment. In particular, there is a large installed base of the Ethernet technology, upon which the Internet Protocol itself has flourished, which must not be ignored. Similarly, the next generation of IP, version 6, is now a proposed standard. Although not a currently ready for public release, at least one implementation of IPv6 is known to exist, and the Internet Engineering Task Force has asked for people to begin experimenting with its use on wider scales while more implementations are developed. While acknowledging problems of heterogeneity, the solution set must address several technical challenges. In support of complex distributed simulation systems, the Next Generation Internet should incorporate: (1) Guaranteed Resources Reservation. The technology should support guaranteed quality of service across all connected systems in a flexible and expeditious manner. (2) IP Multi-cast. The new network system should support true multi-cast. (3) Reliable Real-time Multi-cast. The multi-cast solution should allows for reliable real-time delivery. (4) Minimal Latency, Jitter, and Variance. The network performance should be predictable and sufficient to support distributed simulation designs. (5) Heterogeneous Network Support. The solution should allow for incorporation of legacy networks.(Ethernet, FDDI, ATM, etc.) Proposed Research Efforts Experiments need to be done within the existing Internet infrastructure to determine the scalability of the various transport layer protocols needed by HLA. If new network protocols can be derived from existing protocols, they will not only continue to support existing equipment (a must), but can also inherit their most desirable characteristics (best effort delivery, reliable transport, LAN emulation with arbitrary configuration, low latencies across ATM sub-net boundaries, and guaranteed quality of service across heterogeneous networks, to name a few examples). In the conduct of research into the requirements of distributed simulation on the Next Generation Internet we propose to: 1) Develop and execute a series of distributed simulation experiments which explore the limitations of the current Internet infrastructure. This work will characterize the scalability of the existing commodity Internet infrastructure to support, for example, Large Scale Multi-cast Applications (LSMA) such as HLA federations. 2) Identify a potential solution set in the form of current hardware and software which partially support the stated requirements. 3) Support the Internet Research Task Force. The focal point for much of the data obtained in the investigation will be the IRTF. Various working groups will also be delivered data and findings. The existence of the Reliable Multi-cast Research Group is a partial acknowledgment of our application problem statement. We would support the RMRG in the form of briefings and papers. 4) Develop prototypes of possible solutions. In collaboration with other research in advanced distributed simulation, experiments can be performed to facilitate advancements in DIS, while refining potential solutions for the NGI. and, finally, as the requirements and specifications for the NGI take shape, 5) Work with leading distributed simulation programs to implement the new protocols in a way supportive of the NGI development. This final point of research would ensure that the Distributed Simulation community could immediately benefit from the NGI initiative. This would also ensure an important DoD and International support base for NGI well in advance of its large scale introduction as well as an important source of feedback. Participating Institutions The three institutions involved, University of Texas at Austin, University of Central Florida, and the Georgia Institute of Technology are current members of Internet 2, and are all heavily involved in the Simulation Interoperability Standards Organization and standards development. Research in the areas highlighted above should provide a basis to take advantage of the speed and bandwidth of the Next Generation Internet in support of improved distributed simulations both for DoD and many other standard, interactive distributed simulations **************************************************************************** ******* --end of White Paper +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Angus L. M. Thom McLean III Research Scientist II Georgia Tech Research Institute Phn 404/894-7486 Information Technology and Telecommunications Lab Fax 404/894-9081 347 Ferst Dr., Atlanta, Georgia 30332 Email: thom.mclean@gtri.gatech.edu +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++