The complexity and dynamics of the Internet are driving the demand for
scalable and effective network simulation. This talk describes a collaborative
on-line simulation scheme to provide real-time network simulators.
We present a method in which a large network is decomposed into parts and each
part is simulated independently and concurrently with the others, exchanging
periodically information about the delays in paths of each part. The
decomposed parts are connected to each other through links that represent
network connection. Parts iterate over the selected simulated time period
until the exchanged delay information converges to a constant value within the
prescribed error metrics. We present a method for measuring the convergence of
the iterations to the fixed point solution.
Each decomposed part may represent a subnet or a subdomain of the entire
network, thereby mirroring the network structure in the simulation design.
The proposed method is independent of the specific simulator technique
employed to run simulators of the parts of the decomposed network. In this
talk, we provide a description of the proposed method and its implementation
based on ns and ssfnet simulators. We also present the simulation times and
test results of simulation of the sample communication networks.
About Boleslaw K. Szymanski
Dr. Boleslaw K. Szymanski is an Associated Dean for Information Technology
and a Professor at the Department of Computer Science and a co-founder of the
Scientific Computation Research Center at Rensselaer Polytechnic Institute.
He received his Ph.D. in Computer Science from National Academy of Sciences
in Warsaw, Poland, in 1976. He was a visiting professor at the Aberdeen
University in Scotland and a faculty at the University of Pennsylvania. He is
an author and co-author of more than 150 scientific publications and an editor
of three books. Dr. Szymanski is also the Editor-in-Chief of Scientific
Computing and a member of the editorial board of Parallel and Distributed
Computing Practices. He is an IEEE Fellow and was a National Lecturer for
the Association for Computing Machinery.
His research interests focus on parallel processing and simulations and
networking. His work has been supported by NSF, NASA, ONR, DARPA, ARO and
industry. He has been recipient of IBM Faculty Award. He is also a co-founder
of Premonitia, Inc, a network management startup in Acton, MA.