Proceedings of the 2007 Winter Simulation Conference
S. G. Henderson, B. Biller, M.-H. Hsieh, J. Shortle, J. D. Tew, and R. R. Barton, eds.
OPTIMISTIC PARALLEL DISCRETE EVENT SIMULATION OF THE EVENT-BASED
TRANSMISSION LINE MATRIX METHOD
David W. Bauer Jr.
Ernest H. Page
The MITRE Corporation
7525 Colshire Drive
McLean, VA 22102, U.S.A.
ABSTRACT
In this paper we describe a technique for efficient paralleliza-
tion of digital wave guide network (DWN) models based
on an interpretation of the finite difference time domain
(FDTD) method for discrete event simulation. Modeling
methodologies based on FDTD approaches are typically
constrained in both the spatial and time domains. This
interpretation for discrete event simulation allows us to in-
vestigate the performance of DWN models in the context
of optimistic parallel discrete event simulation employing
reverse computation for rollback support. We present par-
allel performance results for a large-scale simulation of a
3D battlefield scenario, 100km
2
and at a height of 100m
with a resolution of 100m in the X-, Y-planes, and 10m in
the Z-plane for 754 simultaneous radio wave transmissions.
1 INTRODUCTION
Parallel discrete event simulation (PDES) technology has
been employed successfully over the last 30 years to im-
prove the performance of many modeling methodologies.
Recently, researchers in this field have begun to investigate
the efficacy of PDES as applied to modeling physical sys-
tems. Common approaches to modeling physical systems
include, but are not limited to, ray-tracing and the finite
difference time domain (FDTD) methods. However these
methods traditionally have not benefitted from discrete event
simulation because of constraints in the spatial and time
domains, related to each method. For example, the FDTD
method is limited by the Courant-Lewy-Friedrichs (CFL)
condition (Courant, Friedrichs, and Lewy 1928, Courant,
Friedrichs, and Lewy 1967).
In the past few years, researchers have begun to adapt
these methods to the discrete event paradigm. Notably, a
2005 study applied discrete event simulation to the particle-
in-cell (PIC) method (Karimabadi et al. 2005) and achieved
a two order of magnitude increase in performance. The PIC
method has a long history, reviewed in Birdsall (1991), and
is commonly used in the area of plasma physics. The
break-through in this approach was the removal of the CFL
constraint, which allowed for a two order of magnitude
improvement of the runtime. This interpretation was then
studied in the context of parallel discrete event simulation
by Tang et al. (2006) for a 1D spacecraft model, which
improved the performance by an additional order of mag-
nitude.
In 2006, James Nutaro published a study adapting the
FDTD method with respect to digital wave guide network
wave simulation to the discrete event paradigm. Here,
the focus was on the propagation of electromagnetic waves
through a complex 3D environment. A formal description of
the algorithm was defined for discrete event systems. Nutaro
indicated greater than an order of magnitude improvement
in the cost of computation over the FDTD method, for
high resolution 3D models of Digital Waveguide Networks
(DWN).
In this paper we apply the formal model outlined in
Nutaro (2006) to the PDES paradigm, specifically optimistic
synchronization enabled by reverse computation. We chose
this method to study because the algorithm allowed for reso-
lutions independent of the wavelength of the electromagnetic
waves modeled. Our challenge is to apply this method to
a DWN physical simulation for battlefield scenarios where
the scale of the environment (100km long X 100km wide
X 100m in height) is large and where the number of radio
wave transmissions is large, in this case 754 simultaneous
wave transmissions.
The main contribution of our work is the application of
the Event-Based Transmission Line Matrix (ETLM) mod-
eling method to the area of PDES known as optimistic
simulation. Optimistic synchronization was first proposed
by Jefferson (1985) and allows as fast as possible event
execution where violations of the causality constraint may
occur. Where violations occur, the causality constraint is
then preserved by rolling back improperly processed events,
676 1-4244-1306-0/07/$25.00 ©2007 IEEE