The Subject Area of Advanced Applications is intended to
cover topics that arise when constructing simulations of
complex applications such as computer systems, communication systems,
manufacturing systems, health care systems, and transportation systems
(this list is representative, not exhaustive. Other types of
systems are of interest). It is frequently the case
that problem characteristics specific to an application can
be exploited to enhance the size, complexity, or speed of
a simulation. For instance, direct-execution simulation accelerates
computer cache simulations; fluid-flow models accelerate network
simulations. We solicit submissions that develop novel simulation techniques,
that report on tools embodying novel techniques, and that
analyze the performance benefits accrued by advanced methods.
Distributed interactive simulation is concerned with the interoperation of
diverse, geographically distributed simulations, often operating with
real-time constraints and with humans, hardware, and software in the loop.
It is used extensively for training, test and evaluation, and engineering
analysis. Requirements for distributed simulation, in contrast with those
for parallel simulation, often place more emphasis on consistency and
coherency than on high performance. The United States Department of
Defense (Dod) is the prime mover in developing technology to support distributed
simulations. To date, two major DoD approaches have matured: Distributed
Interactive Simulation Protocol (DIS) and Aggregate Level Simulation
Protocol (ALSP). Currently, there is a DoD effort to establish a common
technical framework which includes a common semantic model, data standards,
and a distributed simulation architecture: the DoD High Level Architecture.
Fundamental research issues abound, including challenges relating to time
management, variable resolution modeling, semantic models, communications
and computational latency management, interoperability, object management
and sharing, data visualization, system survivability, multi-level security
and software reuse. These challenges are by no means unique to DoD-related
activities: they apply as well to modeling and simulation in the natural
and physical sciences, engineering, socio-economics and the humanities.
Extending distributed simulation technology outside of DoD has begun
in the areas of emergency management and air traffic control.
The model execution area handles manuscripts dealing with advances concerning
the EXECUTION of discrete event simulation programs on digital computers.
This includes execution on sequential, parallel (multiprocessor and
multicomputers), and distributed (LAN-based and WAN-based) computing
systems. Papers concerning novel techniques and algorithms for executing
simulations (e.g., to improve runtime efficiency), analyses of existing
techniques and algorithms, and methodologies for execution on parallel
and distributed systems whose central contribution concerns the relationship
with the simulation execution mechanism are appropriate for this area.
Other aspects of simulation model development may also be relevant to this
area, if the central contribution relates to model execution (e.g., simulation
languages for execution on multiprocessor computers). A major emphasis of
this area is papers concerning PARALLEL or DISTRIBUTED execution of
simulation programs.
This area focuses on the creation and manipulation of models over
the lifetime of the system study. Conceptual models represent non-
executable models which serve to organize static and dynamic system
components, whereas executable models are of the following types:
declarative (such as automata and event graphs); functional (such as
queuing models and control engineering diagrams); constraint (such
as equations); spatial (such as PDEs and cellular automata); and
multimodel (integrating more than one model type). Models are created in
deductive or inductive approaches. The deductive approach suggests
model creation from earlier models or specifications. Inductive
techniques include system identification and parameter estimation.
Current areas of interest to modeling methodology include --but
are not limited to-- new modeling techniques; model engineering
approaches; object-oriented modeling methodology; hybrid and
hierarchical modeling; metamodeling; model theory and visual modeling.
This area covers all aspects of the generation, analysis and modeling
of random, or pseudorandom, phenomena for use in simulation models.
Specific topics of interest are random number generators, random number testing,
low-discrepancy sequences, random variate transformations,
stochastic process and random object generators, statistical distribution fitting,
data modeling and correlated sequences.
The Simulation Analysis area seeks to publish high-quality papers
analyzing simulation methodology. Topics of particular relevance
are output analysis, simulation-based optimization,
variance reduction techniques, Markov chain Monte Carlo,
and the interface of simulation, random search, and global
optimization. A paper's contribution may
lie in the development of a new technique, in advancing
underlying theory, or in a novel application of an existing
technique. Recent papers in this area have addressed,
for example, the study of initialization bias
through regenerative process theory, the application
of importance sampling to telecommunications problems,
interactions between metamodel estimation and variance
reduction, gradient estimation, and optimal selection.
Papers making substantive theoretical contributions
or addressing important applications are particularly welcome.
This area of TOMACS covers the assessment of accuracy of simulation models.
The accuracy is assessed through a variety of activities such as
verification, validation, and accreditation. Verification deals with
substantiating that a model is transformed from one form into another, as
intended, with sufficient accuracy. Validation deals with substantiating
that a model, within its domain of applicability, behaves with satisfactory
accuracy consistent with the study objectives. Accreditation deals with the
official determination of model credibility. Verification, validation, or
accreditation is conducted by performing testing. |
