Introduction to Simulation
Defines simulation, its applications, and the benefits
derived from using the technology. Compares simulation
to related activities in analysis and gaming.
- DOD Overview.
Explains the simulation perspective and categorization
of the US Department of Defense.
- Training, Analysis,
and Gaming. Provides a general delineation
between these three categories of simulation.
Describes the fundamental components that are found
in most military simulations.
Describes the basic differences between functional
and object oriented designs for a simulation system.
Emphasizes the importance of providing an infrastructure
to support all simulation models, tools, and functionality.
Describes the newest implementation of an infrastructure
in the form of an object oriented framework from
which simulation capability is inherited.
Interoperability initially meant constructing a
dedicated method for joining two simulations for
a specific purpose.
- DIS. The
virtual simulation community developed this method
to allow vehicle simulators to interact in a small,
The constructive, staff training community developed
this method to allow specific simulation systems
to interact with each other in a single joint training
This program was developed to replace and, to a
degree, unify the virtual and constructive efforts
The primary method for executing simulations has
been various forms of queues for ordering and releasing
- Trees. Basic
queues are being supplanted by techniques such as
Red-Black and Splay trees which allow the simulation
store, process, and review events more efficiently
than their predecessors.
- Event Ownership.
Events can be owned and processed in different ways.
Today's preference for object oriented representations
leads to vehicle and unit ownership of events, rather
than the previous techniques of managing them from
a central executive.
Single processor simulations made use of a single
clocking mechanism to control all events in a simulation.
This was extended to the idea of a "master
clock" during initial distributed simulations,
but is being replaced with more advanced techniques
in current distributed simulation.
The "master clock" too often lead to poor
performance and required a great deal of cross-simulation
data exchange. Researchers in the Parallel
Distributed Simulation community provided several
techniques that are being used in today's training
& Optimistic. The most notable time management
techniques are conservative synchronization developed
by Chandy, Misra, and Bryant, and optimistic synchronization
(or Time Warp) developed by David Jefferson.
In addition to being synchronized across a distributed
computing environment, many of today's simulators
must also perform as real-time systems. These
operate under the additional duress of staying synchronized
with the human or system clock perception of time.
- Object Interaction.
Military object modeling is be divided into two
pieces, the physical and the behavioral. Object
interactions, which are often viewed as 'physics
based', characterize the physical models.
Military objects are often very mobile and a great
deal of effort can be given to the correct movement
of ground, air, sea, and space vehicles across different
forms of terrain or through various forms of ether.
- Sensor Detection.
Military object are also very eager to interact
with each other in both peaceful and violent ways.
But, before they can do this they must be able to
perceive each other through the use of human and
Encounters with objects of a different affiliation
often require the application of combat engagement
algorithms. There are a rich set of these
available to the modeler, and new ones are continually
Object and unit attrition may be synonymous with
engagement in the real world, but when implemented
in a computer environment they must be separated
to allow fair combat exchanges. Distributed
simulation systems are more closely replicating
real world activities than did their older functional/sequential
ancestors, but the distinction between engagement
and attrition are still important.
The modern battlefield is characterized as much
by communication and information exchange as it
is by movement and engagement. This dimension
of the battlefield has been largely ignored in previous
simulations, but is being addressed in the new systems
under development today.
- More. Activities
on the battlefield are extremely rich and varied.
The models described in this section represent some
of the most fundamental and important, but they
are only a small fraction of the detail that can
be included in a model.
Military simulations have historically included
very crude representations of human and group decision
making. One of the first real needs for representing
the human in the model was to create a unique perception
of the battlefield for each group, unit, or individual.
Battlefield objects or units need to be able to
react realistically to various combat environments.
These allow the simulation to handle many situations
without the explicit intervention of a human operator.
Today we look for intelligent behavior from simulated
objects. Once form of intelligence is found
in allowing models to plan the details of a general
operational combat order, or to formulate a method
for extracting itself for a difficult situation.
Early reactive and planning models did not include
the capability to learn from experience. Algorithms
can be built which allow units to become more effective
as they become more experienced. They also learn
the best methods for operating on a specific battlefield
or under specific conditions.
- Artificial Intelligence.
Behavioral modeling can benefit from the research
and experience of the AI community. Techniques
of value include: Intelligent Agents, Finite State
Machines, Petri Nets, Expert and Knowledge-based
Systems, Case Based Reasoning, Genetic Algorithms,
Neural Networks, Constraint Satisfaction, Fuzzy
Logic, and Adaptive Behavior. An introduction
is given to each of these along with potential applications
in the military environment.
Military objects are heavily dependent upon the
environment in which they operate. The representation
of terrain has been of primary concern because of
its importance and the difficulty of managing the
amount of data required. Triangulated Irregular
Networks (TINs) are one of the newer techniques
for managing this problem.
The atmosphere plays an important role in modeling
air, space, and electronic warfare. The effects
of cloud cover, precipitation, daylight, ambient
noise, electronic jamming, temperature, and wind
can all have significant effects on battlefield
- Sea. The
surface of the ocean is nearly as important to naval
operations as is terrain to army operations. Sub-surface
and ocean floor representations are also essential
for submarine warfare and the employment of SONAR
for vehicle detection and engagement.
Many representations of all of these environments
have been developed. Unfortunately, not all
of these have been compatible and significant effort
is being given to a common standard for supporting
all simulations. Synthetic Environment Data Representation
and Interchange Specification (SEDRIS) is the most
prominent of these standardization efforts.
Military commanders have always dealt with the battlefield
in an aggregate form. This has carried forward
into simulations which operate at this same level,
omitting many of the details of specific battlefield
objects and events.
Recent efforts to join constructive and virtual
simulations have required the implementation of
techniques for cross the boundary between these
two levels of representation. Disaggregation
attempts to generate an entity level representation
from the aggregate level by adding information.
Conversely, aggregation attempts to create the constructive
from the virtual by removing information.
It is commonly accepted that interoperability in
these situations is best achieved though disaggregation
to the lowest level of representation of the models
involved. In any form the patchwork battlefield
seldom supports the same level of interoperability
across model levels as is found within models at
the same level of resolution.
Models are abstractions of the real world generated
to address a specific problem. Since all problems
are not defined at the same level of physical representation,
the models built to address them will be at different
levels. The modeling an simulation problem
domain is too rich to ever expect all models to
operate at the same level. Multi-Resolution
Modeling and techniques to provide interoperability
among them are inevitable.
Simulation systems and the models within them are
conceptual representations of the real world.
By their very nature these models are partially
accurate and partially inaccurate. Therefore,
it is essential that we be able to verify that the
model constructed accurately represents the important
parts of the real world we are try to study or emulate.
The conceptual model of the real world is converted
into a software program. This conversion has
the potential to introduce errors or inaccurately
represent the conceptual model. Validation
ensures that the software program accurately reflects
the conceptual model.
Since all models only partially represent the real
world, they all have limited application for training
and analysis. Accreditation defines the domains
and conditions under which a particular model can
be reliably used.
- VV&A Principles.
The Department of Defense has established specific
guidelines for conducting VV&A. Simulation
researchers have also defined fundamental principles
that are important for this activity.