Training drills conducted by the U.S. military services increasingly
feature more digital simulations, said Army Col. Wm. Forrest Crain,
director of the Defense Modeling and Simulation Office. He also
noted that the simulation technology available today falls short
in many areas.
“The number of simulation-supported training events is steadily
growing,” he said. This fiscal year, the Defense Department
conducted 22 simulation-driven multi-service exercises. But the
use of computer models in military planning does not always work
as advertised. “Computers in 1964 predicted the United States
would win Vietnam War,” Crain said in an interview in Tysons
Corner, Va. “In 1990, computers predicted there would be tens
of thousands of casualties in the Gulf War.”
The reason those models didn’t work was because they couldn’t
properly simulate human behavior, he explained. The next generation
of simulations will need to address the “representation”
of human behavior, Crain said. “We’ve barely scratched
the surface.”
The Defense Modeling and Simulation Office (DMSO), which reports
to the Pentagon’s director of research and engineering, is
responsible for advocating investments by the military services
in simulations technologies, which are used for everything from
weapon design, to troop training and war planning.
“The proper role that DMSO has is to support the community
at large, rather than a specific program,” said Crain. One
exception to that policy has been DMSO’s direct involvement
in the troubled JSIMS (joint simulation system) program. The project
started in 1994, and was conceived as the “flagship”
modeling and simulation technology that eventually would replace
outdated legacy systems. JSIMS would be used in distributed training,
mission planning and mission rehearsal. But JSIMS has been plagued
by delays and budget cuts, and, currently, is scheduled to become
operational in late 2002. DMSO officials were asked to become more
closely involved with JSIMS, in order to save the program and the
Defense Department’s large investment.
“The cost of the program exceeded $1 billion, to develop
a ‘simulation of simulations’ to support the training
of all services,” said Crain. But he believes the expense
will be justified, if JSIMS can replace the dozens of simulations
that the individual services have developed. “At least 30
or more models and simulations were used in Korea for the Ulchi
Focus Lens exercise” last year, said Crain. “Those models
and simulations represent various pieces within the services—air
defense, air campaign, ground fighting simulations. JSIMS is replacing
all that.”
Another joint simulation program under way, called JMASS, aims
to provide a common repository of digital models, which would facilitate
“simulation-based acquisition,” said Cindy Porubcansky,
program manager of the joint modeling and simulation system (JMASS).
“We are building tools and services to build models. We are
not a simulation. We are a toolkit,” she explained.
DMSO will be taking an active role in simulation-based acquisition,
said Crain. This capability allows weapon developers to design and
build systems entirely in digital environments. The technology has
been used in programs such as the Joint Strike Fighter and in commercial
Boeing jets. Air Force Lt. Col. Eileen Borjman is DMSO’s liaison
to the Pentagon for simulation-based acquisition (SBA) projects.
“I found that when you don’t have someone with an assigned
responsibility, no one is held responsible to coordinate,”
Crain said.
SBA, however, has been hampered by the lack of model sharing among
the services and by unresolved issues such as protection of intellectual
property and technical data rights. “I don’t know that
we are far enough along in SBA right now to say that it will or
will not work,” said Crain. “We are looking at the databases
of each service. ... It’s important for them to have access
to each other’s databases to fight in a joint environment.
There is hesitation sometimes to release the database, because it
may or may not be used for what it was originally intended.”
Michael F. Bauman, director of the Army’s Training and Doctrine
Command’s analysis center, said SBA only will work if the
military services and the industry revamp the conventional ways
of doing business.
“We have some real institutional roadblocks in our business,”
he said. “I call it the ‘dirty little secret of our
business.’ We are unable to share sensitive data,” both
proprietary system data and “plain old intelligence data,
which are the basis to our scenarios,” said Bauman. “I
have enormous amounts of data that I would like to share with industry
and I am frustrated that I am not able to share it. ... We have
not fixed this problem satisfactorily to enable collaboration.”
SBA, he added, “isn’t going to happen until some of
these impediments are removed. We are not going to have a sufficiently
realistic data structure underpinning all that, to allow collaboration.
“People don’t want to give data away, because it can
be used against them. We see that all the time. ... You have to
be able to share data and accept the risk that there will be people
out there who will be trying to use it to their own advantage,”
Bauman said. “The downside does not justify lack of sharing.
The upside is much greater.”
High Level Architecture
DMSO, meanwhile, is in charge of the so-called “HLA transition”
that was mandated for all Defense Department simulations. HLA stands
for “high level architecture,” and is the standard with
which all simulations had to comply by October 1, 2000. HLA was
conceived as a way to make simulations compatible with each other,
so they can be networked and shared among military services and
allied forces.
Crain did not expect that every simulation would be HLA-compliant
by the deadline. “There are still some loose ends that we
have to face,” he said. “There is no guarantee that
there’ll be enough money to convert all models to HLA standard
for interoperability.
“This is not to be interpreted as a ‘backing off,’”
he said. “Interoperability is still the goal. HLA is still
the standard for interoperability.”
The October 1 deadline was set about four years ago by the Pentagon,
as the “no-can-play date,” said Crain. The plan was
for HLA to drive the development of new simulations. Legacy simulations
would have to be migrated to the new standard. Earlier this summer,
DMSO predicted that more than 200 simulations would not be able
to be HLA compliant by October 1. The agency also estimated that
the HLA transition for each legacy simulation would cost $50,000.
For 200 simulations, the expense would be $10 million. “But
the more significant problem is the skills and technical expertise
that are needed to develop for all those simulations,” Crain
said. “We don’t have enough around to do all those models
and simulations. So it’s the people skills and the money that
are factors” causing the delays.
“That is not to say that we are going to slide back the date,”
he cautioned. “We have stood up an HLA task force headed up
by DMSO’s deputy director, Navy Capt. Dave Johnson. It has
representatives from all the services and joint community. They
are looking at what should be the next step.
“It’s possible that some may perceive this as a ‘backing
off’ from the original goal. That is not the case. ... We
just have to recognize that we will not be there on time.”
The October 1 deadline did not apply to non-U.S. NATO simulations
and models.
In the long term, Crain also aims to improve the quality of simulations
used for urban warfare training. Most simulations available today
represent areas such as jungles, deserts and woods. But few offer
high-fidelity, realistic models of cities, he explained. Urban combat
modeling is difficult, said Crain, because each building is different
from the other, and there are many components to each building.
Not only does the simulation have to replicate the inside of a building,
but also the sewer systems, the railways, the subway. “You
have a multidimensional environment, rather than rolling in the
open field.
“You encounter other factors we are not prepared very well
to model,” he said, such as the “impact of firing weapon
systems through a wall, the impact on communications, if buildings
interrupt the direct line of sight.
“Now, you are dealing with cases when you are fighting up
and down, from one floor to the next, instead of laterally. It is
a significant problem. When you are trying to do that level of modeling,
where do you go to get the data that tells you how that building
is built?” Cost is another problem. The higher the resolution,
the more it costs to develop models, said Crain. Currently, “the
best option we have [for urban training] is live training facilities.”
Bauman believes more money needs to be invested in “acquisition
models that will tell us how to acquire targets in those clutter
environments.
“I would bet that Sprint and AT&T have better models
of communications in urban and complex areas. I can’t prove
that. [But] we don’t have models to tell us how these systems
work in these kinds of environments.”
He agreed with Crain’s comments that models of human behavior
remain wanting. “We don’t have the models. I know we
can create the digital environments, but we want to create the models
of what is going on in those environments. ... We cannot represent
how humans make decisions.”
Behavioral psychologists claim that they are doing “wonderful
studies” about the effects of sleep deprivation, fatigue and
stress. “But that is not what I’m talking about.”
These psychologists “don’t really understand how humans
make decisions.
“If we understood how people make decisions, we could tailor
simulations and training to enhance people’s abilities. If
you could make a digital model of Saddam Hussein, I could take it
to the basement of the White House and help [officials learn how
to] negotiate with [him] and respond in a realistic way.
“Until we do this well,” said Bauman, “we have
to continue to figure out how to integrate war fighters into our
simulations.”