Seven years after the Defense Department embarked on an ambitious program to
develop a computer-based multi-service combat simulation, the project is on
its last legs, having failed to meet expectations and exhausted the patience
of Pentagon acquisition officials.
The Joint Simulation System (JSIMS) is a federation of computerized combat
simulations designed to train commanders and war planners. After years of delays
and cost overruns, the program appeared to be getting back on track in late
2002. But, in December, the department pulled the plug on JSIMS and directed
the program manager to close the office by September 30, 2003.
Ironically, at the same time the Defense Department was issuing its directive
to cancel JSIMS, the program was delivering the first version of the software,
called block 1.
The program manager, Army Brig. Gen. Stephen Seay, said he will carry out the
orders from the Pentagon, but he also conjectured that JSIMS might be brought
back to life if software tests this summer are successful and a senior review
panel decides that no alternative technologies exist today to replace JSIMS.
Meanwhile, Seay already has been working on a “JSIMS exit strategy”
that may still undergo revisions between now and the end of September.
When the program got under way in 1996, the plan was to make JSIMS the premier
training tool for joint-force commanders. It was to integrate real-world and
simulated military assets on a virtual battlefield.
By the mid-1990s, each service already had begun to develop its own tactical
simulations. The Army was the farthest along, with a program called WARSIM.
But the Pentagon didn’t want the services to have separate, non-interoperable
systems, so JSIMS was conceived as a joint architecture.
Upon hearing about the cancellation of JSIMS, one program source said, in hindsight,
the program probably was doomed from the get-go. “When a program begins
with ‘joint,’ you already start out with three strikes against you.”
JSIMS Version 1 software was delivered in December 2002 to the Joint Forces
Command Joint Warfighting Center, where it will undergo testing this summer.
“We have installed the software in our simulations facility to begin
validation testing,” said a JFCOM spokesman. “The validation process
will take the remainder of the calendar year as we install the software, train
personnel to use JSIMS, and then conduct a three part validation process, which
will culminate with a full systems verification and validation test. ... The
Joint Warfighting Center, as the primary user, will work closely with the JSIMS
program manager to fix problems encountered.”
Seay took over the troubled program in October 2001. Since then, he said in
an interview, “we restructured the program [after] it had missed several
milestones.” He described JSIMS as “the largest constructive simulation
ever attempted by the Defense Department.” The price tag for developing
the system had been estimated at $1.5 billion. The project so far has cost nearly
a billion dollars.
As program executive officer for Simulation, Training and Instrumentation,
Seay manages 160 Army projects. But JSIMS is a special case. It does not fall
under the same management structure as other PEO STRI programs, because it’s
joint. As the manager for JSIMS, Seay not only reports to the Department of
the Army, but also to the Pentagon’s acquisition chief, Edward “Pete”
Aldridge Jr.
When he took charge of JSIMS, Seay said, he was surprised to see that the program
had not been managed like most acquisition projects of that magnitude.
“It had never been to a review to get an approved cost baseline, which
is one of the fundamental things we have to do in acquisition,” said Seay.
“This one, for one reason or another, had not.”
Further, JSIMS is “one of the most difficult programs to manage, because
I don’t have any control over the dollars. One [budget] line out of nine
belongs to the PM. The others belong to the services and development agencies.
... If the [non-Army] budget lines are short, it potentially slows down the
development of the entire alliance. ... When I got the program, there were things
that were wrong, and I had to restructure it.”
In fiscal year 2002, the JSIMS budget was $29 million. It dropped to $19 million
in 2003. The 2004 budget request had no money for JSIMS.
Money problems have plagued the program for years, noted the JFCOM spokesman.
“The services have not adequately funded nor updated their legacy constructive
simulation systems.”
Senior Army and Defense Department panels reviewed the program last year, Seay
said. “Cost performance and schedule were approved in July. We said we
would deliver software in December. We did that.”
That was not enough to save JSIMS, however. “Because of the program’s
checkered past and the idea that we are now doing a training transformation,”
top Pentagon officials decided that it was time to cut the losses and search
for alternative technologies.
According to Seay, however, the directive that zeroes the funding for JSIMS
also asks that the “system verification and validation” be comple-ted
by the first quarter of fiscal year 2004. He is not sure how he will do that
if the program office shuts down in September. “To some degree, there
is some conflicting guidance in the PDM [program decision memo-randum],”
said Seay.
The PDM also mandates an “analysis of alternatives” to consider
whether other technologies could replace JSIMS.
The analysis of alternatives will be run by the Pentagon’s office of
program analysis and evaluation, known as PA&E. Sources said that the AOA
is likely to recommend that JSIMS cease to be a joint program and be parsed
into separate systems for each service.
The AOA could take up to a year to complete, Seay noted. He said he is not
convinced that any other technology can meet the requirements set for JSIMS.
“The technology we are using now is the industry’s best to accomplish
the mission,” said Seay. “We think that, by delivering Version 1,
we’ve shown that the modeling is here, the technology is achievable now.”
Much of the trouble experienced in JSIMS, he explained, stems from the complexity
of trying to meet unique and diverse service needs in a single simulation.
“JSIMS is a federation of unique service models, based on a high-level
architecture, with common standards and protocols ... so you can operate in
different scenarios and levels,” he explained. Problems arise, however,
because each service has different definitions for the “entities”
involved in a war game. JSIMS relies on “entity-based models.” But
each service defines an entity in a different way. “The Air Force and
the Navy would classify an entity-based model differently than the ground forces
would,” said Seay. In the Army or Marine Corps, an entity may be a soldier,
a fire team, a squad or platoon. “In a battlefield, there are thousands
of these entities,” which creates computer processing challenges. “It
requires a lot of power to process all these entities and play in real time.”
In the Navy, a ship may have hundreds or thousands of sailors aboard, but for
modeling purposes, a ship is a single entity. “Different services fight
at different levels of entities, which drives the performance of your program,”
said Seay.
JSIMS critics question whether trying to run such diverse entities in the same
model is a recipe for failure. But Seay remains optimistic. “I think that
the technology we have, although rather young, is here.”
He speculated that, if the upcoming tests are successful and if the AOA concludes
that the JSIMS technology is mature, “we can go back and compete for resources
to keep the program alive.”
JSIMS advocates assert that the decision to cancel the program was based on
politics and not on actual program data.
Retired Air Force Lt. Gen. Dave Vesely is on the JSIMS senior leader advisory
board. He said that the termination of JSIMS took everybody by surprise, including
Seay.
“There was a body of detractors in the analytical community at OSD who
did not think that JSIMS was working,” Vesely said. “They based
that on inadequate data and just jumped at that decision based on what they
felt rather than what they knew.”
An industry source said that JSIMS was stopped because “OSD felt that
an excessive amount of resources have been spent already and additional requirements
would be needed.”
In 2000, JSIMS was spiraling into an “unrecoverable death,” said
a Forecast International analyst. That was when the program underwent a major
restructuring.
The Pentagon’s director of operational test and evaluation said in his
annual report that “The JSIMS program announced a schedule breach in June
2002 that resulted in slipping the initial operational test and evaluation from
fiscal year 2003 to 2004.”
The report noted that the initial performance runs of the system “have
demonstrated the difficulty of initializing a large federation representative
of [a] training exercise configuration, the difficulty of operating in a classified
configuration using the common security services, and the lack of stability
across the network and with individual federates.”
The industry source said that JSIMS was ill-fated once the users realized how
complex it was to operate and to develop training scenarios.
The JSIMS software sent to the Joint Warfighting Center for testing may languish
there, predicted Vesely, because nobody there is interested in using it.
The JFCOM spokesman disagreed. “U.S. JFCOM remains a supporter of the
need to produce a JSIMS capability as soon as possible,” he said. “We
view JSIMS as the bedrock for joint training. ... It will provide the bulk of
constructive simulation capability and infrastructure for an integrated live,
virtual, constructive training environment.” Further, he said, the command
is “developing a plan to establish a software support facility for continued
life cycle maintenance support for JSIMS software.” Whatever technology
is chosen to replace JSIMS, he noted, “JFCOM will be a part of planning
for any future software development.”
Vesely said he believed that JSIMS was cancelled for more than just technology-related
problems. Other factors were cost overruns and management “disagreements
on the architecture.”
If JSIMS goes away, he said, the military services and other organizations
involved, such as the National Security Agency, are going to meet their requirements
one way or another, “but they are not going to do it in a coordinated
fashion as it would be with JSIMS. ... We are going to end up with stove-piped
systems that are not going to work together.”
The biggest contributor to JSIMS, the Army, is going forward with its piece
of the system called WARSIM. Seay said that WARSIM is now the most important
project on his agenda this year.
WARSIM is a corps-level battle simulation designed to support the training
of unit headquarters and command posts. The program is in the engineering and
manufacturing development phase, under prime contractor Lockheed Martin Information
Systems.
A parallel effort is a constructive simulation for lower-end tactical war fighting
model, called ONE SAF (One Semi-Automated Force). ONE SAF and WARSIM, said Seay,
“are the models that the Army uses as training and leader development
tools.”
Seay’s organization, PEO STRI, used to be the Army Simulation, Training
and Instrumentation Command, a subordinate to the Army Materiel Command. STRICOM
was restructured into a program executive office and now reports to the assistant
secretary of the Army for acquisition, technology and logistics.
PEO STRI is “on the fast track” to decide what the organization
is going to look like, said Seay. He noted that, previously, STRICOM was fragmented
and training was developed in isolation from operational concerns. He emphasized
that PEO STRI is going to converge the training and operational concerns.
“I think, quite frankly, training solutions that we can develop here
will give a better picture of what operational combat is going to look like,”
he said. In the long run, PEO STRI increasingly will join forces with other
program offices. “You will find that we will gain more business over the
long-term, because of our ability to make things interoperable,” he noted.
Despite the fast advances in simulation technology, Seay said, “there
is some resistance to the use of simulation, because it has not been proven.”
Simulations, however, are the key to effective and low-cost training. “We
can make the images look and smell like they are real,” he said.
He believes that simulation can cut down on live-fire testing and training
even all the way to down to basic marksmanship.
“We are changing the way we do things,” he said. “We train
to a higher level of confidence.” Soldiers cannot just acquire one set
of skills that may no longer prove useful with the changing nature of the battlefield.
“Every place we go to is different, and that is why you need simulation.
It is the way they learn. ... It is the way [they] pick up the skills.”