The Joint Training Experimentation Program:
Lessons Learned from the First Demonstration
Reginald Ford, John Shockley, Michael Beebe, Mark Faust, Gerald Lucha
SRI International
333 Ravenswood Avenue
Menlo Park, CA 94025
650-859-4375, 650-859-4165, 650-859-6734, 650-859-2188, 650-859-4434
reginald.ford@sri.com, john.shockley@sri.com, mike.beebe@sri.com, mark.faust@sri.com, gerald.lucha@sri.com
Mark Johnson
SRI International
4111 Broad Street; Suite A-7
San Luis Obispo, CA 93401
805-542-9330
mark.johnson@sri.com
COL (Ret.) John C. Bernatz
Office of the Adjutant General
9800 Goethe Road
Sacramento, CA 95826
916-854-3676, DSN: 466-3676
John.Bernatz@ca.ngb.army.mil
Keywords: LVC Training Systems; Live-Constructive Interaction
ABSTRACT: The Joint Training Experimentation Program (JTEP) is a multiphase, multiyear effort to develop a
distributed training capability for the California National Guard that includes live, virtual, and constructive training
simulations. JTEP will use existing and readily available systems to address the unique training needs of the Guard, i.e.,
limited training time and units distributed around the state. The current, initial, phase of JTEP is to develop a battalionlevel training capability for ground forces training. The next phase will expand on this capability to provide training for
brigade-level exercises and training for the Guard’s Military Support for Civil Authority (MSCA) mission. Future phases
will continue this expansion in scope to include Air Guard assets and the distribution of JTEP capabilities to other states.
As part of this initial phase, the first JTEP demonstration, a live-constructive (LC) system linkage, was conducted in May
2003. This demonstration linked the Deployable Force-on-Force Instrumented Range System (DFIRSTTM), a live
instrumented training system at Camp Roberts, CA, and the Joint Combat and Tactical Simulation (JCATS), a constructive
simulation located at Camp San Luis Obispo, CA, about 50 miles away. The demonstration scenario centered on an
instrumented platoon of live forces engaging a live scout element. Constructive forces expanded the scope of the exercise to
a company-level scenario. The demonstration successfully incorporated a number of functional goals of JTEP, including
(1) low-cost multi-echelon training over distributed training sites, (2) the use of existing Guard training and networking
assets, and (3) the expansion of the training range to include off-range areas for the constructive forces. Key technical
accomplishments included (1) successful indirect and direct fire engagements between live and constructive entities, using
pop-up targets correlated with JCATS entities as visual stimulation for the live forces; (2) data and voice connectivity over
GuardNet, a non-dedicated network; (3) the use of correlated synthetic terrain for better LC engagement simulation
fidelity; and (4) the demonstration of a low-cost instrumentation set to emulate tank maneuver training with HMMWVs.
This paper provides an overview of JTEP and describes the results and lessons learned from this first LC demonstration.
Additionally, it addresses how these results will impact future JTEP demonstrations and how they might be applicable to
other live, virtual, and constructive (LVC) integration efforts.
1. Introduction
2. JTEP from a Trainer’s Perspective
JTEP overview. The California National Guard (CANG)
currently uses advanced live, virtual, and constructive
(LVC) systems1 to support training, but each system is
stand-alone. JTEP was conceived to bring to the California
Guard the benefits of integrating existing or readily
available training environments, and to enable LVC
interaction over non-dedicated wide-area networks
(WANs).
JTEP is a technology experimentation program whose
purpose is to provide a new training capability to National
Guard soldiers and units. The capability is new in that it
will allow National Guard units to train at the level to
which they are organized for pennies on the dollar. JTEP
leverages existing National Guard communications
capabilities and training systems to effect this capability.
National Guard units will train on equipment and systems
already familiar and in inventory, but now linked together
electronically in real time, over extended distances, and on
the same exercise terrain.
JTEP is an experimentation program that will leverage the
integration successes of other programs whenever possible,
but will also advance the state of the art in system and
simulation interoperability as needed to meet California
Guard training needs. For each of the major program
phases, JTEP progresses through three activities: (1) an
initial study to determine which candidate systems and
integration mechanisms will achieve the greatest impact;
(2) a series of proof-of-concept technology and training
demonstrations or experiments; and (3) the final selection
of systems and the design of demonstrations with the
eventual goal of providing a leave-behind capability
suitable for routine usage in training.
JTEP schedule. JTEP is a multiyear program. The first
(current) year addresses battalion-level training. The second
year will address Military Support for Civilian Authority
(MSCA), additional sites, and brigade-level training. The
third and subsequent years will add MSCA and combat
training functions, and expand the program scope to Air
Guard, additional states, etc.
Scope and organization of this paper. This paper provides
an overview of JTEP LVC integration challenges and early
progress, in particular the initial LC demonstration
conducted on 3 May 2003. Sections 2 and 3 explain the
program’s goals and methodology. Section 4 summarizes
the demonstration’s architecture, technical and operational
achievements, and lessons learned. Sections 5 and 6
describe future plans. As JTEP evolves there will be
additional demonstrations of expanded scope. We envision
submitting papers that address specific technical issues that
arise during the development and execution of the
demonstrations.
1
A live “simulation” comprises real people, real vehicles, real
environment, and simulated weapons. A virtual simulation
comprises real people, simulated vehicles, simulated environment,
and simulated weapons. A constructive simulation comprises some
real people, some simulated people, simulated vehicles, simulated
environment, and simulated weapons.
Training augmentation. JTEP LVC integration improves
training realism by making it easier to combine training
activities that are commonly conducted separately, e.g., live
maneuvers and simulated fire support and close air support
(CAS). It supports larger-scale and more realistic training
scenarios: e.g., one armor company conducting live
maneuvers on an instrumented range can train jointly with
additional companies that are using virtual and/or
constructive systems.
Limited training time, budget, and facilities. JTEP
creates more opportunities for the Guard to train closer to
home, converting travel time to training time. It enables
more advanced training at lower cost, e.g., creating
battalion-size events with company-level operational
investment, i.e., OPTEMPO dollars. It enables the Guard to
conduct exercises with larger units than existing ranges can
support. It allows scenarios on live ranges to extend beyond
range boundaries.
Geographically distributed forces. California Guard
training sites are widely dispersed. Current Homeland
Security missions (e.g., out-of-state activations) exacerbate
the situation. JTEP allows geographically separated units to
train together and to experience a unified battlefield
environment despite physical separation. Data and voice
communications over a WAN enable forces using the live
instrumentation and simulation systems located at each site
to coordinate their actions within the common environment.
Distributed AAR. In the second year we will add a
Distributed After Action Review (AAR) at the conclusion
of the training event. The distributed AAR will leverage
and expand the capabilities of the existing National Guard
Distance Learning Classrooms. Commanders will have the
capability to ensure that all participants at all locations are
able to engage in the highly effective discovery learning
process of the AAR.
MSCA. In the preceding decade, the California National
Guard has conducted approximately 50 percent of all the
nation’s National Guard-provided MSCA missions and
activities. This is a particularly important issue for
California. Our state remains prone to an array of natural
disasters, including earthquake and wildfire. Additionally,
in the post-9/11 world, all states are vulnerable to terrorist
activity and Weapons of Mass Destruction (WMD) events.
The need for meaningful training in the emergency
management and response field has never been more acute
than it is now. Starting with the second phase, JTEP will
include training opportunities for the emergency
management and response community, similar to the
Army's Battle Command Training Program.
The real beneficiaries. JTEP will provide National Guard
soldiers and airmen, who often receive little in the way of
training resources, a capability to engage in multi-echelon
training calculated to make units and soldiers significantly
better prepared to conduct their assigned combat jobs.
Additionally, the National Guard has the mission to protect
the lives and property of the citizens of their states when
called on by their Governors. JTEP will enhance the
capability of National Guard units to accomplish this
mission and will provide significantly enhanced training
and preparation for the civilian community we support.
Experimentation and transition. JTEP creates a
mechanism for the smooth introduction of new technologies
into California Guard training. Technologies are integrated,
developed, and tested in the testbed. Demonstrations
provide an opportunity for Guard personnel to evaluate new
capabilities and to provide feedback to developers. This
cycle facilitates the introduction of advanced LVC training
capabilities into the standard training regimen with
minimum disruption.
3. JTEP Development
Sponsorship. JTEP is being developed by SRI International
under contract to the CANG and National Guard Bureau
(NGB). Direct program oversight is provided by the
California National Guard Joint Staff.
System Analysis. JTEP commenced in September 2002
with a trade-off study of candidate systems and LVC
integration mechanisms. Selected references include [1] –
[26].
Because of the Guard’s training and acquisition
requirements, preference was given to systems currently
used or readily available for use by the CANG to train
armored forces, but other systems were examined for
suitability and affordability. The study was completed in
January 2003. The selected systems are as follows:

Constructive: Joint Conflict and Tactical Simulation
(JCATS). The CANG currently used Janus-T, but the
lack of an interoperability interface mandated the
selection of an alternative.

Virtual: Close Combat Tactical Trainer (CCTT). The
CANG has a platoon-size mobile suite.

Live: Deployable Force-on-Force Instrumented
Training System (DFIRSTTM). The CANG has a
company-size system.

Interoperability: Distributed Interactive Simulation
(DIS). Although the High-Level Architecture (HLA)
was preferred for technical, performance, and future
growth reasons, DIS was selected because it is already
fully supported by all of the selected LVC systems, and
is being used by the Digital Battlestaff Sustainment
Trainer (DBST) for a related JCATS-CCTT integration
task.

Network: GuardNet. Sufficient bandwidth was
available to support the initial scenario for the JTEP
LC demonstration, but bandwidth conservation
presents a long-term challenge that will be addressed in
subsequent JTEP development.
JTEP testbed. SRI developed a testbed to support both
system analysis and system integration. Analysis areas
include system interfaces, interoperability adaptation
mechanisms, performance, and the conformance of
integrated systems, components, scenarios, terrain, etc.,
with “fair fight” requirements. A key aspect of the testbed
approach is that all of the candidate JTEP components are
integrated in a single logical location. Typically, LVC
integration efforts have involved bringing the system
components together during scheduled integration events
prior to an actual exercise. With the JTEP testbed,
integration is ongoing throughout the development process
leading up to the actual demonstration. The testbed
configuration then becomes the basis for the actual
demonstration system. We believe that this approach has
many advantages in identifying technical integration issues
early on, and was a major factor in the overall success of
the demonstration.
A notable feature of the testbed is its distribution between
the SRI Menlo Park facility (SRI-MP) and San Luis Obispo
office (SRI-SLO), which are 200 miles apart. JCATS is
installed in both locations, but SLO is the primary
integration site. DFIRST playback is available at both sites,
but MP alone supports live operations testing. It is
anticipated that CCTT test stations will be installed
exclusively in SLO. This arrangement enforces daily
experience with the challenges of physical and network
distribution, and therefore facilitates a smooth transition
from lab to field environments. The distributed testbed has
also forced us to use non-dedicated networks in all of our
integration testing. The use of these networks in the testbed
configuration uncovered issues that eventually appeared in
the operational GuardNet. The early identification and
resolution of these issues contributed to our success.

DIS 2.0.4 (IEEE 1278.1), the integration protocol.

JCATS 4.0, the constructive simulation.

DFIRST 2.0.6,
instrumentation.

Customized DIS Radio software using EmDee
DisComm6 technology. SRI developed hardware
interfaces to the live radio components (handsets,
SINCGARS radios). Note: ASTi DIS radio systems
were considered, but their acquisition lead-time
precluded their use for an April or May demonstration.

SINCGARS radios interfaced to the DIS Radio
processor in the DFIRST base station, which provided
tactical communications with live armored crews.

SRI Data Forwarder 1.0, which provided reliable multisite distribution of best-effort broadcast DIS PDUs
over GuardNet.

DFIRST
prototype
HUM-1
and
HUM-72
instrumentation, which allowed the battle to be fought
with HMMWVs operating as M1A1s, T-72s, and BTR60s.

MAK VR-Link 3.7.3-ngc libraries, used in the DIS
gateway.

MAK Stealth Observer 4.2-ngc, which provided a 3D
view of the full battlefield at Camp SLO and Camp
Roberts.

CISCO PIX501, which provided security isolation
between GuardNet and the DFIRST wireless LAN.
LVC integration sequencing. At program inception, it was
anticipated that the first demonstration would be virtual
constructive (VC) and the second LVC. However, after the
system analysis study was completed in January 2003, SRI
and the California Guard determined that a DFIRSTJCATS LC integration was more feasible for an April or
May demonstration than a CCTT-JCATS VC integration. A
DFIRST-CCTT-JCATS LVC demonstration is planned for
October 2003.
4. JTEP Live-Constructive Demonstration
Demonstration overview and date. A technology and
training demonstration of the DFIRST-JCATS LC
integration was conducted on 3 May 2003. JCATS
supported command activities at Camp San Luis Obispo,
and DFIRST supported armored unit maneuver training 50
miles away at Camp Roberts. Figure 1 shows the
geographic distribution of the demonstration.
the
live
maneuver
training
4.2 Demonstration Scenario
The demonstration scenario comprised live and constructive
Friendly and OPFOR forces with operation orders as
follows:
Friendly: 1 Tank Company with support elements

Figure 1. Geographic Distribution of Demonstration
4.1 Demonstration Components
The components used to support the LC integration
included
Live: 1 Tank Platoon (4 HUM-1s). Depart assembly
area (AA) vic FQ 995603; proceed along Route Merlin
and destroy scout element with T-72 support at
Objective (OBJ) Wizard vic GQ045595; set up Hasty
Defense and prepare to support by fire for the main
effort on OBJ Magic.Constructive: 2 Tank Platoons
with company headquarters and support elements. 10
M1A1s. 3 M113 Armored Personnel Carriers (1 First
Sergeant, 1 Maintenance, 1 Medics). 1 M88A1
Tracked Recovery Vehicle. 1 Mortar Section (2 Mortar
tracks 120mm). Depart Assembly Area (AA) vic FQ
995603; proceed along Route Merlin by passing
Objective (OBJ) Wizard vic GQ045595 and
proceeding for Deliberate Attack on OBJ Show vic GQ
057592; once OBJ Show is cleared set up Hasty
Defense.
OPFOR: 1 Tank Platoon and support elements

Live: 1 Tank (HUM-72), 2 APC (HMMWVs “guised”
as BTR-60s). Defend vic GQ045595.

Constructive: 4 T-72 Tanks. 2 BMP Armored
Personnel Carriers. 1 Mortar Section (2 Mortar
Tracks). 2 BMP on target lifters for play. Defend vic
GQ 057592.
Figure 2. Demonstration Scenario
The scenario is depicted graphically in Figure 2. Key
elements of the scenario included a design that separated
the majority of live and constructive forces, and specifically
designated constructive units that were mimicked as live
pop-up targets. Because live force elements generally
cannot see constructive force elements, a wraparound
scenario was designed so that constructive forces would not
engage live forces in direct fire (except for the designated
pop-up targets). Constructive Red elements were kept
beyond the range of live Blue, and constructive Blue forces
did not move forward until live Blue tanks had destroyed
live Red elements.
4.3 Demonstration Architecture
Overall network. The distribution of components and the
communications between the Camp SLO and Camp
Roberts sites is shown in Figure 3. Command elements and
constructive OPFOR elements were at Camp SLO, while
live Friendly, live OPFOR, and the pop-up targets
(mimicking JCATS entities) were at Camp Roberts.
The scenario also included two types of engagement
interaction between JCATS and DFIRST. Command
elements used JCATS to initiate indirect fire artillery
missions against DFIRST live entities. More importantly,
JCATS-DFIRST and DFIRST-JCATS direct fire
engagements occurred between JCATS pop-up targets and
live DFIRST entities. Two JCATS entities mimicked live
pop-up targets. The targets were entities “owned” by the
Camp SLO JCATS simulation, but they were located
physically on the Camp Roberts DFIRST range. The popup targets provided visual stimulation to the live players.
Engagement interactions between the targets and the live
entities involved both JCATS and DFIRST. In the JTEP LC
demonstration, this approach enabled a fair fight between
live and constructive elements, but it can also be used
generally in live exercises to enable pop-up targets to return
fire.
Figure 3. Distributed Components and Communications
Camp SLO components and network. The internal Camp
SLO components and network are shown in Figure 4.

The JCATS DIS bridge broadcast JCATS entity state,
weapons fire, and detonation data as DIS PDUs, and
received incoming DFIRST entity state, fire, and
detonation PDUs.

Standard tactical handsets were connected to the DIS
radio interface boxes and software. Voice from the
handsets was broadcast as DIS radio PDUs. Incoming
DIS radio PDUs from DFIRST were transmitted as
voice on the handsets.

Stealth Observer received all JCATS and DFIRST DIS
entity state and detonation PDUs and displayed them
on 3D terrain.

The Data Forwarder was the server node that accepted
connections from remote client nodes. It listened on the
local network for UDP-broadcast DIS PDUs and sent
them to remote clients over GuardNet as a TCP stream.
It also received TCP streams from remote clients and
broadcast them on the local network as UDP DIS
PDUs.
Each friendly and OPFOR command element used a
JCATS station to plan and modify the movements of semiautomated forces and to initiate artillery fire missions. Each
station was able to see own-force JCATS and DFIRST
entities, plus the opposing force entities acquired by ownforce elements (as determined by JCATS algorithms). DIS
radios located at Blue and Red stations allowed command
elements to communicate with live crews located in
vehicles at Camp Roberts.
White force elements saw a combined view of JCATS and
DFIRST entities and engagements on the JCATS White
station and on Stealth Observer. JCATS and Stealth
Observer terrains were correlated, so that the 3D view
faithfully reflected the terrain used by JCATS algorithms to
determine the line-of sight (LOS) among entities. Both
white station and JCATS stations used the JIDPS terrain
data set distributed by JFCOM.
Figure 4. Camp SLO Components and Network
Camp Roberts components and network. The internal
Camp Roberts components and network are shown in
Figure 5.

Because an Ethernet connection to GuardNet was not
available at the DFIRST Base Station location, a
wireless LAN was used to send and receive JTEP
network traffic.

A PIX501 Firewall was used to secure GuardNet
against possible access intrusion from a wireless LAN
eavesdropper.

The DFIRST DIS gate broadcast DFIRST entity state,
weapons fire, and detonation data as DIS PDUs, and
received incoming JCATS entity state, fire, and
detonation PDUs.

Tactical SINCGARS radios were connected to the DIS
radio interface boxes and software. Voice from the
radios was broadcast on the local network as DIS
PDUs. Incoming radio PDUs from JCATS were
broadcast as voice on the SINCGARS radios.

DFIRST software mediated indirect and direct fire
engagements between DFIRST and JCATS. Since popup targets were modeled as JCATS entities, target up
and down commands were associated with data sent
from JCATS to DFIRST, i.e., a target emerging from
defilade was raised, and a killed target was dropped.
Commands were issued to personnel controlling the
target via voice radio.

Stealth Observer received all JCATS and DFIRST DIS
entity state and detonation PDUs and displayed them
on high resolution 3D terrain.

The Data Forwarder was a client node connected to the
Camp SLO server node. It listened on the local
network for UDP broadcast DIS PDUs and sent them
to the server over GuardNet as TCP streams. It also
received TCP streams from the server and broadcast
them on the local network as UDP DIS PDUs.
The observer controller (OC) in the DFIRST Base Station
saw a combined view of JCATS and DFIRST entities and
engagements on the DFIRST 2D map display and the 3D
Stealth Observer. Stealth Observer terrain was derived from
high-resolution terrain data, so that the 3D view closely
correlated with the actual terrain seen by the crews in the
vehicles. It should be noted that the terrain database used by
Stealth Observer in Camp Roberts was different from the
database used in Camp SLO, since Camp Roberts observers
needed a view that matched ground truth as closely as
possible, whereas Camp SLO observers needed a view that
matched the lower-resolution terrain used by JCATS.
4.4 Key Functional Accomplishments
Although this was the first JTEP demonstration, it achieved
a number of functional accomplishments that are key for
effective training. While some are common for LVC
integration, others satisfy unique Guard training needs.
Connecting two training sites. During the JTEP
demonstration, the TOC was at Camp SLO and the tank
crews at Camp Roberts. This distribution of activities
demonstrates JTEP’s potential to convert travel time to
training time. Though not a new development in the
simulation interoperability community, this is a significant
development for the Guard, since it demonstrates JTEP’s
capability to meet the Guard’s training requirements as an
eventual leave-behind system.
Expanded training-site opportunities. The JTEP JCATS
instrumentation comprises laptop computers, DIS radios,
and cables; thus it could be used at any NGB site that has a
GuardNet connection. Because Camp Roberts and NTC are
the only California ranges for armored maneuver training,
there is normally very limited opportunity for tank crews to
train close to home. However, the DFIRST HUM-1/HUM72 instrumentation enables maneuver training in any
location where HMMWVs are available. Also, for Camp
Roberts it extends the training season, since portions of the
range that are closed to tanks due to winter river conditions
remain open to HMWVVs.
Low-cost training using simulation augmentation. The
JTEP demonstration created a company-size event with
platoon-level OPTEMPO dollars; i.e., ths cost of operating
a platoon in a training exercise.
Low-cost training using HUM-1/HUM-72. Substituting
HMMVWs for tanks reduced the operational cost of JTEP
LC demonstration, and it demonstrated the feasibility of
“JEEP-EXs”; i.e., using HMMWVs as tank surrogates for
basic maneuver training. A JEEP-EX capability drastically
reduces cost and time to draw vehicles.
Expansion of real training area. Live maneuver training
is necessarily constrained to stay within range boundaries
and to avoid dangerous or environmentally sensitive areas.
However, in the JTEP training scenario the battle extended
beyond range boundaries to a private farm. No complaints
were received from the farm’s owner because this part of
the battlefield was populated solely by JCATS semiautomated forces
Figure 5. Camp Roberts Components and Network
4.5 Key Technical Accomplishments
We recognize that many of the technical accomplishments
cited below have been reported for a number of previous
LVC exercises and demonstrations. However, they are key
in the JTEP context because of their applicability to Guard
training and the JTEP emphasis on developing a functional
leave-behind training capability.
LC direct-fire interaction. Live DFIRST vehicles
(HMMWVs playing as M1A1s) engaged and killed
constructive JCATS vehicles (BMP-2s), and the JCATS
entities returned fire. The BMP2 weapons were able to hit
the M1A1s but not to damage them before the BMP-2s
were destroyed. The JCATS entities were represented in the
live domain by pop-up targets (typically used for live fire
target practice). The targets were placed in the real terrain,
and then the JCATS entities were assigned to fixed
defensive positions at the locations of the live targets. Each
system reported shots by its entities to the other system as
DIS fire and detonation PDUs and then adjudicated hits to
its own entities. To give the attacking live crews visual
feedback on the effects of their shots, an observer with a
remote control device dropped each target when JCATS
ruled that its associated BMP had been killed.
LC indirect fire interaction. Both sides of the live fight
received indirect fire support from constructive JCATS 120
mm mortar carriers. The live commanders requested fire
missions from a human participant role-playing the fire
direction center over voice communications; the role player
then directed the JCATS entities to fire. JCATS then sent
fire PDUs to DFIRST, and DFIRST assessed damage to the
live vehicles and provided incoming fire notification to the
live crews.
Connectivity: DIS Radios. Tactical voice from the
headquarters element at Camp SLO was connected to live
(DFIRST) entity/vehicle commanders using DIS Radios.
The linkage went as follows: Camp SLO Voice <--> DIS
PDUs <--> GuardNet <--> DIS PDUs <--> SINGCARS
Radios <--> RF <--> VRC-12 Radios <--> In-vehicle Live
Voice. COTS software and GOTS radios were used with
SRI-developed H/W and S/W interfaces.
Connectivity: nondedicated link. Most examples of LVC
interoperation found in the literature were implemented
using dedicated networks. In keeping with the JTEP
objectives, the demonstration utilized a nondedicated thirdparty network for multisite DIS data connectivity. An SRIdeveloped Data Forwarder application bridged the
physically separated LANs over a WAN. During integration
testing, the public Internet was used for communications.
During the demonstration, GuardNet was used.
Terrain toolkit. Terrain data comes in many formats,
resolutions, and coordinate systems. Data may also have
internal correlation flaws and gaps in coverage. To create
high-quality terrain databases from a variety of source data,
we assembled a terrain toolkit comprising Terrex’s
TerraVista, Dart, and SAFUSA; ESRI’s ArcGIS, 3D
Analyst, and Spatial Analyst; Quantum3D’s Audition;
Adobe’s Photoshop; and MAK’s Stealth Observer. For the
LC demonstration, we used this toolkit to correct source
data and to create, annotate, and verify terrain databases.
Correlated terrain. Terrain incompatibilities between
simulation systems are a major impediment to direct-fire
interactions between entities belonging to different systems.
Two terrain data sources for Camp Roberts were available
to us: a JCATS terrain database distributed by the Joint
Forces Command (JFCOM) Joint Integrated Database
Production Service (JIDPS), and high-resolution GIS
source data provided by NGB. Because the NGB terrain
data very closely represents the real terrain viewed by the
live players, it was suitable for correlation testing with the
JCATS terrain. To compare the terrains, we synthesized
OpenFlight databases from the NGB data and from the
JCATS database for use in Stealth Observer. The LOS fans
of the pop-up targets on the two terrains proved to be well
correlated. If the terrains had not been well correlated, we
would have modified the JCATS terrain with the JCATS
Terrain Editor.
Annotated terrain. In DFIRST, company graphics are
created and displayed on a 2D map. We used the rapid
terrain generation capability of our toolkit to replicate these
graphics on the 3D terrain.
HUM-1/HUM-72. DFIRST prototype HUM-1/HUM-72
instrumentation allowed HMMWVs to play as M1A1s and
T-72s in the JTEP scenario. Although a HMMVW
equipped with a surrogate weapon on its roof has obvious
physical and optical differences from an M1A1 or T-72, the
Guard tank crews who participated in the exercise reported
favorably on the value of the instrumentation for maneuver
training.
4.6 Lessons Learned
The preparation leading up to the demonstration provided
most of the lessons learned. The demonstration itself went
very much according to plan.
System analysis. The JTEP System Analysis identified and
evaluated alternatives for LVC systems and integration
mechanisms. The analysis started in September 2002 and
was presented 5 months later on 12 February 2003. At that
time, the California National Guard gave approval to some
recommendations and selected others from among
alternatives. The hands-on integration work for the first
demonstration was completed in the following 2.5 months.
The literature study succeeded in identifying the key
interoperability issues for testbed focus, so we were able to
proceed rapidly and with little wasted effort. We also
benefited from on-site visits and discussions with many
people who are currently working on interoperability
projects, including the National Simulation Center (NSC),
the CCTT program office at PEO-STRI, the Ft Knox
Simulation Center, and TRAC-WSMR. Visits to DBST at
Ft. Hood in November and December 2002 were an
especially rich source of information about real-world
simulation interoperability issues and solutions.
JTEP testbed. We developed a testbed that allowed all
systems and components to be tested together, so that onsite setup during the week before the demonstration
presented few surprises. Dividing the testbed between SRI
Menlo Park and SRI San Luis Obispo made a particularly
valuable contribution to preparing, both technically and
logistically, for distributed operations.
Small scope. The JTEP program plan calls for a series of
demonstrations, each building on the others in modest
increments. The scope of the first demonstration was
limited to two major systems, JCATS and DFIRST, and
two ancillary components, DIS radios and Stealth Observer.
This scope was small enough to enable us to focus on the
key technical issues of LC interaction, data connectivity,
and terrain with sufficient depth to achieve meaningful and
nearly trouble-free interoperation on the first try. JTEP is
certainly not unique in its approach, but the value of
starting small and then building on a stable foundation is
worth reiterating.
Short development time with DIS. It is well known that
DIS did not achieve the nirvana of simple plug-and-play,
and that interface commonality is not equivalent to
interoperability. When two or more simulation systems
come together, the universality of DIS PDUs and
enumerations does not eliminate the need to adjust for
mismatches in the particular entity types, weapons systems,
environmental effects, etc., that each supports; in
acquisition, hit, and lethality algorithms; and in terr*ain
databases and modeling methods. HLA formalizes a
methodology for making adjustments, but is a starting point
rather than an end point. Nevertheless, DIS solves enough
problems that a 3-month integration effort could achieve
meaningful interoperation between a constructive system
and a live system that were developed independently to
meet the needs of very different aspects of the training
domain.
Network throughput and timing. Predeployment network
testing indicated, and demonstration operations verified,
that GuardNet throughput and latency were adequate for a
small-scale exercise. Peak loading for 25 live and
constructive vehicles and two voice channels was under 180
kbits/s. The DIS radios dominated the network traffic.
Scaling up to a battalion-size exercise in the next JTEP
demonstration will present throughput challenges (see
section 6).
Fidelity and training. The fidelity of the training exercise
was clearly an important consideration in the design of the
demonstration system and the training scenario.
Nevertheless, we concentrated on achieving sufficient
fidelity to achieve the Guard’s training and demonstration
objectives for this particular demonstration. For example,
we judged the JCATS/real world line of sight between the
live and pop-up entities to be sufficiently well matched for
the level of training being demonstrated, and so did not
expend the considerable effort that would be needed to
achieve perfect LOS consistency. Additionally, there were
delays in obtaining specific JCATS hit/lethality tables, so
we approximated these effects between JCATS and
DFIRST systems to yield expected results for M1A1 vs.
BMP-2 engagements. As the scope and training objectives
of subsequent demonstrations expand, we will direct
additional attention to fidelity issues as the scenarios
warrant.
Dead reckoning parameters and live instrumentation.
DIS dead reckoning parameters that are suitable in a
simulation context are not suitable for the live domain. The
MAK VR-Link DIS libraries used in the DFIRST gateway
set 0.5 m as the default dead reckoning for entity state
updates. This value works well for the smooth trajectories
of simulated entities, but it is inside the 2–3 m noise level
experienced in live entity tracking using differentially
corrected GPS instrumentation. Before the parameter was
adjusted, most DFIRST tracks were updated at an excessive
1 Hz or greater rate even under low dynamics.
Voice compression. The DIS radios had several voice
compression algorithm choices, including CVSD at 25 kbps
and muLaw at 64 kbps. Because the 24 kbps compression
used in DFIRST recording of tactical voice nets results in
no noticeable degradation, we hoped that the 25 kbps DIS
recording would be satisfactory even though 64 kbps is
commonly used in the DIS community. However, we found
that SINCGARS voice sent over the DIS radios was
significantly degraded at 25 kbps and therefore we were
forced to operate at the 64 kbps rate.
Avoidance of single-point failures. Every experiment is
susceptible to catastrophic failure in some element of
hardware, software, or coordination. Our past experience
has taught us that, in addition to thorough testing, time
spent obsessing over potential disasters and planning
fallbacks and workarounds is time well spent. During the
JTEP demonstration week, we experienced intermittent
failure in the Camp Roberts network communication
through the wireless LAN/PIX501 firewall. When
troubleshooting under field conditions failed to resolve the
problem, we fell back to a preplanned alternative, i.e.,
transmitting the data between Range Control and DFIRST
via two Freewave Ethernet radios. This method reduced our
network throughput somewhat, but caused no noticeable
degradation to integrated JCATS/DFIRST operations. The
failure was diagnosed and resolved after the demonstration.
database will be stitched to the Camp Roberts data to create
a combined playbox.
Software robustness. New environments are stressful for
software and are likely to uncover vulnerabilities not found
in previous testing and usage. We discovered that the
JCATS DIS bridge, and occasionally the server software,
crashed when it received some types of unexpected data,
e.g., a change in an entity’s vehicle type. Most simulation
systems have no reason to change an entity’s type after
startup, but DFIRST does so when a real vehicle, e.g., an
M1A1, is equipped with its “native” instrumentation set,
and then is guised by a command from the Base Station to
play as another vehicle type, e.g., a T-72. We were
fortunate in that the DFIRST software was under our
control, so that we could easily program a workaround.
However, if we had found a similar problem integrating two
third-party systems, e.g., JCATS and CCTT, obtaining a
quick-turnaround fix would have been more problematic. It
should be noted that DBST uses NSC’s SimC4I Interchange
Module for Plans, Logistics, and Exercises (SIMPLE) to
make just this kind of adaptation.
Correlated terrain. A SEDRIS version of the CCTT
terrain database and the high-resolution Camp Roberts
terrain database will be exported to JCATS format to create
correlated terrain. JTEP has acquired a beta version of the
SAFUSA extension to Terrex’s TerraVista that has the
capability to export terrain to native JCATS format.
Cooperation. Anyone who has participated in a
multisystem, multisite, multiorganization integration project
knows that the logistical challenges of the project are on the
same scale as the technical challenges. We have been
impressed by the professional responsibility and personal
helpfulness we encountered everywhere. But never
underestimate the value of having fun. After completing the
two planned mission rehearsals on the day prior to the
demonstration, some of the SRI team were wishing for a
third run to iron out a couple of kinks in execution. Before
we had a chance to request support from the tank crews,
they asked us if they could go out again after dinner.
5. Next Steps for JTEP
LVC Demonstration. The second JTEP demonstration is a
battalion-size training event scheduled for December 2003.
It is planned to include all elements and sites of the first
demonstration, plus a mobile CCTT platoon suite located in
Los Alamitos, California, and a distributed AAR,
comprising the three training sites and the Office of the
Adjutant General (OTAG) in Sacramento.
CCTT integration. The CCTT mobile suites are scheduled
for continuous deployment during the integration period.
The JTEP testbed will use M2SAF desktop trainers and
CCTT SAF as surrogates supporting integration. We have
been granted on-site access to the deployed CCTT mobiles
for targeted connectivity and milestone tests.
Stitched-terrain playbox. Camp Roberts is not among the
seven current CCTT terrain databases. To create a
contiguous playbox, one the CCTT Grafenfels (P6)
Gateway engagement adaptations. To overcome the
effects of terrain and coordinate system differences in
dissimilar
simulations
(CCTT,
JCATS,
live
instrumentation), we plan to investigate an alternative
engagement resolution approach. Instead of modeling
ballistics and detonation location on target vehicles to
compute damage effects, this approach first assesses these
factors in the gateway. Impact location reported by the
shooter’s simulation system is used only to determine the
intended target (engagement pairing). Then hit outcome is
assessed using probabilistic tables from the Army Systems
Analysis Activity (AMSAA). The gateway reissues the
detonation PDU at the location of the intended target in the
coordinate frame of the target’s “owning” simulation
system. This method of engagement resolution is consistent
with that used in DFIRST and other live systems and may
prove surperior to other means of compensating for terrain
and coordinate mismatches, such as ground clamping of
entities and/or detonations.
Expanded voice communications network. The next
demonstration
will
include
a
Battalion
TOC
communications capability, which will expand the number
of voice channels to approximately 10.
Network bandwidth conservation. GuardNet has many
users and uses, e.g., distance learning activities. Preliminary
analysis indicates that average bandwidth usage during a
JTEP battalion-sized exercise will fall within the amount
that can reasonably be allocated to JTEP, but peak demands
will exceed that amount. Candidate methods for conserving
bandwidth include data compression (e.g., generic, or
voice-data specific), prioritization based on tolerance for
latency (e.g., engagement-related messages and maneuvers
tolerate less latency than ordinary maneuvers), and
receiver-side data caching (e.g., to reduce the size of the
message needed to effect an entity state “heartbeat”).
Distributed AAR capability. During the next
demonstration, command elements at each site will
participate in a distributed AAR. The AAR common view
will include standard video teleconferencing, with
whiteboard, and a 2D/3D playback of the exercise
synchronized with recorded command voice nets. GuardNet
will be the channel for communicating the common views.
Candidate methods for ensuring a high-quality common
view at each site, while staying within the limited GuardNet
bandwidth allocation, are currently being analyzed.
Analysis of LC engagements. The next demonstration
should generate more engagements between more evenly
matched opponents who will be better able to use the full
potential of their combat systems. We expect this change to
result in a useful data set that we can analyze to verify the
accuracy of the engagements and check for “unfair fight”
conditions such as asymmetric lines of site or movement
restrictions.
Upgrade to JCATS 4.1.0. The next demonstration will
utilize JCATS 4.1.0 with the latest patches from JFCOM
applied. This version includes enhanced handling of DIS
version information [27]. Vetted JCATS PH/PK tables will
be used to ensure realistic engagements.
6. Possible Near-Future LVC Integration and
Adaptation Mechanisms
DBST. The incorporation of Digital Battlestaff Sustainment
Trainer (DBST) components and technologies will be
evaluated as a means of extending the system with C4I
capacities.
Entity Attribute Ownership Transfer. It is possible that
no matter how well correlated the underlying terrain
databases become and how cleverly the various gateways
use incoming messages, fundamental differences in the
needs of the different systems may make fair fights between
them impossible. In that case, one solution may be to
temporarily transfer the ownership of entities as needed to
the system best able to resolve interactions, e.g., to generate
a computer-controlled CCTT tank to represent a JCATS
tank in a fight with manned CCTT pods. Such a step is not
feasible for the next demonstration because it would require
software changes in both CCTT and JCATS. In any case,
the incremental-step philosophy followed by JTEP demands
that we first see how far correlated terrain and gateway
engagement simulation adaptations can take us.
7. Summary and Conclusions
JTEP is a multiyear, multiphase program to develop an
LVC-based training system to support California National
Guard training in combat as well as MSCA functions.
Currently in the first phase of its initial year, JTEP has
completed its first demonstration, successfully linking live
and constructive training systems distributed between two
training areas. In addition to achieving the primary JTEP
objectives for this demonstration, we demonstrated LC
direct and indirect fire engagements, the use of a
nondedicated network for overall connectivity, the use of
LC integration and low-cost training instrumentation to
drastically reduce training costs, and the use of virtual
private land as part of a training exercise. Future JTEP
demonstrations will expand on this initial demonstration to
include a virtual system and battalion-level training in the
next phase, and MSCA and joint assets in subsequent
phases. As these phases and demonstrations are complete,
we envision additional lessons learned and opportunities to
build on JTEP as a leave-behind LVC training capability
for the National Guard.
8. References
[1]
Synthetic Theatre of War–Architecture (STOW-A)
1.6 S/W BaseLine Upgrade for Lesson Learned
Report, ADST-II-CDRL-STOWA-900102A, 1999.
[2]
Tiernan, T.R., Synthetic Theatre of War–Europe
(STOW-E) Final Report, Naval Command Control
and Ocean Surveillance Center, 1995.
[3]
National Simulation Center, LVC
Environment Master Plan, draft, 2002.
[4]
Common Training Instrumentation Architecture,
Version 1.0, STRICOM, 2002.
[5]
Powell, Edward, Range System Interoperability
using TENA and the IKE 2 Middleware, CTEIP
Foundation Initiative 2010, 2002.
[6]
Bowers, Andy and David L. Prochnow, JTLSJCATS: Design of a Multi-Resolution Federation for
Multi-Level Training, MITRE, 2002.
[7]
Conflict Simulation Laboratory, JCATS Algorithm
Manual, Lawrence Livermore National Laboratory,
draft, 01 March 2003.
[8]
U.S. Army, M&S System Summary–DBST, 2002.
[9]
McHale, Peter, and Starks, Eric, CCTT Gateways,
Lockheed Martin Information Systems, 2002.
[10]
Callahan, Bob, Interoperability Challenges Between
CCTT and JCATS: A Theoretical Study, White
Paper to Project Manager Combined Arms Tactical
Trainer, 2/26/2003.
Training
[11] IEEE Standards Committee on Interactive Simulation,
IEEE Standard for Distributed Interactive
Simulation–Application Protocols, IEEE Std 1278.11995.
[12] IEEE Standards Committee on Interactive Simulation,
IEEE Standard for Distributed Interactive Simulation
― Application Protocols, IEEE Std 1278.1a-1998
(Supplement to IEEE Std 1278.1-1995).
[13]
Defense Modeling Simulation Office, High-Level
Architecture Interface Specification, Version 1.3,
1998.
[14]
EPFL, Geneva, IIS, Nottingham, Thomson, TNO,
Review of DIS and HLA Techniques for COVEN,
ACTS Project N. AC040, 1997.
[15]
Dingel, Juergen, David Garlan, and Craig Damon,
Bridging the HLA: Problems and Solutions, Sixth
IEEE Workshop on Distributed Simulation and Real
Time Applications, 2002.
[16]
Pullen, M., M. Myjak, and C. Bowens, Limitations
of Internet Protocol Suite for Distributed Simulation
in the Large Multicast Environment, The Internet
Society Network Working Group, RFC 2502.
[17]
TRADOC, US Army, DIS Versions of Janus, White
Sands Missile Range, 1998.
[18]
Burks, Terrell, Alexander, Tom, Lessmann, Kurt,
LeSueur, Kenneth, Latency Performance of Various
HLA RTI Implementations, 2002.
[19]
Noseworthy, J. Russell, IKE2—Implementing the
Stateful Distributed Object Paradigm, 5th IEEE
Symposium
on
Object-Oriented
Real-Time
Distributed Computing, 2002.
[20]
Wood, Douglas, Implementation of DDM in the
MAK High Performance RTI, MAK Technologies,
2002.
[21]
Myjack, Michael, Lake, Tom, and Roberts, David,
Timing: Mechanisms for Ownership Transfer, 1999.
[22]
LaVine, Nils, Kehlet, Robert, O’Connor, Michael,
and Jones, Dennis, Transferring Ownership of
ModSAF Behavioral Attributes.
[23]
Richbourg, Robert F., Robert J. Graebener, Tim
Stone, and Keith Green, Verification And Validation
(V&V)
of
Federation
Synthetic
Natural
Environments, IITSEC 2001.
[24]
Fortin, Michael Rita Simons, and Michael
Butterworth, Database Interoperability–The CCTT to
AVCATT
Conversion
Experience,
Interservice/Industry
Training
Systems
and
Education Conference, Nov 2002.
[25]
Ceranowicz, Andy, Torpey, Mark, Helfinstine, Bill,
Evans, John, and Hines, Jack, Reflections on the
Joint Experimental Federation, 2002.
[26]
Balci, Osman, “Verification, Validation and
Accreditation of Simulation Models,” Proceedings of
the 1997 Winter Simulation Conference (Atlanta,
GA,
7-10
December).
IEEE,
1997.[27]
JCATS Release Notes Version 4.1.0, Lawrence
Livermore National Laboratory Conflict Simulation
Laboratory, 1 March 2003.
Author Biographies
REGINALD FORD, Software Development Manager at
SRI International, has 23 years of experience in test and
training range instrumentation systems for the Army, Navy,
Air Force, and Marine Corps. He helped establish SRI’s
Software Engineering and Development Center. He
manages DFIRST software development and the integration
of JTEP software systems and components.
JOHN SHOCKLEY, Senior Research Engineer at SRI
International, has 19 years of experience in test and training
range instrumentation systems for the Army, Navy, and Air
Force. He began working on modeling and simulation
aspects of these systems and has since participated in
DIS/HLA standards development activities for over 11
years, concentrating on integrating live and virtual systems.
He is the JTEP project leader.
MICHAEL BEEBE, Research Engineer at SRI
International, has experience in systems engineering,
remote sensing, GIS, and terrain database creation. In
addition to being active in the development of DFIRST for
the National Guard, he has worked with the Special
Operations community. His primary JTEP responsibility is
synthetic natural environments.
COL (Ret.) JOHN BERNATZ, JTEP Program Manager
for the California National Guard, is a retired armor officer
who has commanded armor and cavalry units through
brigade level. He has over 33 years of experience as an
Army trainer and training manager and was the Operations
Officer for the 40th Infantry Division, Mechanized, during
the Los Angeles riots of 1992 and the Northridge
earthquake of 1994.
MARK FAUST, Research Engineer at SRI International,
has been working on test and training range instrumentation
systems for the Army, Navy, and Marine Corps for 10
years. He is a lead DFIRST software engineer and has
extensive experience testing radio data links. His JTEP
work includes developing software for DFIRST
interoperability with external simulation systems.
MARK JOHNSON, Senior Software Engineer at SRI
International, has over 20 years of experience in
engineering and simulation and is a member of SRI's
Software Engineering and Development Center. On the
JTEP project he concentrates on JCATS, CCTT and the
interoperation of these systems.
GERALD V. LUCHA, Principal Engineer at SRI
International, has worked for 28 years on a wide variety of
SRI projects related to instrumented test and training ranges
of the Army, Navy, and Air Force. His experience includes
not only studies and analysis of range requirements,
instrumentation concepts, and performance, but also on-site
assessments of numerous land, air, and sea combat training
ranges in the United States and abroad. He currently serves
as the DFIRST Chief Engineer and was responsible for
assuring network connectivity in the field for JTEP as well
as the Camp Roberts end of the demonstration.