Executive Summary

Transformations for the Future – The United States is being driven to transform its military forces to take advantage of new intelligence technologies. The U.S. must also be prepared to deal with emerging asymmetric warfare threats and many other non-traditional types of aggression. This major transformation requires new analytical capabilities to explore new concepts. These capabilities would identify those programs that would provide breakthrough capabilities as well as those that would manifest debilitating problems in future operations.

We have learned from costly experience that our new technologies tend to produce such large amounts of data the an improved ability to filter and fuse the data is required to give the user just that data needed without overloading that user.

Experimentation into new intelligence technologies requires simulation because it is not practical to conduct live exercises before even the early prototypes of the new intelligence devices are available. Even when these prototypes do become available, it is demonstrably not prudent to expend lives, material, and operational funds to use live forces in all of the experiments. Live experiments also tend to be fettered by constraints that curtail the scope of the activity and preclude sufficient repetitions.

Repetitions of the same scenario in the proposed computer simulations produce more valid and extensive distributions of possible outcomes. Slightly altering situations or reactions in subsequent exercises gives the experimenter a way of assessing the best course of action for varied situations. Without these new analytical capabilities, many of these necessary data sources will be lost and policy decisions will be much more difficult to make.

Although many kinds of simulations are important, Human-in-the-Loop (HITL) simulations are particularly important because they permit humans (the warfighters) to interact with the future via a computer-generated virtual reality, a synthetic battle space. It is also important that HITL simulation be of high fidelity and resolution, i.e. represent combat actions at a realistic entity-level of detail. This fidelity and resolution is necessary because we need to see the individual interactions of the new technologies, new platforms, and new concepts with discrete battlefield entities such as tanks, vehicles, aircraft and even individual personnel.

The HITL virtual reality simulation will require the improved simulation of human behavior for the combat entities, the intelligence operators, and for the Command and Control (C2) nodes. In a virtual, large-scale battlespace, there will be too many C2 nodes for humans to economically staff all of the nodes, so we need to model the behaviors of many of these nodes in software. However, many simulation professionals would observe that, as of today, this is a capability that needs to be enhanced if it is to be useful.

Any significant delay in resolving these issues may be incur catastrophic losses on the battlefield later. Thus, it is important to identify the technologies available for the simulation of the new intelligence capabilities and of human behaviors as soon as possible. This should allow decision makers in the government to make the most effective choice of technologies to pursue and to have documented support for their choices.

Impact of New Technology – These are not merely theoretical issues. Large sums of money, the lives of our citizens, and the power of the United States are all dependent on our adopting the new technologies and using them well.

The impact of the new style of combat, called Network Centric Warfare (NCW), is described by Vice Admiral Cebrowski as “... a revolution in military affairs (RMA) unlike any seen since the Napoleonic Age,...” (Cebrowski & Garstka, 1999.) Both Joint Vision 2010 (CJCS, 1996) and Joint Vision 2020 (CJCS, 2000) cite the necessity for Information Superiority to enable the achievement of Dominant Maneuver, Precision Engagement, Full-Dimensional Protection, and Focused Logistics envisioned by those documents. Supporting publications further illuminate that concept, especially in terms of joint operations. (CJWC, 1997)

Virtual reality and simulations have the capability of giving the U.S. an experience with this otherwise unknown future combat environment. Envisioning the future is a province for both technical and creative personnel. Military history seems to support the thesis that ascendancy comes more to the group that accepts and masters the new technologies than to the first to acquire it, e.g. the French had more tanks than the Germans at the beginning of World War II, but the Germans had learned to use them more effectively.

DoD Needs – The optimal use of newly-developing intelligence capabilities in future Joint Operations and in Network Centric Warfare battlespace depends on new technology, new tactics, and new operation’s concepts. It will challenge the ability of humans to work in an information-rich environment.

Recent experience in the Gulf War showed that there was much more information available to the warfighters than could be internalized by them at all levels of command. Technology had created a “fire hose” of information that made it difficult for those needing the information to ferret out the facts they needed.

Simply adding new sensors and systems without analyzing how these new sources of information will fit into the duties of the future military leaders will not provide the benefits sought. A well-designed and rigorously validated simulation provides an excellent test-bed on which to analyze the value of the new sensor or system. It also would be useful in early planning and training for the use of these system in combat.

War Games and Simulation History - This is not the first time in our history that we have been faced with the problem of how to adapt to new technology. The rush to accept new technology during times of war has been contrasted with the invariable reticence to adopt untried systems between major conflicts. Often, these shortsighted peacetime policies were very costly later when the new technologies were not available to combat personnel at the beginnings of the next conflicts.

The U. S. defense establishment has previously benefited from war games, simulation, and experimentation to create anticipated environments before they came to exist in reality. Many of the experiments were “live,” field exercises. The service war colleges conducted some significant number of the experiments as “board” exercises. These war games and their progeny were proven useful in examining many well-known military issues in the past. Well-documented efforts were conducted to plan, analyze, and train for those past era’s future operations. We can, and we must, learn from the successes and failures of these past activities.

Current Requirements - New capabilities and changing enemies present a completely new environment in which our armed forces will be forced to operate. Intelligence activities in Joint Operations and NCW present a number of challenges and issues, many of which appear to be amenable to effective resolution by the experimentation we are proposing.

While very important to success, emulation of Command and Control entities is still in the early stages of development. Some innovative and promising work has been done by a group of SOAR programmers at the University of Southern California. (Hill, 2000) Both the capability to generate the appropriate command decisions and to model higher-level headquarters (battalion and above) require dramatic advances to meet DoD and other government intelligence needs.

Human Behavior Research – Lately, significant new research has been done on modeling the human behavior of decision-makers. Research efforts in the behavioral sciences, education, computer science, and computational science all contribute to new capabilities that should enable types and sophistication of experimentation not yet implemented. Recognizing the most promising techniques and then identifying the most appropriate approach would allow DoD analysts, trainers, and commanders to make a significant improvement in future research and development.

Supercomputer Capabilities - To the computational scientists at their research supercomputing centers, these profuse and multifaceted battlespace simulation would seem similar to the characteristics in the complex physical science problems that frequently are brought to them for analysis or modeling. The personnel at these centers have years of experience in successfully adapting existing serial computer programs to high performance parallel computers. (Fox, et al., 1994)

This experience is an invaluable asset for the U.S., if we are to direct our resources and limited time to the best advantage. It is clear that experiments that simulate an entire Theater-of-War are especially useful when analyzing new sensor systems intended for use on that scale. Simulation at that scale requires the power of SPP supercomputing. (Messina, et al., 1997)

For the DoD to effectively simulate new sensors, platforms, techniques, and human behaviors in large-scale experiments, a number of issues present themselves. For these simulations to be suitable for use by the intelligence community, by the sensor researchers, and by the nations’ leaders, we must quickly identify which of the issues need to be resolved. The resolutions of the critical issues will allow us to ascertain if these needed analytical capabilities are all achievable by the use of today’s emerging technologies. We can then establish the most promising development path for their effective implementation. Their resolution will allow us to calculate the cost-savings and defense-enhancing capabilities of the NISC.

Below, several important issues are suggested for consideration:

Available Now- The immediate need to gain insight into the anticipated command and intelligence activities in the NCW battlespace precludes the launching of a new multi-year search for the mere promise of an improved analytical tool. Commanders and policy-makers need to know, now, what they may face on the next battlefield. Analysts need access to joint experimentation to best prepare for heretofore-unexperienced contingencies. Trainers for intelligence personnel need to experience the future reality as soon as possible, so that they can design the best training for the young men and women who will be sent “in harms way.”

What is the best, most reliable, and quickest path for producing useful analyses for DoD decision-makers?

Entity level – Entity level experimentation provides a number of desirable capabilities that will be required for high-resolution, computer-generated models of vehicles, sensors, and human behaviors. Many experts would argue that adequate experimentation is impossible without the ability to simulate and control these individual entities.

Alternatively, constructive or aggregate simulations present an additional layer of abstraction between the sensors and their performance in a simulated battlefield environment, with the heightened likelihood that important impacts will be over-looked. It is therefore desirable to implement current entity-level simulation technology on a large-scale battlefield test-bed.

The Artificial Intelligence (A/I) entities utilize Monte Carlo techniques for outcome determination. That means that the computer uses a random assignment of outcomes within the identified range of possibilities, e.g. if actual combat experience indicates that 50% of landmine hits result in broken tank treads, when a simulated tank runs over a simulated mine, the computer will randomly indicate a broken tread 50% of the time.

The program and the entities must also provide an easily-accessed and credible interface for the HITL players. The analysts and the exercise participants typically have several alternate views of the battlefield from which to chose.

Typical combat scene from a synthetic battlespace.

As the visualization is driven by the each entity’s position and status information, the entities also must have the ability to effectively communicate all of the required visualization input data.

What is the nature and quantification of the desirability of using entity-level, HITL simulations and how can existing and anticipated computational science serve these needs?

Realistic Communications’ Environment – Existing work on the automatic generation of communication traffic will need to be expanded and current techniques will need to be improved to create a realistic information-space in which future intelligence and command behaviors can be simulated and assessed. The command nodes and intelligence cells contemplated for this use function almost entirely from communicated intelligence and status reports.

Can an improved simulation environment effectively generate, simulate, archive, and analyze communications traffic in a Joint Operation on an NCW battlespace?

New Battlefield Effects – We would prefer that our command nodes and intelligence sensors function flawlessly in the battlespace of the future. In preparing for this future, the simulated capabilities of anticipated new sensors and simulated effects of the new weapons have a significant impact on the utility of the experimentation. The requirement for an increasing number of complex simulated battlefield effects will be of great importance.

Sensors with new capabilities will also be faced with new limitations and new environmental conditions that degrade their performance, e.g. Synthetic Aperture Radar (SAR) may see through clouds, but certain types of smoke may obscure the image at certain frequencies. Additional programming of the relevant modules of the existing entity-level simulations should be able model these environments.

One of the aspects of human behavior that is of the greatest potential interest is the reaction of commanders at both the senior level and the intelligence-cell level to this radically changed battlefield. The effects that the HITL participant experiences must be realistic and engage the participants at an emotional level approaching the actual experience.

How can we best simulate asymmetric attacks, information warfare, Weapons of Mass Destruction (WMD), sensors, and other new developments to effectively analyze sensor performance and human behaviors on future battlefields?

New Sensors Models – Sensors are expected to play an increasingly decisive role for commanders on the battlefield of the future. Simulated sensors need to have a range of realistic simulated targets that are operating on appropriate terrain databases. The simulated command nodes need to input data from these simulated sensors. Either HITL participants or A/I modules will needed to be control the intelligence cells.

What are the potential new system-wide requirements to effectively simulate the sensors and command entities and to simulate the interfaces resulting from the not-yet-conceived sensors/command nodes?

Scalable - Vice Admiral Cebrowski notes that “Specific top-down experimentation will be required, because of cost and size, to establish overarching priorities, but these are expected to spawn experiments from the bottom up and facilitate cultural and organizational changes.” (Cebrowski & Garstka, 1999.)

The simulation system needs the ability to easily represent the smallest military units and then to seamlessly scale up to represent Theater of War scenarios. Demonstrable scalability should be a key element in the design, including the ability to allow a small experiment to participate in an on-going, larger evolution, so that small experiments are not being conducted in an operational vacuum.

This map view of a simulation demonstrates the ability to represent small-scale actions.

To make the experiment useful, sufficient entities must be available to effectively model higher-level command nodes, intermediate command organizations, new wide-area sensors, intelligence cells, and individual combat units. To make the dual approach argued for by Adm. Cebrowski economically feasible and technically practical and adequately scalable, NISC must have easily structured SPP computer simulations and capabilities for even larger metacomputing to provide experiments to cover the ranges of needs of the users.

Can the scalability concepts already spawned and nurtured in the U. S. research university supercomputing centers be adapted and transferred to military simulations without “reinventing the wheel?”

Scale of NISC Operations – If it were a combined military and civilian effort, the NISC envisioned by this paper might be well served by an assigned military staff of approximately four officers and six enlisted personnel. The larger civilian staff would plausibly be divided into three divisions: research, simulation operations, and administration/logistics. This civilian staff would be effective at about twenty-eight members, with at least six of these being Ph.D.s.

For a physical plant, we would anticipate needing a facility on the order of 25,000 square feet, with advanced computers, display devices, offices, conference rooms and operational spaces. Close proximity to adequate “surge housing” for exercise participants would be desirable. This housing could be either nearby military bases or civilian accommodations to support exercise participants needs.

In computing needs, we would want adequate state-of-the-art visualization and high band-width, secure communications capabilities. However, virtually all of the development and small scale computing could be accomplished on PC’s and using PC cluster technology.

What are the best choices in terms of staffing, facility design, location, computer equipment, and organizational makeup?

There are a number of major universities, Federally Funded Research and Development Centers (FFRDCs,) and other research centers that have developed capabilities that would be useful for NISC. Collaborative research is a virtual sine qua non of effective research in the United States today. The journal articles and other research cited at the end of this paper are from researchers from many of the institutions upon which NISC would rely: Caltech, USC, JPL, Institute for Defense Analyses, NRAD, Information Sciences Institute, University of Michigan, Institute for Creative Technologies, and others.

One of these research centers that has an abiding interest in the future development of a national intelligence test bed is the Institute for Creative Technologies (ICT) at the University of Southern California, in Marina del Rey. ICT is a University Affiliated Research Center (UARC) and it is primarily focused on training for the battlefield of the future. Importantly, it has a special approach to projecting the future of technology and how this will impact combat. ICT was formed to “… advance the state of immersive training simulation.”

It is ostensibly unique because it is based on a “hybrid approach” that synthesizes the capabilities of the U.S. entertainment community and the academic computing community to create realistic and futuristic combat and non-combat training environments with high emotional engagement for the user.

Using this mix of factually based creativity and visionary science, ICT has embarked upon a course of creating a virtual reality representing the battlefield of 2020 and beyond to use for training personnel.

ICT conceptualizations of future warfare are based on input from the Army.

Researchers - Many of ICT’s researchers came from its sister research center at the University of Southern California, the Information Sciences Institute (ISI), where they had worked on the ModSAF, SF Express, and SOAR projects. These successful efforts included developing Command FORces (CFOR) modules (Hill, 2000). This collection of researchers has documented experience and proven effectiveness in this field. They literally are an invaluable asset for attacking and resolving the problem of the ever-increasing sophistication needed to adequately model higher-level command functions. Personality modeling is a specialty of the Psychology Department at USC and a collaboration to pursue their insights is also under way.

In addition, at ICT, there is resident expertise in large scale computing. This is sometimes referred to as Scalable Parallel Processor (SPP) supercomputing and it utilizes thousands of processors to enable simulations of more than 100,000 battlefield entities (Messina, et al., 1998). This expertise would be useful during these formative stages to ensure simulation design decisions made at the small unit level do not unduly inhibit the subsequently necessary parallelization of that design at the Division or Theater of War level. The ability of the large supercomputing centers to lead the workstation capabilities in power has been estimated at equal to about ten years. This is a critical decade when viewing the stakes of optimizing our defense strengths rather than relying on the sacrifice of the warfighters.

ICT is also home for specialists in the field of ethnography, the study of human behavior in specific cultural settings. These techniques have been effectively used in studying comparable military situation such as small unit command issues. (Ben-Ari, 1998) Some non-military studies are also insightful, such as a similar civilian environment like air traffic control (Sanne, 1999).

One recent innovation may greatly reduce knowledge acquisition time. It is a technique incorporating observation and explicit goal determination into the important process of documenting and understanding the behavior of intelligence personnel in the simulated combat environment (van Lent, 1998.)

The staff members at ICT have also established a good relationship with the Naval Postgraduate School (NPS), in Monterey California. We intend to collaborate with members of the faculty and graduate students in Operations Research and other departments to apply their techniques to this work. These NPS graduate students are mid-grade officers from all of the services and they will bring a “war fighter’s” perspective to the research. Further, they will carry their understanding and enthusiasm for this approach back to the services for the rest of their active duty careers.

For all of these reasons, a National Intelligence Simulation Center should be funded with sufficient monies to establish a facility of its own. It should be provided with a suitably senior military commander to act as champion and it should be provisioned with a research staff of world-class scientists. Such a center would undoubtedly become an incubator of technologies that would evolve and transfer into the civilian sector with resulting economic benefits. NISC could easily become the beacon for international interest in large-scale computer simulation and experimentation in human behavior and organizational management.

One attractive configuration would allow for a continuous simulation with the scale of a Theater of War battlefield, into which subscriber researchers, analysts, and policy makers could remotely log and make assessments on new concepts and models. NISC would do the programming to provide the models requested by the user or alternately provide guidance and support so the user could program their own compatible models. Specific scenarios and conditions could be scheduled at the users discretion and expense. Data transfers from and to users in these modes would not require unusually large bandwidths, but would often require very high levels of security.

We would argue that this activity should be a new, independent, and joint-service activity. If this activity is attached to an existing command or office, much of its impact will be lost in the existing focus on their on-going research and the day-to-day demands of the other activities at that command or office. If it is attached to an existing research initiative, then vested and installed interests there will preclude achieving the most innovative and focused approach that is desirable for the simulation of future battlefields for use by intelligence technology developers. If the activities that NISC should host are instead “farmed out” to whatever command will accept them in any given year, much of the energy of the personnel will be expended in finding and preparing new space and organizing the logistics required for such experimentation.

The major criteria of the site selection for the NISC should include ease of travel, provision for good security, a staff retention-enhancing environment, and adequate support of all kinds. It would profitably be co-located with, but not subservient to, one of the significant high-performance computing facilities. Large-scale simulation and experimentation will require some of the best supercomputing available in the U.S., as the NISC research, development, and training endeavors are very well suited to SPP supercomputers.

Institutionally, it would seem logical to have the Center be part of the Director of Research and Engineering (DDR&E) establishment, with a close and collaborative relationship with the Defense Advanced Research Projects Agency (DARPA). The commander’s seniority should be sufficient to meet the goals of the NISC.

With outer space, new technologies, and new threats posing many questions for defense and legislative decision makers a more focused simulation effort is called for. The issues, opportunities, and perils of the use of space for defense purposes are very complex. NISC would provide an aid to better visualization of the future and would allow this nation’s leaders to focus on the broader issues.

The questions to be addressed are characterized by their importance and by their immediacy:

· Which sensors systems?

· How many and how effective?

· Which platforms are best?

· At what cost?

· What support is needed?

· Can they be effectively used?

The intelligence community and this country need a National Intelligence Simulation Center.

Author

Dan M. Davis, is the Director of the Joint Experimentation on Scalable Parallel Processors for the Joint Forces Command. He was previously the Director on the staff of the Maui Supercomputing Center, University of Hawai'i. Prior to that, he was an Associate Director, Information Sciences Institute, University of Southern California, and he has been active in large-scale distributed simulations for the DoD and in researching supercomputer uses in education. While he was the Assistant Director of the Center for Advanced Computing Research at the California Institute of Technology, he managed Synthetic Forces Express, a multi-year simulation project of Caltech, JPL and ISI. An active duty Marine Cryptologist, he currently holds a U.S.N.R. commission as a Commander. He has served as the Chairman of the Coalition of Academic Supercomputing Centers and the Coalition for Academic Scientific Computation. He received a B.A. in 1973 and a J.D. in 1975, both from the University of Colorado in Boulder. dandavis@isi.edu, (310) 448-8434.

Definitions

- Joint Force is a “... general term applied to a force composed of significant elements, assigned or attached, of two of more Military Departments, operating under a single joint force commander.” (Joint Pub 1-02, 2000)

- Metacomputing is the networking of supercomputers together using high performance networks. This requires a set of efficient communications management programs. (Foster & Kesselman, 1997)

- Network Centric Warfare is defined as “...networking sensors, decision makers, and shooters to achieve shared awareness, increased speed of command, higher tempo of operations, greater lethality, increased survivability, and a degree of self-synchronization. ... by effectively linking knowledgeable entities in the battlespace.” (Alberts, Garstka & Stein, 1999)

- Scalable Parallel Processor (SPP) computers are that class of supercomputer that use many, sometimes thousands, of high performance processors to accomplish an otherwise unobtainable compute power. Scalability is achieved through sophisticated hardware and software that allows the user to get a proportional increase in compute capability with each additional processor used. (Fox, Messina & Williams, 1994)

References

Alberts, D.S., Garstka, J.J., & Stein, F.P., (1999), Network Centric Warfare, 2^nd Ed., DoD C4ISR Cooperative Research Program: Washington, D.C.

Ben-Ari, E. (1998). Mastering Soldiers: Conflict, Emotions and the Enemy in an Israeli Military Unit. New Directions in Anthropology, V. 10. Oxford: Berghahn Books.

Cebrowski, A.K., & Garstka, J.J., (1998), Network Centric Warfare: Its Origin and Future, Naval Institute Proceedings, 124/1, 28-35.

CJCS, (1996), Joint Vision 2010, Chairman of the Joint Chiefs of Staff, Washington, D.C.

CJCS, (2000), Joint Vision 2020, Director for Strategic Plans and Policy, J5: Strategy Division, Washington, D.C.: U.S. Government Printing Office

CJWC, (1997), Concept for Future Joint Operations, Commander, Joint Warfighting Center, Fort Monroe, VA.

Foster, I. & Kesselman C., (1997), Globus: A Metacomputing Infrastructure Toolkit, Intl J. Supercomputer Applications, 11(2): 115 –128

Fox, G., Messina, P.C. & Williams, R. D., (1994), Parallel Computing Works!, New York: Morgan Kaufmann Publishers.

Hill, R. W., Gratch, J., & Rosenbloom, P.S., (June, 2000). Flexible Group Behavior; Virtual Commanders for Synthetic Battlespaces. Proceedings of the Fourth International Conference on Autonomous Agents, Barcelona, Spain.

Joint Pub 1-02, (2000), Department of Defense Dictionary of Military and Associated Terms, Chairman of the Joint Chiefs of Staff, Washington, D.C.

Messina, P. C., Brunett, S., Davis, D. M., Gottschalk, T. D., (1997, April) Distributed Interactive Simulation for Synthetic Forces, In J. Antonio, (Chair), Mapping and Scheduling Systems, International Parallel Processing Symposium, Geneva, Switzerland.

Sanne, J. (1999). Creating Safety in Air Traffic Control. Unpublished doctoral dissertation, Institute of Tema Research, Linköping University, S-581 83 Linköping, Sweden.

van Lent, M. C. & Laird, K. E., (1998). Learning by Observation in a Complex Domain. Proceedings of the Knowledge Acquisition Workshop, Banff, Canada.

A National Intelligence Simulation Center

University of Southern California

Executive Summary

Immediate Concerns and Needs

Modeling and Simulation

Issues for Effective Modeling and Simulation for Intelligence Analyses

Research Institution Background

Conclusion