Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Military Missions and How IOS Models Can Help C omputational modeling and simulation (M&S) technology have long been useful military tools, although these models have focused primarily on physical effects, such as the predicted capabilities of sensors or weapons systems. Today, the changing nature of military mis- sions is driving the need for new types of computational models that focus on human behavior, specifically on human behavior in social units, such as organizations and societies. The military has traditionally made use of computational modeling in three broad areas of activity: 1. Analysis and forecasting for planning. Models are used for the fusion of fragmented and incomplete information about enemy activities and capabilities. For example, models of enemy equip- ment can be used to interpret fragmentary data on the performance of that equipment (e.g., what capabilities in the equipment could have resulted in the observed performance). Forecasting models are used to develop courses of action (COAs) based on the desired out- comes and their estimated likelihood of achieving those outcomes. At a simple level, for example, models are used to forecast the effectiveness of different types of weapons against different kinds of targets. 2. Simulation for training and rehearsal. Models are used in simula- tions that create training and rehearsal environments. For example, pilots practice complex and dangerous combat maneuvers in simu- lators before encountering them in exercises or combat, and tank 23
24 BEHAVIORAL MODELING AND SIMULATION commanders practice ground combat missions before an actual engagement. In both situations, considerable effort goes into modelÂ ing the environment (e.g., aerodynamics and terrain), simulating the dynamics of the friendly and enemy sensors and weapons Âsystems, and providing the critical performance feedback to Âtrainees needed for skill improvement and âlearning to criterion.â 3. Design and evaluation for acquisition. When a system is designed, built, and acquired, models are used throughout the process to pre- dict performance and make design decisions based on cost-benefit trade-offs. For example, detailed physical and electronic models can be used to predict the additional range of a sensor accruing from a proposed enhancement (and increased cost), to support a cost-benefit trade-off. In this chapter we argue that the successful performance of all three of these activities in todayâs military environment requires not only the traditional set of physically based models and simulations now used, but also computational models of human behavior, particularly computa- tional models of human behavior in social units. We begin by describing todayâs changing military missions in order to explain whyâin the current e Â nvironmentâanalysis, planning, training, and acquisition require models of human behavior at many levels: at the individual level, at the team or organizational level, and at the societal level. We then give specific examples of how these individual, organizational, and societal (IOS) models could be used by the military. Finally, we briefly review current military IOS modeling efforts and summarize the major challenges involved in meeting current needs. Subsequent chapters provide a broader review of state-of- the-art IOS behavioral modeling approaches, assess the extent to which those approaches have the potential to meet military needs, identify major shortfalls and gaps, and recommend a plan of action to address them. Military Missions Now and Into the Future This section reviews the changing nature of todayâs military missions to explain why effective forecasting, training, and acquisition require com- putational IOS models. Overarching Strategy and Operational Enablers The changing nature of current and future military missions is made quite explicit in the Department of Defenseâs (DoD) Quadrennial Defense Review (U.S. Department of Defense, 2006). Coming out of a long tradition of âattrition-basedâ conventional warfare and backed ultimately by nuclear-
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 25 based mutual assured destruction (MAD), DoD is now undergoing a shift of tectonic proportions to operationalize the National Defense Strategy of âfighting the long warâ and has identified five critical operational enablers: 1. Defeating multinational multiethnic terrorist networks that âseek to break the will of nations that have joined the fight alongside the United States by attacking their populationsâ and âuse intimi- dation, propaganda and indiscriminate violence in an attempt to subjugate the Muslim world under a radical theocratic tyrannyâ (U.S. Department of Defense, 2006, p. 13). 2. Defending the homeland in depth against both terrorist networks and hostile states with weapons of mass destruction (WMD) capa- bilities. Globalization enables âthe spread of extremist ideologiesâ and âthe movement of terroristsâ and âempowers small groups and individualsâ with the result that ânation-states no longer have a monopoly over the catastrophic use of violenceâ (U.S. Department of Defense, 2006, p. 36). 3. Shaping the choices of countries at strategic crossroads to protect the âfuture strategic position and freedom of action of the United States, its allies and partnersâ by shaping the choices of âmajor and emerging powers . . . in ways that foster cooperation and mutual security interestsâ (U.S. Department of Defense, 2006, p. 39). In addition to the Middle Eastern region, countries of particular con- cern are India, China, and Russia. 4. Preventing the acquisition or use of WMD by hostile states (e.g., Iran) or nonstate actors (e.g., Osama bin Laden). âBased on the demonstrated ease with which uncooperative states and non-state actors can conceal WMD programs and related activities, [we] must expect further intelligence gaps and surprisesâ (U.S. Depart- ment of Defense, 2006, p. 45). 5. Refining DoDâs force planning construct for wartime to move grad- ually from a two-front conventional campaign capability to more loosely defined âdistributed, long-duration operations, including unconventional warfare, foreign internal defense, counterterror- ism, counterinsurgency, and stabilization and reconstruction opera- tionsâ (U.S. Department of Defense, 2006, p. 36). This is a remarkable shift in emphasis since the terrorist attacks in the United States on September 11, 2001, and may very well be a turning point away from more than 50 years of conventional force planning (backed by MAD) and the start of a much more agile and indigenously sensitive force. The United States is no longer fighting nation-states using conventional weapons but instead is fighting a very different kind of organizationâ
26 BEHAVIORAL MODELING AND SIMULATION t Â errorist networksâin a battlespace in which effects may be defined by the attitudes and behaviors of civilian noncombatants rather than by bombs on targets. In order to analyze, plan, train, and acquire effective technology for this new battlespace, models are needed to help people understand and interpret fragmentary information about terrorist activities and understand the likely effects of U.S. actions on the attitudes and behaviors of diverse multicultural civilian populations. People need to understand the forces that drive individuals to join terrorist organizations, how these organizations function, and how they organize action. People need to understand the factors that contribute to the stability of neighborhoods and regions and how military actions as well as political, diplomatic, and economic actions contribute to that stability. People need to understand complex shifting cultural allegiances and how U.S. actions affect those allegiances. Models of sensor and weapons systems are not adequate tools for fighting this long war. The nationâs defense planners need IOS models that capture the rich- ness of individual, team, organizational, societal, and cultural influences that can help to address the key dimensions of the new battlespace. Dimensions of the New Battlespace In this section we examine some of the drivers of the changing DoD mission to gain insight into what this shift in mission means for IOS model- ing requirements. The Impact of Urbanization One of the key drivers in this shift has been the growing recognition that fundamental world demographics are changing: âThe worldâs urban population reached 2.9 billion in 2000 and is expected to rise to 5 billion by 2030. Whereas 30 per cent of the world population lived in urban areas in 1950, the proportion of urban dwellers rose to 47 per cent by 2000 and is projected to attain 60 per cent by 2030. . . . At current rates of change, the number of urban dwellers will equal the number of rural dwellers in the world in 2007â (United Nations, 2002, Part I, p. 5). The military implica- â note of caution is appropriate here. Although it is true that at the time of this writing the A United States is not engaged in a conventional war, that is not to say that it will not be engaged in one at some point in the future. Thus, there is always the danger that the Ânation will be âpreparing for the last warâ (e.g., todayâs Afghanistan and Iraq campaigns) via a wholesale shift in focus to nonconventional strategies, tactics, and weapons systems. DoD recognizes this, as noted in the fifth âoperational enablerâ cited above (U.S. Department of Defense, 2006, p.Â 36), identifying the desire to âmove gradually [emphasis added] from a two-front conventional campaign capability. . . .â Clearly, the operative issue is how long this transition takes and to what extent it transforms the servicesâ force structure.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 27 tions of this fact are explored in depth in two recent RAND studies (Glenn, 2000; Vick et al., 2002). Key issues and implications that emerge from these studies and others include: â¢ Most, if not all, of the future conflicts the nation will face will have an urban component, based both on historic precedent and on the fact that the adversaries are no match for U.S. forces in âopen fieldâ engagements. â¢ It will no longer be sufficient to avoid urban and surrounding built-up areas during military operations, as has so long been U.S. doctrine. According to a 2002 Joint Chiefs of Staff report, âurban areas are the natural battleground for terrorists: the effects of ter- rorist acts are greater and more noticeable and the terrorist groups more difficult to locate and identifyâ (Joint Chiefs of Staff, 2002, p. III-27). From a âhearts and mindsâ standpoint, there is also a clear political advantage of having a close connection with the noncombatant urban population. â¢ Urban operations are extremely difficult, with the operational environment characterized by high densities and tempos, inherent complexity, and constraints. The battle tempo can be extremely high, forcing rapid assessments, decisions, and actions. Collateral damage issues covering critical infrastructure losses, damage to symbolic edifices, and noncombatant loss of life are critical. Urban operations are also complicated by the fact that mission objec- tives can vary dramatically in both time and space, running from all-out conflict to infrastructure rehabilitation. This spatiotemporal nonuniformity has been referred to as the âthree-block warâ by the former commandant of the Marine Corps, General Charles C. Krulak: âIn one moment in time, our service members will be feeding and clothing displaced refugees, provid- ing humanitarian assistance. In the next moment, they will be holding two warring tribes apartâconducting peacekeeping operationsâand, finally, they will be fighting a highly lethal mid-intensity battleâall on the same day . . . all within three city blocks. It will be what we call the âthree block warââ (Krulak, 1997, p. 139). In these stability and support operations (SASO) stages, it becomes increasingly important to interact with and not alienate the local popula- tion, get their support to identify social networks of adversaries (and poten- tial allies), and anticipate first- and second-order effects (i.e., Â unintended consequences) of actions that are within the scope of the unitâs capabili- ties (i.e., executing a search-and-destroy mission) but that may be highly counterÂproductive in the long run. It also follows that as the mission becomes dictated less by military objectives than by social and political
28 BEHAVIORAL MODELING AND SIMULATION objectives, there is a need to ensure greater interaction with other orga- nizations outside the local unitâs normal sphere of interest. Not only does this imply a greater reliance on joint operations (coordinating the sister services), and increasingly a reliance on coalition (non-U.S.) partners, but it also implies greater interagency coordination, both national (e.g., the State Department, the intelligence agencies, the organs of public diplomacy, U.S.-based nongovernmental organizations or NGOs), international (e.g., sister intelligence services, non-U.S. NGOs), and private-sector economic interests. As a consequence, in order to address and achieve the peacemak- ing objectives in the new theaters of war, planners must somehow consider and assess the aggregated complex interactions of entire social systems, both regional in behaviors and global in influence, at resolutions of fidelity neither needed nor attempted in prior military history. The objectives and technologies of peacemaking in this environment are very different from those of conventional warfare, most notably, a sub- stantially increased emphasis on peacekeeping, disaster relief, and nation- state building (see, for example, the Urban Sunrise study of the Air Force Research Laboratory, 2004). The urban operational environment serves to transform what was once viewed as a strictly military (and tactically diffi- cult) engagement into something that is now considerably more holistic and focuses primarily on social, organizational, and cultural factors involving key individuals, nonmilitary groups, local crowds, and indigenous popula- tions, all within a rich tapestry of a complex local infrastructure overlaid by local, national, and transnational economic markets, organizational and social structures, traditions, cultures, and religious beliefs. The Growing Importance of Pre- and Postconflict Operations The changing nature of military missions is putting increasing focus on operations that occur before and after periods of overt conflict. These pre- and postconflict operations may persist much longer than the conflict itself, as is all too well illustrated by the current situation in Iraq. In the doctrine for Joint Urban Operations (JUO) (Joint Chiefs of Staff, 2002) five phases are recognizedâunderstand, shape, engage, consolidate, and transition (USECT, emphasis added): â hile W the military is the branch of the U.S. government having primary responsibility for projecting U.S. power overseas, it may be a classic case of âmission creepâ for the military to be taking a leading role in economic development, political reconstruction, diplomacy, disaster relief, and intercultural communication. But this is exactly what is happening in todayâs con- flicts, with young servicemen serving effectively as âmayorsâ of Iraqi villages, see http://www. washingtonpost.com/wp-dyn/content/article/2007/01/11/AR2007011101576.html. And this is likely to remain the case until other U.S. agencies or NGOs can take the lead, or the United States successfully transitions these functions back to the local population.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 29 1. Understand: âThe JFC [joint forces commander] evaluates the urban battlespace, including the urban triad [the physical terrain, the urban infrastructure, and the population] and the threat, to determine the implications for military operations. This evalua- tion extends from complex terrain considerations to the even more complex impact of the sheer number of actors operating in an urban battlespace. On one hand there may be adversary military troops, criminal gangs, vigilantes, and paramilitary factions oper- ating among the noncombatant population. On the other hand, especially in MOOTW [military operations other than war], the situation may be further complicated by the presence of nonÂmilitary government departments and agencies, to include intelligence, law enforcement, and other specialized entitiesâ (Joint Chiefs of Staff, 2002, Chapter II, pp. 8-9). 2. Shape: âShaping includes all actions that the JFC takes to seize the initiative and set the conditions for decisive operations to begin. The JFC shapes the battlespace to best suit operational objectives by exerting appropriate influence on adversary forces, friendly forces, the information environment, and particularly the elements of the urban triad. Methods of shaping may include . . . the phased deployment and employment of joint forces. Rather than deploying combat forces initially, the JFC may, in many cases, need to deploy noncombat forces early, such as civil affairs (CA), Â public affairs (PA), medical support, and psychological operations (PSYOP) units. . . . Critical to shaping operations is the isolation of the urban area to support the campaignâ along physical, informational, and moral dimensions (Joint Chiefs of Staff, 2002, Chapter II, p. 11). 3. Engage: âTo engage, the JFC brings the full dimensional capabilities of the force to bear in order to accomplish operational objectives. Engagement can range from full combat in war to FHA [foreign humanitarian assistance] and logistic support for disaster relief operations. It consists of those actions taken by the JFC against a hostile force, a political situation, or a natural or humanitarian predicament that will most directly accomplish the mission. In all cases, the speed and precision with which the JFC engages will largely determine any degree of success. . . . [S]uccessful engage- ment requires . . . the seizure, disruption, control, or destruction of the adversaryâs critical factors,â which include their âcapabilities, requirements, and vulnerabilitiesâ and may include o âtangible components of the infrastructure such as power grids, communications centers, transportation hubs, or basic services.â
30 BEHAVIORAL MODELING AND SIMULATION o âintangible socio-economic or political factors such as financial centers and capabilities, particular demographic groups and sites, and cultural sensitivities.â In addition, âboth offensive and defensive JUOs will probably entail heavy use of IO [information operations] and CMO [civil military operations]â (Joint Chiefs of Staff, 2002, Chapter II, p. 12). 4. Consolidate: âIn war and MOOTW, the focus of consolidation is not just on protecting what has been gained, but also retaining the initiative to disorganize the adversary in depth. . . . Consolida- tion may place heavy emphasis on logistic support and CMO. The nature of the urban triad ensures that the JFC will have to contend with issues concerning physical damage, noncombatants, and infra- structure as part of consolidation. CMO and PSYOP units may continue to be especially critical in this aspect, as well as engineer- ing efforts ranging from destruction to repairs to new construction. Equally important are the expected issues of infrastructure collapse and the tasks of FHA and disaster reliefâ (Joint Chiefs of Staff, 2002, Chapter II, pp. 12-13). 5. Transition: âIn general, the end state of JUOs is the termination of operations after strategic and operational objectives have been achieved. This may include the transfer of routine responsibilities over the urban area from military to civilian authorities, another military force, or regional or international organizations. . . . In JUOs, transition may occur in one part of an urban area while engagement still is going on in another [three-block war]â (Joint Chiefs of Staff, 2002, Chapter II, p. 13). Note the overall emphasis on the social and organizational interactions of a diverse set of actors, including noncombatants, noncombat forces, and local and multinational civilian agencies. There is also a focus on the effects of informational, socioeconomic, and political factors on attitudes and behaviors in the urban battlespace. Changes in the Nature and Scale of Intervention Operations Urbanization and the broader view of military USECT interventions yield a dramatic expansion of considerations of scale, in both spatial and temporal dimensions, as well as an expansion in the nature and types of intervention to be considered. In the spatial dimension, urban operations demand a much finer view of the battlespace: it is no longer sufficient to consider high-level aggregates of large units and large geographic areas of responsibility, such as one might do
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 31 in planning conventional operations at the division level and above. Instead, the urban domain demands a block-by-block (if not building-by-building) geographic focus, at squad-level units consisting of only a few individual s Â oldiers. At the other end of the spectrum, the broad considerations of USECT phasing of an engagement call for understanding wide-ranging geopolitical factors, including the nation-states involved, and the associated ethnic, cultural, religious, and economic factors in the region. These are typi- cally not small or geographically focused but may in fact encompass huge spatial overlay regions of the potential battlespace (e.g., the Middle East). As a consequence, there are simultaneous demands to have a very fine spatial focus (at, say, the building level) while simultaneously being highly sensitive to the very large regional characteristics of the battlespace. In the temporal dimension, a similar situation exists. The fine-scale urban focus, with its short âinteraction distances,â typified by an impro- vised explosive device (IED) or a rocket-propelled grenade, demand a very fine-grained temporal view of events for assessment, planning, and execu- tion. Planning horizons are short, and urban operations demand a high temporal resolution of activities if operations are to succeed. The time available to plan operations is likewise compressed, and planning windows are compressed, often down to minutes. At the other end of the spectrum, USECT phases can take months or years to accomplish and are often char- acterized by considerably slower temporal dynamics and windows, in both the planning and the execution of activities. Thus, as in the situation with the spatial dimension, there is a simultaneous stretching of the temporal dimension from both ends, from very quickly occurring events at a high temporal resolution (e.g., building clearing), to activities that evolve at a considerably slower pace, demanding low temporal resolution but long time horizons (e.g., nation building). A key issue for modeling IOS behavior is the spatiotemporal âcover- ageâ that must be accommodated in models. One can clearly no longer expect that a high-level aggregate model of, say, an armored division cov- ering miles of open plain will be up to the challenge of anticipating the outcome of a fast-paced short-range small-unit urban engagement. Nor will the small-unit model be any indicator of overall outcome in the big picture of the overall military engagement. And neither is up to the challenge of anticipating outcomes in the larger USECT tableau, with its many other dimensions beyond the application of military force. Growth of the spatiotemporal scale is also accompanied by an expan- sion of intervention options available in urban operations over the several USECT phases. This is a natural consequence of the additional dimensions â This is perhaps best illustrated with the detailed step-by-step choreography that goes into the planning of a simple room clearing by a four-man squad.
32 BEHAVIORAL MODELING AND SIMULATION and structures that make up the urban environment and its indigenous population, as illustrated in the deliberately simplified three-layer struc- ture of Figure 2-1. Shown here is the conventional physical structure (and infrastructure) that is the focus of traditional military campaigns, on which is superimposed an information structure associated with elements of the underlying physical entities, in turn superimposed by a cognitive struc- ture characterized by individual and group perceptions, beliefs, intentions, plans, and actions (Air Force Research Laboratory, 2004). The focus for planning military operations is increasingly on under- standing and forecasting ânonkineticâ effects. Kinetic effects are associ- ated with the use of âkinetic weaponsââconventional bullets and bombs. Nonkinetic weapons and defenses are associated primarily with IO, which include the triad of electronic warfare, computer network operations (both defensive and offensive), and influence operations, which include PSYOPS, military deception, and operations security (OPSEC). Nonkinetic options also include the use of nonlethal weapons at the individual or crowd level (e.g., high-powered microwaves) and at the population level (e.g., disabling or destroying one or more components of, say, an urban infrastructure). In this expanded battlespace, planning and executing effects-based oper- ations (McCrabb, 2001) require analysis of the potential effects that a given set of diplomatic, information, military, and economic (DIME) actions will have across the full range of the political, military, economic, social, infor- mation, and infrastructure (PMESII) context. To be useful for analysis and planning, behavioral models must capture not only the separate effects of each action in each of these areas but also the interactions of these factors. HOW IOS behavioral Models CAN HELP THE MILITARY The changing nature of DoDâs mission has greatly increased the need for IOS models that capture the cognitive, organizational, societal, and cultural factors that are critical in the urban battlespace. IOS models are needed across the full spectrum of operations, particularly during urban â e W introduce the term âforecastingâ here, in place of predicting, to reemphasize the difficult problem of anticipating individual or organizational behavior (see Chapter 1), in comparison to that of anticipating the consequences of well-understood physical or engineering laws, the latter operating under conditions in which there is neither agency nor feedback involved (e.g., when you swing a hammer, the hammer does not deliberately try to avoid the nail in order to dissuade you from further swinging, so that your dynamic model of the muscle-hammer system is reasonably âpredictiveâ). The term âforecastingâ is also loaded with weather analogies, serving to remind us of how weather âpoint predictionsâ (in time and space) are almost always wrong and how âbounding envelope forecastsâ are much more likely to capture the future trajectory of the weather, especially as the spatial and temporal resolution grows more coarse (i.e., with larger geographic areas covering âclimate zonesâ and longer time windows covering âseasonal variationsâ). See also the extensive discussion of forecasting in Chapter 8.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 33 Pools of beliefs, intents, plans, and COAs Cognitive Cultural Structure Intelligence Domain Inform ation S tructu re GeoSpatial Intelligence Domain Phy sica l Str uctu re FIGURE 2-1â Heterogeneous structures that must be represented in the urban environment. SOURCE: Air Force Research Laboratory (2004, p. 10). operations, as indicated by the number one recommendation of the recent Joint Urban Operations Workshop: âEmploy high-resolution modeling, simulations, and other decision support tools that incorporate friendly, enemy, and neutral forces, plus the urban population in order to conduct rehearsals, assess courses of action, and make better decisions faster than the enemy in an urban operationâ (Mahoney, 2005). This section reviews how IOS models can contribute to todayâs missions in the three broad areas: (1) analysis and forecasting for planning, (2) train- ing and rehearsal, and (3) design and evaluation for acquisition. Another view of such applications is found in Axelrod (2004).
34 BEHAVIORAL MODELING AND SIMULATION Potential Use of IOS Models for Analysis, Forecasting, and Planning In general military operations, COA development and planning has been traditionally a completely manual operation, with a heavy reliance on staff experience and seat-of-the-pants âmental modelsâ of the adversary and its likely response to potential military activities. Consequently, it is often the case that only a few COAs are generated, evaluated, and planned forâoften with only minimal computer-based support. Figure 2-2 illustrates the essential closed-loop nature of military plan- ning and operations and indicates where models could be of use. A walk around the loop begins with the external world or battlespace shown at the bottom of the figure. Many actors populate the space, including blue (friendly) forces, red (adversary) forces, and a range of others depending on the particular environment (e.g., whether it is urban or not). Some of the blue assets include sensors and data collection systems that pick up incom- plete and uncertain information about the battlespace and, via associated communications assets, transmit it to a variety of data processing facilities and data storage centers. Some support relatively short-term data needs (Intelligence [INTEL] data) for current operations, while others may sup- port long-term development of background data and knowledge bases. The INTEL data support âinner loopâ situation assessmentâthat is, short-term assessment of the state of the battlespaceâto estimate the cur- rent situation in the face of collected information that is incomplete, noisy, and stale, which may also be compromised by reporting errors, communi- cations failures, and deliberate disinformation on the part of the adversary. This is clearly a complex estimation process. Given this estimate of the cur- rent situation, decisions can be made and orders/requests can be modified or generated, triggering a set of general action requirements defining how to use a range of blue assets (data collection systems, weapons platforms, etc.), thus closing the loop. Also shown in the figure is the use of background data and long-term knowledge bases to support âouter loopâ situation forecasting of the future evolution of the battlespace, based on the inner loopâs current assess- ment of the situation and any available behavioral, cultural, or historical knowledge pertinent to the conflict, geography, and population. This is clearly another complex process fraught with uncertainty, both because of the attempt to forecast into the future based on knowledge of the current situation and the reliance on uncertain information stored in the knowledge â his is essentially a more detailed version of the OODA (observe, orient, decide, act) loop T of Colonel John Boyd, USAF (Ret.). For more information about John Boyd and his writings, see Defense and the National Interest (2007). â he term âpredictionâ is shown in the figure to be consistent with the original USAF study T from which the figure is adapted.
Commanderâs â Intent Knowledge Future Bases Estimated future situation Behavioral Situation Planner Cultural Cultural Cultural Cultural Predictor Historical MI, OPS, LOG, â¦ Background Data Initial Updated COAs COAs Estimated Situation current situation Order Action Intel Data Assessor and alerts Generator requirements Databases Current Asset Manager Models for Behavior Forecasting System Model Development Process taskings Collected Data 2-2.eps Blue and Coalition Assets Blue & Coalition Assets Potential Potential Red Data Collection Red Red Red Systems Systems Allies Allies Assets Assets Weapons Platforms . . Neutrals Neutrals . Non- Non- Non- Logistics editable but rocky, made from wmf Combatants Combatants Potential Potential Blue Blue Allies Allies Battlespace FIGURE 2-2â Models for behavior forecasting are fundamental to battlespace assessÂment, forecasting, and management. 35 SOURCE: Adapted from U.S. Air Force Scientific Advisory Board Study (2002a).
36 BEHAVIORAL MODELING AND SIMULATION bases and generated by the inner loop situation assessment activity. The objective is to generate an estimate of the future situation (or envelope of future situations) at different time scales and geographic resolution, so as to be able to plan accordingly, across a range of time horizons, areas of responsibility, and military functional specialties (INTEL, OPS, logistics, etc.). The net results of this process are COAs and plans generated at all echelons, to support the inner loop action generation activities, as indicated in the figure. Whether focusing on the inner loop or outer loop activities, it should be clear that current estimates and future forecasts must naturally rely on IOS behavior models of some sort. They may be implicit seat-of-the-pants mental models held by the personnel performing the intelligence, planning, and operational functions, or they may be explicit and simplified computa- tional models (possibly instantiated at vastly different time scales or spatial resolution), but they all implicitly attempt to forecast behavior by using a model as an âextrapolation engineâ operating on the current assessed state of the situation, using the best available information and knowledge col- lected from the battlespace, and knowing what future blue asset activities are likely to be. IOS behavior models, their associated simulations, and model-derived tools are needed to track, identify, and target critical individuals and resources and to assess the relative ability of various courses of action to influence adversary behavior and to win the hearts and minds of the indig- enous population. Whether the issue is mapping the human terrain (Kipp, Grau, Prinslow, and Smith, 2006; Schaffer, 2005), or understanding the atmospherics, evaluating the impact of interventions to promote or inhibit state failure, forecasting hot spots of activities in urban settings, or provid- ing more cultural and cognitive situation awareness, IOS behavior models and their derivatives (simulations and tools) are clearly needed. Models for Understanding, Forecasting, Shaping, and Responding to Adversary Behavior Reliable anticipation and forecasting of individual human and col- lective organizational behavior on the part of the adversary is the highest goal of all military commanders. This view is embraced by the Army and the Marine Corps, in their call for doctrine and tools that enable âpredic- â his last component supporting the forecasting process assumes that blue assets behave T according to plan and is predicated on the notion that âthe best way to predict the future is to create itâ (The Drucker School, Claremont Graduate University, 2008). See http://www. cgu.edu/pages/4181.asp. â ee http://www.army.mil/professionalwriting/volumes/volume4/december_2006/12_06_2.html. S
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 37 tive analysisâ (Kasales, 2002), especially as potential future engagements become more asymmetric and urban, less influenced by traditional for- malisms of conventional military doctrine, and more determined by the contextual influences of society: political and military organizations, ethnic groups, national cultures, and transnational religious organizations (Brown, 1997; Lwin, 1997; Staten, 1998). A similar view is held by a host of mili- tary and other groups: â¢ The Air Force Scientific Advisory Boardâs (SABâs) Predictive BattleÂ space Awareness Study (U.S. Air Force Scientific Advisory Board, 2002a, 2002b) â¢ A more recent SAB study on the need for behavior modeling in urban operations (U.S. Air Force Scientific Advisory Board, 2005) â¢ Ongoing science and technology efforts being conducted by the Defense Advanced Research Projects Agency (DARPA) (e.g., the RAID, Integrated Battle Command, and Integrated Crisis Early Warning System programs) â¢ The Office of Naval Researchâs Affordable Human Behavior ModelÂ ing program (see http://www.onr.navy.mil/sci_tech/personnel/342/ training_afford.asp) â¢ The USAF Commanderâs Predictive Environment program (see http://www.wintersim.org/abstracts06/Mil.htm) â¢ JFCOMâs Urban Resolve program (see http://www.jfcom.mil/about/ experiments/uresolve.htm and the like) â¢ Conferences and workshops focused on the problem (e.g., the 2003 DMSO-sponsored conference on organizational simulation, see https://www.dmso.mil/public/) â¢ The 2003 Army Research Institute-sponsored Workshop on Cogni- tion and Multi-Agent Interaction, the Air Force Office of Scientific Research (AFOSR) Workshop on Culture and Adversary Modeling, 2005, 2006 â¢ The Annual International Conference on Complex Systems (see http://necsi.org/events/conferences.html) Before and during military operations, IOS models can serve as decision aids and as guides for data collection. Models can, have been, and should be developed to support tactical, operational, and strategic missions. Key uses for IOS models include model-based INTEL fusion and situation assessment, forecasting (projecting), planning (COA development and assessment), mis- sion rehearsal, execution monitoring, and postexecution assessment. IOS models can be used to gather information (âseeâ); assess or evaluate the cur- rent state (âidentifyâ); explain, understand, and forecast behavior (âthinkâ); shape, manage, and disrupt oneself or the enemy (âdoâ); and aid in decision
38 BEHAVIORAL MODELING AND SIMULATION making or strategizing (âreflectâ). IOS models hold the promise of aiding the war-fighter by providing a better toolkit for knowing the enemy. From the tactical to the strategic levels, there is a need to forecast adversarial reasoning. IOS models can be used to provide guidance on the space of actions that the adversary might take and why, thereby reducing surprise. Moreover, these models can suggest what actions are the most probable and provide insight into the general order of actions. Such models, however, do not and should not be expected to provide guidance on exactly what action will be taken when (as we discuss later). The deployment of IOS models needs to be accompanied by training in their appropriate use and in the interpretation of the results generated. Models for Understanding, Forecasting, and Shaping Societal Behavior Increasing military involvement in military operations, peacemaking, and peacekeeping is creating a need for the military to understand, forecast, shape, and respond to the larger context of societal norms, expectations, perceptions, and behavior. Examples abound as to this need: â¢ Understanding the local society, its history, and its current overÂ lapping networks increases the likelihood that one might be able to identify those who would harbor terrorists or turn to terrorism. â¢ Understanding the local culture and its homogeneity, or lack thereof, is necessary for planning effective PSYOPS campaigns, and assessing the impact afterward. â¢ Shore leave has repercussions for the local population, including increasing the monetary inflow that can stabilize some local busi- nesses while leading lead to an increase in corruption and a change in the power base. IOS models, in general, hold the promise of enabling the identification of geographic locations where and periods when threats are likely to emerge as a function of current events and action (or inaction) by U.S. and coali- tion forces. Two notable DARPA-sponsored programs have focused on the identification of potential global hot spots and the forecasting of their likely evolution over time: the Pre-Conflict Anticipation and Shaping (PCAS) pro- gram (described later in this chapter), directed at forecasting the likelihood of a nation-state collapse (Popp et al., 2006), and the Âfollow-on Integrated Crisis Early Warning System, which has as its goal âthe development of state-of-the-art computational modeling capabilities that can monitor, assess, and forecast, in near-real time, a variety of Â phenomena associ- ated with country instabilityâ (see http://www.darpa.mil/ipto/Âsolicitations/ open/07-10_PIP.pdf).
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 39 Multiagent models and system dynamic simulations that take social and cultural factors into account could be used to assess the likely conse- quences of the COAs executed by the U.S. military and coalition forces on state Âstability. They could be used to assess the potential impact of a multiÂ pronged initiative with DIME, and to assess the consequences in PMESII. This is exactly the focus of an ongoing DARPA-sponsored program in Inte- grated Battle Command (see http://www.darpa.mil/sto/solicitations/IBC/). IOS models could also be used for identifying what COAs will have the least negative or most positive effects on civilians and neutrals. In all cases, the value of such models and simulations could be enhanced if informa- tion on the underlying social and organizational networks and available resources were taken into account and if the models were combined with effects-based operations models. Models could also be used to determine the effects of U.S. information operations activities designed to influence attitudes and behaviors of individuals in different cultures. In this case, the effectiveness of these information activities is likely to be enhanced by link- ing social network models with psychological profile information, cultural models, and psychosocial models. Models for Understanding Enemy Command and Control Structures Understanding enemy command and control structures through IOS behavioral models enables the identification of vulnerabilities and strengths before planning friendly activities. This was pointed out in a previous study on human behavioral modeling (National Research Council, 1998) but has only recently been acted on because of the post-9/11 focus on counterÂterrorism and the rapid development and dissemination of support- ing tools. In particular, support tools, such as text and data mining facilities, are beginning to be used for extracting information from open sources (e.g., news articles, websites, etc.) to identify events and structures that enable the detection and recognition of terrorist and insurgent networks and orga- nizational structures. Much of this work is classified, but significant insight can be gained from parallel efforts ongoing in the commercial world (see, for example, the growing conference on âtext analyticsâ at http://www. textanalyticsnews.com/usa/program.shtml). However, the strength of these tools could be considerably increased if they were combined with social and dynamic network modeling techniques, to enable a model-based approach to text and data mining and information fusion. Such tools could then be used in an âalertâ mode to identify what data should be collected, as well as in an âevaluateâ mode to suggest the impact of various courses of action. Given an understanding of the enemyâs command and control structure, targets could be identified for disrupting
40 BEHAVIORAL MODELING AND SIMULATION the enemy, and various courses of action developed to achieve those goals. COAs that are intended to demoralize, disrupt, or inhibit action or recruit- ment on the part of the enemy could then be evaluated if behavioral models incorporated realistic affective, social, and cultural influences. These models could be made even more effective if they were placed in data-farming envi- ronments so that the space of COAs could be more effectively mapped. Models for Training and Mission Rehearsal IOS behavior models have the potential for providing significant benefit to U.S. and coalition forces in training (of general skill sets) and mission rehearsal (for mission-specific expertise), based on years of experience of simulation-based training in more conventional areas (e.g., flight training, tank tactics training, etc.), the critical dependence of learning on perfor- mance feedback and âafter action reviewâ insight, and, perhaps most importantly, the opportunity to learn from errors that might be devastat- ingly fatal in the real world. Key uses of models for training include model- based simulation of virtual actors (including simulated entities, such as teammates, adversaries, and noncombatants), games to provide immersive experiences, and models to preassess potential new training tools. Model- based training simulations and systems can provide training that: â¢ supports a number of activities, such as teaching individuals how to be more culturally aware, training teams how to coordinate and fight as a unit, training commanders how to evaluate the organiza- tional health of their battalion; â¢ enables live, large-scale war-gaming with truly dynamic enemiesâ these training systems can be constructive or done using virtual reality or gaming systems; â¢ takes into account social, cultural, or organizational factors and can be used for realistic training of individuals, teams, or organiza- tions; and â¢ crosses operational activities and enables joint or coalition training. Training at all levels, up to and including higher headquarters staff, is vital to ensuring successful joint and coalition operations. New demands are being placed on training and rehearsal systems that increase the need for modeling to support training. Effective training and rehearsal systems immerse the trainee in realistic scenarios, provide information about roles and responsibilities, enable the development of technical skills, and provide experience working in joint or coalition task forces, facing new, dynamic, and culturally distinct enemies. The problem is providing such an immersive training environment in less time, for less money, for more personnel, using
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 41 models and simulations that can be rapidly adapted to changing missions and new adversaries. The changing nature of military missions and the increased emphasis on peacekeeping, disaster relief, and nation-state building means that training systems are needed for more than conventional training in weapons usage, war-fighting tactics, and basic survival. Indeed, there is now a need for training in cultural awareness, crowd behavior, negotiation, management, and city planning. Training systems need to be more flexible to capture the changing nature of the enemy from nation-states to insurgents, from one monolithic actor to federations of loosely aligned tribes, and from large-scale weapons to improvised explosive devices. Finally, new training technologies are required due to changes in military staffing, such as fewer, more Â computer-savvy recruits, increased use of reserves, and just-in-time training. Massively multiplayer online games (MMOGs) can, in principle, pro- vide such an immersive environment. The value of such systems for mis- sion rehearsal will be increased if the cultural and social models embedded are more socially realistic than those in current games. Dynamic network m Â odels hold the promise of providing a dynamic adversary for war-gaming, and the value of such systems will be increased if links can be made between the network models and models of action, planning, and goal attainment. Training can be provided for teams and larger units by populating scenarios with socially and culturally realistic artificial actors as team mem- bers, but there is a need for simulation infrastructure to rapidly develop socially realistic and culturally differentiated artificial actors. Training for future scenarios can be provided by using system dynamic and multiagent computational models that allow the user to look ahead and do what-if analysis of alternate scenarios, such as the impact of tsunamis or hurricanes on various regions of the world or the impact of avian flu on military personnel. The value of these systems will be increased if they move beyond military and economic factors to consider social, political, diplomatic, and information factors. Expert systems and cultural models can be used to increase cultural awareness and train military personnel in crowd control behavior. The value of these systems will be increased if they can be rapidly populated with data as new adversaries arise. In general, IOS behavioral models can be used effectively for improved training, but more realistic models of actors, groups, and nation-states are needed. A key aspect of the current training and rehearsal process is that, during training, military personnel are provided access to people (or their model-based surrogates) with whom they will be working in the field. For example, at joint war games, Air Force, Army, Navy, and Marine personnel meet, plan, and execute together. This increases their transactive knowledge
42 BEHAVIORAL MODELING AND SIMULATION of who knows what and who can do what, which in turn improves group performance. If IOS models can help support this function, war games can be reduced in size, conducted more frequently, and tuned to specific indi- viduals and organizations needing specific training or rehearsal. Models for Military Systems Development, Evaluation, and Acquisition DoD is in the midst of two revolutionary changes (Frost, 1998): a revo- lution in military affairs (RMA) and a revolution in business affairs (RBA). The RMA involves the military requirements and concepts envisioned in light of the threat environment and advances in technology (Joint Chiefs of Staff, 2000). The RBA addresses how to leverage technology and com- mercial business processes to the how-to-buy problem. The use of M&S has been recognized as a key facilitator to addressing military training and education and the acquisition of military systems (Office of the Secretary of Defense, 1996). For example, the Air Forceâs Modeling and Simulation Strategic Plan (Johnson, 2004) spells out a number of points of focus to meeting the challenges of RMA and RBA by providing a persistent synthetic battlespace infrastructure to support the exploration, design, development, analysis, and testing of new war-fighting systems and concepts (as well as more conventional military training and mission rehearsal activities). Inherent in the necessary M&S infrastructure to support system acqui- sition is the requirement to provide realistic representations of battlespace entities (blue, red, and neutral), natural and cultural features (terrain, locale), and physics-based effects (sensor processing, missile flyout, etc.). Such an infrastructure must provide a framework to support the rapid integration of these synthetic representations across all simulation levels (i.e., campaign, mission, engagement, or engineering). In the Air Force, this M&S capability has direct relevance to important acquisition programs, such as the Joint Strike Fighter, the Multi-sensor Command and Control Constellation, the Joint Distributed Engineering Plant, and the Distributed Mission Operations/Mission Rehearsal initiative. The other services have their own acquisition programs in which M&S capabilities are directly relevant to effective acquisition. As a consequence, a substantive effort is needed to develop the requisite M&S components to address the evolving threat environment. IOS behavior models that could support automated means to generate and adapt red strategies/tactics in line with asymmetric warfare can be a key enabling component for meeting the objectives needed to support acquisi- tion. Key uses here are to preevaluate the value of new technology in a vari- ety of scenarios that are both physically and culturally accurate, assess the need for particular skills in soldiers, and assess how generational changes in soldiers may lead them to need or utilize technologies differently from their
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 43 predecessors. Realistic IOS models could also be put to good use while sys- tems are under development, especially as more acquisition programs adopt the âspiral developmentâ paradigm, in which the requirements for each new spiral are determined not only by the evaluation results of the past spiral, but also by the changing battlespace requirements generated by an adaptive and resourceful adversary. (The counter-IED development work is an excel- lent example of rapidly changing tactics and countertactics; see http://www. nationaldefensemagazine.org/issues/2006/jan/adaptive_foe.htm.) Models for Enabling Command and Control Weapons Systems In the current network-centric operating environment (Alberts and Hayes, 2003), command and control (C2) organizations and the informa- tion infrastructure that supports them are becoming increasingly important. One could say that C2 organizations, their information architectures, and their underlying communication infrastructures have in fact become the new weapons. This was underlined in 2004 when General Jumper, then Air Force chief of staff, officially designated the Air and Space Operations Center as a weapons system. And it has been formally recognized with the newly formed Cyber Command in the Air Force. As stated by the organiza- tionâs first commander, General Elder: âThe Air Force now recognizes that cyberspace ops is a potential center of gravity for the United States and, much like air and space superiority, cyberspace superiority is a prerequisite for effective operations in all warfighting domainsâ (Wait, 2007). The Army and the Navy similarly recognize the leverage obtainable from effective C2 systems, especially given the Armyâs commitment to the networked Future Combat System program and the Navyâs effective inven- tion of the term ânetwork-centric warfareâ through Admiral Cebrowskiâs leadership,10 but it is fair to say that all three services have tended to focus on the hardware and software infrastructure (communications pipes, fusion algorithms, decision aids, visualization techniques, etc.), with less emphasis on the human and organizational component of effective C2. What is becoming increasingly clear, especially in light of the current conflicts in Afghanistan and Iraq, is that there is a critical need for the rapid design and redesign of military units, including the architecture of their C2 systems, to meet changes in missions and to respond to innovations in enemy activities. Related to this is the need to be able to identify vulner- abilities in current C2 structures. IOS models have significant potential for assessing, designing, and evaluating the impact of new technologies or new C2 procedures on potential vulnerabilities, strengths, shared situ- â See http://www.army.mil/fcs/. 10â ee S http://www.oft.osd.mil/biographies/cebrowski_with_pic.cfm.
44 BEHAVIORAL MODELING AND SIMULATION ation awareness, work distribution, and Â adaptability to enhance friendly operational effectiveness, while defending against enemy actions, across a full spectrum of cultures, nations, and nonnation-state actors. Using such models has the potential for moving the military beyond logistics planning to organizational planning, facilitating improved recruitment strategies, and enabling just-in-time team design. Military personnel often remark that, as soon as they get to the field, the plan goes out the window. One way of building in flexibility (or resil- ience) is to support the design and redesign of the various units to take into account changes in mission, changes in technology, attrition, rotation, or incorporation of joint and coalition forces. Many times such design needs to be made on the fly as the situation changes. In this case, the commander is faced with the problem of identifying experts quickly and incorporating them in a âtiger team.â A related problem is assessment of the unitâs orga- nizational health, its vulnerabilities, its shared situation awareness, and its overall war-fighting effectiveness. A traditional approach to organizational design has been to identify structures that are optimized to meet some organizational criteria. This approach is insufficiently flexible in many cases, as the military needs to operate in a responsive and adaptive mode. Criteria for designing adaptive units are under investigation, and emerging behavior modeling efforts are beginning to afford new possibilities for organizational design (ÂLevchuk, Yu, Levchuk, and Pattipati, 2006, 2004; Pattipati, Meirina, Pete, ÂLevchuk, and Kleinman, 2002; Neal Reilly, 2006; Levchuk, Levchuk, Meirina, ÂPattipati, and Kleinman, 2004; Levchuk, Levchuk, Luo, Pattipati, and Kleinman, 2002a, 2002b; Entin, 1999). IOS models, simulations, and assessment tools could be used to pre- evaluate the impact of new technology on the unit, identifying potential ways in which the unitâs structure should change in response to this inser- tion. Dynamic network models linked to various databases with streaming information on personnel could enable real-time assessment of shared situ- ation awareness and organizational health. Text mining tools and shared mental model assessment tools could be used to improve information flow and rapidly process incoming data. IOS models of unit needs could be used to form a âsmartâ command center that could be used to push information to people only when they need it. Design tools and smart command center tools are well within the reach of current technology. The key problems are those of scalability, handling streaming data, and linkage of noninvasively collected data to dynamic network metrics of organizational health and text-mining evaluations of information flows.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 45 Representative Model-Addressable Problems in a Scenario Context We now illustrate how different behavioral models might be used to address specific questions raised by the commander and his staff during the course of operations, how models might be used to train for particular skill sets and missions, and how models might be used to help specify an âoptimalâ organizational design for a given mission. The intent is not to be exhaustive but merely illustrative, with the goal of motivating a closer look at the use of detailed behavioral models in a range of military activities. To orient our analysis and review of IOS modeling approaches, we developed scenario elements that are derivative of the one detailed in TRADOC PAM 525-3-90 O&O 22 JUL 2002 (U.S. Army, 2002), which describes a mission in the trans-Caucasus region. Three vignettes are devel- oped to provide a construct for the purpose of addressing the potential of behavioral models supporting operations of a brigade combat team as part of a joint campaign. These vignettes center around â¢ tactical operations in entry operation (entry), â¢ operational maneuver by air, combined arms operation for urban warfare (transition), and â¢ secure portion of a major urban area (JUO). Details of the scenarios and vignettes are given in Appendix B. In conjunction with an Army subject matter expert, we have specified repre- sentative model-addressable questions for portions of three vignettes. We think that these vignettes and the associated model-addressable questions only begin to scratch the surface in terms of providing suitable challenge problems to stimulate the modeling community and to provide a common reference frame for discussing alternative approaches to the same problem. In fact, we propose, in Chapter 11, that an initial effort in a large-scale, multiyear research programâfocusing on comparing and integrating dif- ferent disciplines, perspectives, and levels of detailâbe dedicated to the definition of a number of well-defined and highly focused challenge prob- lems that can serve as a common basis for comparing and contrasting dif- ferent approaches. If the vignettes and questions presented here can serve as a launching point, some effort might be saved in the long term, but the primary purpose of presenting these in this study has been to focus the com- mittee on relevant military problems and to provide the reader with some sense of the broad range of challenges that exist in the military domain. Box 2-1 shows the resulting representative high-level model-Âaddressable questions. Given these representative problems and issues, a number of more specific questions were generated, to illustrate the kinds of specific questions that might be asked during the unfolding of the vignettes.
46 BEHAVIORAL MODELING AND SIMULATION BOX 2-1 Representative ModelâAddressable Problems and Issues of Interest to the Commander (âWeâ in this box refers to the commander and his forces.) Analysis and Forecasting for Planning Disrupt terrorist networks. Fuse uncertain and partial information from multiple sources to identify the dynamic network structure of a terrorist organization. How can we best disrupt those networks? â¢ Tribal leader Muhkta is on the fence about whether or not to support the intervention. Which is likely to be the most effective way of gaining his s Â upportâovert recognition, overt financial reward, covert financial reward, covert protection of family, or a combination of methods? â¢ We need to disable/disrupt the clan of followers of Sheik Mustafa while our troops are moving toward the city. If we ensure he is disconnected from his clan during this phase of the operations, is it likely to degrade the clanâs decision making as related to their willingness to conduct offensive military operations? â¢ In order to reduce IED attacks, are the terrorist networks with their support base in our target city more vulnerable to selective attacks on their leadership or interruption of their recruitment programs? â¢ Abdul X is the leader of a terrorist network. Mohamed is on the network council and more radical than Abdul X. If Abdul X is killed, how likely is it that Mohamed will become the leader of the network? Forecast adversary response to COAs. In an urban operation, forecast the likely response of local insurgents to friendly force movements, basing, and logistics. Identify likely counters to proposed COAs and identify early harbingers of those counters. â¢ What will impact the local economy the least: denial of transportation fuels or denial of electricity? â¢ The JTF can plan on placing its logistics support base either within the bounds of the city or in the adjacent countryside. Which population in the area, urban or rural, will be less hostile to the presence of the logistics base? â¢ To establish crowd control early in the urban environment, is controlling an area, like the civilian neighborhood, or a point of special interest, like a mosque, more likely to mitigate crowd behavior? â¢ In neighborhoods not committed to radicalism, what is the most influential means to insert forces: in combat vehicles or on foot? â¢ JTF wants to use disinformation to partially protect our intentions of moving from forward operating base (FOB) to the city. Is the most effective point of insertion of the disinformation the few public media outlets or the informal rumor mill/tribal network?
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 47 BOX 2-1 Continued Societal forecasting. Forecast the effects of alternative diplomatic, infrastructure, military, economic courses of action on attitudes and behaviors of residents in a region of interest. Assess the likelihood of state failure and identify actions that will lead to escalation of violence. â¢ Troops give a lot of meals ready to eat (MREs) to locals. Considering the items in MREs and the local culture, will MREs be a better giveaway than basic grains and cooking oil? â¢ Entry phase combat will be kept at the lowest level possible. Given local condi- tions and the impacts of the blockade, which will the locals respond better to initially: engineers/civil works or medical response teams? â¢ Considering the effect of the blockade, which will have the psychological effect most supportive of our mission end state: overwhelming force or âhelping handâ intervention? â¢ Which approach will least offend locals as we travel from the initial entry area to the city: keeping civilian vehicles in a separate convoy or infusing them into tactical convoys? â¢ Can we forecast the response by the local religious leaders to the presence of female soldiers on the streets of the city? â¢ A specific Mosque is known to be the headquarters of a particular militia. Joint forces will destroy the mosque in order to deny access by the militia. Which will produce the least negative impact in the neighborhood: announcing our intentions to destroy the mosque or destroying it unannounced? â¢ How do attitudes differ between the tribal regions of the country and the urban area we are targeting? â¢ What is the formal communication dynamic between the host national gov- ernment (HNG) and the population? What is the informal communication dy- namic? (How do people get information on a day-to-day basisâcoffeehouses, religious structures, etc.?) How great is the delta between formal and informal communication dynamics? â¢ What are the expectations of the population about the governmentâs ability to provide services? â¢ Is the HNG a government on the road to collapse? â¢ Are there indicators of popular support for the alternative power structure? Are they reflected in local media and among the local intelligentsia? Training and Rehearsal Crowd control training. Create an immersive virtual training environment in which soldiers can learn to take appropriate action based on the correct interpre- tation of the behavior of small groups of citizens and understand the triggering mechanisms for violent responses by the crowd. continued
48 BEHAVIORAL MODELING AND SIMULATION BOX 2-1 Continued â¢ To effectively control crowds we need to know where the leaders are. In this setting, are crowd leaders more likely to be leading from the front, urging from the rear, or not on site? Given the answer, should we use information opera- tions or force to control the crowds? â¢ Given the nature of the small villages along the route from our FOB to the city, is it likely there will be crowds along the route, are they likely to be friendly or hostile, and in either case will stopping to interact with them be likely to alter their feelings? Design and Evaluation for Acquisition Organizational design: force composition and command and control archi- tecture. The Army is moving toward modular forces focused on joint and expedi- tionary capabilities. These units of action will be rapidly reconfigured and equipped for specific mission requirements. The Navy is fielding expeditionary strike groups that include marine expeditionary units capable of amphibious operations at- tached to Navy ships. The Navy and the Marines follow different doctrine and are in the process of defining flexible supporting and supported relationships that allow them to function effectively as a combined fighting unit. â¢ Develop a recommended force composition (systems, equipment, units and personnel) for a humanitarian assistance mission. â¢ What command and control architecture will be most effective for this mission? â¢ What are the appropriate organizational coordination points for most effectively working with NGOs during the humanitarian assistance mission? â¢ Is the force composition structure recently used for a humanitarian assistance mission appropriate for a disaster relief operation that requires immediate deployment? â¢ Are new roles needed to take advantage of the information-rich network-Âcentric environment? For example, would an information commander/coordinator role result in more effective mission performance? OVERVIEW OF current dOd IOS MODELING efforts In this section we briefly review major IOS behavioral modeling efforts under way to address military questions such as those described above, pointing out some of the major challenges that confront these efforts. The DMSO Master Plan for Modeling and Simulation In 1995, DoD published a master plan for M&S, in an attempt to unify efforts across all services, identify needed areas of development (gaps), and
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 49 minimize duplication of efforts (overlaps). The plan was âthe Department of Defenseâs first step in directing, organizing, and concentrating its M&S capabilities and efforts on resolving commonly shared problemsâ (U.S. Department of Defense, 1995, p. i). The Defense Modeling and Simulation Office (DMSO) was given six major objectives under this plan, including âprovide authoritative representations of human behaviorâ (U.S. Depart- ment of Defense, 1995). The DoD M&S master plan also specified a set of more detailed subobjectives for achieving these goals, as well as a detailed timetable for initiating and concluding some of these activities. The lofty goals and aggressive timelines of the DMSO master plan have not been achieved, 10 years after they were first promulgated.11 A quick review of DMSOâs Modeling and Simulation Resource Repository (MSRR) at http://www.msrr.dmso.mil/ would appear to demonstrate this. The MSRR system is maintained by the Modeling and Simulation Information Analy- sis Center (MSIAC). It includes five nodes representing the three services (Army, Navy, and Air Force), the DoD system, and the Defense Intelligence Agency and provides âretrieval of metadata descriptions of modeling and simulation resourcesâ (Defense Modeling and Simulation Office, 2007), including models, simulations, frameworks/toolkits, background reference material, and the like. The following bullets summarize the results of a recent (December 2006) search of the three service nodes: â¢ The Army node (see http://www.msrr.army.mil/) indexes 926 Âmodels, simulations, and simulators. Of these, fewer than 20 relate to individual human cognition, behavior, or performance. Of those, four focus on human visual performance (e.g., VISEO), three on human-in-the-loop (HIL) simulators, one on anthropometry (Jack), and the remaining few on four distinct behavior models: IMPRINT, IUSS, MATREX, and OneSAF (more on these later). Of the same 926Â modeling resources, only three relate to group or organizational modeling: C3GRID (built on MATREX), a crowd model based on diffusion kinetics (RDEBBSM), and a software tool for building an organizational model (C3TRACE). Searching for Â models associ- ated with the keywords âculture/cultural,â Ââeconomic,â âethnic,â 11â t I is beyond the scope of this study to attempt to do a forensic analysis of DMSO per- formance in this area. A number of factors may have contributed: a problem scope that was s Â imply âtoo bigâ for the funding and personnel resources available to DMSO; a science and technology portÂfolio decision that emphasized simulation engineering issues over basic science and technology; a political/economic environment that pitted DMSO against the entrenched M&S agencies in the services (Army, Navy, Air Force) and other agencies; etc. But it certainly would be worth revisiting the officeâs past history, should recommendations be made to reju- venate the office or to create a new one with similar responsibilities.
50 BEHAVIORAL MODELING AND SIMULATION âpolitical,â âreligion/religious,â and âsocialâ yielded no hits on the database.12 â¢ The Air Force node (see http://afmsrr.afams.af.mil/) indexes 54 m Â odels, 39 simulations, and 26 simulators. Of these, less than a dozen relate to individual human cognition, behavior, or per- formance, and of these, two refer to HIL simulators, one to a Â nthropometry (INTERMEDIATE), one to decision aiding (for target prioritization), and the remaining few on generic frame- works (DIAS, FLAMES, ICET) or distinct behavior models (CART/ IMPRINT, JSAF, OMAR, and STELLA). Searching for models associated with the keywords âpoliticalâ and âsocialâ called up the DIAS generic framework and the IO suite for command and con- trol warfare, both developed by the Air Force Agency for ModelÂ ing and Simulation (AFAMS), but neither explicitly representing human behavior. Models associated with economic features were focused on acquisition, and no models were associated with the terms Ââculture/cultural,â âethnic,â or âreligion/religious.â â¢ The Navy node (see http://nmso.navy.mil/) indexes 832 models and simulations. Of these, fewer than a dozen relate to individual human cognition, behavior, or performance, with most focusing on HIL simulations or human visual performance. Only one dis- tinct (cognitive) behavioral model is called out: the Air Defense Commander simulation (full name Autonomous Agent-Based Simulation of an AEGIS Cruiser Combat Information Center Performing Battle Group Air-Defense Commander Operations), which models small-team performance in C2 (Navy Modeling and Simulation Office, 2004). Searching for models associated with the keywords âculture/cultural,â âeconomic,â âethnic,â âorganization/Âorganizational,â âpolitical,â âreligion/religious,â or âsocialâ yielded no hits on the database.13 One might be led to conclude on the basis of these results that the M&S community is not active in developing models of individual and group behavior. This is not the case. Rather, MSIAC is simply not keeping pace with the explosive development and application of behavioral models that 12â he T term âeconomicâ did identify two tools not related to human economic behaviors, and the âsocialâ search term did identify the SPECTRUM facility at the National Simulation Center at Ft. Leavenworth, which claims to âuse a subject matter expert developed database to describe the political, economic, and social characteristics of the region being simulatedâ for use in HIL war-gaming simulations. The SPECTRUM description was last updated in 1998. 13â he âorganization/organizationalâ keyword did identify several organization-level Âtrainers T used by the Navy and dependent on HIL operation but not organization-level behavior models.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 51 started in the mid-1990s and continues to grow today, both inside DoD and in the behavioral research and computational modeling community. And no one else is keeping pace with the M&S effort either, even within DoD. There simply is no comprehensive archive or summary of all human behavioral models developed for or applied by DoD, although several organizations have specialized âsnapshotsâ with associated information on their state of technical readiness, limitations, availability, etc. Clearly, an across-DoD survey, maintained in a regularly updated fashion, would be particularly valuable, especially if it went beyond a simple verbal descrip- tion and attempted to describe each M&S resource in a common ontology or framework, so that comparisons could be made across models and simulations. Selected Current DoD Behavioral Modeling Efforts A complete survey of the state of IOS model development and appli- cation (both inside and outside DoD but having potential for use in DoD applications) goes beyond the charge of this committee. We can, however, give a brief overview of some of the more visible efforts14 on the basis of surveys conducted outside MSIAC and on the basis of the committeeâs knowledge of the domain. It is appropriate to note that our focus here is not on the traditional M&S tools used by the military operations research and training communities (e.g., AASPEM, CASTFOREM, CBS, CCTT, CSSTSS, EAAGLES, EADSIM, EAGLE, JANUS, JCATS, JCM, JWARS, MTWS, TACBRAWLER, TACSIM, WARSIM2000), which focus on the physical aspects of the battlefield and the associated sensor/weapon/C2 sys- tems. Instead, it is on the complex behaviors generated by the indiÂviduals, teams, and organizations of people populating the battlespace.15 In fact, few IOS models are being used on a daily basis by war-fighters, military planners, or military trainers. Most existing models accredited16 for use by military personnel are large-scale models of physical systems that do not take social, cultural, organizational, or affective factors into account. Key exceptions are identified in Appendix Table 2-A1 at the end of this chap- ter, which tabulates some of the major current efforts in this latter area of 14â any are classified or simply buried in organizational stovepipes, leading to significant M overlap or duplicative activity. 15â brief overview of these military simulations is given in National Research Council A (1998, pp. 33â50). 16â erification, validation, and accreditation is a well-defined DoD process; an overview of V this overall âcertificationâ process is given in the section entitled âMilitary Â Approaches to Verification and Validationâ in Chapter 8. In simple terms, accreditation Â occurs when the accrediting agency (the owner of the simulation) places its stamp of approval on the valida- tion results.
52 BEHAVIORAL MODELING AND SIMULATION militarily relevant IOS behavioral modeling. In the paragraphs that follow, we describe only a few of these activities to provide a general sense of the overall effortâsince a full review is beyond the scope of this study. OneSAF Family of Models and Simulations The OneSAF family of simulations includes One Semi-Automated Forces (OneSAF); OneSAF Objective System (OOS); OneSAF Testbed (OTB); Joint Semi-Automated Forces (JSAF); and ModSAF (Modular Semi-Automated Forces) and provides some capabilities for modeling human behaviors that vary by culture. Underlying this is the OneSAF Test Bed, which is a model derived from ModSAF (Parsons and ÂWittman, 2004). OneSAF has behavior representations that are effectively implemented in code and facilities that enable the user to rapidly develop instantiated models of new groups, com- munities, etc., considering a set of social, economic, and political factors. Although OneSAF is more flexible and provides better culturally sensitive modeling than was previously possible, it still has limitations. One is that OneSAF is still under development but is nearing government acceptance testing for the initial operating capability. A second limitation is that it is not clear at this time how alternative models could be linked to or federated with OneSAF. Finally, the structure by which cultural variables are included in OneSAF may limit the type of cultural factors that can be included. Task Network Models and Tools Task network models describe actorsâ behaviors in terms of interÂ dependent tasks to be accomplished in order to achieve an overall goal. These models have their foundations in the Navyâs PERT17 chart devel- opment in the early 1950s and owe their popularity to the ease of con- structing them and the clear visualization they afford in terms of task ÂnterÂdependencies and task completion progress. MicroSAINT18 popular- i ized their use in the 1970s in modeling human performance in tasks via task networks by (1) adding simple human performance parameters to each block in the network (the likelihood of correct task completion, time to complete, etc.); (2) making graphical task network construction easy to do by the nonspecialist; and (3) providing a discrete-event standalone simula- tion environment for exercising the model over time. Many task network models have been developed for simulating military tasks, and the basic MicroSAINT language has been extended by develop- 17â ERT stands for Program Evaluation Review Technique, a methodology closely related to P the Critical Path Method used to identify bottlenecks in overall task progress. 18â ee http://www.adeptscience.co.uk/products/mathsim/microsaint/. S
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 53 ment supported by ARL under the IMPRINT program, as well as by sub- sequent extensions by the Air Force Research Laboratory (AFRL) under the CART program, to support embedding into other simulation environments. Several derivatives have been developed by the behavior modeling commu- nity, including C3HPM, which builds on IMPRINT, and C3TRACE and HOS, which build directly on MicroSAINT. More sophisticated researchers, particularly from the ACT-R community, have made efforts to integrate MicroSAINT models with more traditional cognitive architectures. Cognitive and Cognitive-Affective Architectures and Models A wide variety of cognitive and cognitive-affective architectures and models are represented in Appendix Table 2-A1. Although their history may not be as long as the task network models (going back perhaps 25Â years to the pioneering work of Anderson, 1983), they are remarkably diverse in their underlying structures, their associated computational implementations and development tools, and their applications, both military and nonÂmilitary. This includes the âpure cognitiveâ architectures/models, which tend to be standalone and used within their own communities (ACTâR, CLARION, COGNET, EPIC, OMAR and D-OMAR, SAMPLE, and Soar); the âhybrid cognitiveâ architectures/models, which bridge the gap between communi- ties by combining models (EPIC-ACT-R, IMPRINT-ACT-R, Soar-EPIC, and others); and the cognitive-affective architectures, which extend the pure cognitive into the affective domain (MAMID, MINDS, and PMFServe). One major commonality among all of the architectures/models is that they were developedâinitially at leastâwith the goal of modeling the individual human faced with dealing with some sort of cognitive task. That focus on the individual has been maintained while extensions have been made in many different directions (perception, motor control, affect, m Â emory, multitasking, among others). It is only recently that significant effort has begun to be devoted to dealing with modeling groups of indi- viduals, from small teams to large organizations. As described in the agent- based modeling (ABM) section of Chapter 6, one of the primary barriers to representing the behaviors larger groups of individuals using cogni- tive and cognitive-affective models is the computational constraints: These m Â odels tend to be very fine-grained, and running a large number of them on a single host quickly brings the simulation to a grinding halt. However, this is expected to be less of a problem as the hardwareâs computational speed increases, and better use can also be made of parallelism across mul- tiple platforms. But a more fundamental problem exists: the lack of social knowledge in most of these representations. Cognitive modelers are keenly aware of the need to incorporate mental models of the environment they are interacting with, but they seem to be less so inclined regarding the mental
54 BEHAVIORAL MODELING AND SIMULATION models of the other agents they are interacting with, perhaps because of the infinite regress involved. This is clearly a needed direction for further research if this category of ABM is to succeed in modeling larger collections of cognitive and cognitive-affective agents. Multiagent Systems ABM environments and multiagent systems trade off the complexity of individual cognitive-affective agents for an increase in the sheer num- ber of agents and a concomitant increase in the complexities in interagent i Ânteractions. These are described in more detail in Chapter 6, but it is worth commenting briefly on the three multiagent models highlighted in Appendix Table 2-A1 and how they have been extended and applied to DoD questions of interest. Construct is a multiagent network simulation framework that supports the modeling and analysis of dynamic agent networks that evolve over time as a function of agent-to-agent interactions, and it clearly has direct applicability to the growth of terrorist networks. CORES is a multiÂ agent environment that supports the inclusion of DIME/PMESII factors in the agent interactions, to support understanding of broader contextual fac- tors in agent and network behaviors. ÂBioWar combines multiagent Âmodels of social Â networks, disease models, and population demographics into a single integrated model of the impact of a biological warfare attack on a city. Additional multiagent models and frameworks developed at ÂCarnegie Mellon Universityâs Center for Computational Analysis of Social and Organizational Systems include DyNet, NetWatch, OrgSim, and VISTA, and the reader is referred there for further information (see http://www.casos.cs.cmu.edu/). For truly large-scale multiagent model development efforts, a number of communities are developing domain-free MAS frameworks and toolkits. These include SWARM, developed in the Center for the Study of Complex Systems (see http://www.cscs.umich.edu) at the University of Michigan; the Java-based REPAST agent simulation environment (North et al., 2005; Tatara et al., 2006); and MASON, another Java-based multiagent simula- tion environment, developed at George Mason University (see http://cs.gmu. edu/~eclab/projects/mason/). At the time of this writing, it is unclear what, if any, inroads have been made into the DoD M&S community. Massively Multiplayer Online Gaming Americaâs Army is an MMOG developed by and for the Army (Zyda, Mayberry, McCree, and Davis, 2005). The game was designed as a recruit- ing (Belanich, Sibley, and Orvis, 2004) and training (Farrell, Klimack, and Jacquet, 2003) tool to paint a realistic portrait of combat in the U.S. Army. The game falls into a first person shooter (FPS) game genre, and all the
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 55 game features are based on the real world. However, it goes well beyond ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ï¿½ being an FPS game (Nieborg, 2004), since social and cultural factors are increasingly being embedded in both the scenarios and the attributes of the roles that the players can take on. Additional information on Americaâs Army is provided in Chapter 7. DIME/PMESII Models A number of behavior modeling efforts aimed at understanding large- scale behaviorsâat the societal and nation-state levelsâare under way to explore the effects that DIME actions will have across the full range of the PMESII context. These include DARPAâs IBC, PCAS, and ICEWS programs, the Air Forceâs SROM effort, and JFCOMâs SEAS program. The Integrated Battle Command (IBC) program (Allen, 2004) empha- sizes linked and networked behavior models that can support military plan- ning and decision making for dealing with asymmetric threats embedded in an urban environment. The approach clearly recognizes the importance of obtaining and maintaining a clear understanding of the complex socioÂ political context. In terms of planning and executing effects-based opera- tions (McCrabb, 2001), this translates into the analysis of the potential effects that a given set of DIME actions will have across the full range of PMESII variables. The key to successfully executing such encompassing analyses lies in the development of the embedded behavior models repre- senting the full range of PMESII variables and how they can be individually and collectively affected by specific DIME actions. A conceptual representation19 of the model âspaceâ is shown in Fig- ureÂ 2-3, in which the dimensions are the DIME dimensions, the PMESII dimensions, and the modeling paradigms themselves, this last shown as modeling âfamilies.â As noted in the program description (Allen, 2004; see http://www.afcea.org/events/pastevents/documents/AFCEAIICPanel.ppt): Each model in the family may represent its portion of the domain in a manner and level of fidelity quite different from other models. . . . The Modeling Paradigms include techniques such as: concept maps, social network models, influence diagrams, differential equations, causal models, Bayesian networks, Petri nets, event-based simulation, and agent based 19â learly, this is not intended to represent modeling ârealityâ in any sense but is merely an C attempt to illustrate (1) the concept of different modeling paradigms/families covering differ- ent portions of the DIME/PMESII modeling space; (2) the potential for their interacting, e.g., outputs of one driving the inputs of another; and (3) the possibility of uncovering âunintended consequencesâ through these interactions. But it must be recognized that, fundamentally, the figure is merely an illustration of the concept of multiple models interacting at multiple levels and nothing more.
56 2-3.eps bitmap image B FIGURE 2-3â IBC modeling space. SOURCE: Adapted from Allen (2004).
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 57 simulation. The need for a variety of modeling paradigms also stems from the fact that the different domains of knowledge do not lend themselves to being represented by one common paradigm such as an influence network. Also, human subject matter experts have preferences in the use of different paradigms and different paradigms fit different styles of thought.20 The figure also illustrates how different models in different families interact via their interconnections (inputs, outputs, and state interactions). An analyst may investigate the impact of a DIME action and a model may forecast a primary PMESII outcome, but that effect may also âstimulate another model that predicts an effect that stimulates another model and in a cascade manner, the family of models, in a symbiotic manner, may predict another effect. Such cascading can produce astonishing results because, while a human may grasp and master a single model, it is unlikely that a human can predict the complex interactions between models!â (Allen, 2004; see http://www.afcea.org/events/pastevents/documents/ÂAFCEAIICPanel.ppt). As noted earlier, DARPA has sponsored two other programs focused on the identification of potential global hot spots and the forecasting of their likely evolution over time: the PCAS program, directed at Â forecasting the likelihood of a nation-state collapse (Popp et al., 2006), and the Âfollow-on I Â ntegrated Crisis Early Warning System, which has as its goal âthe develop- ment of state-of-the-art computational modeling capabilities that can Âmonitor, assess, and forecast, in near-real time, a variety of phenomena associated with country instability.â The latter program is in its early stages of development (see http://www.darpa.mil/ipto/solicitations/open/07â10_PIP.pdf). The Air Forceâs Stabilization and Reconstruction Operations Model (SROM) (Robbins, Deckro, and Wiley, 2005) analyzes the organizational hierarchy, dependencies, interdependencies, exogenous drivers, strengths, and weaknesses of a countryâs PMESII systems using a complex set of interdependent system dynamics representations. SROM models a country system in a lumped-parameter fashion as a national model (NM), which is then defined in terms of its n regional submodels that interact with each other and the NM. Each regional submodel contains six functional submodels: the demographics submodel, the insurgent and coalition mili- tary submodel, critical infrastructure, law enforcement, indigenous security 20â he T two assertions made here are based on the program managerâs long experience in the M&S world and generally match what the modeling community has long known, namely, that (1) different domains often call for different modeling paradigms (e.g., modeling a Âsocial network is probably better represented by network modeling methods, than, say, by an argumentation framework) and (2) different domain experts have different preferences for representing their knowledge to others (e.g., some may be more expressive with a declarative expert system approach while others may be more facile with a graphically based Bayesian network formalism).
58 BEHAVIORAL MODELING AND SIMULATION institutions, and public opinion. The utility of SROM has been demon- strated using Operation Iraqi Freedom as a case study. Simulation Frameworks and Tools In addition to these domain-focused modeling efforts, there are many efforts devoted to the development of general-purpose frameworks that make the modelerâs job easier. As noted in Table 2-A1, these include the C2 modeling framework C3GRID, the team/organizational modeling frame- work DDD, the generic M&S frameworks FLAMES, ICET, and MATREX, the social network analysis tool ORA, and the collaboration decision aid framework SIAM. Many others exist in or are in development inside DoD, as well as outside in the academic and commercial worlds. Of particular note are the multiagent development and simulation environments com- mented on earlier (e.g., MASON, REPAST, and SWARM). Other Efforts In addition to these large efforts, there are hundreds of development efforts either recently concluded or just under way, varying dramatically in scale and focus, to produce representative and useful IOS models for the military. At one end is the spectacularly unsuccessful and terminated Joint Simulation Systems (JSIMS) effort, which attempted to be all things to all people, serving as DoDâs general M&S environment. At the other is the IUSS/IWARS M&S program, which is successfully focusing on small-team behavior at the squad level. In between are efforts like the now concluded MIDAS effort of the National Aeronautics and Space Administration to build an âend-to-endâ model of the human operator (of rotorcraft), and DARPAâs RAID program, aimed at forecasting adversary behavior in the urban environment. However, there is no general inventory of what models exist and at what level of technical readiness. As a result there is duplication of efforts and too little effort at making these existing models interÂoperable. Furthermore, there is a trend for well-educated military personnel with some computational training to develop small, special-purpose IOS models that meet specific needs. A little programming training, however, does not make a good modeler, especially when that modeler is unaware of the importance of cognitive, affective, organizational, social, and cultural factors. Major Challenges for Development of IOS Models for Military Applications The current status of IOS modeling in DoD is the result of the funding profile for M&S in the last 10-15 years. Beginning in 1995, DMSO began
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 59 centralizing funding and development efforts in M&S. The High Level Architecture (HLA) and the JSIMS systems are archetypes of efforts funded and managed by DMSO during that decade, with funding to other systems at the service level cut to focus on these centralized efforts. In the end, after spending some $1.8 billion in development funds, JSIMS was canceled and the services all had to try to recover from the loss of JSIMS plus the loss of development time, effort, and funding on service-specific M&S efforts. During this same decade, however, there was an increase in the fund- ing of models at the basic and applied research levels. This led to the development of a large number of models that are particularly relevant to the military and may even be used at commands, but that are gener- ally not yet accredited. Examples of tools that evolved in this period are SIAM, ORA, and SEAS. In addition, work in this period gave rise to the model-driven experimentation paradigm (MacMillan, Diedrich, Entin, and Serfaty, 2005). Interoperability Challenges While some utility has been derived from HLA, its requirement that everything be statically defined ahead of time and its reliance on interÂ operability at the source code change level mean that the interoperability of defense simulations and their ability to change as threats rapidly change are greatly diminished. Had a bit of time been spent in the mid-1990s to design a dynamically extensible, semantically interoperable simulation infraÂ structure, defense M&S interoperability would now be more advanced. Furthermore, such an effort would have paved the way for incorporation of some of the IOS models now emerging. Another difficulty with the centralized approach is that it assumes that modeling needs can be predefined. It is apparent that, as the military mission changes, M&S needs change, and new models are often needed immediately. Hence, an alternative distributed paradigm is needed that enables rapid access to new models and enables the military to make use of the increasing number of models that exist even when they were not devel- oped expressly for military purposes. A possible alternative paradigm has a plug-and-play distributed infrastructure with data distributed across sets of servers with appropriate access controls; multiple models and simulations for different purposes with appropriate access control; and documentation, intelligent tools for aiding the user in determining which tools can be used with which data, and web enablement. In this way, any developer could place a model in the distributed system.
60 BEHAVIORAL MODELING AND SIMULATION Data Collection and Validation Challenges As noted in the Urban Sunrise report (Air Force Research Laboratory, 2004), most models to be used in real-world settings need to be tied to data. For example, models of insurgents often need as a basis data on the insurgency, such as the number of insurgents, modus operandi, sources of support, means of interaction, weapons, and location of activities. Data collection, however, is often done piecemeal by relying on subject matter experts to go out and collect data after a need for those data has been demonstrated, or opportunistically, as when a soldier, adversary, or civil- ian provides unsolicited intelligence (e.g., when an insurgent group posts a video of an IED attack on the web). This means that new ways of thinking about validation are needed, and it means that the models need to operate with uncertain and incomplete data, but the science of model creation and validation with incomplete and uncertain data does not exist. The nature of the data that are, or can be, collected is often not consistent with the data requirements of the existing models. For example, a model may require data on who actually interacts with whom when all that is available is who is known to have participated in what events. Applications-focused tools do not exist, such as expert systems for identifying for the user what models in their arsenal can be used given their data. The data are streaming, and time and location information is critical for data to inform action. For example, knowing the location and time of IED attacks is critical to identifying courses of action to protect U.S. Âsoldiers from future attacks. However, in many cases, databases do not contain the time and location data. Moreover, even when the data exist, many types of models cannot make use of that information. Because important data are classified, many models are developed in a vacuum, without access to the real data. Representative and unclassified data would be highly valuable and would get a wider range of model developers involved. However, the disadvantage is that the models are often tested and validated using proxy data that are conceptually different from, and may not even have the same data fields as, the classified data. This can result in erro- neous assumptions of model validity at the classified level and in erroneous assumptions by the modelers about what needs to be modeled. In addition, there are across-the-board needs for better modeling infra- structure, methods to link models to streaming data, and improved model visualization systems. Finally, there is a need for socially intelligent tools for collecting and interpreting intelligence information, particularly on insur- gents and terrorists. IOS model-based fusion and data collection manage- ment techniques are needed. IOS models for identifying potential missing or erroneous data should also be developed.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 61 A key to successful IOS models in this area is the development of measures and procedures that are robust with respect to scale and missing data. For example, IOS models of the adversary may need to be able to scale to 107 (ten million) actors. Even when there are massive amounts of missing and erroneous data, IOS measures need to be robust, appropriate, and meaningful. Adversarial models need to be sensitive to cultural factors, particularly to alternative goals, preferences for actions, and gender roles. CONCLUSION Current military missions and todayâs operating environment have cre- ated a pervasive need for models that can capture and forecast the behavior of humans acting in social units, ranging from small groups and teams to neighborhoods, cultural and ethnic groups, and entire societies. IOS Âmodels are needed to understand adversary and nonadversary behavior and to forecast the effects of alternative courses of action on that behavior. Todayâs broader missions focus not just on COAs for conventional combat with well-identified adversaries, but also on COAs for influencing the attitudes and behaviors of noncombatants at levels of detail ranging from block-by- block urban operations to the stability of nation-states. The COAs to be analyzed include not just military actions but the broader DIME/PMESII dimensions that may influence behavior. IOS models are also needed for training and rehearsal, to create realistic environments in which the mili- tary may test planned COAs and learn new skills associated with cultural awareness, joint and coalition operations, and stability and support opera- tions. IOS models are valuable for design, evaluation, and acquisition as well. They can support the evaluation of potential contributions of new technologies to effective operations as well as the design of command and control organizational architectures that are effective for rapidly changing missions and new environments. Efforts are under way to meet the militaryâs needs for IOS models, but they are fragmented and uncoordinated, with no central direction, little information sharing, and no mechanisms to guard against duplication of effort in multiple locations. All of the current efforts face challenges for interoperability, with models developed from different perspectives unable to communicate in any meaningful way. Models also face data collection and validation challenges, with data collection efforts often piecemeal and unrelated to modeling requirements, and validation strategies frequently absent altogether. The chapters in Part II review the state of the art in IOS modeling to evaluate the extent to which current approaches can meet military require- ments as outlined above. On the basis of that review, we analyze where broad gaps exist and recommend a plan of action to fill those gaps.
62 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Selected IOS Models Acronym Acronym Expansion Description ACT-R Adaptive Control A cognitive architecture in which a neural network of Thought activation system controls the activation of a rule- based production system model for simulating and understanding detailed human cognition. ACT-R continues to evolve to perform the full range of human perceptual, cognitive, and motor tasks, supported largely only in the academic community. Has been hybridized with other models, notably IMPRINT and Soar. ADC Air Defense Models the operations of an AEGIS Cruiser Combat Commander Information Center (CIC) team performing air defense duties in a battle group using multiagent system (MAS) technology implemented in the Java programming language. Americaâs Â Massively multiplayer online game (MMOG), Army starting as a first-person shooter game and now evolving to more complex environments and tasks and used as a recruiting tool. BioWar Combines computational models of social networks, communication media, disease models, demographically accurate agent models, wind dispersion models, and a diagnostic error model into a single integrated model of the impact of a biological warfare attack on a city. BioWar moves beyond existing epidemiological models by accounting for the heterogeneity of social networks and the geographical distribution of people when forecasting disease outbreaks. C3GRID Command, Parametric C4ISR modeling capacity for network- Control, centric warfare. Provides the capability to simulate Communication the common operating picture management for a Grid Model given force structure at the platform level. C3HPM C3 Human Provides high-resolution modeling of individual Performance Model human operators in terms of task performance and human decision processes in the execution of combat tactics, techniques, and procedures. Built on top of IMPRINT and operates in the MATREX simulation environment.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 63 Sponsor/ Category Research Center Reference/Website Cognitive ACT-R Research http://act-r.psy.cmu.edu/ architecture and Group at Carnegie modeling framework Mellon University Model SPAWARSYSCOM http://www.movesinstitute.org/~shcalfee/ index.html Real-time game U.S. Army http://www.americasarmy.com/ environment Hybrid model Carnegie Mellon http://www.casos.cs.cmu.edu/projects/ incorporating social University, DARPA, biowar/index.html networks, disease CDC, NSF models, dispersion models Network modeling U.S. Army, http://www.msrr.army.mil/index. tool RDECOM cfm?RID=MNS_A_1001514 Modeling framework Army Research http://www.arl.army.mil/ARL-directorates/ Laboratory (ARL) HRED/imb/imprint/References.pdf continued
64 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description C3TRACE Command, Network modeling tool to support the evaluation Control, and of different organizational structures and Communicationsâ communications network topologies to evaluate Techniques for overall C3 system performance. Built on top of the Reliable MicroSAINT. Assessment of Concept Execution CART Combat Network modeling tool initially applied to single- Automation pilot model operating JSF and subsequently applied Requirements to a nine-member time critical targeting (TCT) cell Testbed in an air operations center (AOC). Built on top of IMPRINT, it enabled one of the first instances of integrating MicroSAINT with an external world model simulation. CLARION Connectionist Cognitive architecture for connectionist/neural Learning with representation of implicit (subsymbolic or neural Adoptive Rule network) knowledge and semantic representation Induction ON-line of explicit (symbolic chunks and rules) knowledge. Provides for explicit representation of static knowledge as well as acquisition of subsymbolic knowledge through learning over time. COGNET, Cognition as a COGNET is an executable cognitive architecture and iGEN Network of Tasks iGEN is the associated development environment. Both have been applied in a number of DoD- sponsored modeling exercises, most notably in the AFRL/HE AMBR air traffic control human behavior modeling and simulation program and in the Navy TADMUS antiaircraft defense modeling effort. Little or no technical literature appears to be available describing the technical details and therefore used little outside CHI Systems, its commercial developer. Construct A multiagent model of group and organizational behavior in which the agents communicate, learn, and make decisions in a continuous cycle, dependent on the perceptions and goals of the individual and the goals and culture of the group. When agents interact they communicate and learn both task knowledge and cognitive knowledge. These dynamic relationships are grounded in structuration theory, which is the notion of construction and reconstruction of the social system through human interaction based on rules and resources.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 65 Sponsor/ Category Research Center Reference/Website Modeling framework ARL www.hfes.org/web/BulletinPdf/ bulletin0405.pdf Modeling framework USAF AFRL/HE http://www.maad.com/MaadWeb/ ongoing_projects/onprojma.htm#Combat Cognitive Dept. of Cognitive http://www.cogsci.rpi. architecture for Science, Rensselaer edu/~rsun/clarion-ub.html individual entity Polytechnic Institute; modeling Army Research Institute (ARI) Cognitive USAF AFRL/HE and http://www.chisystems.com/ architecture and Navy SPAWAR model development environment Multiagent dynamic Carnegie Mellon http://www.casos.cs.cmu.edu/projects/ network model University, DARPA, construct/index.html ONR continued
66 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description CORES Complex A multiagent network simulation model that uses Organizational organizational, social, political, and economic Reasoning System dynamics to generate forecasts of the likely actions and responses of adversarial actors. Scenarios are represented in a framework consisting of actors, resources, goals, actions, effects, and relations. Based on these entities, the model generates forecasts of likely actions and responses of actors. Potential areas of application include military intelligence and learning, political and corporate negotiation, disaster relief and crisis management, and business intelligence. DDD Distributed Focuses on team functions that drive performance, Dynamic such as communications and coordination. Model Decision-making users can specify allocations of people, equipment, and material, and specify performance objectives/ constraints, such as job/mission objectives, timing, and coordination requirements. DDD models the resultant team/environment interactions based on empirically observed team/organization interactions and provides a simulation environment for calculating team performance metrics, based on a team performance model embedded in the simulator. DIAS Dynamic Object-oriented framework for integrating disparate Information multidisciplinary simulation models, supporting Architecture legacy code reuse, and modeling of cooperative System behaviors of agents. EPIC Executive-Process/ Cognitive modeling architecture for human Interactive Control information processing that accurately accounts for the detailed timing of parallel human perceptual, cognitive, and motor activity, in multitasking situations. Primarily an academic tool for researchers interested in fine-level details of perception and cognition. Applied to operator-centered design of undersea ship systems and many other systems. FLAMES Flexible Analysis A framework for developing constructive simulations Modeling and and interfaces between constructive, virtual, and live Exercise System simulations. It has applications that support scenario definition, scenario execution, scenario postprocessing and scenario visualization.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 67 Sponsor/ Category Research Center Reference/Website Multiagent network Carnegie Mellon http://www.casos.cs.cmu.edu/ model, incorporating University, DARPA, (Kowalchuck, Singh, and Carley, 2004) DIME/PMESII NSF factors Team/organization Aptima; AFOSR, http://www.aptima.com/a-sim.php performance AFRL, ARL, DOT, modeling NASA, NavAir, tool/environment Office of Naval Research (ONR) Generic simulation Argonne National http://www.dis.anl.gov/DIAS/ framework Lab, DIS Division Cognitive University of http://www.eecs.umich.edu/~kieras/ architecture for Michigan, ONR epic.html individual entity modeling Generic framework USAF AFRL/MN and http://www.ternion.com NAIC continued
68 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description HOS Human Operator HOS V is a MicroSAINT-based task-level simulation Simulator language for the individual human operator. Invokes micro models for primitives like perception, decision, and action. The output of HOS V consists of task performance timelines, errors, user-defined system performance measures, and component, person, and other resource utilization. Has been incorporated into COGNET to support modeling of low-level operator activities. IBC Integrated Battle The IBC framework provides a means of integrating Command disparate environmental and IOS behavior models to support the analysis of the potential effects that a given set of DIME actions will have across the full range of PMESII variables, at the nation-state level. Each model in IBC may represent its portion of the domain in a manner and level of fidelity quite different from other models. Modeling paradigms include such techniques as concept maps, social network models, influence diagrams, differential equations, causal models, Bayesian networks, Petri nets, event-based simulation, and agent-based simulation. ICET Integrated Concept Addresses modeling, simulation, and analysis of Evaluation Tool advanced cross-weapons communications concepts. Built on top of FLAMES. ICEWS Integrated Crisis Goal is âthe development of state-of-the-art Early Warning computational modeling capabilities that can System monitor, assess, and forecast, in near-real time, a variety of phenomena associated with country instability.â This is a relatively recent start with no publications as of this date. IMPRINT Improved A stochastic task network modeling tool for the Performance individual soldier. Task analysis is used as a starting Research point to assess the interaction of soldier and system Integration Tool performance. A network is constructed representing the flow and performance time and accuracy for operational and maintenance missions. Workload profiles for crew members are generated so the workload distribution and peaks and valleys can be examined. The underlying engine is the MicroSAINT task network modeling environment.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 69 Sponsor/ Category Research Center Reference/Website Language for ARL http://www.dtic.mil/dticasd/ddsm/closed/ individual operator ddsm0023.html task modeling Framework for DARPA http://www.darpa.mil/sto/solicitations/IBC/ integrating different index.htm (Allen, 2004) DIME/PMESII models Generic framework USAF AFRL/MN and http://www.ternion.com NAIC Decision aid with DARPA IXO http://www.darpa.mil/ipto/solicitations/ embedded models open/07-10_PIP.pdf of nation-state behaviors Human task ARL http://www.arl.army.mil/ARL-Directorates/ modeling HRED/imb/imprint/imprint.htm environment continued
70 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description IUSS, Integrated Unit Constructive, force-on-force model for assessing the IWARS Simulation System, combat worth of systems and subsystems for both Infantry Warrior individuals and small-unit dismounted war-fighters Simulation in high-resolution combat operations. Early versions were labeled IUSS. Current version, IWARS, is being built on IUSS Version 4 and the AMSAA Infantry MOUT Simulation (AIMS) models. JSAF Joint Semi- With its roots in the DARPA Synthetic Theater of Automated Force War (STOW) program and derived from ModSAF, the JSAF simulation provides entity-level simulation of air, ground, and maritime forces in support of command and staff training and mission rehearsal. The JSAF federation provides a distributed modeling and simulation (M&S) framework composed of multiple federates to represent a realistic synthetic environment; model C2, logistics, and weapon effects; provide automated reasoning of entities via simple task behaviors and more advanced pilot behavior modeling via TacAir Soar; and interface with simulation and real-world systems (e.g., DIS, HLA, C4I Gateways). Based on technology developed prior to OneSAF. JSIMS Joint Simulation A federation of service-unique models of service- Systems specific entities, based on a high-level architecture, common standards, and common protocols. JSIMS was going to be the primary M&S tool to support future joint and service training, education, doctrine development, and mission rehearsal for the Army, Air Force, Navy, DIA, DISA, NASM, TRANSCOM, and SOCOM. JSIMS was going to be progressively developed into a robust, interactive joint synthetic battlespace (JSB) for training strategic national joint tasks and joint and service tactical tasks in all phases of operations (mobilization, deployment, employment, sustainment, and redeployment). After nearly 7 years and $2 billion of investment, it was cancelled in 2004.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 71 Sponsor/ Category Research Center Reference/Website Modeling framework Army Natick Soldier http://nsc.natick.army.mil/media/fact/ss&t/ and specific small Center/AMSAA IWARS.PDF unit models for infantry behaviors Computer-generated Joint Forces http://afmsrr.afams.af.mil/ force (CGF) Command (JFCOM) index.cfm?RID=MDL_AF_1000066 application for Training and Analysis simulating a wide Center range of cross-service military entities M&S environment JFCOM N/A for all DoD needs continued
72 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description MAMID Methodology An integrated symbolic architecture, models high- for Analysis level decision making, with focus on the role of and Modeling affective factors (emotions and traits). MAMID of Individual models the cognitive appraisal process to dynamically Differences generate emotions in response to incoming stimuli and models the subsequent effects of these emotions on distinct stages of decision making. Its parametric methodology supports the modeling of multiple, interacting individual differences and facilitates the rapid creation of distinct agent profiles. MATREX Modeling The RDECOM-supported MATREX STO (science Architecture and technology objective) enables the integration for Technology of interoperable component engineering-level Research & simulations and models that conform to a common Experimentation architecture specification. MATREX is a framework, not a model, designed to integrate existing models into a robust representation of the battlespace (terrain, dynamic environmental effects, and physics- based modeling). It will be used to support and augment testing and training in either human-in-the- loop or constructive simulations. It will also support the integration of human behavioral models, such as IWARS, but does not support the direct construction of such models. MicroSAINT Microprocessor- A discrete-event network simulation language based Systems for developing task network models of humans Analysis of performing well-defined sequential tasks. It combines Integrated the operator with the external world model entities Networks (e.g., airplanes), making plug-in operator models difficult to implement. Many models have been developed for military simulations, and the basic language has been extended by development supported by ARL under the IMPRINT program and by subsequent extensions by AFRL under the CART program. The language is particularly popular with modelers having little background in human perceptual or cognitive processes because of its ease of use. More sophisticated researchers, particularly from the ACT-R community, have made efforts to integrate MicroSAINT models with more traditional cognitive architectures.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 73 Sponsor/ Category Research Center Reference/Website Cognitive Army Research http://www.psychometrixassociates.com/ architecture for Institute (ARI, hudl_mamid.pdf individual entity NASA, AFOSR) modeling M&S environment Army RDECOM N/A for all Army needs Simulation language Micro Analysis and http://www.maad.com/index.pl/ and tools for Design, ARL micro_saint developing task- network models of human behavior continued
74 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description MIDAS Man-machine Developed to support helicopter cockpit design for Integration Design the Army, with a primary focus on anthropometry, and Analysis physical layout of instrumentation, and operator System workload. Primitive sensory models drive a production rule system to drive activity selection effected by simple motor models of the operatorâs limbs. A Z-scheduler handles rule collisions, but its psychological basis is unclear. MIDAS was a standalone system with a single instantiation at NASA Ames, and little public documentation is available regarding the detailed cognitive structures employed. MINDS Modeling The MINDS Behavior Moderator Engine (BME) Individual has been developed as a plug-in for other cognitive Differences and architectures (e.g., ACT-R, OMAR, SAMPLE, Soar), Stressors as a means for generating personality- or stress- based moderators that can moderate structures or parameters of the target cognitive architecture, to emulate, for example, the effect of fatigue level on perception or fear on cognitive task performance. MINDS has been integrated with the SAMPLE cognitive architecture and embedded in the IWARS simulation environment to model infantry squad leader decision making. ModSAF Modular Semi- An outgrowth of the early semi-automated force Automated Forces (SAF) program to simulate red ground force entities (e.g., tanks) executing basic maneuvers and missions (attack, defend, etc.) while engaging blue forces commanding simulated ground force entities (e.g., tanks) in the simulation network (SIMNET) environment developed during the 1980s. The modular SAF (modSAF) was developed to support composable (red) SAF behaviors, to minimize recoding efforts needed for training under different battle conditions, tactics, etc. SAF behaviors can be operated by behind-the-scenes red entity operators.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 75 Sponsor/ Category Research Center Reference/Website End-to-end Army Aeroflight http://caffeine.arc.nasa.gov/midas/ workstation for the Dynamics New_MIDAS_Design.html design of multicrew Laboratory, NASA helicopter cockpits, Ames Research with embedded Center operator models Generic behavior Army Natick Soldier (Neal Reilly, Bachman, Harper, Marotta, moderator engine Systems Center, ONR and Pfautz, 2007)Â for use in individual (Neal Reilly, Harper, and Marotta, 2007) entity cognitive architectures Computer- Army Simulation, N/A generated force Training, and (CGF) application Instrumentation for simulating Command maneuvering ground (STRICOM) entities continued
76 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description OMAR, Operator Model OMAR is a cognitive architecture and simulation D-OMAR Architecture, environment to develop models of human operators Distributed OMAR interacting with a variety of other operators and nonhuman entities. The basic components are a production rule-based cognitive processor driven by inputs from production memory, long-term memory, and working memory, this last driven by auditory and visual inputs. The architecture relies heavily on a centralized, synchronous production rule framework. An initial version was programmed in LISP, limiting its usability; a more recent version is implemented in Java. OMAR has been used in a number of simulations, including military air traffic control. OneSAF, One Semi- OneSAF is a constructive modeling and simulation OOS, OTB Automated Forces, environment intended to replace entity-based OneSAF Objective simulations. OneSAF is designed for numerous System, M&S domain applications, including research, OneSAF Testbed experimentation, training, COA analysis, and mission planning. OneSAF models automated and semi-automated behaviors for entities and units up to the brigade level and supports the full spectrum of military operations, including urban missions. Designed as an extensible architecture, the OneSAF distribution includes tools for creating new components and behaviors to meet future modeling and simulation requirements. OOS was the predecessor system for OneSAF. OTB was the predecessor program for developing new technologies for OOS, focusing on test, integration, and user feedback. ModSAF was an earlier predecessor of all the programs. ORA Organizational A risk assessment tool for locating individuals Risk Analyzer or groups that are potential risks given social, knowledge, and task network information. After building the network by connecting the nodes (people) via links (relationships) to other nodes (people), ORA conducts a form of social network analysis (SNA) to assess risk of individuals in the network. ORA is essentially a network development and analysis tool.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 77 Sponsor/ Category Research Center Reference/Website Cognitive USAF AFRL/HE http://omar.bbn.com/manual/index.html, architecture for http://omar.bbn.com/ individual entity modeling Computer-generated Armyâs Program http://www.onesaf.org/, force (CGF) Executive Office for http://www.onesaf.net/ application for Simulation, Training, simulating a wide and Instrumentation range of military (PEO STRI) entities Social network model ONR, DARPA, ARL, http://www.casos.cs.cmu.edu/projects/ora/ building and analysis NSF, AFOSR software.html tool continued
78 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description PCAS Pre-Conflict A recently concluded DARPA program to investigate Anticipation and the effectiveness of different computational social Shaping science approaches to support forecasting the likelihood of a nation-state failure (e.g., Sudan). The PCAS architecture consists of four modules for data collection, modeling, gaming/shaping tools, and decision support tools. Computational modeling approaches include system dynamics, multiagent systems, Bayesian influence models, diffusion models, and regression modes. PMFServ Performance An integrated framework that permits one to Moderator examine the impacts of stress, culture, and emotion Function Server on decision making. PMFServ has been used to create and simulate the people and objects of a number of scenarios, including crowd scenes (civil unrest in the United States, urban conflict in the Mideast), asymmetric threat leaders and followers, the Black Hawk Down recreation in the UnrealTournamentâ¢ game engine, and world leader modeling in a diplomacy and strategy game. Over the past 5 years the instructor has been sponsored by DMSO, ONR, IDA, GM, Army, DARPA, JFCOM, and others. RAID Real-time Supports real-time forecast analysis of probable Adversarial enemy actions in urban operations against irregular. Intelligence and RAID leverages novel approximate game-theoretic Decision-making and deception-sensitive algorithms to continuously identify and update forecasts of likely enemy actions while continuously estimating likely deceptions in the available battlefield information. Significant effort in the program is being applied to evaluating the programâs performance relative to that of human analysts unaided by RAID. SAMPLE, Situation SAMPLE is a cognitive architecture comprised of GRADE Awareness Model modules for fuzzy rule-based perception, Bayesian for Person in the belief network-based situation awareness, and Loop Evaluation, production rule-based decision making. GRADE is an Graphical Agent agent development environment for rapidly creating Development SAMPLE models for different domains/tasks. Both Environment have been used with JSAF, IWARS, the EAAGLES air combat simulation, the FACET ATM simulation, and the UnrealTournamentâ¢ gaming engine.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 79 Sponsor/ Category Research Center Reference/Website Nation-state DIME/ DARPA/IXO (Popp et al., 2006) PMESII modeling methodologies Cognitive University of http://www.seas.upenn.edu/~barryg/ architecture for Pennsylvania, HBMR.html individual entity DMSO, ONR, IDA, modeling and U.S. Army, DARPA, associated agent JFCOM development environment Decision aiding tool Â DARPA IIXO http://dtsn.darpa.mil/IXO/ with game-theoretic programs.asp?id=43 model for adversary (Kott and Ownby, 2005) behavior forecasting Cognitive AFOSR, AFRL, http://www.cra.com architecture for ARL, DARPA, NRC, (Harper, Ton, Jacobs, Hess, and individual entity NSSC, ONR Zacharias, 2001) modeling and associated agent development environment continued
80 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description SEAS Synthetic An agent-based software development environment Environments that incorporates seven behavioral primitives: for Analysis and initiate, search, decide, execute, communicate, Simulation update, terminate. No attempt is made to model fundamental cognitive or social behavioral models, but a capability is provided for representing entities at the individual, organizational, and institutional level. Developers claim that the SEAS environment integrates multiple theories from various disciplines to program behaviorally accurate agents, but little has been available in peer-reviewed journals to substantiate that claim. JFCOM has been a strong supporter, especially in the attempts to model large- scale, nation-state-level projections (DIME/PMESII input-output forecasts) and COA assessments. SIAM Situational A collaborative decision aiding tool to help multiple Influence analysts and experts decompose and analyze complex Assessment Model problems. It consists of a user-friendly graphical interface that supports the development and exercising of influence networks, a utility function decision-theoretic approach that builds on belief networks. SIAM allows each factor or influencing relationship affecting a decision to be examined separately, yet it optimizes understanding of the overall impact of, and the interrelationships among, the contributing factors. Soar, Simulation of Soar is an operator modeling production rule system Soar-EPIC Adaptive Resource, in which existing rules propose potential operators Executive-Process/ that might be used to solve the current goal or Interactive Control problem. It is focused on problem solving and has its roots in GOMS. Its lack of a perceptual front end and motor back end has motivated hybridization with EPIC to provide these services. Although its psychological basis is less well-developed than other research-oriented models, Soar has been applied to a number of military systems modeling efforts (notably TacAir Soar).
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 81 Sponsor/ Category Research Center Reference/Website Software agent- Simulex, JFCOM http://www.simulexinc. based development com/products/case_studies/#seas-vis environment Visualization and SAIC http://www.saic.com/business/technologies/ decision aiding tool license/it/siam.Pdf Operator modeling University of http://sitemaker.umich.edu/soar/home, production rule Michigan, SoarTech http://www.soartech.com system continued
82 BEHAVIORAL MODELING AND SIMULATION Appendix TABLE 2-A1â Continued Acronym Acronym Expansion Description SPECTRUM Provides an environment with multicolored, multisided icons in an effort to simulate realistic situations that is conducive to MOOTW (SASO). SPECTRUM portrays the graphics and terrain of this environment and adds the human dimension, to account for the impact of economics, politics, regional populations, nongovernmental agencies (NGOs), and humanitarian relief agencies. SROM Stabilization and Analyzes the organizational hierarchy, dependencies, Reconstruction interdependencies, exogenous drivers, strengths, and Operations Model weaknesses of a countryâs PMESII systems using systems dynamics modeling techniques. SROM models a country in a holistic lumped parameter manner as a national submodel, which is then defined in terms of its n regions as a system of systems. Each regional submodel itself contains six functional submodels: demographics submodel, insurgent and coalition military submodel, critical infrastructure, law enforcement, indigenous security institutions, and public opinion. STELLA A simulation-based training environment to train soldiers in information operations. A cognitive model was constructed using Bayes inference nets and neural nets to guide combat models based on internal logic. At the time of this review, fuzzy set theory was being contemplated for modeling the propagation of rumors, and a mathematical submodel of IW was being developed using q-analysis and Boolean nets to study the structure and dynamics of IW.
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 83 Sponsor/ Category Research Center Reference/Website Sociocultural training National Simulation http://www.msrr.army.mil/ system Center (NSC) DIME/PMESII USAF AFRL/IF (Robbins, Deckro, and Wiley, 2005) regional or nation- state modeling environment Information warfare DISA, AFAMS http://www.disa.mil/ training system
84 BEHAVIORAL MODELING AND SIMULATION References Air Force Research Laboratory. (2004). Urban sunrise. AFRL-IF-RS-TR-2004-22, Final tech- nical report. Rome, NY: Author. Alberts, D.S., and Hayes, R.E. (2003). Power to the edge: Command and control in the infor- mation age. Washington, DC: Command and Control Research Press. Allen, J.G. (2004). Commanderâs automated decision support tools. Briefing to Âproposersâ sym- posium for DARPAâs Integrated Battle Command program Defense Advanced ÂResearch Projects Agency (DARPA). Available: http://www.afcea.org/events/pastevents/documents/ AFCEAIICPanel.ppt [accessed Feb. 2008]. Anderson, J.R. (1983). The architecture of cognition. Cambridge, MA: Harvard University Press. Axelrod, R. (2004). Modeling security issues of Central Asia. Prepared for CMT International, U.S. Govt. Contract # 2003*H513400*000. Available: http://www.personal.umich.edu/ ~axe/research/Security_Central_Asia.pdf [accessed Feb. 2008]. Belanich, J., Sibley, D.E., and Orvis, K.L. (2004). Instructional characteristics and motivational features of a PC-based game, research report 1822. Alexandria, VA: U.S. Army Research Institute for the Behavioral and Social Sciences. Brown, M.T. (1997). Terrorist use of weapons of mass destruction within the United States: Asymmetric warfare paradigm in the 21st century. Carlisle Barracks, PA: U.S. Army War College. Available: http://stinet.dtic.mil/cgi-bin/GetTRDoc?AD=ADA326609&Location= U2&doc=GetTRDoc.pdf [accessed Feb. 2008]. Defense and the National Interest. (2007). Boyd and military strategy. Available: http://www. d-n-i.net/second_level/boyd_military.htm#discourse [accessed July 2007]. Defense Modeling and Simulation Office. (2007). Modeling and simulation resource repository (MSRR). Available: http://www.msrr.dmso.mil/ [accessed April 2007]. The Drucker School, Claremont Graduate University. (2008). Perspectives on Peter. Available: http://www.cgu.edu/pages/4181.asp [accessed Feb. 2008]. Entin, E.E. (1999). Optimized command and control architectures for improved process and performance. In Proceedings of the 1999 Command and Control Research and TechÂ nology Symposium, Newport, RI. Farrell, C.M., Klimack, W.K., and Jacquet, C.R. (2003). Employing interactive multimedia instruction in military science education at the U.S. Military Academy. Presented at the 2003 Interservice/Industry Training, Simulation, and Education (I/ITSEC 2003) Confer- ence, Dec. 1â4, Orlando, FL. Frost, R. (1998). Simulation based acquisition. Washington, DC: Office of the Secretary of Defense. Glenn, R.W. (2000). Heavy matter: Urban operationsâ density of challenges. (RAND Mono- graph Report.) Santa Monica, CA: RAND. Harper, K.A., Ton, N., Jacobs, K., Hess, J., and Zacharias, G.L. (2001). Graphical agent de- velopment environment for human behavior representation. In Proceedings of the 10th Conference on Computer Generated Forces and Behavioral Representation, Orlando, FL: Simulation Interoperability Standards Organization. Johnson, D. (2004). Air Force modeling and simulation: âNew Vectorâ updates. Intercom, 45(12). Joint Chiefs of Staff. (2000). Joint vision 2020: Americaâs military: Preparing for tomorrow. Washington, DC: U.S. Government Printing Office. Joint Chiefs of Staff. (2002). Joint publication 3-06: Doctrine for joint urban operations. Washington, DC: U.S. Government Printing Office. Kasales, M.C. (2002). The reconnaissance squadron and ISR operations. Military Review, MayâJune, 52â58 [English version].
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 85 Kipp, J., Grau, L., Prinslow, K., and Smith, D. (2006). The human terrain system: A CORDS for the 21st century. Military Review, LXXXVI(5), 8â15. Available: http://www.army. mil/professionalwriting/volumes/volume4/december_2006/12_06_2.html [accessed Feb. 2008]. Kott, A., and Ownby, M. (2005). Tools for real-time anticipation of enemy actions in tactical ground operations. Paper presented at Simulation Interoperability Standards Organiza- tionâs Tenth Conference on Computer Generated Forces and Behavioral Representation, Orlando, FL. Kowalchuck, M., Singh, S., and Carley, K. (2004). CORESâComplex organizational Â easoning system. (Report No. CASOS Technical Report CMU-ISRI-04-131.) Pittsburgh, r PA: Center for Computational Analysis of Social and Organizational Systems, Carnegie Mellon University. Krulak, C.C. (1997). The three block war: Fighting in urban areas. Vital Speeches of the Day, 64(5), 139â141. Levchuk, G.M., Levchuk, Y.N., Luo, J., Pattipati, K.R., and Kleinman, D.L. (2002a). Norma- tive design of organizationsâPart I: Mission planning. IEEE Transactions on Systems, Man, and Cybernetics. Part A: Systems and Humans, 32(3), 346â359. Levchuk, G.M., Levchuk, Y.N., Luo, J., Pattipati, K.R., and Kleinman, D.L. (2002b). Norma- tive design of organizationsâPart II: Organizational structure. IEEE Transactions on Systems, Man, and Cybernetics. Part A: Systems and Humans, 32(3), 360â375. Levchuk, G.M., Levchuk, Y.N., Meirina, C., Pattipati, K.R., and Kleinman, D.L. (2004). Normative design of project-based organizationsâPart III: Modeling congruent, robust, and adaptive organizations. IEEE Transactions on Systems, Man, and Cybernetics. Part A: Systems and Humans, 34(3), 337â350. Levchuk, G.M., Yu, F., Levchuk, Y., and Pattipati, K.R. (2004). Networks of decision-making and communicating agents: A new methodology for design and evaluation of organizational strategies and heterarchical structures. In Proceedings of the 2004 International Command and Control Research and Technology Symposium, June 15â17, San Diego, CA. Levchuk, G.M., Yu, F., Levchuk, Y., and Pattipati, K. (2006). PERSUADE: Modeling frame- work for the design of modular Army organizations. In Proceedings of the 2006 Com- mand and Control Research and Technology Symposium, June 22â22, San Diego, CA. Lwin, M.R. (1997). Great powers, weak states and asymmetric strategies. Monterey, CA: Naval Postgraduate School. Available: http://stinet.dtic.mil/cgi-bin/GetTRDoc?AD=AD A340989&Location=U2&doc=GetTRDoc.pdf [accessed Feb. 2008]. MacMillan, J., Diedrich, F.J., Entin, E.E., and Serfaty, D. (2005). How well did it work? Measuring organizational performance in simulation environments. In W.B. Rouse and K.R. Boff (Eds.), Organizational simulation (pp. 253â272). Hoboken, NJ: John Wiley & Sons. Mahoney, R. (2005). Joint urban operations workshop on command and control and net centric environmental science and technology, executive summary. (Report No. IDA Doc. D-3132.) Alexandria, VA: Institute for Defense Analyses. McCrabb, M. (2001). Explaining effects: A theory for an effects-based approach to planning, executing, and assessing operations, version 2.0. Available: http://www.sci.fi/~fta/EBO. htm [accessed Feb. 2008]. National Research Council. (1998). Modeling human and organizational behavior: Application to military simulations. Richard W. Pew and Anne S. Mavor (Eds.). Panel on ModelingÂ Human Behavior and Command Decision Making: Representations for Military Simula- tions. Commission on Behavioral and Social Sciences and Education. Washington, DC: National Academy Press.
86 BEHAVIORAL MODELING AND SIMULATION Navy Modeling and Simulation Office. (2004). Autonomous agent-based simulation of an AEGIS cruiser combat information center performing battle group air-defense com- mander operations. S.H. Calfee, Masterâs thesis, Naval Postgraduate School, Monterey, CA. Available: http://stinet.dtic.mil/cgi-bin/GetTRDoc?AD=ADA414842&Location=U2 &doc=GetTRDoc.pdf [accessed Feb. 2008]. Neal Reilly, W.S. (2006). Modeling what happens between emotional antecedents and emo- tional consequents. Paper presented at the Agent Construction and Emotions (ACE 2006) Conference, April 18â19, Vienna, Austria. Neal Reilly, W.S., Bachman, J., Harper, K.A, Marotta, S., and Pfautz, J. (2007). Modeling the effects of behavior moderators for simulation-based human-factors design. In Pro- ceedings of the 51st Annual Meeting of the Human Factors and Ergonomics Society, Baltimore, MD. Neal Reilly, W.S., Harper, K.A., and Marotta, S. (2007). Modeling concurrent interacting behavior moderators for simulation-based acquisition tasks. Paper presented at Spring Simulation Multiconference (SpringSim â07), Norfolk, VA. Nieborg, D.B. (2004). Americaâs Army: More than a game. Bridging the gap: Transforming Knowledge into Action through Gaming and Simulation. In Proceedings of the 35th Conference of the International Simulation and Gaming Association (ISAGA), Munich, Germany: SAGSAGA. North, M.J., Howe, T.R. Collier, N.T., and Vos, R.J. (2005). The repast simphony runtime system. Paper presented at the Agent 2005 Conference on Generative Social Processes, Models, and Mechanisms, Argonne National Laboratory, October, Argonne, IL. Avail- able: http://repast.sourceforge.net/papers/Agent_2005_Repast_Runtime.pdf [accessed Feb. 2008]. Office of the Secretary of Defense. (1996). Study on the effectiveness of modeling and simula- tion in the weapon system acquisition process. Washington, DC: Author. Parsons, D., and Wittman, R. (2004). OneSAF: Tools and processes supporting a distributed development environment for a multi-domain modeling and simulation community. Presented at the European Simulation Interoperability Workshop (Euro SIW 04), June 28âJuly 1, Edinburgh, Scotland. Pattipati, K.R., Meirina, C., Pete, A., Levchuk, G., and Kleinman, D.L. (2002). Decision networks and command organizations. In A.P. Sage (Ed.), Systems engineering and management for sustainable development, Encyclopedia of life support systems. Oxford, England: UNESCOâEOLSS. Popp, R., Kaisler, S.H., Allen, D., Cioffi-Revilla, C., Carley, K.M., Azam, M., Russell, A., Choucri, N., and Kugler, J. (2006). Assessing nation-state instability and failure. Pre- sented at the Aerospace Conference 2006 IEEE, Boston, MA. Robbins, M., Deckro, R.F., and Wiley, V.D. (2005). Stabilization and reconstruction opera- tions model (SROM). Paper presented at the Center for Multisource Information Fusion Fourth Workshop on Critical Issues in Information Fusion: The Role of Higher Level Information Fusion Systems Across the Services, University of Buffalo. Available: http:// www.infofusion.buffalo.edu/ [accessed Feb. 2008]. Schaffer, P. (2005). Mapping the human terrain: GIS in support of cultural intelligence, Track: Defense and intelligence. Redlands, CA: ESRI GIS Mapping and Software. Tatara, E., North, M.J., Howe, T.R., Collier, N.T., and Vos, J.R. (2006). An introduction to repast modeling by using a simple predator-prey example. In Proceedings of the Agent 2006 Conference on Social Agents: Results and Prospects, Argonne National ÂLaboratory, A Â rgonne, IL. Available: http://repast.sourceforge.net/papers/Agent_2006_ÂRepast_ÂTutorial. pdf [accessed April 2008].
MILITARY MISSIONS AND HOW IOS MODELS CAN HELP 87 United Nations. (2002). World urbanization prospects: The 2001 revision (ST/ESA/SER. A/216). Department of Economic and Social Affairs Population Division. New York: Author. U.S. Air Force Scientific Advisory Board. (2002a). Predictive battlespace awareness to improve military effectiveness, volume 1: Summary. (Report #SAB-TR-02-01âauthors: G.B. Harrison, R.O. Johnson, M.R. OâNeill, M.W. Ganz, and T.F. Saunders.) Washington, DC: Author. U.S. Air Force Scientific Advisory Board. (2002b). Predictive battlespace awareness to improve military effectiveness, volume 2: Panel reports. (Report #SAB-TR-02-01âauthors: G.B. Harrison, R.O. Johnson, M.R. OâNeill, M.W. Ganz, and T.F. Saunders.) Washington, DC: Author. U.S. Air Force Scientific Advisory Board. (2005). Report on Air Force operations in urban environments, volume 1: Executive summary and annotated brief. (Report #SAB-TR- 05-01âCommittee on Air Force Operations in Urban Environments.) Washington, DC: Author. U.S. Army. (2002). U.S. Army Training and Doctrine Command (TRADOC) pamphlet (Pam) #525-3-90, The United States Army objective force operational and organizational plan for maneuver unit of action. Washington, DC: U.S. Army Training and Doctrine Command. U.S. Department of Defense. (1995). Modeling and simulation (M&S) master plan. DoD #5000.59-P. Washington, DC: Office of the Undersecretary of Defense for Acquisition and Technology. U.S. Department of Defense. (2006). Quadrennial defense review report. Washington, DC: Author. Vick, A.J., Stillion, J., Frelinger, D.R., Kvitky, J., Lambeth, B.S., Marquis, J.P., and Waxman, M. (2002). Aerospace operations in the urban environment. (RAND monograph report #MR-1187.) Santa Monica, CA: RAND. Wait, P. (2007). Defense domain, civilian awareness. Elder, Garcia walk two sides of the cyberÂ security beat. Government Computer News, January 22. Available: http://integrator. hanscom.af.mil/2007/January/01252007/01252007-25.htm [accessed March 2008]. Zyda, M., Mayberry, A., McCree, J., and Davis, M. (2005). From Viz-Sim to VR to games: How we built a hit game-based simulation. In W.B. Rouse and K.R. Boff (Eds.), Organi- zational simulation (pp. 553â590). Hoboken, NJ: John Wiley & Sons.