Current and future military missions will increasingly involve multinational forces. 1,2
Ideally, such forces should be rapidly drawn together, flexibly led, and responsively deployed. Moreover, these multinational forces will need to address a wide range of dynamically evolving tasks to support defensive, policing, or humanitarian missions, both locally and throughout the world. Such missions often involve objectives set by consensus in the international community (for example, through the United Nations). These missions require agility and effective use of limited resources to achieve complex and multiple objectives. Participants will need to assemble and maintain a "virtual organization" reflecting the interests of the participating countries and nongovernmental organizations and their capabilities.
The situation in these missions is typically fluid, resulting in frequent changes in strategies and objectives. Commanders require effective means to clearly define and relay the mission objectives to planning and logistics staff, and from there to coalition partners and field personnel. They also need means to inform the public and media and to counter misinformation.
In this context, academic, government, and industry researchers worldwide are working together to develop knowledge systems to support such coalition operations. Topics of interest to this growing community include
• Innovative theory and techniques for forming coalitions and supporting similar virtual organizations
• Applications and requirements for knowledge-based coalition planning and operations management
• Knowledge-based approaches to command and control
• Knowledge-based approaches to coalition logistics
• Knowledge-based approaches to Operations Other Than War, such as peacekeeping missions and other humanitarian operations
• The theory, principles, and practice of using multiagent systems in coalition operations
• Tools and techniques for knowledge-based simulation and modeling of coalition operations
• Security and maintenance of private information or knowledge in coalition operations
• Autonomous versus centrally managed coalition operations
The first Knowledge Systems for Coalition Operations meeting took place in Edinburgh, Scotland, in May 1999; it focused on knowledge-based planning for coalition operations. 3
At that meeting, some participants formed an international working group to encourage international collaboration on KSCO.
The KSCO 2002 conference, held in Toulouse, France, in April 2002, was the second such meeting. (The sidebar
lists URLs for the KSCO conferences and other related Web sites.) It brought together practitioners and key decision makers in coalition operation management with researchers from knowledge representation and reasoning, planning, multiagent systems, and related areas to exchange experiences and ideas, share inspiration, and suggest novel concepts. Practitioners could meet face-to-face and learn and appreciate the value of recent research achievements. Researchers could share their latest ideas with a sympathetic, informed audience and receive feedback, and could develop or deepen links to potential end users of these ideas. The conference covered a wide range of topics, including
• Organization of coalitions
• Shared models or standards
• Agents for coalition
• Coalition architectures
• Functional and operational areas
• Coalition research programs
This special issue of IEEE Intelligent Systems
presents articles based on a small selection of the papers from the KSCO 2002 proceedings. 4
It gives a flavor of the breadth of research being undertaken and the intelligent systems technologies being drawn upon to support coalition operations.
In "A Knowledge-Based Approach to Coalition Formation," Michal P
ek, Vladimír Ma
ík, and Jaroslav Bárta focus on the complex problem of planning humanitarian-relief operations. A key challenge of this domain is that participants in such efforts resist central-planning approaches that require all parties to share detailed information about goals and resources. In contrast to these approaches, the authors combine classical negotiation mechanisms with acquaintance models and social-knowledge techniques to reduce communication traffic and confine most information sharing to subcommunities of agents in voluntarily formed alliances.
In "Coalition Agents Experiment: Multiagent Cooperation in International Coalitions," David Allsopp and his colleagues describe their Coalition Agents Experiment. CoAX is an international collaboration to investigate and demonstrate agent-based services to support coalition operations in realistic scenarios involving actual military systems. In CoAX, the DARPA CoABS Grid ( http://coabs.globalinfotek.com) provides interoperability between agent systems. CoAX uses this interoperability to structure agents into domains reflecting organizational, functional, and national boundaries; security and information access policies govern these domains. A series of demonstrations show how agents support the rapid, coordinated construction of coalition command structures for intelligence gathering, visualization, planning, and execution. In addition, they show how humans manage the flow of using advanced process and task management tools.
In "Force Templates: A Blueprint for Coalition Interaction within an Infosphere," Robert Marmelstein shows how to integrate an information system with coalition operations. The Joint Battlespace Infosphere organizes and distributes combat information to all levels of participants. In the JBI, force templates are the principal mechanisms for rapidly integrating battlespace entities and clients. A force template contains the information that lets participating entities "plug in" to the JBI: identity, requirements, services, and operation mode. It also includes the context and policy that define an entity's contract with the JBI. So, the JBI can achieve the flexibility it needs to rapidly and reliably share information among coalition members.
In "Dynamic Coalition Formation among Rational Agents," Matthias Klusch and Andreas Gerber survey results from the traditional field of static-coalition formation and explain the specific properties of dynamic coalitions. They also offer a novel scheme that enables rational agents to react to events that occur dynamically during coalition formation. An agent designated as the coalition leader continually attempts to improve its coalition's value, simulating possible coalition reconfigurations and informing members about the resulting alternatives. In turn, the other agents report their estimates of the quality of relevant services. By combining insights from several different approaches and by careful experimentation, Klusch and Gerber are making steady progress in this important research area.
Since the first KSCO meeting, much progress has occurred in theory, research, and experimentation related to knowledge systems that support international military and humanitarian-relief operations. A large multinational project exploring systems for the Commander in Chief of the 21st century has begun a coalition element in its work—see the sidebar
for related URLs. Nevertheless, much work remains, both on the research front to generalize and formalize results to date and on the application front to test and deploy the most mature of these results in realistic settings.
We are grateful to the reviewers for this special issue: Robert Brennan, Marco Carvalho, Joerg Denzinger, Jim Doran, Martyn Fletcher, Jan Jelinek, Matthias Klusch, Shri Kulkarni, Michael Luck, Vladimír Ma
ík, Jörg Müller, Paolo Petta, Mihaela Ulieru, Rainer Unland, and Andrzej Uszok.
is the technical director of the Artificial Intelligence Applications Institute and holds the Personal Chair in Knowledge-Based Systems at the University of Edinburgh's Centre for Intelligent Systems and Their Applications. He works on knowledge-based planning and activity support systems and is involved with industrial and governmental organizations deploying AI technology in the UK, Europe, Japan, and the US. He graduated in computer studies from the University of Lancaster and received his PhD in machine intelligence at the University of Edinburgh. Contact him at the Artificial Intelligence Applications Inst., Univ. of Edinburgh, 80 South Bridge, Edinburgh EH1 1HN, UK; firstname.lastname@example.org; www.aiai.ed.ac.uk/people/staff/bat.html.
Jeffrey M. Bradshaw
is a research scientist at the University of West Florida's Institute for Human and Machine Cognition. A Fulbright Senior Scholar, he is former chair of ACM SIGART and a member of the NASA Ames RIACS Science Council and of the Autonomous Agents Steering Committee. Among other publications, he coedited Knowledge Acquisition as a Modeling Activity
(John Wiley & Sons, 1993), Software Agents
(AAAI Press, 1997), Software Agents for the Warfighter
(ITAC, 2002), and the forthcoming Handbook of Agent Technology
(AAAI Press). He received his PhD in cognitive science from the University of Washington. Contact him at the Inst. for Human and Machine Cognition, Univ. of West Florida, 40 S. Alcaniz, Pensacola, FL 32501; email@example.com; www.coginst.uwf.edu/users/?jbradshaw.
Michal P chou ek
is an assistant professor in artificial intelligence at the Czech Technical University's Department of Cybernetics. He also heads the Agent Technology Group at the Gerstner Laboratory and is a senior consultant with the software company CertiCon. His research focuses on agent technologies, multiagent systems, coalition formation, and production planning. He received his Dipl. Ing. in technical cybernetics from the Czech Technical University, his MSc in information technology from the University of Edinburgh, and his doctorate in artificial intelligence and biocybernetics from the Czech Technical University. Contact him at the Czech Technical Univ., Faculty of Electrical Eng., Dept. of Cybernetics, Technicka 2, 166 27 Prague 6, Czech Republic; firstname.lastname@example.org; http://labe.felk.cvut.cz/~pechouc.