Small & Simple: Application-Specific Multi-Agent Systems

Small & Simple: Application-Specific Multi-Agent Systems

Michael Barlow (University of New South Wales, Australia) and Robert Cox (University of Canberra, Australia)
DOI: 10.4018/978-1-59140-640-2.ch002
OnDemand PDF Download:
No Current Special Offers


This chapter discusses the use of abstract multi-agent models of conflict — ABDs (Agent-Based Distillation) — for security and defense purposes. The chapter promotes a complex-systems, bottom-up approach to modeling conflict in a highly abstracted manner. Three ABDs drawn from three application areas – Operations Analysis, Wargaming (decision support) and Counter-Terrorist Activity — are used as illustrations. The history, philosophy and design approach of ABDs are discussed, including the key features of abstraction, bottom-up modeling, ease of use and rapid exploration of the parameter space of a problem. Challenges facing the successful design, implementation and deployment of ABDs are discussed, including appropriate level of abstraction, validation, acceptance, parameter setting and understanding of the range of outcomes generated. The Conceptual Research Oriented Combat Agent Distillation Implemented in the Littoral Environment (CROCADILE) ABD is used to illustrate the design of an ABD, highlighting that they are not complex pieces ofsoftware by military simulation standards. A patrol scenario is constructed in which a team of patrolling agents seeks to intercept an infiltrator crossing a border. The relative benefits of increasing sensor and communication capabilities of the patrol are explored in terms of the patrol’s ability to successfully intercept the infiltrator. The Tactical Decision Support System (TDSS) ABD shows a different application of the technology — in this case, wargaming Courses Of Action (COA). An experiment is described utilizing officers from the Australian Royal Military College to wargame COA using both traditional and TDSS approaches. For the TDSS approach, significant improvement in outcomes across a range of criteria was found by both the officers using the system and the evaluating staff. Finally, the System for Life and Intelligence Modeling (SLIM) ABD is used to explore the consequences of a government’s arrest policies in a homeland defense scenario. The model incorporates different geographic regions (homeland, foreign deployment, etc.), borders, terrorists, civilians, police, military, forensic evidence and bomb attacks. The consequences of lowering or raising the arrest threshold in terms of the amount of evidence required against an individual agent are shown in terms of bomb attacks, deaths, terrorists arrested and civilians wrongly arrested.

Complete Chapter List

Search this Book: