Control Aspects in Multiagent Systems

The work described in this chapter develops a control framework formodelling, analysis and execution of parallel/distributed time-dependent multi-agentsystems. The goal is to clearly separate agent behaviours from crosscutting controlconcerns which in general are orthogonal to a specific application and transparentlyaffect and regulate its evolution. The approach centres on a minimal computationalmodel based on actors with asynchronous message-passing and actions. Actors arethe basic building blocks for modelling the business logic of an application. Actionsmodel activities needed by actors, which have a time duration and require specificcomputing resources to be executed. Action execution can be either preemptableor not preemptable. Actions are the only abstraction units which have to be reifiedwhen passing from model analysis to model implementation. Therefore, the useof actions favours model continuity, i.e., a seamless transformation from modelanalysis by simulation to model implementation and real time execution. Differentpluggable control strategies ranging from pure concurrent to time sensitive (real-timeor simulated-time) were implemented. Control strategies are compliant with agentmobility and resource availability. For demonstration purposes, the realized controlframework was tailored to the JADE distributed agent infrastructure. This chapterfirst describes the control framework and its prototyping in JADE. Then presents twocase studies. The first one is devoted to a thorough assessment of the timing behaviourand performance of a company help desk system. The second one is concerned withthe schedulability analysis of a real-time tasking set. Finally, directions of on-goingand future work are drawn in the conclusions.