Agent Execution

• An agent is created by extending the Agent class, implementing various behaviors which are descendats of Behavior, then instantiating them and adding them to the agent once it starts up.
• The calls are addBehavior and removeBehavior.
• There is one thread per agent.
• Within an agent the behaviors use cooperative multitasking. That is, one behavior must give up control before the next one can start. This was used to avoid synchronized methods which (they claim) are 100 times slower.

José M. Vidal .

7 of 21