The deployment configuration diagram describes where the agents are located and which different elaborating units need to communicate in order to permit the communications among the agents. As usual, elaborating units are shown as 3-D boxes (see next figure). Agents are shown as components; their name is in the form agent-name: agent-class. Communications among agents are represented by dashed lines with the communicate stereotype, directed as in the R.D. diagram. For each communication described in the R.D. diagram occurring between agents in different elaborating units, a dashed line is drawn. The receiving agent has an interface to show that it is capable of dealing with that communication. (I.e. It understands the protocol used.) An extension of the UML syntax is used in order to deal with mobile agents moving from one computer to another. A dashed line with the move_to stereotype represents it.
In this diagram it is also possible to specify the hardware devices used by the agents (sensors and effectors) and the modes of communication among agents in different elaborating units (traditional/wireless networks, for example). If two agents in different elaboration nodes need to communicate (as stated in the previous phases of the design), a path of connection should be provided between the two nodes.
These constraints about the connections could also be dynamic. In fact, if agent A needs to communicate with agent C, but moves across the network in figure, we need to introduce the connection constraints as dependent on agent A’s position. We introduce an OCL constraint in all the needed connections for this specific purpose.
Next phase: Code Reuse
Previous phase: Single-Agent Behavior Description
Home: PASSI 2 homepage