View composition in multiagent architectures

Experience with a multiagent architecture for an industrial Automatic Guided Vehicle Transportation System (AGVTS) reveals a lack of support for relating and composing multiple views in architectural descriptions. This prevents separating concerns while maintaining a good overview of the architecture as a whole, hampering changeability. We argue for explicit support for relating and composing views. Concretely, we propose three types of relations that allow specifying and composing structural views. As validation, we extend xADL with concepts of relations and compositions. The usability and consequences of this extension are assessed in the context of the multiagent architecture for the AGVTS. We learn that the approach improves changeability by offering explicit support for (1) separating concerns in separate views and (2) composing the overall architecture. There is an additional cost of documenting relations. Nevertheless, documenting view relations gives rise to architectural descriptions that are more expressive and less ambiguous.status: publishe