Approach for Model Driven Development of Multi-agent Systems for Ambient Intelligence

Abstract

Ambient Intelligence systems are smart environments, which are reactive and proactive to people making their actions safer and more efficient. In respect to the relation with users, they should be non-obtrusive, context aware, personalized, adaptive and anticipatory. Reliable way to achieve these main features is the choice of appropriate development approach that will ensure the reusability, portability, scalability and interoperability of ambient intelligence applications. The model driven development is a promising approach for development of different software applications using models at different levels of abstraction and applying model transformation to code generation. In this paper an extension to the traditional model driven development approach is suggested based on model driven architecture and compositional models as core elements for automation of expert activities, enabling the context-awareness of application. The ambient intelligence systems are represented as layer-based multi agent systems that enable the division of system elements into levels, reduce the coupling between modules and facilitates abstractions as well as the distributions of responsibilities. The suggested approach is illustrated with an example from the field of health and medicine, namely the development of Holter monitoring system using context based on O-MaSE methodology and agentTool III.