Model-driven development of Particle System Families

Ambient-intelligence systems are characterized by a shift away from desktop computers to a variety of devices, which are unobtrusively embedded in the user environment. The usage of small computing nodes (particles) is one of the most efficient ways of their implementation. Such nodes are often installed in different appliances within an Ambient-intelligence system. In each case there is a joint particle infrastructure, which is reused as-is, but there are varying sensors, actuators and application logic deployed with the particle. This leads to a significant amount of software variability developers are confronted with. The combination of this variability with the complexity of the particle implementation burdens software development activities considerably. In this paper an approach is proposed that reduces the effort of these activities by bringing together model-driven development with product line engineering.