Task-Driven Automated Component Deployment for Ambient Intelligence Environments

Abstract

This article presents a strategy to deploy component-based applications gradually in order to match the functionality of pervasive computing applications onto the current needs of the user. We establish this deployment strategy by linking component composition models with task models at design-time, from which a runtime deployment plan is deduced. Enhanced with a Markov model, this deployment plan is able to drive a component life cycle manager to anticipate future deployments. The result is a seamless integration of pervasive computing applications with the user's tasks, guaranteeing the availability of the required functionality without wasting computing resources on components that are not currently needed.