Dynamic planning and weaving of dependability concerns for self-adaptive ubiquitous services

Abstract

Ubiquitous computing and service-oriented computing enable the development of a new trend of applications that can opportunely interact with services discovered in the surrounding landscape. Al-though sporadic, this type of interaction requires the deployment of dependable mechanisms to ensure the correct completion of the in-teractions. However, the integration and the configuration of these mechanisms depends not only on the type of service accessed, but also on the surrounding environment. Such a variability requires an extensive effort of the developers to support the alternative mech-anisms. Thus, to reduce this effort, we propose to integrate the Aspect-Oriented Programming (AOP) principles into the MUSIC planning-based adaptation middleware in order to dynamically plan and weave dependability concerns into the application depending on the execution context. In particular, this paper introduces our continuous support for AOP, which includes i) a uniform model for describing the dependable application configurations and ii) a modular middleware platform for weaving and configuring the de-pendability concerns when necessary.