Modeling Service Composition Reliability in Pervasive Computing
Klaus Marius Hansen, Jeppe Brønsted
Science Institute, University of Iceland,
Dunhaga 3
IS-107 Reykjavik, Iceland
Email: kmh@hi.is
Abstract
In pervasive computing, new functionality is often realized by recombining existing functionality in the form of service compositions. As pervasive environments inhabit even greater parts of our lifes, it becomes important to which degree we can justifiably trust service compositions. In this report, we address an important component of trust, namely reliability of service compositions. In particular, we present an approach to model and estimate the reliability of service compositions in terms of reliability of services and implement the approach in an existing, REST-based service framework for pervasive computing. Our evaluations through implementation and simulation show that our approach can model useful service composition problems, is implementable on resource-constrained devices, and is sufficiently accurate in reliability estimation.