Alan Davy

Task Driven Service Composition for Pervasive computing environments

Telecommunications Software and Systems Group (TSSG)
Waterford Institute of Technology (WIT)

Pervasive computing environments, such as smart spaces, offer a wide variety of heterogeneous services to users. These environments are becoming more and more service oriented and modular. With the availability of so many services to perform diverse individual tasks within the smart space, users find themselves having to design custom-built applications to perform certain tasks, or invoke each service individually to achieve required tasks. This is a time consuming activity and usually means that setting up and becoming familiar with the environment takes longer than performing the intended tasks. This also means the user must have a reasonably in-depth knowledge of how to configure services based on his/her requirements.

This situation will not be acceptable in the future. The fact that the environment is pervasive should also mean that the services it provides are pervasive. Services are being designed to interact with each other, not just with the end users. The user must be able to take advantage of this, without being aware of how these services interact. Task-driven computing (Zhenyu Wang & David Garlan 2000) aims to translate the users computing intention i.e. task, into a collection of dynamically organised services interacting to satisfy the requirements of the users task.

In order for a system to be able to translate user intent, or a task definition into a composition of services, the services must first have well defined interfaces and more importantly they must be accurately described not only by functional but non-functional information such as the semantic of the service (Justin O'Sullivan, David Edmond, & Arthur Ter Hofstede). In the Web Service arena, semantic web languages, such as the Web Ontology Language (OWL)(Dean et al. 2002) or DAML+OIL (Horrocks et al. 2001), provide the foundations for such sufficiently rich descriptions.

The aim of this research work is to develop a process for the composition and management of services within a smart space environment. The following sections outline current technologies that are being employed in this area, an overview of what tasks are and how they may be represented to a system, an overview of what service semantics are and how they are required for service composition, and a section on the process and requirements of service composition for pervasive computing environments. Finally the paper outlines a Policy management system for service, in use at WIT for the Mzones program, and how service composition will be integrated into this platform.