Pervasive computing presents an exciting realm where intelligent devices interact within the background of our environments to create a more intuitive experience for their human users. Context-awareness is a key requirement in a pervasive environment because it enables an application to adapt to the current situation. Context-awareness is best facilitated by a context management system that supports the automatic discovery, retrieval and exchange of context information by devices. Such a system must perform its functions in a pervasive computing environment that involves heterogeneous mobile devices which may experience intermittent connectivity and resource and power constraints. The objective of the chapter is to describe a robust and adaptable context management system. We achieve an adaptable context management system by adopting the autonomic computing paradigm, which supports systems that are aware of their surroundings and that can automatically react to changes in them. A robust context management system is achieved with an implementation based on widely accepted standards, specifically Web services and the Web Services Distributed Management (WSDM) standard.
TopBackground
The notion of pervasive, or ubiquitous, computing goes back to the seminal writings of Weiser (1991). He defines it to be a state where computing devices are so pervasive and critical to our activities that they are taken for granted and effectively disappear into the background. Recent progress in several areas, including the development of smaller and more powerful computing and communication devices, the connectivity in both wired and wireless networks and the emergence of accepted standards for data transfer and presentation (for example HTTP, XML and WAP), are bringing the vision of pervasive computing closer to reality.
Context-awareness is one of the cornerstones of the pervasive computing paradigm. A context-awareness framework provides mechanisms to support context-aware applications. Satyanarayanan (2001) points out that a context-awareness framework must address a number of issues:
- •
How is context represented and stored?
- •
How is the stored context accessed?
- •
What are the minimum services needed to make context-awareness feasible?
- •
How is context acquired? Context may be part of a user’s personal computing space or may have to be sensed in real-time from the environment.
- •
What are the relative merits of different location-sensing technologies?
A number of frameworks to support the development of context-aware applications have been proposed. The Context Toolkit (Dey, Abowd & Salber, 1999) consists of context widgets and a distributed infrastructure to host the widgets. Context widgets encapsulate context information and hide the details of context sensing. The infrastructure includes services to store, share and protect context. SOCAM (Gu, Pung & Zhang, 2005) is a service-oriented middleware to support context-aware applications. It is based on Web services and provides services for service discovery and context storage, provision and interpretation. The Java Context Awareness Framework (JCAF) (Bardram, 2005) is a service-oriented infrastructure that provided context acquisition, management and distribution through a network of cooperating context services. Context services are Java entities that provide a well-defined API.
A context management system addresses the first two issues raised by Satyanarayanan. It supports the discovery and understanding of local services, devices and environmental constraints (daCosta, Yamin & Geyer, 2008). It is typically made up of a context model and a context manager middleware that implements the model. The context model provides the overall structure of the framework, and specifies how interactions take place between devices. The context manager middleware is the software that implements the interaction specified by the model.
Strang and Linnhoff-Popien (2004) present a survey of context models. They identify several main types of models:
- •
Key-value models represent context values as (variable, value) pairs and are frequently used in distributed services frameworks.
- •
Markup scheme models integrate the model schema and values using markup languages such as XML (Bray, Paoli, Sperberg-McQueen, Maler, & Yergeau, 2006).
- •
Graphical models have been derived from generic modeling methods such as UML and ORM.
- •
Object-oriented models exploit the encapsulation and reusability present in an object-oriented approach. The details of context processing are encapsulated at the object level and access to context information is only through specified interfaces.
- •
Logic-based models formulate the context as a set of facts, expressions and rules.
- •
Ontology-based models provide a uniform way of specifying a model’s core concepts as well as an arbitrary amount of subconcepts and facts, which facilitates sharing and reuse of contextual knowledge.