Context-aware computing is one technology to support wireless ubiquitous computing as it enables services and applications in the wireless communication ...
W. Kellerer, A. Tarlano: Context Aware Wireless Ubiquitous Computing, in Proc. 10th WWRF, New York, NY, USA, October 27-28, 2003.
Context-Aware Wireless Ubiquitous Computing Wolfgang Kellerer, Anthony Tarlano Future Networking Lab DoCoMo Communications Laboratories Europe GmbH Landsbergerstr. 312 80687 Munich, Germany {kellerer, tarlano}@docomolab-euro.com Abstract Marc Weiser's vision [1] on ubiquitous computing includes the demand for wireless solutions towards a large scale inter-connected environment where all participating nodes are acting autonomously, but transparent to the user. Context-aware computing is one technology to support wireless ubiquitous computing as it enables services and applications in the wireless communication environment to benefit from environmental information. In particular, context aware computing describes a paradigm, where an application uses knowledge related to a set of environmental states and settings to determine and adapt the applications behavior. The work presented in this paper describes an architecture supporting context aware computing in wireless ubiquitous computing environments based on the concept of context spaces as virtual meeting points for communities of ubiquitous computing peers. Keywords Context awareness, ubiquitous computing, community building, peer-to-peer communication, service platform enabling technologies
1 Introduction There has long been interest in the wireless services research community to enable applications to deliver services, to users, benefiting from increased access to information regarding the context surrounding the user. A wireless computing environment is especially characterized by dynamic device mobility of wireless devices. We believe that Weiser's 1991 vision [1], of a ubiquitously enabled computing background is built on the realization of a very loosely structured computing environmen, was the catalyst for recent ubiquitous computing research. We regard wireless technology as the main enabler for such environment. Having the definition of ubiquitous computing as a convergence of mobile and pervasive computing in mind [2], we therefore address wireless ubiquitous computing (or short: wireless computing) in this paper. In this sense, Weiser envisioned large numbers of embedded wireless devices would create a ubiquitous execution environment for applications. Currently at a top-level context-awareness can be characterized by the paradigm of context-aware computing [3]. It is our view that beyond the single paradigm of context-aware computing, many other technology building blocks will be required to realize truly usable context-awareness. Context-Aware Computing, refers to a paradigm where an application uses knowledge related to a set of environmental states and settings to determine and adapt the applications behavior. Context perception in a dynamic ubiquitous environment will inevitably be more successful with a communication enabled environment. That means, in order for applications to have context-awareness, devices such as active and passive sensors, would need to expose the state of the environment using a flexible communication paradigm. The peer-to-peer messaging concept is regarded as a promising candidate for an application-layer communication protocol for sensor devices. In general, a Peer-to-Peer (P2P) system, refers to a networking paradigm and a class of systems, using an overlay addressing scheme, to create a dynamic non-centralized environment for connected peers to share messages. Peers can offer and consume resources from connected peers. Recent P2P
systems have enabled peers, with individual interests, to cooperate and collaborate with other peers within the concept of a peer-group to advertise and discover any kind of digital content. Context information in our opinion is an emerging P2P content. Bringing the concept of peer groups a step further, ideas such as network communities [7] introduce human social behavior to technical systems. Network communities are interest groups having a relationship with joint interest, which are modeled after human societies. Two import concepts of communities are that they may overlap and that they are dynamic with respect to size. The dynamic nature of the community is due to the fact that as members join and leave a community the structure of the community is emergent. Autonomous Software Agents are capable of proactive behavior, reactive behavior, and goal driven behavior. Agents, deployed in a multi-agent society scenario, can function together by using community mechanisms. In a multi-agent society the social communication environment described in [4] is one of the basic concepts resembling a human social community. In this paper we propose to facilitate context-awareness in a dynamic wireless environment based on the above described concepts.
2 Scenario and Requirements A straight-forward context-aware computing scenario is illustrated in Figure 1. Tina, who is commuting from her home town to Munich every day, has just left her office to get back home to prepare dinner since she has invited her sister Anne tonight. When she enters her car at the parking lot, she activates her personal assistant application to support menu preparation. This context-aware application is checking her electronic household system, which registers any discovered item via RFIDs that come with each product. The electronic household system indicates that there are not enough ingredients in her refrigerator, thus the application suggests a menu and a related shopping list. The shopping list is not only based on the contents of the refrigerator, but it also takes into account her favorite recipes. Since, the fish shop, that Tina usually passes on her way home, has a special offer for fresh salmon today, it is suggested as the main dish and included in the shopping list. The shopping list automatically checks the availability and price of products in the supermarkets along the route and selects one or two markets, which offer all the products in order to avoid several stops. Related recipes are automatically transferred to and displayed in her kitchen. What kind of food does Tina like?
Context awareness What is in Tina's refrigerator now?
What possibilities does Tina have for shopping right now? Showing suggested menu and shopping list automatically to her
Figure 1: Context-aware computing scenario
This scenario reveals a number of functional requirements for such a context-aware ubiquitous computing system. •
Context information has to be made available by sensors such as RFIDs at the products and respective readers
•
In addition to the pure sensor information, e.g., availability and price of a product, metadata information is needed, e.g., to determine if the location of the sensor is in a shop along the route home
•
Mechanisms are needed to cope with the vast amount of context information in the communication system and to preprocess it accordingly to allow the application to quickly find relevant information
•
The application has to be able to specify its request according to its interest in a certain part (sub-group) of the available context information
•
Communication mechanisms are needed in order to exchange the context information between the context sensors and to the application
In the following we describe the building blocks of an architecture facilitating the above described scenario while fulfilling the requirements.
3 Architectural Framework To support applications for context-aware computing a new service platform middleware architecture is needed that allows an application to receive relevant context information. We refer to the dynamic nature of wireless computing and to the need for efficient context information handling avoiding information overload. The example scenario, Figure 2, illustrates our concept of context spaces [5] as a realization of community building among peers. Such peers can provide context information, request context information, or both to support context-aware applications. Our community building principle forms a hierarchical architecture of context spaces, each serving as a virtual meeting point for ubiquitous computing peers. The basic building blocks of such architectural framework for a service support platform middleware are •
The integration of a wireless P2P technology platform for communication between peers
•
An overlay communication mechanism thereupon providing community building means in order to group peers in context spaces
•
A mechanism to express the interest of a peer to join or build a community with other peers [6]
•
An addressing scheme for context spaces
•
Interfaces to allow the search for and the exchange of context information between the peers in a context space, whereas peers could represent sensors or applications
•
A mechanism to build a hierarchical space of context spaces in order to reduce overall the information overflow for applications, to be able to select specific context spaces of interest
Coming back to our example scenario, Figure 2 shows two context spaces: Tina's 'home environment' comprising context information about her kitchen and Tina's 'way to work' comprising context information provided by sensors on her preferred route. Both are represented in a hierarchically higher layer context space. Participating in this context space is also a peer acting as an agent that represents the application that retrieves the context information in order to provide a menu and a shopping list to Tina.
Push of result to the application
PU Application Peer, e.g., User Agent of Tina Pij Peers as objects In the context space P*ij Peer representing an object space
PU P*i
P*i
Pi
Pi
Pi Pi
Pi
Pi
Context Space of Tina's home environment
Sx
Pi
Sx Context Sensors Sx Sx
Context Space of Tina's nearby context (location, route)
Figure 2: Context spaces architectural framework
This architectural framework adapts to the dynamics of the wireless computing environment as it does not rely on a predetermined relationship between the application and specific context sensors that determines the application process. Moreover, the context spaces serve as containers of currently available, addressable, grouped context information topics. Their arrangement may vary according to the current availability of the information. Furthermore, there is no adaptation of the application program needed to reflect a change in context information.
4 Conclusion The work presented in this paper described an architecture supporting context aware computing in wireless ubiquitous computing environments based on the concept of context spaces as virtual meeting points for communities of ubiquitous computing peers. References [1]
M. Weiser, The Computer for the Twenty-First Century, Scientific American, pages 94-104, September 1991.
[2]
K. Lyytinen, Y. Yoo, Issues and Challenges in Ubiquitous Computing, Communications of the ACM, pages 63-65, December 2002.
[3]
G. Chen, D. Kotz, A Survey of Context-Aware Mobile Computing Research, Dartmouth Computer Science TR2000-381, 2001.
[4]
J. Odell, et al., Modeling Agents and their Environment, Proceedings of the AOSE Workshop at AAMAS 2002, Bologna, Italy, July 2002.
[5]
A. Tarlano. Context Spaces Agent Service Architecture. In Proceedings EURESCOM Summit 2002, Heidelberg, Germany, October 21 - 24, 2002.
[6]
A. Tarlano, W. Kellerer, Agent Context Interest Communities for Wireless Computing. In Proceedings IAESTED conference on Wireless and Optical Computing, Banff, Canada, 2003.
[7]
E.D. Mynatt, A. Adler, M. Ito, V.L. O’Day, Design For Network Communities, Proc. of ACM SIGCHI, 1997.
