User-centric services provisioning in wireless environments

130 communications of the acm | november 2008 | vol. 51 | no. 11 The proliferaTion of mobile dev ices (for example, PDAs, 3G mobile phones) and the deployment of more sophisticated wireless communication infrastructures are empowering the Web with the ability to deliver data and services to mobile users. New mobile applications will take advantage of ubiquitous wireless networking to create virtual worlds, with which we can interact while walking, driving our cars, or riding public transport. Moreover, like stationary users, mobile users also require integrated access to relevant services to achieve complicated goals. For example, a class assistance service is relevant to college students, which helps them manage their class activities using mobile devices, by integrating services like attendance reminder and question post. However, several obstacles still hinder the seamless provisioning of services in wireless environments. On the one hand, current Web service provisioning techniques that assume a strong Internet access (for example, fast, low latency, reliable and durable network connections), are inappropriate because wireless environments possess distinguishing features and inherent limitations such as low throughput and poor connectivity of wireless networks, limited computing resources, and frequent disconnections of mobile devices. On the other hand, the environments that mobile users interact with are generally highly dynamic. The variability in computing resources, display terminals, communication channels, and user conditions and preferences require applications to be context aware so that they can adapt to rapidly changing conditions. To date, services provided for mobile users are still hard to build. Two main challenges need to be considered. The first challenge is about the personalized access to services. Personalized support becomes even more crucial, when access of services takes place in wireless environments. For example, the access to services by mobile users tends to be time and location sensitive, meaning that mobile users might need to invoke particular services in a certain period of time and/or a certain place. In addition, mobile users require integrated access to relevant services because many of people’s daily activities are not independent. A service provisioning environment allowing personalized discovery, selection, and composition of services is therefore needed. The second challenge is about handling limited resources of mobile devices. Mobile devices posses, to a certain extent, limited resources (for example, battery power and input capabilities). Therefore, mobile devices better act as passive listeners (for example, receiving the results) than as active tools for service invocation, so that the computational power and battery life of the devices can be extended. We developed a multi-agent based architecture that aims at providing a distributed, adaptive, and contextaware platform for personalized service doi: 10.1145/1400214.1400241