A multi-resolution agent for service-oriented situations in ubiquitous domains

In recent years, the situation-aware system field, with service-oriented technologies, has experienced extraordinary development. However, situation-aware systems have yet to achieve widespread use for managing complex situations in ubiquitous domains. Most research focuses on developing context modeling, event recognition, mobility detection, and intelligent reasoning, which are only applicable to static applications in a specific small-scale domain such as a home network. Consequently, a user may experience undesired and unpredictable service interruptions owing to the system's failure to manage service conflicts due to the complex situations which involve a larger number of inter-dependent dynamic objects in multi-domains. A situation management system is required to support users in seamlessly receiving the best services available, while concurrently resolving possible conflicts through a combination of effective methods for situation modeling, recognition, prediction and decision. We propose a multi-resolution agent which provides service prediction and service convergence for enhanced user service reception. This is accomplished by using a dynamic policy modification and adaptive situation learning with activity patterns, through information fusion. The system recognizes possible service conflicts and predicts the adaptive services on complex, often incomplete and unpredictable, dynamic situations, when users roam among different domains requesting new services. We also present the performance results with a new metric, service satisfaction, to show that the proposed solution can support guaranteed services in unique situation management examples, in which possible conflicts are smoothly resolved in complex situations involving nomadic users.