I. Introduction
With the rapid advancements in sensor and wireless communication technologies, various novel pervasive computing applications are emerging such as humanoid robot, wireless body area network [1], [2], smart house and intelligent vehicle. An important function of these applications is to provide a close monitoring on the status of an operation environment so that timely and appropriate responses are generated to react to the critical events occurring in the environment. For example, in a humanoid robot, it needs to continuously collect information about its working environment to decide what to do; In a smart house system, various sensors work together for tracking moving objects and detecting intruders. To support the real-time monitoring functions efficiently, periodic application transactions may be defined according to the specific requirements of the monitoring tasks. Each invocation of an application transaction has a hard deadline on its completion time, and failing to meet the deadlines can result in serious problem and even be catastrophic, e.g., in health monitoring. How to schedule a set of periodic application transactions to meet their deadlines is a typical real-time scheduling problem [3]. However, simply satisfying the real-time constraints of the application transactions are not sufficient to achieve the real-time monitoring functions effectively. Another important issue is how to maintain the temporal validity of the realtime data objects installed in the system so that the application transactions only access to data objects with sufficiently up-to-date quality.