RFID composite event definition and detection

Abstract

RFID systems generate large volume of data about localization of people and objects; this information is filtered by a middleware and sent to upper level applications so they can detect events happening in the environment and react properly. This approach implies two drawbacks: firstly, developers invest time programming how to analyze data; secondly, network resources could be unnecessarily wasted when middleware sends data which is irrelevant for the application. To overcome these drawbacks, we present an approach where applications define composite events of interest through a XML-based language, and filtered information is analyzed by a new layer in order to notify applications only when interesting events occur. We present our language called RFID-CEDL for defining interesting events using RFID data and we describe the mechanism used to recognize such events. As demonstration of our approach, we present examples for a hospital environment.