ABSTRACT
In this paper, we propose an RFID coding method to detect events from RFID time series data and a stream processing algorithm for real-time event detection. (k,n)-threshold method enables stable event detection by allowing events to be extracted when k or more tags are observed among n tags with the same group ID. We designed Local Event Handling Coding (LEHC), which is a 96-bit coding system that conforms to the EPCGlobal standard. LEHC is a coding method that specializes in extracting events from changes in the interrelationships of multiple tags. Events such as EMERGE, CROSS, DROP, MERGE, and DIVIDE have been defined. Also, by encoding the event ID into tag itself, it is possible to detect the event locally. Furthermore, we developed a stream processing algorithm for detecting events in real time and evaluated its effectiveness. As an application example, we introduce activities of daily living (ADL) recognition in facilities for the elderly. In dense sensing environment where RFID tags are installed on people and tools in the facility, movements of people and things were extracted as events, ADL was recognized from them, and functional independence measurement (FIM) was measured.
