Selectivity of EPC Data for Continuous Query Processing on RFID Streaming Events

Abstract

EPCIS as a component of EPCglobal Architecture Framework provides store and search for RFID event data such as EPC, time and location information. EPCIS provides onetime query (poll) interface and continuous query (subscribe) interface. An existing sequential matching technique for continuous query processing makes a heavy load when the number of registered queries is being increased. To solve this problem, multi-dimensional query index technique was been proposed. But it is not appropriate for a 13-dimensional domain condition of EPCIS, an optional condition of event domains, and an optional condition of query domains. So this paper proposes a scheme of query index with multiple 1-dimensional indexes as a basic approach. However, the basic approach also makes a heavy load of processing when wrong query execution procedure is being executed. Therefore, this paper proposes dynamic query execution plan technique considering selectivity of each domain. Through the proposed technique, we measured performance improvement of continuous query processing.