Research On Cleaning Approach Based On Non-deterministic RFID Composite Event

In recent years, with the micro-electronics, computer and network technology, RFID technology has been widely used. However, there are a lot of negative reading and positive reading and other abnormal conditions in the RFID data collection process, reducing the accuracy of query results in the RFID (Radio Frequency Identification) applications, greatly restricting the further development of RFID technology. At present the processing of the RFID original data on the unreliability and non-deterministic mainly using data cleansing techniques, Most of the existing cleaning methods are directing at the basic flow of events'cleaning to eliminate these anomalies. But because these methods are based on inference method, such as sliding window and so on, resulting in any way they can not achieve 100% cleaning levels, therefore, the dirty basic flow of events because of the halfway cleaning will be brought to the RFID middleware system. The composite event query result based on the dirty basic flow of events will show missed and false detection events. So how to effectively clean the indeterminacy of these complex events to bring it closer to the truth becomes a research hotspot.This paper intensively studies the RFID data cleaning and non-deterministic characteristics of RFID data, the work are as follows:This paper proposed a data cleaning mechanism based on RFID composite event - SOCC. SOCC was first proposed expanding the particle size, abstracting RFID data from the basic detection level to the logic area level with the semantic. Through the basic semantic translation data abstraction algorithm, we make the RFID data from the form of original three tuples turn to a new four tuples with a semantic, making the raw data turning into a smooth and non-deterministic basic event flow of RFID.Consistent with the order of operations for RFID complex event of non-deterministic problem, we conducted in-depth research. For undetected problem of the composite events that meet the group moving Characteristics, we produce the FZQ algorithm to deal with the problem of drilling; For undetected problem of the composite events that meet the ungroup moving Characteristics, we produce the QP algorithm and the XQJ algorithm to handle the undetected problems.Consistent with the non-order of operations for RFID complex event of non-deterministic problem, we mainly study two aspects of missed detection processing and error detection processing. Because the classification of non-sequential composite event are too numerous, we only study the non-deterministic processing method of complex event that meet the conjunctive operation and the disjunction operation.In this paper, we proved the SOCC data cleaning mechanism and other method that we proposed is reasonable and effective through experiments, a large number of experiments show that our methods have obvious performance advantages, the efficiency and cleanliness both higher than the existing strategy, Effectively implementation cleaning of the RFID data.