Research On Main Memory Index For Rfid Data Stream

RFID (Radio Frequency Identification) technology is a fast, non-contact identification technology. Based on its low-cost, easy to use, characteristics of mature technology, RFID technology has been widely applied to various fields. A middleware platform is the current trend of development for RFID applications, RFID middleware systems are an important component of the RFID technology. RFID middleware builds bridges between the various application systems and the physical world of tags, thus the of RFID middleware technology has come.In the application of RFID technology, the system will generate a large number of real-time RFID data, the raw data is acquired by the RFID readers then sent to the RFID middleware. RFID middleware, on the one hand have to store and access a large number of data; on the other hand, in order to get the accurate data, we have to process the raw data stream. These two aspects are key technologies for RFID middleware system, but also the current problems to be solved.In order to achieve high-performance RFID middleware system, this article focuses on the massive label records storage and quick access to efficient research, the existing mass data management techniques are based on the B-tree or B + tree with disk indexing technology. Disk index for the large high-throughput real-time RFID middleware is unable to meet demand for its high-speed access, so this thesis use of memory as data storage and indexing of the media, in-depth labeling composition, structural characteristics, as well as the application of the memory after the design of a Suffix_T index structure, it will split label record store, the first to use the characteristics of the prefix public label, so that RFID tags record of massive space greatly compressed, and then use the T-tree structure to store records suffix. This index structure in terms of update rate, query speed, or in space, both the occupancy rate than the traditional index structure AVL tree, B-tree, T trees and other significant improvements, the final conclusion proved through experiments.In addition, data obtained from the RFID reader and deposited into the main memory database is the original RFID data stream, which is also mixed with a variety of erroneous flow data. Before the further application, RFID middleware system have to process the data stream. This paper details the variety of errors in data stream and the reasons caused by. We made a In-depth study the negative readings which contributes the most unreliability of the raw RFID data stream. Then we analyze the adaptive cleaning algorithm and find out several points which are imperfections and unreasonable, So we designed an improved algorithm ISMURF, and the accuracy of the improved algorithm has obviously improved to the original one in our experimental verification.