Existing methods for the management of multidimensional data typically do not scale well with an increased number of dimensions or require the unsupported augmentation of the kernel. However, the use of multidimensional data continues to grow in modern database applications, specifically in spatio-temporal databases. These systems produce vast volumes of multidimensional data, and as such, data is stored in commercial RDBMS. Therefore, the efficient management of such multidimensional data is crucial. Despite it being applicable to any multidimensional vector data, we consider Radio Frequency Identifications (RFID) systems in this work. Due to RFID’s acceptance and rapid growth into new and complex applications, together with the fact that, as with commercial applications, its data is stored within commercial RDBMS, we have chosen RFID as a pertinent testbed. We show that its data can be represented as vectors in multidimensional space and that the VG-curve combined with Multidimensional Dynamic Clustering Primary Index, which can be integrated into commercial RDBMS, can be used to efficiently access such data. In an empirical study conducted on three, five and nine dimensional RFID data we show that the presented concept outperforms available off-the-shelf options with a fraction of the required space.
|Cite as: Terry, J., Stantic, B. and Sattar, A. (2012). Indexing RFID data using the VG-curve. In Proc. Australasian Database Conference (ADC 2012) Melbourne, Australia. CRPIT, 124. Zhang, R. and Zhang, Y. Eds., ACS. 117-126 |