Supporting RFID-based item tracking applications in Oracle DBMS using a bitmap datatype

  • Authors:
  • Ying Hu;Seema Sundara;Timothy Chorma;Jagannathan Srinivasan

  • Affiliations:
  • Oracle, One Oracle Drive, Nashua, NH;Oracle, One Oracle Drive, Nashua, NH;Oracle, One Oracle Drive, Nashua, NH;Oracle, One Oracle Drive, Nashua, NH

  • Venue:
  • VLDB '05 Proceedings of the 31st international conference on Very large data bases
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Radio Frequency Identification (RFID) based item-level tracking holds the promise of revolutionizing supply-chain, retail store, and asset management applications. However, the high volume of data generated by item-level tracking poses challenges to the applications as well as to backend databases. This paper addresses the problem of efficiently modeling identifier collections occurring in RFID-based item-tracking applications and databases. Specifically, 1) a bitmap datatype is introduced to compactly represent a collection of identifiers, and 2) a set of bitmap access and manipulation routines is provided. The proposed bitmap datatype can model a collection of generic identifiers, including 64-bit, 96-bit, and 256-bit Electronic Product Codes™ (EPCs), and it can be used to represent both transient and persistent identifier collections. Persistent identifier collections can be stored in a table as a column of bitmap datatype. An efficient primary B+- tree-based storage scheme is proposed for such columns. The bitmap datatype can be easily implemented by leveraging the DBMS bitmap index implementation, which typically manages bitmaps of table row identifiers. This paper presents the bitmap datatype and related functionality, illustrates its usage in supporting RFID-based item-tracking applications, describes its prototype implementation in Oracle DBMS, and gives a performance study that characterizes the benefits of the bitmap datatype.