Supporting real-time supply chain decisions based on RFID data streams

While RFID technology has been widely praised for its ability to streamline supply chain processes, little attention has been given to its unique data capturing characteristics to support real-time decision making. Being able to efficiently perform complex real-time analysis on top of RFID event streams is a key challenge for modern applications. This provides management with a novel data analysis mechanism to allow better, tactical, on time, well-informed decisions. The two main issues in RFID data management (RFDM) concern expressibility (how to simply and concisely express stream queries) and performance (how to efficiently evaluate stream queries). In this paper we claim that a spreadsheet-like query model, where formulation is done in a column-wise fashion, can express intuitively a large class of useful and practical RFDM queries. We propose a simple SQL extension to do that and show how these queries can be evaluated efficiently. We finally discuss a prototype called COSTES (COntinuous SpreadsheeT-likE computations), which implements our SQL extensions and evaluation algorithms. Presentation takes place within the context of two representative RFID applications, namely shelf availability and in-store sales promotions.