COSTES: Continuous spreadsheet-like computations

Abstract

AutoID technology has been extensively used for a diversity of applications ranging from access control systems to airport baggage handling, livestock management systems, automated toll collection, theft-prevention, and automated production systems. Being able to efficiently perform complex real-time analysis on top of streams of RFID events is a key challenge. This will provide management with a novel data analysis mechanism to allow better, tactical, on time, wellinformed decision-making. Challenges involve both syntactic and evaluation issues. In this paper we propose a set of linguistic requirements for RFID data management (RFDM) queries and describe an SQL-like construct to address these. We argue that a large and useful class of (potentially complex) RFDM queries can be easily expressed and efficiently evaluated using this formality: start from a relational expression and successively extend it with columns representing aggregates over dynamically updated ordered sets. This incremental construction of a query is similar to formula definitions in spreadsheets. We present a fully functional prototype called COSTES (COntinuous SpreadsheeTlikE computationS) implementing spreadsheet-like queries and discuss its architecture and performance. We describe our system in the context of a supply chain management application.