Processing Sliding Windows over Disordered Streams

Abstract

Bursty and out-of-order tuple arrivals complicate the process of determining the content and boundary of sliding windows. To process windows over such streams, two issues need to be addressed: how to sort input tuples efficiently and how to estimate punctuations. In this paper, we focus on these issues to process sliding windows efficiently and accurately over disordered streams. Regarding the first, we propose an order- preserving hash method to sort input tuples in constant time. Regarding the second, we present an estimation method based on the maximum distance between input tuples and their means. Based on these proposed methods, we finally provide a structure of window operators.