Scalable efficient composite event detection

  • Authors:
  • K. R. Jayaram;Patrick Eugster

  • Affiliations:
  • Department of Computer Science, Purdue University;Department of Computer Science, Purdue University

  • Venue:
  • COORDINATION'10 Proceedings of the 12th international conference on Coordination Models and Languages
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Composite event detection (CED) is the task of identifying combinations of events which are meaningful with respect to program-defined patterns. Recent research in event-based programming has focused on language design (in different paradigms), leading to a wealth of prototype programming models and languages. However, implementing CED in an efficient and scalable manner remains an under-addressed problem. In fact, the lack of scalable algorithms is the main roadblock to incorporating support for more expressive event patterns into prominent event-based programming languages. This lack of scalable algorithms is a particularly acute problem in event stream processing, where event patterns can additionally be specified over time windows. In this paper we describe GenTrie, a deterministic trie-based algorithm for CED. We describe how complex event patterns are split, how each sub-pattern maps to a node in the trie, and demonstrate through empirical evaluation that GenTrie has higher throughput than current implementations of related languages.