Semantic Complex Event Processing over End-to-End Data Flows

Other · Sun Apr 01 00:00:00 EDT 2012

OSTI ID:1290123

Zhou, Qunzhi ^[1]; Simmhan, Yogesh; Prasanna, Viktor K.

---

1. University of Southern California; Power System Information & Advanced Technologies LADWP Power System Engineering Division

Emerging Complex Event Processing (CEP) applications in cyber physical systems like SmartPower Grids present novel challenges for end-to-end analysis over events, flowing from heterogeneous information sources to persistent knowledge repositories. CEP for these applications must support two distinctive features - easy specification patterns over diverse information streams, and integrated pattern detection over realtime and historical events. Existing work on CEP has been limited to relational query patterns, and engines that match events arriving after the query has been registered. We propose SCEPter, a semantic complex event processing framework which uniformly processes queries over continuous and archived events. SCEPteris built around an existing CEP engine with innovative support for semantic event pattern specification and allows their seamless detection over past, present and future events. Specifically, we describe a unified semantic query model that can operate over data flowing through event streams to event repositories. Compile-time and runtime semantic patterns are distinguished and addressed separately for efficiency. Query rewriting is examined and analyzed in the context of temporal boundaries that exist between event streams and their repository to avoid duplicate or missing results. The design and prototype implementation of SCEPterare analyzed using latency and throughput metrics for scenarios from the Smart Grid domain.

• Research Organization: City of Los Angeles Department
• Sponsoring Organization: USDOE
• DOE Contract Number: OE0000192
• OSTI ID: 1290123
• Report Number(s): DOE-USC-00192-116
• Country of Publication: United States
• Language: English