zur Startseite

CEPiL Internationale Spitzenforschung II

CEP in the Large
ProjekttypInternationale Spitzenforschung II
Gefördert durch Baden-Württemberg Stiftung gGmbH
Beginn 2010/10/01
Leiter Prof. Dr. rer. nat. Dr. h. c. Kurt Rothermel
Prof. Dr. Umakishore Ramachandran
Mitarbeiter Koldehofe, Boris
Ottenwälder, Beate
Mayer, Ruben
Ansprechpartner Koldehofe, Boris
Ottenwälder, Beate
Mayer, Ruben
Kooperationspartner Georgia Institute of Technology (Georgia Tech)

Context and objective

"CEP in the large" is a cooperative research project between the Universität Stuttgart and the Georgia Institute of Technology (Georgia Tech) funded by the research program "Internationale Spitzenforschung II” of the Baden-Württemberg Stiftung gGmbH. It involves researchers from the Institute of Parallel and Distributed Systems (IPVS) of the Universität Stuttgart, under direction of Prof. Kurt Rothermel, as well as Prof. Umakishore Ramachandran from Georgia Tech. The main objective is to investigate new methods for robust, scalable and secure complex event processing (CEP) and their integration into applications in large-scale scenarios. Hence the IPVS can integrate its strong background in optimized operator placement, event correlation as well as event processing, while the Georgia Tech strengthens the project with their expertise in stream processing, e.g. through the Stampede system.

Complex Event Processing

Complex event processing is a key technology in order to analyze high rate event streams and integrate sensor data into applications. It is already applied in many domains such as business process management, manufacturing surveillance or item tracking within logistics chains. CEP enables developers to define and determine high-level events of relevance to the applications which are composed by sequences of low-level sensor events. This is done on the basis of powerful query languages. Sharing the query results between multiple entities already helps to reduce the overall data volume as the transmission of high-level events typically consume far less bandwidth than the original sensor data. Higher scalability can be achieved by a distributed approach towards CEP which allows balancing the load in forwarding and detecting complex events over multiple processing nodes. Additional optimizations are employed to organize the flow of data streams to ensure performance characteristics such as very high data rates or good end-to-end latency.

Project Scope

Nowadays, CEP operates on highly robust communication infrastructures, although the nodes on which correlation can be done may be located in different networks, e.g. in infrastructure based networks as well as wireless mesh networks with mobile nodes. Moreover, if mobile devices contribute to event processing resource consumption is also of great importance. In current solutions, reliability and performance can be severely affected by dynamics, whereas in distributed CEP approaches failures of components can affect the availability of complete situations. Finally, most current approaches completely lack of support for security, or security significantly reduces the scalability.

The goal of this project is to enable highly scalable, robust, and secure CEP. We aim at exploring the concepts along various performance dimensions, such as energy, latency, and bandwidth constraints. Since large-scale CEP may comprise a large number of network technologies we aim to explore two important network technologies with high relevance to the applications: infrastructure based networks and mesh networks.