24-09-2012, 12:39 PM
Event Stream Processing (ESP)
Event Stream.docx (Size: 12.88 KB / Downloads: 34)
Abstract
This technical report is about Event Stream Processing (ESP). It describes two of our seminar topics, Event Processing Engines and the HiFi approach, and states their relations to ESP. There is a lot of research going on right now and some of the papers were published
just recently. ESP handles multiple data streams as inputs and tries to detect meaningful events out of these streams. There is the need of an event specification language to define the events and to put queries into the system. A stream processor basically scans the incoming data streams of primitive data, consisting of a tag and a timestamp, to match the defined queries. If the query processor finds a match then it can return results or even trigger an action. This sounds pretty simple, but there are many open questions. Such systems usually produce a huge amount of raw data. There
can be many incoming data streams which have to be processed in real-time. In ESP there is the possibility to define complex events. This means that simple events can be reused and composed to build a complex event. Often in ESP time windows are for example used to restrict a query to a certain time span. The papers in the seminar listed a lot of application scenarios. We heard about supply chains, RFID supported libraries, smart.