|
Obtaining context sensitive information about the environment can be
of great value for business applications to monitor, analyse and
control their processes in real-time. Areas of application are for
example logistics, manufacturing, retailing or power plants.
Emerging technologies, like RFID, allow even small items of interest
to be equipped with digital information. Sensors measure
environmental conditions like temperature, pressure or pollution and
GPS can be used for locating moveable objects. The generated data
has to be processed by an information system. The concept of complex
event processing (CEP) allows to do this asynchronously.
Furthermore, events can be correlated to derive knowledge about
situations of interest. However, centralized CEP systems have only
limited scalability what impedes their use in distributed systems
with many event input sources. Also a large variety of CEP systems
exists on the market so that systems belonging to dierent business
domains in most cases cannot interact. To overcome these drawbacks
of centralized CEP systems a distributed CEP system is introduced
that makes use of available, heterogeneous CEP technologies. This
diploma thesis is part of the Distributed Heterogeneous Event
Processing (DHEP) project involving IBMR Research & Development
Boblingen and the University of Stuttgart. The project aims for
providing a framework for distributed, heterogeneous event
processing with focus on enterprise systems. The speci c
contribution of this thesis is to integrate the context information
of a business domain into the DHEP system. That means a data model
must be available that describes the environment (e.g. information
about location, people, goods, tools, enterprise structure) and
enables to interpret the data from input sources in the current
context. For the description of such data models industry standards
exist in dierent domains that use concepts from complex object
modelling languages, like UML or OWL for example. Based on such
descriptions corresponding executable code and data structures can
be generated. These artefacts are used to enrich primitive events
from sensors or other input sources with higher level information
and make it accessible for distributed, heterogeneous CEP.
|