Abstract:
Information generating environments produce large volumes of data which
serves the needs of millions of users around the world. These users prefer only the
relevant information out of the data. However the process of extracting only the
relevant information out of this data is a tedious job. So they delegate the
responsibility of information extraction to systems which extract and transmit the
relevant information to the interested users. Publish-subscribe systems (pub-sub)
provide a basic architecture of such systems which receive data from information
generating bodies and transmit the relevant information to the users according to
their requirements (subscription) without any external invocation. Pub-sub help in
reducing the efforts required at the users end to retrieve the relevant information.
Their performance depends upon the mechanism acquired to: receive data from
information sources, manage the requirements of users (subscription management),
find out which incoming data satisfies users requirement (event-subscription
matching) and then deliver (event dispatch) this information to the interested users.
Different techniques are adopted to perform these tasks in the existing
work; however they lack a technique which consumes less time to perform the
above mentioned tasks and reduces the resource consumption at the systems end.
This research provides an open architecture i.e. Enhanced Publish
Subscribe Middleware (EPSM). EPSM proposes Hierarchal Grouping
Methodology (HGM) to positively affect the performance of pub-sub making it
suitable for time critical environments. HGM is based upon the fact that users
within a particular domain bear similar data requirements i.e. subscriptions. This
technique is called hierarchal as it establishes that apart from being similar, user
2
requirements are related through a parent child relationship. As some users pose
subscriptions which are very complex requiring complex computations while
others are generalized version of the complex subscriptions. HGM first groups
together subscriptions on the basis of similarity and then relates them in the parent
child relationship.
Thus, EPSM provides an effective mechanism to improve the performance
of a pub-sub in the areas of subscription management, the process of eventsubscription
matching and event dispatch.