Caching Context Information in Pervasive Systems
Speaker: Myilone Anandarajah, ITEE
When: 2006-06-21 10:00:00
Venue: 78-420
Host: A/Prof Jadwiga Indulska
Abstract:Context-aware systems (CASs) are systems that use context
information to adapt their behaviour or the content they
provide. The role of context management systems (CMSs) is to store,
retrieve and evaluate context information. CMSs store context
information in repositories, which can be distributed. This project
addresses the problem of disconnections in context-aware
systems. Disconnections in a CAS may occur because of node mobility,
network failures or node failures. A research opportunity lies in
improving the robustness of CASs to disconnections. While
traditional distributed systems methods of improving robustness in
the face of disconnections can be applied to CASs, the additional
metadata available to CASs may be leveraged to provide smarter
caching algorithms. This research will test the above hypothesis by
investigating "smart" caching algorithms for context-aware
applications.
Context information models describe context information of
applications and the relationships between context information
types. We base our research on the Context Modelling Language (CML)
which provides a sophisticated fact-based context information model.
CML allows a definition of metadata for context information facts
that describe the freshness, standard error or any other metric that
is deemed important by an application designer. CML can also model
temporal and alternative fact types. These features of CML can be
used to perform caching of context information.
This seminar will provide background information on existing context
management systems and context modelling languages. Distributing
context information on static and ad hoc infrastructure will also be
discussed as well as current caching techniques used on static and
ad hoc infrastructure. The seminar will also present an overview of
planned future work which includes: (i) identifying quality metrics
which can be used to proactively cache context information, (ii)
defining a process that designers of context-aware applications can
follow to identify which quality metrics should be included in the
context information models they design, (iii) researching algorithms
for determining which context information facts to cache, and (iv)
researching algorithms which purge context information facts based
on values of metadata defined in CML.
Biography:(biography unavailable)
Type: M.Phil confirmation
Contact:A/Prof Jadwiga Indulska, seminar host (jaga@itee.uq.edu.au)
or Guido Governatori (ITEE seminar co-ordinator)
(guido@itee.uq.edu.au)