Transcript CASS * Middleware for Mobile Context-Aware Applications

CASS – Middleware for Mobile Context-Aware
Applications
Patrick Fahy
Siobhan Clarke
Trinity College Dublin, Ireland
2008.07.16
Summarized by Babar Tareen, IDS Lab., Seoul National University
Introduction

CASS (Context-awareness sub-structure) is server based
middleware

Provides support for Context aware applications on hand-held
and other mobile devices

Supports high level context data abstraction

Separation of context based inferences and behaviors from
application code

Allow an application’s context reasoning and behaviors to be
changed without re-compilation
Copyright  2007 by CEBT
2
Requirements for CASS
 Support a large number of context sources
 Provision for context history
 Support for context interpretation

For example raw temperature reading to reading in degree Celsiu
s
 Support High-level abstraction of contexts
 Middleware must be event based so that application does not
need to poll for context changes
 Middleware must be extensible
 Support transparent use of distributed sources of context
 Support the separation of application procedure from contextawareness rules and behaviors
Copyright  2007 by CEBT
3
Architecture

Sensor nodes are computers wit
h sensors attached and may be
mobile or static

Middleware being server based,
allows use of a database and ar
tificial intelligence components
as required

Applications need not to comm
unicate with each individual sou
rce of context directly but only
with the middleware
Copyright  2007 by CEBT
4
Implementation Details
 CASS middleware uses a database for persistent data storage
 Advantages of using database

Data can be read and manipulated at a high level of abstraction
using Structured Query Language (SQL)

The database can be used to store, not only context, application
and user data but also domain knowledge in the form of context
-awareness rules and behaviors relevant to specific applications

By providing a suitable interface to the database containing the c
ontext inference rules, these rules could then be updated and cha
nged by users rather than programmers
 CASS inference engine uses forward chaining
 CASS stores application knowledge base as database tables
Copyright  2007 by CEBT
5
Applications
 Two applications under development

MALLET (Maintenance Assignment Listing Lightweight Electronic T
ool)
–

Allows domestic and building maintenance tasks to be organized in c
ontext-aware way
STONE (Start ON Entrance)
–
Designed to avoid the awkward start to presentations and lectures in
volving slideshows
Copyright  2007 by CEBT
6
Paper Evaluation
 Good points

Moving context sensors to separate computers and allowing the
sensors to communicate with the middleware allows any type of
device, with or without context sensing capabilities, to run contex
t-aware applications

Because devices acquire context data from middleware, context d
ata can be shared among multiple devices
 Bad Points

Only getting context related data may limit the possible applicati
on scenarios for context-aware services. For example an applicati
on will be more interested in local temperature around the devic
e and not the temperature at some point where sensing node is
deployed
Copyright  2007 by CEBT
7