CS 525s - Sensor and Stream Data Management - Fall 2006
Location at http://web.cs.wpi.edu/~cs525/f06s-EAR/
Fuller Labs 311, Thursdays at 6:00pm - 8:50pm
Thursdays right after class (8:50pm onwards)
Mondays FL 238 at 12:00-1:00 noon (check : changes announced here)
If you have any questions that cannot wait, please stop by my office or
send electronic mail. In the later case,
I'll do my best to answer your email within a 24 hour time window.
Mailing List for Course
Note that a email mailing list
for this course has been set up.
Details to be announced in class.
Topic of Course
In this course we will study the emerging research area of sensor and
stream data management. This area has recently been receiving a
significant attention within and also beyond the database
community. This novel technology in part is driven by ubiquitous
computing, i.e., by computing through small devices, sensors, cell
phones, PDAs, and the like that are embedded in the
environment. Similarly, many applications ranging from scientific,
energy, weather, environment, and health services in the future will
depend on such technologies. Performing computations on such an
infrastructure cannot achieve its great promise
without developing methods for guaranteeing real-time access to
relevant data. This is one area where data management meets ubiquitous
computing technology. Further, relevant data is typically pushed into
the computing infrastructure for further processing, then termed 'data
streams'. Data streams are prevalent: they are
rapid and unbounded streams of
data such as sensor readings, but also telephone call records,
financial tickers, web usage logs, network packet traces, and so
on. With streams everywhere, a new area, called Stream Data
Management (SDM), has emerged aiming to produce generic software
technology similar to that of Database Management Systems for
streaming data. We will study this novel technology targeting to meet
the need of online monitoring applications, in which continuous
queries operate in near real-time over data streams.
In summary, our objective will be to learn about these novel
advances in the field, and also to gain an understanding of the
fundamental similarities as well as differences between data stream
processing and traditional data management.
The topic is too new and evolving so rapidly
for there to exist a comprehensive and up-to-date text book that
would contain all the material we will study in this course.
Thus instead we will be utilizing a variety of publications
taken directly from
the primary literature. These manuscripts will be linked
indexed by the day on which the material
is to be studied in our course.
In general, there will be a core collection of readings
in selected topics in sensor and stream data management around which
our discussions will be centered.
These topics may include adaptive query processing,
query optimization, in-sensor query processing, plan migration, distributed computing, on-line aggregation and mining, shedding and approximation, constraint-aware computing, and others.
You should have a basic familiarity with relational databases,
equivalent of a beginning course in databases. Instead, if you have
used a database system before in practice, you may also have
sufficient basic understanding required for the course. If you are in
doubt, please talk to the instructor.
Upon completion of this course, students should be able to:
Explain the differences and similarities between traditional database
systems and stream engines.
Describe the different components, including their underlying
principles and algorithms, of a stream engine.
Design and develop a stream- or sensor-related database project.
Demonstrate skills to critically analyze technical literature
and assess technological advances in databases.
Communicate their ideas effectively
in the form of a presentation to a technical audience.
Overall, able to take
first "baby steps" in database research, and perhaps
computer science research in general.
end of syllabus