CS 252 Project Progress Report
                  April 21, 2003

Aggregation Query Under Uncertainty in Sensor Networks
------------------------------------------------------

Yozo   Hida     (yozo@cs.berkeley.edu)
Paul   Huang    (pbhuang@bmrc.berkeley.edu)
Rajesh Nishtala (rajeshn@cs.berkeley.edu)

Project Webpage:  
  http://www.cs.berkeley.edu/~yozo/cs252/project.html

Progress
--------
Annotated bibliography written.  The objective of the
project has become clearer.

There are some mention of outlier detection of sensor networks 
[3], but done as a selection query from the user without any 
aggregation involved.  We intend to integrate the outlier detection 
inside the network to incorporate local connectivity and history 
information (using outlier detection techniques from [2, 19]).  
In order to cope with failing nodes, we implement a simple multipath 
algorithm mentioned in section 2.1 in [13].  We also plan to add a 
failure notification mechanism, whereby nodes periodically reports 
failed nodes (if they are responding) to call attention of the
user.

After the above has been implemented we are going to test on
a dataset to measure its effectiveness on

  (1) random distribution of failures, with varying frequency
  (2) failures near the root node, with varying number of nodes
      failing

Parameters to investigate includes mode of failure (e.g., sending
wrong sensor values, or dropping/corrupting communication from 
beneath), failure frequency, location of failures, and node density.

Working CVS version (3 weeks ago) of the TinyOS/Nido simulator
with TinyDB, and have familiarized with the TinyDB code.  Currently 
working on:

  (1) checking and implementing the simple multipath algorithm
  (2) implementing the spacial outlier detection on each node
  (3) generation of data sets

References: see annotated bibliography available from project webpage.