PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Modeling link qualities in a sensor network
Jure Leskovec, Purnamrita Sarkar and Carlos Guestrin
In: SIKDD 2005 at multiconference IS 2005, 17 Oct 2005, Ljubljana, Slovenia.


Sensor networks are ad-hoc wireless networks of small, low-cost sensors, which can measure characteristics of their environment. Autonomous low-cost sensors often have limited battery life, and are prone to failures and communication losses. It is thus important to devise efficient power usage, communication and message routing schemes. In this work, we concentrate on estimating the link qualities between pairs of sensors in a natural environment. The estimation is a basic component for algorithms that optimize the power of radio transmission signal, communication schedules, and a routing scheme. Our results show that simple regression models give estimates with only 6% error. We also show the dimensionality reduction techniques help us understand the topology of the communication network and identify potential bottlenecks in the network.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Computational, Information-Theoretic Learning with Statistics
Multimodal Integration
ID Code:1205
Deposited By:Blaz Fortuna
Deposited On:24 November 2005