17 December 2013Abstract:
Wireless sensor networks have been deployed outdoors ever since their inception. They have been used in areas such as precision farming, tracking wildlife, and monitoring glaciers. These diverse application areas all have different requirements and constraints, shaping the way in which the sensor network communicates. Yet something they all share is the exposure to an outdoor environment, which at times can be harsh, uncontrolled and difficult to predict. Therefore, understanding the implications of an outdoor environment is an essential step towards reliable wireless sensor network operations.
In this thesis we consider aspects of how the environment influence outdoor wireless sensor networks. Specifically, we experimentally study how meteorological factors impact radio links, and find that temperature is most significant. This motivates us to further study and propose a first order model describing the impact of temperature on wireless sensor nodes. We also analyze transmission errors in an outdoor wireless sensor networks, identifying and explaining patterns in the way data gets corrupted. The findings lead to a design and evaluation of an approach for probabilistic recover of corrupt data in outdoor wireless sensor networks. Apart from the experimental findings we have conducted two different outdoor deployments for which large data sets has been collected, containing both link and meteorological measurements.
Available as PDF (10.77 MB)
Download BibTeX entry.