Experimental Challenges in Wireless Sensor Networks — Environment, Mobility, and Interference

• Main Author: Rensfelt, Olof
• Format: Doctoral Thesis
• Language: English
• Published: Uppsala universitet, Avdelningen för datorteknik 2012
• Subjects: Wireless Sensor Networks; Testbed; Mobility; Interference classification
• Online Access: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-179807; http://nbn-resolving.de/urn:isbn:978-91-554-8448-4

Wireless sensor networks are used to collect sensor data in different applications such as environmental monitoring, smart building control, and health care applications. Wireless sensor nodes used are typically small, low-cost, and battery powered. The nodes are often hard to access after deployment, for example when they are in remote  locations. Another property of wireless sensor networks is that their operation is dependent on the environment they operate in, both due to the specific sensor readings but also due to the effects on communication by factors such as fading and radio interference. This makes it important to evaluate a wireless sensor network in its intendent target environment before final deployment. To enable experiments with wireless sensor networks in their target environment, we have designed and implemented a testbed called Sensei-UU. It is designed to allow WSN experiments to be repeated in different locations, thus exposing effects caused by the environment. To allow this, the testbed is designed to be easily moved between experimental sites. One type of WSN applications Sensei-UU is aimed to evaluate is protocols where nodes are mobile. Mobile testbed nodes are low-cost robots which follow a tape track on the floor. The localization accuracy of the robot approach is evaluated and is accurate enough to expose a protocol to fading phenoma in a repeatable manner. Sensei-UU has helped us develop a lightweight interference classification approach, SoNIC, which runs on standard motes. The approach only use information from a standard cc2420 chipset available when packets are received. We believe that the classification accuracy is good enough to motivate specific transmission techniques avoiding interference. === WISENET