In this thesis, we focus on topics relevant to developing and deploying large-scale wireless sensor network (WSN) applications within real dynamic urban environments. Given few reported experiences in the literature, we designed our own such network to provide a foundation for our research.
The Smart Condo, a well-defined project with the goal of helping people age in place, provided the setting for our WSN that would non-intrusively monitor an occupant and environment. Although we carefully designed, developed, and deployed the network, all of our planning did not prepare us for a key challenge of that environment: significant radio-frequency interference. Most researchers tend to ignore the existence of interference along with its potentially serious implications: beyond impacting network performance, it can lead researchers to misleading or unrealistic conclusions.
Interference is a particularly difficult problem to study because it varies in time, space, and intensity. Other researchers have typically approached the problem by investigating only known interferers. Instead, we approach the problem more generally and consider interference of unknown origins. We envision nodes periodically observing their environment, recognizing patterns in those observations, and responding appropriately, so we use only standard WSN nodes for our data collection. Unfortunately, collecting high-resolution data is difficult using these simple devices, and to the best of our knowledge, other researchers have only used them to collect rather coarse data.
Within the Smart Condo urban environment, we recorded a transceiver's received power level at 5000 Hz, a higher rate than we encountered elsewhere in the literature, using 16 synchronized nodes. We explored traces from 256 channels and observed a number of recurring patterns; we then investigated classifying traces automatically and obtained rather promising results. We focused on the two patterns most detrimental to packet reception rates and further investigated both sampling and classification techniques tailored to them. As part of our work, we extended our simulator, making it capable of generating impulsive interference, and developed a proof-of-concept pattern-aware medium access control (MAC) protocol.
Through experiments using both the simulator and WSN devices, we evaluated the classifier and proof-of-concept MAC. Our results show that impressive gains in the packet reception rates are possible when nodes can recognize and appropriately react to interference. Using our techniques, nodes can communicate more efficiently by reducing the number of failed transmissions and consequently decreasing overall network congestion.
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:AEU.10048/1965 |
| Date | 11 1900 |
| Creators | Boers, Nicholas M. |
| Contributors | Nikolaidis, Ioanis (Computing Science), Gburzynski, Pawel (Computing Science), Lu, Paul (Computing Science), Nascimento, Mario (Computing Science), Cockburn, Bruce (Electrical and Computer Engineering), Lutfiyya, Hanan (Computer Science) |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | 4105062 bytes, application/pdf |
Page generated in 0.0015 seconds