Author

Cheng Zhong

Date of Award

12-2008

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science (MS)

Department

Spatial Information Science and Engineering

Advisor

Michael F. Worboys

Second Committee Member

Silvia Nittel

Third Committee Member

M. Kate Beard-Tisdale

Abstract

Wireless sensor networks provide a new tool that enables researchers and scientists to efficiently monitor dynamic fields. In order to extend the lifetime of the network, it is important for us to minimize network data transmission as much as possible. Previous work proposed many useful aggregation techniques to answer max, min and average questions, and some of them have been employed in real applications. But we cannot get spatial information from these aggregation techniques. This thesis presents an efficient aggregation technique for continuous generation of contour maps for a dynamic field monitored by a wireless sensor network. A contour map is a useful data representation schema that provides an efficient way to visualize an approximation to the monitored field. In this thesis, we discuss an energy-efficient technique, which we call Isovector Aggregation, for generating such contours using an in-network approach. Our technique achieves energy efficiency in two principal ways. Firstly, only a selection of nodes close to contours are chosen to report, and each reported message contains information about a part or all of the contours, rather than any single node’s ID and value pair. Secondly, contours are progressively merged and simplified along the data routing tree, which eliminates many unnecessary contour points from contour vectors before they are transmitted back to the base station. Using Isovector Aggregation, the base station receives a complete representation of the contours that requires no further processing. Analysis shows that for region-related monitoring, Isovector Aggregation is the only technique that has O( n) traffic generation and that considers in-network traffic reduction at the same time. These two factors make Isovector Aggregation highly scalable. Experimental results using simulations also show that Isovector Aggregation involves considerably less data transmission compared to other approaches, such as the no-aggregation approach and the Isolines Aggregation technique, without compromising the accuracy of representations of the baseline maps.

Share