| Wireless sensor networks allow remote and accurate monitoring of physical environment, but these networks are often deployed in inhospitable and nearly inaccessible environments. The lack of accessibility due to harsh environments makes routine maintenance of wireless sensor networks impossible. The networks must then be able to maintain themselves unattended. These accessibility and environmental constraints make resource management an essential requirement of any wireless sensor network in order to ensure continued performance and maintain quality of service to acceptable levels. Intelligent use of resources of wireless sensor networks could be achieved by using modern techniques of knowledge discovery and data mining. This thesis considers a special type of wireless sensor networks named Point-of-Coverage Wireless Sensor Networks. The Thesis proposes the Target-based Association Rules, and associated data preparation mechanisms to improve the performance and quality of service of Point-of-Coverage wireless sensor networks by reducing the energy consumption of the knowledge discovery process. To implement our solution, we exploit the nature of Point-of-Coverage wireless sensor networks and divide sensor nodes into disjoint groups that independently fully cover all of the targets nodes in the field of interest. Data gathering is then performed by each group, alternatively, using the proposed data preparation mechanisms. Then, the gathered data are used by the sink to generate the Target-Based Association rules. Simulation of the proposed data preparation mechanisms shows the benefit of including energy consideration when data mining wireless sensor networks. |
