| Summary: | 博士 === 元智大學 === 資訊工程學系 === 100 === As the advances in micro-sensor and wireless communication technologies, wireless sensor networks (WSNs) become an effective paradigm to monitor critical applications over a large-scale area, such as detections of forest fires or leaks of toxic gases. It is significant to guarantee the quality of surveillance (QoSu) of such applications on WSNs. However, an individual wireless sensor is error-prone due to environmental interference or low-cost hardware design. False positive and false negative are two major fault behavior of a sensor on detection. The K-coverage configuration, which guarantees that each location in an area is covered by at least K active sensors, is widely exploited to enhance the detection quality. However, little work investigates the explicit detection performance of K-coverage configuration. Simply increasing the coverage degree effectively reduces the occurrence of detection miss but generates high false alarm rate. A significant issue arises: How can we determine a sufficient K-coverage configuration to guarantee a desired QoSu while prolonging the system lifetime? In this dissertation, we first diminish both detection miss rate and false alarm rate on K-covered WSNs by performing a data fusion scheme, and then formalize the QoSu of such configuration in terms of explicit metrics. Given a QoSu, the proposed model derives an adequate value of K and data fusion parameters to meet the required QoSu while maximizing the system lifetime. A low-cost sleep scheduling algorithm is further proposed to perform the deployment parameters. We argue that the experimental results facilitate the utilization of K-coverage configuration to monitor critical applications on low-cost WSNs.
|
|---|
