| Summary: | 博士 === 國立清華大學 === 資訊工程學系 === 99 === Monitoring natural environments and disasters using wireless sensors could eliminate the constraints of wires that exist in traditional methods and could expand monitoring regions and increase monitoring efficiency. Currently, most wireless sensor systems are designed with some implicit assumptions, including that sensors are either stationary and installed in the ground or that they are able to move autonomously. In this thesis, we study a Non-Autonomous Mobile Wireless Sensor Network (NAMWSN), i.e., wireless sensor that drift on the surface of a river, sea, or debris flow and are moved passively by currents. The advantage of a NAMWSN is that the wireless sensors can directly and in-situ measure the parameters inside the target phenomena, which existing immobile equipments cannot do.
In this thesis, we design and implement a non-autonomous mobile wireless sensor system to monitor debris flow. Traditional debris flow monitoring systems are stationary, but are expected to monitor moving debris. Obviously, this approach cannot efficiently track debris flows. For this reason, based on the NAMWSN concept, we propose a novel method of in-situ monitoring debris flow by dropping specially designed sensors in a debris flow and letting them move within the flow; thereby making the continuous tracking of debris flow and real-time reporting of data possible. The design considerations are significantly different from an indoor wireless sensor system. We analyze and evaluate the proposed wireless-sensor-based debris flow monitoring system in many aspects, including low power operation, system responsiveness, communication performance and packaging, etc.
In contrast to traditional immobile WSNs, designing a NAMWSN system requires the development of new techniques, such as building a mobility model of the NAMWSN sensors. In this thesis, we collect the drifting trajectories of the wireless sensors using GPS; these trajectories are converted into maps and are used to build the mobility model of drifting objects. Based on the mobility model, an unlimited number of virtual drifting trajectories can be generated and fed into simulators. Therefore, we can analyze the performance of the newly designed NAMWSN system without costly experiments and can improve system performance.
|
|---|
