Driven by new demands in both civil and national-security applications such as environmental monitoring and border control, sensing devices will not only be deployed at home and in urban areas, but also pervade every corner of the world. Remotely deployed systems have to work in an unattended manner, face harsh and complex environments, and rely on unstable energy sources (e.g., solar energy). Therefore, building remote sensing systems and collect data of interests from them are confronted with unique challenges. First, data collection is subject to loss in communication because wireless links that such systems rely on are error-prone by nature. Second, because of the limited connectivity to the outside world via wireless communication, the sensory data have to be stored in the system when the remote deployment is disconnected from the basestation, and thus are facing uncontrolled loss in storage caused by physical dynamics. Third, when remotely-deployed nodes become unresponsive, it is generally hard to determine what caused the anomalous silence and assess the status of the data collection process without sending a person to the field. Furthermore, the dynamic nature of the energy source calls for new system designs. In this thesis, we present a suite of our work on addressing the above challenges. In particular, we propose adaptive schemes to dynamically adjust the coding redundancy used to mitigate the data loss in communication and storage under time varying energy constraints, and we propose a tele-diagnostic tool to automatically infer node states based on its power consumption traces. The proposed work has been evaluated on a real solar-powered sensing testbed that we designed and deployed.
【 预 览 】
附件列表
Files
Size
Format
View
Towards building reliable solar-powered remotely-deployed sensing systems
PDF
download
878987797
028-85220240
