Leveraging Ubiquity: A Big Data Approach to Environmental Observation

点击翻译按钮获取中文摘要

It is increasingly recognized that modern society's practices and lifestyles cannot be sustained indefinitely. This is particularly evident in the case of water resources, where climactic and social uncertainty are impeding a holistic understanding of the complex interaction of drinking water supplies, agriculture, storm/sanitary systems, ecosystems, and extreme events such as droughts and floods. To mitigate the societal stressors imposed by this uncertainty, "intelligent" infrastructure, composed of interconnected data and control systems, has been proposed to facilitate a more adaptive water resource management approach. This infrastructure will be enabled by real-time sensing, modeling, decision support, and actuation, which will permit it to reason about the world around it and to reconfigure to meet changing needs. In the context of complex dynamical systems, such as those that comprise our water resources, adaptation is a feedback process driven by emergent system behaviors that are presented over time. Hence, while adaptive management holds great promise, it is also prone to failure if it is not continually informed by accurate, inclusive, and current information on the state of the system and by projections of future change. No observational technology yet exists that can provide real-time measurements of environmental systems at sufficient spatiotemporal scales to drive the models necessary to enable intelligent infrastructure. This research program address the challenge of providing environmental measurements at sufficient resolution to support the adaptive management of water resources by advancing a new paradigm of ubiquitous environmental sensing. Ubiquitous sensing does not rely solely on the deployment of large networks of dedicated environmental sensors to improve observational resolution, but also opportunistically leverages “pervasive” sensors, such as a cell phone’s integrated humidity sensor. Specifically the proposed research program will be organized around the following question: Can measurements from a currently unmanageably large number of sensors of varying resolution, purpose, and accuracy be combined to create spatiotemporal measurements of the environment with an accuracy and resolution currently unattainable by traditional environmental sensing alone? At the crux of this question is the ability of a sensing system to integrate high-accuracy, low-resolution measurements from dedicated environmental sensors with low-accuracy, high-resolution measurements from repurposed pervasive sensors. However, because of the great accuracy disparity between dedicated environmental sensors and repurposed pervasive sensors and the high rate of sensor malfunction expected from embedded environmental sensors, traditional data integration methods cannot be used. The proposed work will provide benefits to society by providing insights on the effective design and use of ubiquitous sensor networks to improve real-time adaptive management of complex natural and built systems through predictive control and smart infrastructure, a task that will become increasingly important as we try to maintain economic and social security in the face of environmental change. It will also enhance infrastructure for research through the creation of a program for real-time sensing of coupled natural, built, and social systems using ubiquitous sensors. Finally, it will advance discovery and understanding, while promoting teaching and learning by integrating leading-edge research into interdisciplinary natural science and engineering education for Thompson Rivers University’s diverse student population.