Energy-aware Sparse Sensing

无线传感器网络的广泛部署和智能手机的无处不在，使得大规模传感在我们生活的许多方面发挥了关键作用，如环境监测、灾后评估和生物医学传感等。这些应用通常会产生天文数字的数据量，这给数据的高效处理、存储和传输带来了巨大的挑战。该项目将通过开发一种新的能源感知稀疏传感方案来应对这些挑战。建议的方案可以显著减少由极其有限的能源供电的传感器网络中需要收集和处理的数据量，例如从光、振动或热能获取能量的设备。稀疏感知是因为某些信息比其他信息更重要，因此只对最重要的信息进行采样可以显著减少收集的信息量。此外，稀疏感测方案还可以用于从系统已经收集的海量数据中提取有用信息。这对于人群感知应用尤其重要，在这种应用中，大量用户自愿使用智能手机或平板电脑贡献感知数据。拟议的研究将加速大规模参与式传感和能量收集传感器网络的广泛部署。该项目取得的成果可广泛应用于救灾、股市分析、监控、污染监测等领域。这项研究将有助于公共安全，改善国土安全，促进美国的社会和经济发展。该项目的目标是通过能量感知稀疏感知的新范式，通过适应传感器的能量可用性和被监测对象的时变性质，动态地稀疏地对随机场进行采样，从而克服大规模能量受限传感应用中的大数据挑战。能量感知稀疏传感方案可以显著减少采集和处理的数据量，在能源受限的系统中弥合能源供应和能源需求之间的差距，并为大规模传感应用提供节能和可扩展的解决方案。实现这一目标的具体研究方向包括：1)创建一种新的动态稀疏传感范式。动态稀疏感知是通过“信息多样性”实现的，即在不同的空间位置和时间瞬间收集的数据对某个设计目标具有不同的影响。因此，通过仅对信息最丰富的数据进行采样，可以显著减少所需样本的数量，这些数据可以通过学习先前的感测结果来自适应地识别。2)为光伏或热设备等能量采集设备供电的系统设计节能稀疏传感技术。对于具有能量采集设备的系统，由于传感器之间的能量供应不均匀，因此存在能量多样性。能量感知稀疏感知是通过将能量多样性与信息多样性相结合来实现的，从而使有限的能量资源被用于收集最具信息量的数据。

Enabled by the widespread deployment of wireless sensor networks and the ubiquity of smart phones, large-scale sensing has played a critical role in many aspects of our life, such as environment monitoring, post disaster assessment, and biomedical sensing, etc. Such applications usually generate astronomical volumes of data, which impose formidable challenges for efficient processing, storage, and transmission of data. This project will address these challenges by developing a new paradigm of energy-aware sparse sensing scheme. The proposed scheme can significantly reduce the amount of data to be collected and processed in sensor networks powered by extremely limited energy sources, such as devices harvesting energy from light, vibration, or heat. The sparse sensing is enabled by the fact that some information is more important than others, thus the amount of collected information can be significantly reduced by sampling only the most essential information. In addition, the sparse sensing scheme can also be used for extracting useful information from an ocean of data that have already been collected by a system. This is especially important for crowd-sensing applications where a huge number of users voluntarily contribute sensing data with their smart phones or tablet computers. The proposed research will accelerate the wide range deployment of large-scale participatory sensing and energy-harvesting sensor networks. Results obtained from this project can be applied to a wide range of applications, such as disaster relief, stock market analysis, surveillance, and pollution monitoring, etc. This research will contribute to public safety, improve homeland security, and promote the social and economic development of the United States. The goal of this project is to conquer the "big data" challenge in large-scale energy-constrained sensing applications through a new paradigm of energy-aware sparse sensing, which dynamically and sparsely samples a random field by adapting to the energy availability of sensors and the time-varying nature of the monitored objects. The energy-aware sparse sensing scheme can significantly reduce the amount of data to be collected and processed, bridge the gap between energy supplies and energy demands in energy-constrained systems, and provide energy-efficient and scalable solutions to large-scale sensing applications. The specific research thrusts leading to this goal include: 1) Create a new paradigm of dynamic sparse sensing. The dynamic sparse sensing is enabled by "information diversity", i.e., data collected at different space locations and time instants have different impacts on a certain design objective. Thus the number of required samples can be significantly reduced by sampling only the most informative data, which can be adaptively identified by learning through previous sensing results. 2) Design energy-aware sparse sensing techniques for systems powered by energy harvesting devices such as photovoltaic or thermal devices. For systems with energy harvesting devices, there is energy diversity due to non-uniform energy supplies across the sensors. The energy-aware sparse sensing is achieved by aligning energy diversity with information diversity, so that the limited energy resources are used to collect the most informative data.