I-Corps: Artificial Intelligence (AI)-Based Sensing and Data Efficient Sampling, Transmission, Storage, Analysis and Cloud Computing

I-Corps项目更广泛的影响/商业潜力是改善日常生活中无处不在的传感器数据，包括个人设备、智能家居和基础设施、工业机器和工厂、自主机器和自然环境。基于传感器的“大数据”提供了对各种系统的实时可见性。然而，这些数据的数量、速度和种类很快就会超过吸收、提炼和分析所有数据的能力。该项目旨在开发建模方法，以改善涉及物联网（IoT）、太空探索和生物技术等嵌入式系统的各种行业中数据收集、处理和分析的方式。工业机器的远程监测和海洋和生物圈的分布式传感是一些应用领域。这个I-Corps项目是基于一种新型传感器数据建模方法的开发。到2025年，基于传感器的数据生成量将达到73万亿千兆字节。当前的数据收集和分析管道对于这种规模的数据是低效和不经济的。该技术研究了欠采样，在数据收集、传输、存储、云计算和分析方面节省了几个数量级。一些应用领域包括工业物联网（IIoT），环境和海洋监测，以及自主无人机和车辆。后奈奎斯特时代信息提取管道的新方法以欠采样、低维潜在表示和机器学习为中心，在IIoT系统的数据和工程复杂性方面节省了几个数量级。对基于欠采样的低维潜在表征的学习以及使用它们来解决不同的下游任务的见解是本项目的智力贡献。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this I-Corps project is to improve data from sensors that have become ubiquitous in daily life, including personal devices, smart homes and infrastructure, industrial machines and factories, autonomous machines, and the natural environment. Sensor-based ‘big data’ offers real-time visibility into a variety of systems. However, the volume, speed, and variety of this data will soon overwhelm the ability to ingest, refine, and analyze all the data. This project seeks to develop modeling methodologies that will improve the way data is collected, processed, and analyzed in a variety of industries involving embedded systems such as Internet of Things (IoT), space exploration, and biotechnology. Remote monitoring of industrial machines and distributed sensing in the oceans and the biosphere are some application areas.This I-Corps project is based on the development of a novel sensor data modeling methodology. Sensor-based data generation is about to reach 73+ trillion gigabytes by 2025. Contemporary data collection and analysis pipelines are inefficient and uneconomical for this scale of data. This technology investigates undersampling for several orders of magnitude savings in data collection, transmission, storage, cloud computing, and analytics. Some application areas include the Industrial Internet of Things (IIoT), environmental and ocean monitoring, and autonomous drones and vehicles. The novel methodology in the post-Nyquist era information extraction pipeline centered around undersampling, low-dimensional latent representations, and machine learning, with several orders of savings in the data and engineering complexity of IIoT systems. The insight on undersampling-based learning of low-dimensional latent representations and using them to solve different downstream tasks is the intellectual contribution of this project.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.