CAREER: Robust and Adaptive Streaming Analytics for Sensorized Farms: Internet-of-Small-Things to the Rescue

职业：适用于传感农场的稳健且自适应的流分析：小型物联网的救援

• 批准号: 2146449
• 负责人: Somali Chaterji
• 金额: $ 55万
• 依托单位: Purdue University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-02-15 至 2027-01-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2146449&HistoricalAwards=false
• 关键词: CAREER; Robust; Adaptive; Streaming; Analytics

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Today’s increasingly sensorized agricultural farms are composed of sensors and drones generating copious volumes of data. Two trends in computation have catalyzed the “Internet-of-Small-Things”, or IoST, in relation to digital and sustainable agriculture. First, the availability of inexpensive sensors that can withstand the rigors of agriculture. Second, the development of approximation algorithms for on-device computation of data analytics algorithms. In parallel, some demanding algorithms can be opportunistically offloaded to edge devices or to the cloud. There is an increasing trend to leverage the data from these “small” sensor nodes to actuate dependable, prompt, and resilient actions. Dependable means the algorithms need to deal with missing or corrupted data, network disruption, and node failures. Prompt refers to low-latency decisions, which are at par with the needs of the farmers or digital agriculture providers. The proposed project, Sirius, brings together IoST with machine learning (ML), and creates a compute fabric that is adaptive to the cyber and the physical conditions, and provides prompt actuation, resilient to noisy sensor nodes and communication channels.Sirius will achieve, for the first time, in the context of Cyber Physical Systems for digital agriculture: (1) On-device computation that will adapt to the computation capabilities of heterogeneous devices, to the network conditions, and to the contention on the devices due to co-located applications. (2) Approximate computation for heavyweight streaming analytics using a network of inherently unreliable sensor nodes. (3) Leverage the continuum of sensors, edge devices, and cloud to opportunistically adapt the computation to match the user requirements. This will then be extended to the recent server-less computing architectures and to drone swarms for mobile surveillance, sensing, and actuation. The outcomes of this research will have significant societal impacts in the area of sustainable agriculture. It will also propel education and investigation in multiple disciplines. This will range from data science for on-device analytics and actuation, innovative networking applications, approximating computer vision algorithms, and energy-efficient drone surveillance for agricultural applications. Further, this work will use gaming platforms and new data science courses to be offered in HBCUs and over interactive online platforms. This CPS CAREER proposal will also grow the infrastructure of US data scientists, in particular undergraduate and graduate students, including those from underrepresented minorities, and build strong cross-disciplinary expertise in machine learning, computer vision, and digital agriculture. Finally, the datasets that Sirius will create will be carefully curated and used in CPS-themed camps. The project will also leverage the strong industry linkages to facilitate translation of discoveries to usable prototypes.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.

该奖项的全部或部分资金来源于《2021 年美国救援计划法案》（公法 117-2）。当今日益传感化的农业农场由产生大量数据的传感器和无人机组成，计算的两种趋势催化了“ “小物联网”（IoST）与数字化和可持续农业有关，首先是能够承受农业严酷环境的廉价传感器的可用性。同时，一些要求较高的算法可以机会性地卸载到边缘设备或云端，利用这些“小型”传感器节点的数据来驱动可靠的趋势越来越明显。 、迅速且有弹性的行动意味着算法需要处理丢失或损坏的数据、网络中断和节点故障。及时是指低延迟决策，这符合农民的需求。拟议的项目 Sirius 将 IoST 与机器学习 (ML) 结合在一起，创建了一种适应网络和物理条件的计算结构，并提供快速驱动，能够适应嘈杂的传感器节点和通信通道。 Sirius将首次在数字农业网络物理系统的背景下实现：（1）设备上计算将适应异构设备的计算能力、网络条件和竞争在设备上由于(2) 使用本质上不可靠的传感器节点网络进行重量级流分析的近似计算。 (3) 利用传感器、边缘设备和云的连续性来适时调整计算以满足用户需求。然后扩展到最近的无服务器计算架构和用于移动监视、传感和驱动的无人机群。这项研究的成果也将在可持续农业领域产生重大的社会影响。推动多个学科的教育和调查，包括用于设备分析和驱动的数据科学、创新的网络应用、近似计算机视觉算法以及用于农业应用的节能无人机监控。 HBCU 和交互式在线平台将提供新的数据科学课程，该 CPS 职业提案还将发展美国数据科学家的基础设施，特别是本科生和研究生，包括来自代表性不足的少数族裔的学生，并在以下领域建立强大的跨学科专业知识。机器最后，Sirius 将创建的数据集将在 CPS 主题营地中精心策划和使用，该项目还将利用强大的行业联系来促进将发现转化为可用的原型。通过使用基金会的智力价值和更广泛的影响审查标准进行评估，NSF 的法定使命被认为值得支持。

项目成果

ORION and the Three Rights: Sizing, Bundling, and Prewarming for Serverless DAGs

ORION 和三项权利：无服务器 DAG 的规模调整、捆绑和预热

• 发表时间: 2022-07
• 期刊: USENIX OSDI proceedings
• 作者: Mahgoub, Ashraf;Barsallo, Edgardo;Shankar, Karthick;Minocha, Eshaan;Elnikety, Sameh;Bagchi, Saurabh;Chaterji, Somali.
• 通讯作者: Chaterji, Somali.

FLAIR: Defense against Model Poisoning Attack in Federated Learning

FLAIR：联邦学习中模型中毒攻击的防御

• DOI: 10.1145/3579856.3582836
• 发表时间: 2023-07-10
• 期刊: Proceedings of the 2023 ACM Asia Conference on Computer and Communications Security
• 作者: Atul Sharma;Wei Chen;Joshua C. Zhao;Qiang Qiu;S. Bagchi;S. Chaterji
• 通讯作者: S. Chaterji

Vega: Drone-based Multi-Altitude Target Detection for Autonomous Surveillance

Vega：基于无人机的多高度目标检测，用于自主监视

• DOI: 10.1109/dcoss-iot58021.2023.00044
• 发表时间: 2023-06-01
• 期刊: 2023 19th International Conference on Distributed Computing in Smart Systems and the Internet of Things (DCOSS-IoT)
• 作者: Akhil B;arupalli;arupalli;Sarthak Jain;Akash Melachuri;Joseph Pappas;S. Chaterji
• 通讯作者: S. Chaterji

LiteReconfig: cost and content aware reconfiguration of video object detection systems for mobile GPUs

LiteReconfig：移动 GPU 视频对象检测系统的成本和内容感知重新配置

• DOI: 10.1145/3492321.3519577
• 发表时间: 2022-03-28
• 期刊: Proceedings of the Seventeenth European Conference on Computer Systems
• 作者: Ran Xu;Jayoung Lee;Pengcheng Wang;S. Bagchi;Yin Li;S. Chaterji
• 通讯作者: S. Chaterji

Smartadapt: Multi-branch Object Detection Framework for Videos on Mobiles

Smartadapt：移动设备上视频的多分支目标检测框架

• DOI: 10.1109/cvpr52688.2022.00256
• 发表时间: 2022-06-01
• 期刊: 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
• 作者: Ran Xu;Fangzhou Mu;Jayoung Lee;Preeti Mukherjee;S. Chaterji;S. Bagchi;Yin Li
• 通讯作者: Yin Li