CC* Integration-Large: Robust and Predictable Network Infrastructure for Wide-Area Hybrid Sensor Networks

CC* 大型集成：适用于广域混合传感器网络的稳健且可预测的网络基础设施

• 批准号: 2019164
• 负责人: Shamik Sengupta
• 金额: $ 99.86万
• 依托单位: Board of Regents, NSHE, obo University of Nevada, Reno
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-10-01 至 2024-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2019164&HistoricalAwards=false
• 关键词: CC; Integration; Large; Robust; Predictable

Increased use of Internet-of-Things (IoT) sensor devices is revolutionizing science and engineering applications, such as smart cities and environmental hazard monitoring. These sensors are often deployed in remote and distributed environments and rely on complex data networks that use both wired and wireless communication to stream large volumes of data back for analysis and distribution. Design and management of these complex hybrid networks is a daunting task due to network capacity fluctuations and dynamic data flow characteristics. This project develops a new software-driven network infrastructure to help automate network management of these emerging hybrid sensor networks for science and public service.This project develops and deploys an operational Software-Defined Networking (SDN) network management and monitoring infrastructure for hybrid wide-area research networks spanning hundreds of kilometers in Nevada for distributed applications in wildfire, climate, and traffic safety. Current practices of inflexible network setup with limited monitoring capability struggles to satisfy ever-increasing science needs, such as on-demand data pipeline creation and quality-of-service satisfaction. Moreover, the project enhances network transparency through deployment of high-precision (i.e., port-, flow-, and packet-level) network monitoring and performance measurement (i.e., PerfSonar) nodes. The project implements a deep-learning-based anomaly detection mechanism to protect sensitive data from cyber attacks. Integrating SDN with high-precision monitoring into wide-area sensor networks has the potential to accelerate adoption of IoT devices in many science areas by addressing core hybrid WAN (wide-area network) challenges such as routing, troubleshooting, and anomaly detection. Developing these integrations now is critical, because hybrid WAN infrastructures (particularly in non-urban regions) will remain bandwidth-limited relative to data generation devices into the foreseeable future. This project allows University of Nevada, Reno (UNR) to continue leadership in wide-area research IoT systems, expand institutional platforms for hybrid-cloud operations, and scale up key products for communities as part of UNR's land-grant mission.Any data produced in the context of this project will be made available to the public and maintained throughout the duration of the project and beyond. Developed source code will initially be maintained in a private GitHub repository, which will be released at “https://github.com/UNR-HPN/SDNWideArea” periodically when the codebase becomes stable. The repository will be maintained as part of ongoing support operations by UNR cyberinfrastructure personnel assigned to the infrastructure at the close of the project. Performance monitoring in the project will be associated with regional dashboards as best practices dictate.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.

物联网（IoT）传感器设备的使用越来越多，正在彻底改变科学和工程应用，例如智能城市和环境危害监测。这些传感器通常部署在远程和分布式环境中，并依赖于复杂的数据网络，这些网络使用有线和无线通信将大量数据流回进行分析和分发。由于网络容量的波动和动态数据流的特点，这些复杂的混合网络的设计和管理是一项艰巨的任务。该项目开发了一个新的软件驱动的网络基础设施，以帮助这些新兴的混合传感器网络的自动化网络管理，用于科学和公共服务。该项目开发和部署了一个可操作的软件定义网络（SDN）网络管理和监控基础设施，用于内华达州数百公里的混合广域研究网络，用于野火，气候和交通安全的分布式应用。目前的做法是不灵活的网络设置与有限的监控能力，努力满足不断增长的科学需求，如按需数据管道创建和服务质量的满意度。此外，该项目通过部署高精度（即，端口级、流级和分组级）网络监控和性能测量（即，PerfSonar）节点。该项目实现了基于深度学习的异常检测机制，以保护敏感数据免受网络攻击。将SDN与高精度监控集成到广域传感器网络中，通过解决路由、故障排除和异常检测等核心混合WAN（广域网）挑战，有可能加速物联网设备在许多科学领域的采用。现在开发这些集成至关重要，因为在可预见的未来，相对于数据生成设备，混合WAN基础设施（特别是在非城市地区）的带宽仍然有限。该项目使内华达州大学里诺分校（UNR）能够继续在广域研究物联网系统方面保持领先地位，扩展混合云运营的机构平台，并为社区扩展关键产品，这是UNR土地赠与使命的一部分。在该项目背景下产生的任何数据都将向公众提供，并在整个项目期间及之后进行维护。开发的源代码最初将在一个私有的GitHub存储库中维护，当代码库变得稳定时，该存储库将定期在“https://github.com/UNR-HPN/SDNWideArea“上发布。该储存库将作为项目结束时分配给该基础设施的UNR网络基础设施人员持续支持业务的一部分进行维护。该奖项反映了NSF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。

项目成果

Avoiding data loss and corruption for file transfers with Fast Integrity Verification

• DOI: 10.1016/j.jpdc.2021.02.002
• 发表时间: 2021-02
• 期刊: J. Parallel Distributed Comput.
• 作者: Ahmed Alhussen;Engin Arslan

Imposters Among Us: A Supervised Learning Approach to Anomaly Detection in IoT Sensor Data

我们身边的冒名顶替者：物联网传感器数据异常检测的监督学习方法

• 发表时间: 2021
• 作者: Das, T.;Shukla, R.M.;Sengupta, S.
• 通讯作者: Sengupta, S.

Learning Transfers via Transfer Learning

通过迁移学习进行学习迁移

• 发表时间: 2021
• 期刊: Innovating the Network for Data-Intensive Science
• 作者: Arifuzzaman, MD;Arslan, Engin
• 通讯作者: Arslan, Engin

Bandwidth and Congestion Aware Routing for Wide-Area Hybrid Networks

• DOI: 10.1109/lanman54755.2022.9819986
• 发表时间: 2022-07
• 期刊: 2022 IEEE International Symposium on Local and Metropolitan Area Networks (LANMAN)
• 作者: O. A. Hamdan;S. Strachan;Engin Arslan

INT Based Network-Aware Task Scheduling for Edge Computing

• DOI: 10.1109/ipdpsw52791.2021.00131
• 发表时间: 2021-06
• 期刊: 2021 IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW)
• 作者: B. Shrestha;Richard Cziva;Engin Arslan