CRII: CSR: Federated Resource Management in Mobile Edge Computing

CRII：CSR：移动边缘计算中的联合资源管理

• 批准号: 1948387
• 负责人: Hana Khamfroush
• 金额: $ 17.5万
• 依托单位: University of Kentucky Research Foundation
• 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: 2020-03-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1948387&HistoricalAwards=false
• 关键词: CRII; CSR; Federated; Resource; Management

Smart cities are urban areas that use Internet connected devices, called the Internet-of-Things (IoT), to collect and analyze large amounts of data to increase the productivity and efficiency of city resources. Smart home sensors and smartphones are examples of things used in the era of smart cities, each creating a tremendous amount of data that needs to be processed. Edge computing is an emerging technology used to process such large amounts of data in real-time. Edge computing also poses new resource management challenges due to its limited capacity. This project focuses on optimization of resource management in edge and cloud computing frameworks.This project proposes to use concepts of federated learning and reinforcement learning to provide optimal solutions for service placement and request scheduling in multi-tier edge and cloud computing frameworks. This project will focus on new algorithms and system design technologies to trade off the quality-of-service (QoS) and the cost in edge computing systems. Three main thrusts include: 1) finding theoretical bounds through mathematical formulation and algorithm design, 2) finding more practical and adaptive solutions through multi-agent reinforcement learning and the federated learning framework, and 3) comparison of practical solutions with theoretical bounds through numerical analysis and implementations on a mobile edge computing test-bed. This project will have impacts on the future IoT and smart cities applications by improving the cost of providing such applications while maintaining certain QoS levels for users. Algorithms and an optimization framework developed in this project will contribute to the research in design and development of scalable and reliable edge computing systems as building blocks of IoT applications. This research will be integrated into classroom teaching as a special topics graduate level course offered by the investigator. Specific outreach activities for middle school, high school and Berea College students aim to increase participation of underrepresented groups in computer science.The data produced by this project including codes, experimental results, videos of talks, and publications will be shared using the project page that will be maintained during the lifetime of this project. Upon finalizing the two-year term of the project, an archive of the project webpage and information will be digitally stored and made available upon request by any researcher or entity if the webpage is taken down. Link to the project repository: http://www.cs.uky.edu/~khamfroush/fed-mec.php. This project is jointly funded by the Computer Systems Research Program in the Division of Computer and Network Systems and the Established Program to Stimulate Competitive Research (EPSCoR).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.

智慧城市是使用互联网连接设备（称为物联网（IoT））收集和分析大量数据以提高城市资源的生产力和效率的城市地区。智能家居传感器和智能手机是智能城市时代使用的东西的例子，每一个都创造了大量需要处理的数据。边缘计算是一种用于实时处理大量数据的新兴技术。边缘计算由于其有限的容量也带来了新的资源管理挑战。该项目主要研究边缘和云计算框架中的资源管理优化问题。该项目提出使用联邦学习和强化学习的概念，为多层边缘和云计算框架中的服务放置和请求调度提供最佳解决方案。该项目将专注于新算法和系统设计技术，以权衡边缘计算系统的服务质量（QoS）和成本。三个主要目标包括：1）通过数学公式和算法设计找到理论界限，2）通过多智能体强化学习和联邦学习框架找到更实用和自适应的解决方案，以及3）通过数值分析和在移动的边缘计算测试平台上的实现将实际解决方案与理论界限进行比较。该项目将通过提高提供此类应用的成本，同时为用户保持一定的QoS水平，对未来的物联网和智慧城市应用产生影响。该项目中开发的算法和优化框架将有助于设计和开发可扩展和可靠的边缘计算系统，作为物联网应用的构建模块。本研究将整合到课堂教学作为一个专题研究生水平的课程，由调查员提供。针对初中、高中和伯里亚学院学生的具体外展活动旨在提高代表性不足的群体对计算机科学的参与。该项目产生的数据，包括代码、实验结果、演讲视频和出版物，将使用项目页面共享，该页面将在项目生命周期内维护。在该项目的两年期结束后，将以数字方式存储项目网页和信息的档案，并在网页被删除时应任何研究人员或实体的要求提供。链接到项目存储库：http://www.cs.uky.edu/~khamfroush/fed-mec.php。 该项目由计算机和网络系统部门的计算机系统研究计划和刺激竞争研究的既定计划（EPSCoR）共同资助。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

Joint Compression and Offloading Decisions for Deep Learning Services in 3-Tier Edge Systems

三层边缘系统中深度学习服务的联合压缩和卸载决策

• DOI: 10.1109/dyspan53946.2021.9677398
• 发表时间: 2021
• 期刊: 2021 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN
• 作者: Hosseinzadeh, Minoo;Hudson, Nathaniel;Zhao, Xiaobo;Khamfroush, Hana;Lucani, Daniel E.
• 通讯作者: Lucani, Daniel E.

QoS-Aware Priority-Based Task Offloading for Deep Learning Services at the Edge

用于边缘深度学习服务的 QoS 感知、基于优先级的任务卸载

• DOI: 10.1109/ccnc49033.2022.9700676
• 发表时间: 2022
• 期刊: 2022 IEEE 19th Annual Consumer Communications & Networking Conference (CCNC
• 作者: Hosseinzadeh, Minoo;Wachal, Andrew;Khamfroush, Hana;Lucani, Daniel E.
• 通讯作者: Lucani, Daniel E.

Optimal Accuracy-Time Trade-off for Deep Learning Services in Edge Computing Systems

• DOI: 10.1109/icc42927.2021.9500744
• 发表时间: 2020-11
• 期刊: ICC 2021 - IEEE International Conference on Communications
• 作者: Minoo Hosseinzadeh;Andrew Wachal;Hana Khamfroush;D. Lucani

Communication-Loss Trade-Off in Federated Learning: A Distributed Client Selection Algorithm

联邦学习中的通信损失权衡：分布式客户端选择算法

• DOI: 10.1109/ccnc49033.2022.9700601
• 发表时间: 2022
• 期刊: IEEE Communications & Networking (CCNC
• 作者: Hosseinzadeh, Minoo;Hudson, Nathaniel;Heshmati, Sam;Khamfroush, Hana
• 通讯作者: Khamfroush, Hana