CAREER: Towards a Marketplace of Networked Services for the Next Billion Devices

职业：迈向下一个十亿设备的网络服务市场

• 批准号: 1751075
• 负责人: Carlee Joe-Wong
• 金额: $ 59.67万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-06-01 至 2025-05-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1751075&HistoricalAwards=false
• 关键词: CAREER; Towards; Marketplace; Networked; Services

Multitudes of devices like vehicles and sensors have recently acquired both Internet connectivity and significant computing power. Much networking research today aims to harness these devices to enable new, cross-device user applications. For instance, external sensors may use cellular and Wi-Fi networks to send measurements to a group of smartphones running augmented reality applications; these smartphones can then work together to analyze the sensor data and display the right information to their users. However, applications can only make use of multiple devices if the device owners agree to cooperate with each other. The goal of this work is to enable cross-device applications by developing mechanisms that incentivize such cooperation. The findings from this project will be integrated into an education plan that emphasizes the role of user and application needs in shaping the evolution of Internet-based technologies. This plan includes developing new undergraduate and graduate courses and short workshop courses for middle and high school students that encourage girls to enter engineering fields. The researcher will incorporate the findings in online blog posts aimed at the general public, and will pursue industry collaborations and outreach to interdisciplinary research communities.The proposed work has two main objectives: (1) to derive algorithms for devices to optimally buy and sell bandwidth and computing resources, and (2) to analytically characterize the resulting marketplace outcomes in terms of the benefit to participating devices. It will be divided into three research thrusts that each address both objectives. First, a simple type of marketplace is considered, in which devices can be either users or providers. User devices decide whether to subscribe to bandwidth or computing resources from one or more provider devices. Second, a more dynamic marketplace is considered, in which users can dynamically negotiate for bandwidth and computing resources from providers, depending on their specific needs over time. The third research thrust builds on the user and provider strategies from the first two thrusts, and considers devices that can dynamically decide whether to act as users and providers, so as to maximize their benefits from participating in bandwidth and computing resource marketplaces. Each research thrust will result in new models, theorems, and decision algorithms that account for the dynamics and inter-dependencies of users' bandwidth and computing resource needs. Small-scale human subject experiments and large-scale virtual test-bed experiments will be used to validate the theoretical work.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.

众多设备等设备最近获得了互联网连接和重要的计算能力。当今的许多网络研究旨在利用这些设备启用新的跨设备用户应用程序。例如，外部传感器可以使用蜂窝和Wi-Fi网络将测量结果发送到运行增强现实应用程序的一组智能手机；然后，这些智能手机可以一起分析传感器数据并向其用户显示正确的信息。但是，如果设备所有者同意相互合作，则只有应用程序才能使用多个设备。这项工作的目的是通过开发激励这种合作的机制来实现跨设备应用。该项目的发现将集成到一个教育计划中，该计划强调用户和应用需求在塑造基于Internet技术的发展中的作用。该计划包括开发新的本科和研究生课程以及为中学生和高中生的简短讲习班课程，这些课程鼓励女孩进入工程领域。研究人员将将这些发现纳入针对公众的在线博客文章中，并将追求行业合作和向跨学科研究社区进行宣传。拟议的工作具有两个主要目标：（1）为设备提供算法以最佳地购买和销售带宽和计算资源，以及（2）以分析范围来表征成果的范围。它将分为三项研究，每个研究都解决了两个目标。首先，考虑了一种简单的市场类型，其中设备可以是用户或提供商。用户设备决定是从一个或多个提供商设备中订阅带宽还是计算资源。其次，考虑了一个更具动态的市场，其中用户可以根据其特定需求随着时间的推移而动态地谈判带宽和计算资源。第三项研究推力是从前两个推力的用户和提供商策略建立的，并考虑可以动态决定是否充当用户和提供商的设备，从而最大程度地利用参与带宽和计算资源市场的收益。每个研究的推力都将导致新的模型，定理和决策算法，以解释用户带宽和计算资源需求的动态和相互依赖性。小规模的人类主题实验和大规模的虚拟测试床实验将用于验证理论工作。该奖项反映了NSF的法定任务，并认为使用基金会的智力优点和更广泛的影响审查标准，认为值得通过评估来获得支持。

项目成果

AdaCoOpt: Leverage the Interplay of Batch Size and Aggregation Frequency for Federated Learning

• DOI: 10.1109/iwqos57198.2023.10188807
• 发表时间: 2023-06
• 期刊: 2023 IEEE/ACM 31st International Symposium on Quality of Service (IWQoS)
• 作者: Weijie Liu-;Xiaoxi Zhang;Jingpu Duan;Carlee Joe-Wong;Zhi Zhou;Xu Chen

Machine Learning on Volatile Instances

• DOI: 10.1109/infocom41043.2020.9155448
• 发表时间: 2020-03
• 期刊: IEEE INFOCOM 2020 - IEEE Conference on Computer Communications
• 作者: Xiaoxi Zhang;Jianyu Wang;Gauri Joshi;Carlee Joe-Wong

Fog-based Data Offloading in Urban IoT Scenarios

• DOI: 10.1109/infocom.2019.8737503
• 发表时间: 2019-04
• 期刊: IEEE INFOCOM 2019 - IEEE Conference on Computer Communications
• 作者: P. Kortoçi;Liang Zheng;Carlee Joe-Wong;M. D. Francesco;M. Chiang

Machine Learning on Volatile Instances: Convergence, Runtime, and Cost Tradeoffs

• DOI: 10.1109/tnet.2021.3112082
• 发表时间: 2022-02
• 期刊: IEEE/ACM Transactions on Networking
• 作者: Xiaoxi Zhang;Jianyu Wang;Li-Feng Lee;T. Yang;Akansha Kalra;Gauri Joshi;Carlee Joe-Wong

A Community Platform for Research on Pricing and Distributed Machine Learning

定价和分布式机器学习研究的社区平台

• DOI: 10.1109/icdcs47774.2020.00117
• 发表时间: 2020
• 期刊: 2020 IEEE 40th International Conference on Distributed Computing Systems (ICDCS
• 作者: Li, Xuanzhe;Gomena, Samuel;Ballard, Logan;Li, Juntao;Aryafar, Ehsan;Joe-Wong, Carlee
• 通讯作者: Joe-Wong, Carlee