CAREER: WolfPack: An Application-Network Co-Design Framework for Performance-Guaranteed Real-time Applications at the Network Edge

分布式实时应用构成了许多改变生活的技术的核心。为了满足这类应用的实时性要求，边缘计算在终端设备和用户附近提供了一个低延迟、高吞吐量的计算环境。然而，现代应用的复杂结构大大阻碍了为这些应用提供端到端的实时保证，特别是面对各种边缘计算网络的复杂结构和动态特性。本项目研究EDGE应用与EDGE网络的联合优化与配置，为现代分布式实时应用提供端到端的性能保障。具体而言，本项目旨在共同设计EDGE计算中计算和网络资源的配置，在一个名为Wolfpack的通用优化框架下为广泛的应用和边缘网络场景提供端到端的性能保证。该项目将集中于三个相辅相成的研究重点：(I)用于任意结构的应用和边缘网络的基于打包的通用优化框架，(Ii)一套用于处理风险并在面临应用和网络动态时提供稳健性的建模和算法工具，以及(Iii)用于激励边缘资源供应并提供性能保证的经济机制。总而言之，这项研究的成果不仅将有助于边缘资源管理和性能保障的核心知识，还将有助于包装/覆盖优化、真实机制设计和基于优化的系统设计的文献。该项目的研究组成部分将支持改变生活的实时应用，造福于公民和社会，包括但不限于智能城市、互联产业、自动驾驶。研究成果将通过期刊/会议出版物、辅导论文、公开演讲、研究研讨会和课程教学广泛传播。研究与教育密切结合，用更新的材料和实用的教学工具推进本科和研究生教育的课程，促进K-12年级学生在研究生阶段积极开展STEM研究，提高女性和代表性不足学生对计算机的认识和参与。该项目的所有研究成果和成果将在项目网站(https://people.engr.ncsu.edu/ryu5/project_wolfpack.php).上公开提供网站将至少在整个项目期间(五年)积极维护，并在项目结束后定期更新，以包括本研究的进一步研究进展。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Distributed real-time applications constitute the core of many life-changing technologies. To satisfy the real-time requirement of such applications, edge computing provides a low-latency and high-throughput computing environment near the end devices and users. However, the complex structures of modern applications significantly hinder providing end-to-end real-time guarantee to these applications, especially in face of the also complex structures and dynamic nature of various edge computing networks. This project investigates the joint optimization and configuration of edge applications and the edge network, for providing end-to-end performance guarantee to modern distributed real-time applications.Specifically, this CAREER project aims to co-design the provisioning of both computing and network resources in edge computing, to provide end-to-end performance guarantee for a wide range of applications and edge network scenarios under a general optimization framework named the WolfPack. The project will focus on three complementary research thrusts: (i) a general packing-based optimization framework for applications and edge networks with arbitrary structures, (ii) a set of modeling and algorithmic tools for handling risks and providing robustness in face of application and network dynamics, and (iii) economic mechanisms for incentivized edge resource provisioning with performance guarantees. Collectively, the outcomes of this research will contribute to not only the core knowledge of edge resource management and performance guarantee, but also the literature of packing/covering optimization, truthful mechanism design, and optimization-based system design.The research component of this project will empower life-changing real-time applications that benefit citizens and society, including but not limited to smart city, connected industry, autonomous driving. The research outcomes will be broadly disseminated through journal/conference publications, tutorial papers, public talks, research seminars, and course teaching. Research is closely integrated with education to advance the curriculum of undergraduate and graduate education with renewed materials and practical teaching tools, promote active STEM research by students in K-12 through postgraduate levels, and increase awareness and participation in computing from female and underrepresented students. All research artifacts and outcomes of this project will be publicly available on the project website (https://people.engr.ncsu.edu/ryu5/project_wolfpack.php). The website will be actively maintained for at least the entire period of this project (five years), and also routinely updated after the end of this project to include further research advances following from this research.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.