边缘云中服务质量感知的可靠资源分配优化算法研究

With the emerging techniques of mobile networks and cloud computing, more and more mobile users are using various cloud services, and Internet-of-Things (IoT) has been widely adopted in many industrial applications. Mobile users usually have ultra-low delay requirements of service responses, while industrial users require more very high reliability of their services. Traditional cloud services are deployed in large-scale data centers that are usually located in remote areas. Accessing such cloud services thus incurs long access delays. Unlike traditional cloud computing technique, mobile edge clouds reduce the long access delays of cloud services in remote data centers via deploying cloud services within the proximity of users. In addition, network function virtualization (NFV) technique provides agile, flexible and low-cost network services via implementing network functions as software running in virtual machines. This project aims to enable low-cost, low-latency, and reliable cloud services in mobile edge clouds via leveraging the technique of NFV. This however faces many challenges including highly dynamic resource availabilities, resource heterogeneity, extra-low delay requirements, and high reliability requirements. This project aims to tackle the mentioned challenges by studying the NFV-enabled resource allocation and scheduling in mobile edge clouds. Joint placement/replication of Virtualized Network Functions (VNFs) and routing user traffic is considered. Expected results include a suite of models, optimization frameworks, algorithms, and methodologies for QoS-aware and reliable resource allocation and scheduling in mobile edge clouds. The impact of these expected results including providing fundamental and theoretical support for 5G mobile networks and IoT applications, via enabling QoS-aware, fast, and reliable network services.

随着移动网络以及云计算的发展，物联网在工业领域被广泛应用，同时也对云服务的响应延迟以及可靠性提出了更高的要求。传统云服务部署在离用户较远的数据中心，访问延迟高。边缘云通过将计算资源部署在距离用户较近的移动接入网络中减少了服务访问的延迟。另一方面，网络功能虚拟化（NFV）技术通过基于软件的网络功能以及灵活的资源分配来提供低成本和可靠网络服务。针对基于NFV边缘云中存在的资源动态性、异质性、极低延迟要求和严格可靠性等挑战，本项目开展边缘云中的资源分配与调度理论研究,围绕虚拟网络功能的放置和复制以及路由两个关键科学问题，设计新颖的资源消耗成本函数以及优化框架、建立新型适合边缘云的可靠性模型、设计可靠虚拟网络功能放置及路由算法以及动态资源管理策略。本项目的预期成果是适用于边缘云的针对NFV的资源分配与调度算法、方法和优化框架，这些成果将为第五代移动通信和物联网的关键应用提供基础理论和关键技术支撑。

本项目以边缘云中的资源分配与调度问题为研究对象，围绕服务质量感知和可靠性的双重要求，针对边缘云中网络功能虚拟化、资源和服务请求动态性及不可预知性引入的资源分配科学问题，分别从服务质量感知的虚拟功能放置和路由优化框架、可靠虚拟功能放置和路由技术、动态资源管理策略和算法设计三方面入手在边缘云中构建服务质量感知的可靠资源分配与调度理论优化框架与方法体系。内容包括：一、资源消耗成本函数与优化框架设计；二、可靠虚拟功能放置、复制算法和路由技术：设计新颖的可靠性模型，优化针对虚拟网络功能的主实例和备份实例的放置问题；三、动态资源管理策略和算法设计：设计高效的资源管理机制与分配算法，从而达到对边缘云中动态资源的接近最优的分配。在国家自然科学基金61802048的支持下，项目负责人在2016-2018年度标注资助号共发表论文25篇、其中一作通信论文19篇，专利3项，培养学生3人。入选大连理工大学“星海骨干”计划、2019年入选大连理工大学“星海优青”计划，2020年入选大连市“本地全职高层次人才（青年才俊）”计划、大连市青年科技之星称号。.