数据驱动的网络多媒体边缘内容分发

Data-driven multimedia edge network and content delivery

近年来,网络多媒体迅猛发展.相较于传统固定网络多媒体内容服务,兴起的移动化多媒体服务中的用户行为发生了根本变化.传统内容分发架构依靠中心化基础设施、全局流行度趋势,无法有效感知和服务移动多媒体内容.利用边缘网络架构结合数据驱动策略服务动态移动多媒体内容,提升服务质量成为一种新的选择.通过数据驱动与跨域协同的研究方法,本文深入研究移动多媒体内容服务规律、系统架构、策略设计等.本文首先探讨数据驱动的大规模移动多媒体内容分发模式挖掘与网络性能分析,揭示边缘多媒体内容分发与传统内容分发的本质区别;进一步,本文给出一般性的边缘多媒体网络与内容分发的研究方法,包括数据驱动与跨域协同的多媒体边缘网络内容分发研究框架.最后,本文给出几种代表性框架,包括基于社交媒体传播预测进行内容部署的SocialCDN,基于内容提供商智能的CPCDN,基于边缘网络设备进行移动多媒体内容分发的EdgeCDN,以及基于用户众筹资源的CrowdCDN.

Recently,network multimedia has undergone rapid developments.Compared with traditional endto-end multimedia content services,user behaviors in emerging mobile multimedia services have experienced fundamental changes.For efficient mobile multimedia content services,the conventional content distribution architecture relies on a centralized infrastructure,in which global popularity trends cannot be effectively obtained.The use of edge network architecture combined with data-driven strategies to aid the dynamic mobile multimedia content has become a new choice for improved service quality.Using data-driven and cross-domain collaboration methods,this paper reveals a set of principles of mobile multimedia content service and system architecture design.This paper first discusses data-driven behavior patterns and network performance analysis.It shows the essential difference between edge multimedia content distribution and traditional content distribution.Further,this paper presents general edge multimedia research methodologies,including a framework of multimedia edge network content distribution using data-driven and cross-domain collaboration methods.Finally,this paper provides several representative frameworks,including Social CDN,which uses social information for content distribution,CPCDN,in which a content-provider-based content delivery framework is proposed,Edge CDN,which is an edge content delivery framework,and finally Crowd CDN,which uses user-based crowdfunding resources for efficient content distribution.