Context-Aware Service Chaining Framework for Over-the-Top Applications in 5G Networks

IEEE INFOCOM 2019 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS)(2019)

引用 7|浏览14
暂无评分
摘要
As over-the-top (OTT) applications such as videos dominate the global mobile traffic, conventional content delivery techniques such as caching no longer suffice to cope with mobile network users' requirements due to e.g., fluctuating radio conditions. In legacy 4G networks, mobile network operator (MNO) and OTT service providers (OSPs) are logically decoupled from each other, hence preventing them to share necessary context and enable in-network context-aware intelligence. The recently standardized 5G network architecture's softwarized and virtualized nature opens up new opportunities for flexible deployment of MNO- and OSP-operated network functions. In this work, we first extend the current 5G standard to enable third-party stakeholders to deploy their own user-plane functions (UPFs) within the MNO infrastructure. Based on this, we propose a service function chaining (SFC) framework within the 5G core network, which allows MNO to dynamically determine the optimal set of UPFs that each flow should traverse based on their real-time contexts. The proposed framework has been implemented in a testbed network. Through realistic experiments, we demonstrate that UPF deployment strategy plays a crucial rule in the resulting SFC performance, and our proposed scheme can achieve performance that is close to the benchmark. Furthermore, we establish recommendations on best practices of UPF deployment strategies in 5G network.
更多
查看译文
关键词
context-aware service chaining framework,5G networks,over-the-top applications,mobile network users,in-network context-aware intelligence,OSP-operated network functions,user-plane functions,UPFs,MNO infrastructure,service function chaining framework,5G core network,UPF deployment strategy,mobile traffic,4G networks,5G standard,content delivery techniques
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要