Virtualized cloud video services

Abstract

We propose new architectures for collaborative video services using software defined networks (SDN) and cloud computing technologies. In the proposed architectures, the video service provider (VSP) is configured as a cloud service. The network service provider (NSP) employs SDN as a cloud service, consisting of a controller cloud and multiple edge network clouds. The proposed centralized and distributed collaborative cloud video services overcome quality of experience (QoE) fluctuations per flow and unfairness between multiple DASH clients employing TCP receive window adaptation over a network slice with reserved capacity as we show that quality of service (QoS) reservation alone is not sufficient to supply fair and steady QoE to all clients. In the centralized approach, the fair-share bitrate of each client is calculated by the traffic engineering module of the NSP, while in the distributed approach the clients collaborate with each other to calculate their own fair shares. Experimental results show that both collaboration architectures perform better than the competitive approach using standard TCP.