Efficient Video Delivery by Leveraging Playback Buffers over Software Defined Networking

Abstract

Adaptive video streaming techniques based on the current network conditions have elevated the overall quality of video streaming over the past few years. Different solutions for providing adaptive video streaming exist in the literature, and one of them is Scalable Video Coding (SVC). A video segment is divided into one base layer and multiple enhancement layers in SVC. Effective buffering at the client and its state is another critical factor for improved viewing quality of the video, which is rarely considered. Timely information regarding client’s buffer state and current network condition is a vital role to adapt with the number of layers in SVC appropriately. Network softwarization through Software Defined Networking (SDN) provides an opportunity to manage and control video streams with better resource management. This paper improves AVS in SDN enabled network through combined measurement of current network bandwidth and client’s buffer state. At the SDN controller, the proposed solution estimates the remaining buffer time and collects the current bandwidth condition of the network to determine the number of enhancement layers for a video segment. SVC video server is extended to transmit a video segment based on the number of enhancement layers decided by the SDN controller. The emulation-based performance evaluation shows that the proposed solution not only improves the video quality against conventional approach under low to moderate network utilization, but also increases the network utilization.