Abstract
Online video streaming services have been gaining rapid growth, and that has led to an increase in Internet traffic and have matched with traditional television video services in terms of content consumption. HTTP Adaptive streaming service has been adopted as the medium for video delivery over the Internet with its variants, HTTP Adaptive Streaming (HLS) and Dynamic Adaptive Streaming over HTTP (MPEG-DASH) supporting all playback devices.
In video streaming domain, the performance is measured in terms of the Quality of Experience (QoE) provided to the user and has been known to be influenced by the bitrate adaptation methodology. The objective scores of QoE are related to the bitrate quality delivered while keeping video stalls and rebuffering to a minimum. These include classes of buffer based and network throughput based adaptive switching mechanisms to optimize the QoE. In our work, we evaluate the performance of both these classes of algorithms over parallel HTTP connections at the client, a proven method to increase network throughput to resultantly increase the bitrate quality over an emulated WAN environment, controlled by actual traces conducted during an extensive measurement study over a distributed global DASH measurement cluster.