Begin with the End in Mind: A Unified End-to-End Quality-of-Experience Monitoring, Optimization and Management Framework

There has been an increasing consensus in the video distribution industry that the design and operation of the full video delivery chain needs to be driven by the quality-of-experience (QoE) appropriate to the end-user. In practice, however, this is an extremely difficult task, largely due to the lack of effective mechanisms to instantaneously measure or predict every viewer's QoE. Moreover, existing quality assurance methods operate independently at different points along the delivery chain, reporting partial, inopportune, and incoherent measurements. This siloed structure leads to a fragmented understanding of the resulting video as it moves through each stage of the delivery network - from the acquisition point and head-end down to the media data center, the network, and eventually the end-user's viewing devices. — Here we propose a new framework that uses a unified end-to-end solution to produce consistent QoE scores at all points along the delivery chain under the same evaluation criterion. This is a framework that produces a clear picture instantaneously to operation engineers, managing executives and content creators about how video QoE degrades along the chain, a framework that allows immediate issue identification, localization and resolution, a framework that enables quality and resource usage optimization of the delivery chain for each of its individual components or as a whole, and a framework that provides reliable predictive metrics for long-term strategic resource and infrastructure allocations. — The main challenge in the implementation of such a framework is to create a unified QoE metric or process by which to measure QoE that not only accurately predicts human QoE, but is also light-weight and versatile, readily plugged into multiple points in the video delivery chain as either a single-ended or double-ended measure, producing real-time QoE scores across a wide range of bitrates, video resolutions, frame rates and dynamic ranges, and combining presentation picture quality with video freezing and adaptive streaming events. We show that the SSIMPLUS metric offers the best promise to such a solution. Once such a framework and the QoE metric is deployed in a video distribution system, many benefits come naturally. We demonstrate the benefit using bandwidth optimization as an example.