Home > Research > Publications & Outputs > Advanced modelling of adaptive bitrate selection

Electronic data

  • 2017saniphd

    Final published version, 2.3 MB, PDF document

    Available under license: CC BY: Creative Commons Attribution 4.0 International License

Text available via DOI:

View graph of relations

Advanced modelling of adaptive bitrate selection

Research output: ThesisDoctoral Thesis

Publication date2017
Number of pages171
Awarding Institution
Thesis sponsors
  • Petroleum Technology Development Fund
  • Lancaster University
<mark>Original language</mark>English


Nowadays, a typical video content provider serves a variety of platforms e.g. smartphones, web browsers, and smart TVs. Each of these platforms has specific requirements with respect to transmission and video quality. Moreover, since these devices are increasingly being used on-the-go, the environment within which most of these video streaming clients operate is both unreliable and time-varying. To cater for these heterogeneous requirements, content providers are increasingly adopting adaptive streaming services. Through such services, the quality of the video content received by a user is adapted to fit its specific requirements and capabilities.

To adapt the video quality, system capabilities such as network capacity and memory have to be continuously monitored and measured, chunk requests have to be scheduled, and then the optimal video rate has to be decided. Each of these tasks is usually managed by a sub-module of the adaptive bitrate selection function. However, these sub-components interact in a non-trivial manner. For example, while on-off chunk scheduling helps to prevent buffer overflow, it negatively affects the TCP throughput. Hence, these complex interactions between these different sub-components of the adaptive streaming algorithm result in unnecessary rebufferings, undesirable variability, and sub-optimal video quality.

To help simplify these interactions, this thesis develops several frameworks and models that define the relationships between the various components of the adaptive bitrate selection system. This includes deriving the valid system state space, which defines the state that an algorithm can be in at any given time, determining the allowable interactions between the various components, and identifying the video quality evolution rules that optimise QoE.

Using this information, some state-of-the-art algorithms are improved and novel ones developed to demonstrate the effectiveness of the proposed approach. The result of extensive evaluations conducted both within a real-world Internet environment and with network trace shows the proposed schemes help in reducing the convergence time, startup delay, and rebuffering events, while at the same time increasing both the average and the stability of the video quality. All this is obtained without any adverse impact on the fairness among the competing players.