Network-Based Management for Optimising Video Delivery
MetadataShow full item record
The past decade has witnessed a massive increase in Internet video traffic. The Cisco Visual Forecast index indicates that, by 2022, (79%) of the world's mobile data traffic will be video traffic. In order to increase the video streaming market revenue, service providers need to provide services to the end-users characterised by high Quality of Experience (QoE). However, delivering good-quality video services is a very challenging task due to the stringent constraints related to bandwidth and latency requirements in video streaming. Among the available streaming services, HTTP adaptive streaming (HAS) has become the de facto standard for multimedia delivery over the Internet. HAS is a pull-based approach, since the video player at the client side is responsible for adapting the requested video based on the estimated network conditions. Furthermore, HAS can traverse any firewall or proxy server that lets through standard HTTP data traffic over content delivery networks. Despite the great benefits HAS solutions bring, they also face challenges relating to quality fluctuations when they compete for a shared link. To overcome these issues, the network and video providers must exchange information and cooperate. In this context, Software Defined Networking (SDN) is an emerging technology used to deploy such architecture by providing centralised control for efficient and flexible network management. One of the first problems addressed in this thesis is that of providing QoE-level fairness for the competing HAS players and efficient resource allocation for the available network resources. This has been achieved by presenting a dynamic programming-based algorithm, based on the concept of Max-Min fairness, to provide QoE-level fairness among the competing HAS players. In order to deploy the proposed algorithm, an SDN-based architecture has been presented, named BBGDASH, that leverages the flexibility of the SDN’s management and control to deploy the proposed algorithm on the application and the network level. Another question answered by this thesis is that of how the proposed guidance approach maintains a balance between stability and scalability. To answer this question, a scalable guidance mechanism has been presented that provides guidance to the client without moving the entire control logic to an additional entity or relying purely on the client-side decision. To do so, the guidance scheme provides each client with the optimal bitrate levels to adapt the requested bitrate within the provided levels. Although the proposed BGGDASH can improve the QoE within a wired network, deploying it in a wireless network environment could result in sub-optimal decisions being made due to the high level of fluctuations in the wireless environment. In order to cope with this issue, two time series-based forecasting approaches have been presented to identify the optimal set of bitrate levels for each client based on the network conditions. Additionally, the implementation of the BBGDASH architecture has been extended by proposing an intelligent streaming architecture (named BBGDASH+). Finally, in order to evaluate the feasibility of deploying the bounding bitrate guidance with different network conditions, it has been evaluated under different network conditions to provide generic evaluations. The results show that the proposed algorithms can significantly improve the end-users QoE compared to other compared approaches.