Traffic-engineered distribution of multimedia content

Author: 
Date created: 
2019-08-15
Identifier: 
etd20430
Keywords: 
Multi-view videos
Adaptive video streaming
Traffic engineering
Telco-CDNs
Multicast forwarding
Efficient data planes
Abstract: 

The amount of video traffic transmitted over the Internet has been steadily increasing over the past decade. This is due to the recent interest in streaming high-definition (HD), 4K, and immersive videos to many users. Streaming such multimedia content at large scale faces multiple challenges. For example, streaming systems need to handle users heterogeneity and interactivities in varying network conditions. In addition, large-scale video streaming stresses the ISP network that carries the video traffic, because the ISP needs to carefully direct traffic flows through its network to satisfy various service level agreements. This increases the complexity of managing the network and system resources. To address these challenges, we first propose a novel client-based rate adaptation algorithm for streaming multiview videos. Our algorithm achieves high video quality, smooth playback and efficient bandwidth utilization. For example, it achieves up to 300% improvement in the average quality compared to the algorithm used by YouTube. Then, we propose an algorithm to efficiently solve the resource management problem in the emerging ISP-managed Content Distribution Networks (CDNs). Our solution achieves up to 64% reduction in the inter-domain traffic. To reduce the network load of live streaming sessions, ISPs use multicast to efficiently carry the traffic through their networks. Finally, we propose a label-based multicast forwarding approach to implement traffic-engineered multicast trees in ISP networks. We prove that the proposed approach is scalable as it imposes minimal state and processing overheads on routers. We implemented the proposed multicast approach in a high-speed network testbed, and our results show that it can support thousands of concurrent multicast sessions.

Document type: 
Thesis
Rights: 
This thesis may be printed or downloaded for non-commercial research and scholarly purposes. Copyright remains with the author.
File(s): 
Supervisor(s): 
Mohamed Hefeeda
Department: 
Applied Sciences: School of Computing Science
Thesis type: 
(Thesis) Ph.D.
Statistics: