Quality Adaptation In Peer-to-Peer Video Streaming: Supporting Heterogeneity and Enhancing Performance using Scalable Video Coding.
[Ph.D. Thesis], (2012)
This is the latest version of this item.
Dissertation Osama Abboud -
(Dissertation Osama Abboud)
Available under Creative Commons Attribution Non-commercial No Derivatives.
Download (5Mb) | Preview
|Item Type:||Ph.D. Thesis|
|Title:||Quality Adaptation In Peer-to-Peer Video Streaming: Supporting Heterogeneity and Enhancing Performance using Scalable Video Coding|
Peer-to-Peer (P2P) techniques for Video-on-Demand (VoD) have attracted a lot of attention recently due to their high potential at improving the performance of today's multimedia systems. Evidently, P2P-based streaming systems already serve thousands of videos to millions of users every day. These large-scale systems are possible because client devices act not only as consumers but also as providers when using P2P.
Nonetheless, current P2P VoD systems still suffer from a major limitation: such systems try to provide the same video quality to all users even if they have different devices with a wide spectrum of resources. We believe that it is of essence that future P2P VoD systems are quality adaptive, meaning that different devices may retrieve different video qualities based on available resources. In this thesis, we develop quality adaptation concepts and algorithms essential for improving the Quality of Service (QoS) of P2P VoD streaming systems.
We proceed towards our goal with four major steps.
1. We design a novel P2P VoD streaming system based on techniques and architectures that help in reducing server resource utilization. This is achieved using distributed peer and block management algorithms that use more information about the neighboring peers, e.g. their bandwidth and playback state. Based on a capacity model, we additionally develop prefetching and upload strategies that help the system adapt to fluctuations in the number of peers and their resources.
2. We develop concepts and mechanisms that enable the use of Scalable Video Coding (SVC) in P2P VoD systems to achieve quality adaptation. Using SVC, we develop a two-stage quality adaptation algorithm that matches the video quality with available local and system resources. Additionally, it adapts to the heterogeneity of Internet devices by considering static and dynamic resources such as screen resolution, throughput, processing power, and availability of video blocks. Extensive evaluations show the superiority of our quality adaptation algorithm compared to classical approaches. Furthermore, we show that shorter playback delays can be achieved in return for reducing the video quality. In other words, we find that the session quality (start-up delay, video stalls) and delivered SVC quality (layer switches, received layers) exhibit a tradeoff.
3. We address quality adaptation at the system level and inside the networks by investigating the potential of making networks media-aware, i.e., managing resources according to the importance of different parts of the SVC video. Subsequently, we develop a media-aware system based on routing elements that allocate resources depending on the SVC video characteristics and video block playback deadlines. Using extensive evaluations, we show that it is possible to reduce playback delay by up to 52% during congestion while also alleviating the side effects of congestion on user perceived quality.
4. Finally, we design mechanisms that use Quality of Experience (QoE) metrics in the P2P VoD system to enhance its performance. The decision of which SVC quality to choose has so far been driven by QoS metrics, such as throughput. In this thesis, we expand the classical selection algorithms to consider the QoE of the different SVC qualities. The SVC video quality is assessed using objective techniques, which are highly scalable in comparison to subjective methods. We show that by making peers favor certain SVC qualities with high objective QoE, it is possible to enhance the performance of the entire system in terms of session and video quality, while content providers are able to reduce up to 60% of their server costs.
|Classification DDC:||000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik
600 Technik, Medizin, angewandte Wissenschaften > 620 Ingenieurwissenschaften
|Divisions:||Fachbereich Elektrotechnik und Informationstechnik > Multimedia Kommunikation|
|Date Deposited:||19 Jun 2012 09:32|
|Last Modified:||07 Dec 2012 12:05|
|License:||Creative Commons: Attribution-Noncommercial-No Derivative Works 3.0|
|Referees:||Steinmetz, Prof. Dr.- Ralf and Effelsberg, Prof. Dr.- Wolfgang|
|Refereed:||30 May 2012|
Available Versions of this Item
- Quality Adaptation In Peer-to-Peer Video Streaming: Supporting Heterogeneity and Enhancing Performance using Scalable Video Coding. (deposited 19 Jun 2012 09:32) [Currently Displayed]
Actions (login required)