Koch, Christian (2018)
Proactive Mechanisms for Video-on-Demand Content Delivery.
Technische Universität Darmstadt
Dissertation, Erstveröffentlichung
Kurzbeschreibung (Abstract)
Video delivery over the Internet is the dominant source of network load all over the world. Especially VoD streaming services such as YouTube, Netflix, and Amazon Video have propelled the proliferation of VoD in many peoples' everyday life. VoD allows watching video from a large quantity of content at any time and on a multitude of devices, including smart TVs, laptops, and smartphones. Studies show that many people under the age of 32 grew up with VoD services and have never subscribed to a traditional cable TV service. This shift in video consumption behavior is continuing with an ever-growing number of users. satisfy this large demand, VoD service providers usually rely on CDN, which make VoD streaming scalable by operating a geographically distributed network of several hundreds of thousands of servers. Thereby, they deliver content from locations close to the users, which keeps traffic local and enables a fast playback start. CDN experience heavy utilization during the day and are usually reactive to the user demand, which is not optimal as it leads to expensive over-provisioning, to cope with traffic peaks, and overreacting content eviction that decreases the CDN's performance. However, to sustain future VoD streaming projections with hundreds of millions of users, new approaches are required to increase the content delivery efficiency. To this end, this thesis identifies three key research areas that have the potential to address the future demand for VoD content. Our first contribution is the design of vFetch, a privacy-preserving prefetching mechanism for mobile devices. It focuses explicitly on OTT VoD providers such as YouTube. vFetch learns the user interest towards different content channels and uses these insights to prefetch content on a user terminal. To do so, it continually monitors the user behavior and the device's mobile connectivity pattern, to allow for resource-efficient download scheduling. Thereby, vFetch illustrates how personalized prefetching can reduce the mobile data volume and alleviate mobile networks by offloading peak-hour traffic. Our second contribution focuses on proactive in-network caching. To this end, we present the design of the ProCache mechanism that divides the available cache storage concerning separate content categories. Thus, the available storage is allocated to these divisions based on their contribution to the overall cache efficiency. We propose a general work-flow that emphasizes multiple categories of a mixed content workload in addition to a work-flow tailored for music video content, the dominant traffic source on YouTube. Thereby, ProCache shows how content-awareness can contribute to efficient in-network caching. Our third contribution targets the application of multicast for VoD scenarios. Many users request popular VoD content with only small differences in their playback start time which offers a potential for multicast. Therefore, we present the design of the VoDCast mechanism that leverages this potential to multicast parts of popular VoD content. Thereby, VoDCast illustrates how ISP can collaborate with CDN to coordinate on content that should be delivered by ISP-internal multicast.
Typ des Eintrags: | Dissertation | ||||
---|---|---|---|---|---|
Erschienen: | 2018 | ||||
Autor(en): | Koch, Christian | ||||
Art des Eintrags: | Erstveröffentlichung | ||||
Titel: | Proactive Mechanisms for Video-on-Demand Content Delivery | ||||
Sprache: | Englisch | ||||
Referenten: | Steinmetz, Prof. Dr. Ralf ; Hausheer, Prof. Dr. David | ||||
Publikationsjahr: | November 2018 | ||||
Ort: | Darmstadt | ||||
Datum der mündlichen Prüfung: | 17 Oktober 2018 | ||||
URL / URN: | https://tuprints.ulb.tu-darmstadt.de/8212 | ||||
Kurzbeschreibung (Abstract): | Video delivery over the Internet is the dominant source of network load all over the world. Especially VoD streaming services such as YouTube, Netflix, and Amazon Video have propelled the proliferation of VoD in many peoples' everyday life. VoD allows watching video from a large quantity of content at any time and on a multitude of devices, including smart TVs, laptops, and smartphones. Studies show that many people under the age of 32 grew up with VoD services and have never subscribed to a traditional cable TV service. This shift in video consumption behavior is continuing with an ever-growing number of users. satisfy this large demand, VoD service providers usually rely on CDN, which make VoD streaming scalable by operating a geographically distributed network of several hundreds of thousands of servers. Thereby, they deliver content from locations close to the users, which keeps traffic local and enables a fast playback start. CDN experience heavy utilization during the day and are usually reactive to the user demand, which is not optimal as it leads to expensive over-provisioning, to cope with traffic peaks, and overreacting content eviction that decreases the CDN's performance. However, to sustain future VoD streaming projections with hundreds of millions of users, new approaches are required to increase the content delivery efficiency. To this end, this thesis identifies three key research areas that have the potential to address the future demand for VoD content. Our first contribution is the design of vFetch, a privacy-preserving prefetching mechanism for mobile devices. It focuses explicitly on OTT VoD providers such as YouTube. vFetch learns the user interest towards different content channels and uses these insights to prefetch content on a user terminal. To do so, it continually monitors the user behavior and the device's mobile connectivity pattern, to allow for resource-efficient download scheduling. Thereby, vFetch illustrates how personalized prefetching can reduce the mobile data volume and alleviate mobile networks by offloading peak-hour traffic. Our second contribution focuses on proactive in-network caching. To this end, we present the design of the ProCache mechanism that divides the available cache storage concerning separate content categories. Thus, the available storage is allocated to these divisions based on their contribution to the overall cache efficiency. We propose a general work-flow that emphasizes multiple categories of a mixed content workload in addition to a work-flow tailored for music video content, the dominant traffic source on YouTube. Thereby, ProCache shows how content-awareness can contribute to efficient in-network caching. Our third contribution targets the application of multicast for VoD scenarios. Many users request popular VoD content with only small differences in their playback start time which offers a potential for multicast. Therefore, we present the design of the VoDCast mechanism that leverages this potential to multicast parts of popular VoD content. Thereby, VoDCast illustrates how ISP can collaborate with CDN to coordinate on content that should be delivered by ISP-internal multicast. |
||||
Alternatives oder übersetztes Abstract: |
|
||||
URN: | urn:nbn:de:tuda-tuprints-82127 | ||||
Sachgruppe der Dewey Dezimalklassifikatin (DDC): | 000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik | ||||
Fachbereich(e)/-gebiet(e): | 18 Fachbereich Elektrotechnik und Informationstechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik > Multimedia Kommunikation DFG-Sonderforschungsbereiche (inkl. Transregio) DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche Profilbereiche Profilbereiche > Internet und Digitalisierung DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche > SFB 1053: MAKI – Multi-Mechanismen-Adaption für das künftige Internet DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche > SFB 1053: MAKI – Multi-Mechanismen-Adaption für das künftige Internet > C: Kommunikationsmechanismen DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche > SFB 1053: MAKI – Multi-Mechanismen-Adaption für das künftige Internet > C: Kommunikationsmechanismen > Teilprojekt C3: Inhaltszentrische Sicht |
||||
Hinterlegungsdatum: | 25 Nov 2018 20:55 | ||||
Letzte Änderung: | 25 Nov 2018 20:55 | ||||
PPN: | |||||
Referenten: | Steinmetz, Prof. Dr. Ralf ; Hausheer, Prof. Dr. David | ||||
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: | 17 Oktober 2018 | ||||
Export: | |||||
Suche nach Titel in: | TUfind oder in Google |
Frage zum Eintrag |
Optionen (nur für Redakteure)
Redaktionelle Details anzeigen |