Please use this identifier to cite or link to this item:
Title: Scalable transmission solutions for media streaming in heterogeneous network environment
Authors: Ho, King-man
Degree: Ph.D.
Issue Date: 2008
Abstract: With the advances in high-performance networks and digital video technology, different Video-on-Demand (VoD) systems have come into practice in recent years. With this technology, users can playback the video content without waiting for the entire file to arrive and also enjoy the flexible control of video playback. However, deployment of a large-scale VoD system requires an enormous amount of server and network resources in order to archive hundreds/thousands of videos for customers and handle an enormous number of concurrent video streams. Thus, one of the most challenging design aspects of a VoD system is how to deliver videos to a large community economically. On the other hand, clients can connect to the network with different communication technologies that the downstream rates vary from 56kbps to 10Mbps or above. Heterogeneity of the network environment is another design issue that should be addressed in order to meet different capabilities of the clients' devices in the system. This thesis presents the results of our work on development and analysis of a VoD system in heterogeneous network environment. In this dissertation, we first investigate a feasible solution for building a unified model for a VoD system in heterogeneous environment that integrates the technologies of proxy caching, video broadcasting, replicated/layering videos. With this model, we study various design options and perform system dimensioning. Then, we extend the framework by developing a complementary approach using both video replication and layering for video streaming such that the system performance can be further improved. On the other hand, we also investigate the benefit of renegotiation about video quality when the system resources cannot satisfy the request from a client. Although the hierarchical architecture approach can greatly improve the system performance, the video server is still the bottleneck in such client-server architecture. To tackle this problem, we consider using the peer-to-peer (P2P) approach to address the issues of system scalability. In this regard, we first propose a peer-to-peer batching (PPB) policy to exploit the multicast/broadcast capability of the network and P2P paradigm to efficiently deliver video data to a large number of clients. The objective of this policy is to consider the trade-off between the network bandwidth requirement for P2P transmission and multicast delivery such that the overall transmission cost can be minimized. In addition, we also develop a fault tolerance and recovery procedure for this policy to take the peer's departure behaviour into account for better management of the system resources. It is found that this policy can leverage the workload of the central server about 50%. To further improve the scalability of the system, we also propose a distributed scheme to disperse the duty of multicast/broadcast delivery from the central server to a number of high-end machines denoted peer servers, which have a higher storage and bandwidth capacity than normal peers. We also show how this framework can be applied to the existing broadcasting protocols to take benefits from the P2P paradigm. To investigate the system behaviour of such framework, another unified model is also developed. With this model, we can have a better understanding of the system dynamics. This model also provides guidelines for efficient management of system resources and realization of VoD services.
Subjects: Hong Kong Polytechnic University -- Dissertations.
Streaming technology (Telecommunications)
Video dial tone.
Internetworking (Telecommunication)
Heterogeneous computing.
Pages: xxiii, 194 p. : ill. (some col.) ; 30 cm.
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of Dec 3, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.