Abstract
Large-scale VOD system is a promising Internet service in current and next generation. However, traditional VOD system is insufficient to satisfy each client’s requirement due to the asynchrony and heterogeneity of client devices and limited system resource for serving large scale of unpredictable requirements. Besides, the multicast service is not globally available due to the deployment of network layer multicast and interoperability. Application layer multicast techniques are proposed to solve the problem. ZIGZAG is a peer-to-peer technique for single-source media streaming which allows the server to distribute the content to many clients by constructing a multicast tree and organizing the clients into an appropriate tree rooted at server. It also avoids the network bottleneck by reducing the processing hops on the delivery path to a client. NICE is a scalable application layer multicast protocol, specially designed for low bandwidth, data streaming applications with large receiver sets, such as real-time news, sports games, on-line radios. e.g... The NICE protocol arranges the peers into a hierarchy and creates a hierarchically-connected control topology. It can reduce the control overhead of the structure over which the data delivery paths are built. PALS is receiver-driven peer-to-peer scheme for on-demand streaming with quality adaptive playback for layer encoded streaming media which allows a receiver to orchestrate the adaptive delivery of stored layer encoded streams from multiple sender peers. Regional proxy caching reduces both network and server loads, as well as client start-up latency by caching video content at proxy servers. Chaining based schemes provide another means of reducing network bandwidth by using client as surrogate servers for parts of videos. In this thesis, we propose the layered chaining scheme based on optimal chaining as well as layered coding techniques. It can not only reduce the server load but also satisfy the clients’ request. Moreover, by adding in the capability of proxy on the streaming delivery platform, every video streaming chain becomes more reliable, available, fully utilized. As shown in the simulation results, our layered chaining scheme achieves better performance in overall reject rate comparing to the non-scalable chaining approach. Moreover, the system resources including server bandwidth as well system buffer are fully utilized and optimized.