Research On Application Layer Multicast Based On Hierarchy And Clustering For Streaming Media

As processing power and storage capacity of the computer and network transmission capacity increasing continuously, streaming media applications such as distance education, video conferencing, multiplayer games, online broadcast/Video-on-Demand are popular with the users. Streaming media technology over Internet has become one of the hotspots in the field of network communications. As the market and technical reasons, IP multicast didn't obtain large-scale promotion, application layer multicast has been proposed as an alternative technology. Application layer multicast may achieve service customized, not require equipment replacement and be easy to deploy. Research institutions at home and abroad have given considerable attention and have achieved some fruits. This paper mainly includes the following content:1. Study the transmission process of streaming media, analyse the video compression and coding, application layer QoS (Quality of Service) control, distribution services, streaming media servers, media synchronization and streaming media protocols, and propose the idea that application layer multicast may rely on network measurement to construct overlay network topology. Multicast network measurement indicators such as delay, bandwidth, packet loss rate, also are studied.2. Based on the principle and method of application layer multicast protocols, such as construction of data topology and control topology and membership management, we compare and analyze similar protocols and obtain the following conclusions:the mesh-based protocols are suitable for medium or small-scale applications for the high control load; tree-based multicast protocols construct shared tree and may achieve data transmission of the high bandwidth requirements; the implicit protocols employ the hierarchical clustering ideas and hash routing and etal., and then have good scalability.3. This paper proposes a cluster-splitting algorithm based on NICE hierarchy and clustering structure. Based on the analysis of NICE construction principle, we optimize the part of the model. The cluster-splitting algorithm reduces the jumpiness of the members between the clusters, enhances the stability of multicast structure and plays an important role for real-time data transmission.4. We design test experiment of the cluster-splitting algorithm. The experiment shows that the algorithm is simple and efficient, and able to adapt real-time streaming media applications very well, which verifies the effectiveness of the algorithm.