Technologies Research On Hybrid Video Multicast Architecture For Large-scale Network

With the widespread penetration of broadband accesses, multimedia services are getting increasingly popular among users and have contributed to a significant amount of today s Internet traffic. A large number of emerging applications, including Internet TV, broadcast of sports events, online games, and distance education, require support for video broadcast, i.e., simultaneous video delivery to a large number of receivers. IPTV not only breaks through the distribution pattern of traditional television but also innovates in the manner and the efficiency of obtaining, spreading and utilizing information.To enhance the carrying capacity of edge network which connects to the core network directly and to meet the controllable, manageable and operational requirements of streaming media traffic,"High Performance Broadband Information Networks"(3Tnet) of the National High-Tech Research and Development Program of China (863 Program) launches the project of"System Performance and Key Technology Researches on the Access and Convergence Router (ACR)". As the core equipment of"China Interactive New Media Network and New Business Projects"which belongs to"11th Five-Year Plan"National High-Tech Development Program, ACR must be able to support P2P technology which is used most extensively on the internet. It aims at supporting P2P live-video multicast. And the premise of this work is to study in depth current peer-to-peer live-video multicast sytems and its architecture. So against the development needs of ACR, the dissertation studied on hybrid video multicast architecture for large-scale network based on ACR project.This dissertation presents new service network architecture hybrid architecture MTreeTV (Mesh-Tree TV) for providing live-video multicast services in the Internet and offers comprehensive solutions to IPTV with good Quality of Experience. It also studied the overlay construction and video segments scheduling algorithm, the effect of key parameters and the placement of the servers which affect the routing performance of video multicast system. This architecture can be applied to other video multicast systems. The dissertation includes the following aspects:1,The dissertation analyzes the architectural choices for supporting Internet multicast. The sparse deployment of IP Multicast that have limited broadcast to only a subset of Internet content publishers. Both CDN-based (CDN, content delivery network) and P2P (Peer-to-Peer) based architectures have their advantages and disadvantages, and each architecture alone does not provide a cost-effective and scalable solution to streaming media distribution. However, one could also conceive of hybrid architectures that seek to use the bandwidth resources of application end-poin ts to the extent possible, but may leverage infrastructure resources where available. Based on the analysis of the architectural choices for supporting Internet multicast, it is concluded that the hybrid architecture faces the overlay construction and scheduling algorithm, the effect of key parameters and the placement of the servers, which affect the user experience of video multicast system. 2,A hybrid video multicast architecture based on physical proximity principle was constructed considering the internet framework and regional distribution of video programs. MTreeTV supports fast channel switching. The video stream will be first pushed to streaming servers; then clients that are interested in the video form a P2P overlay and pull the streaming data from the near peers. Through analysis and simulation, we show that hybrid architectures can have several advantages, it has less overload and is more efficient, stable and adaptable.3,Considering the long path and inefficient routing of traditional P2P overlay, a novel locality-aware overlay construction and scheduling algorithm was proposed. The key idea of the overlay construction algorithm is that the node joins multiple locality-aware trees simultaneously to maintenance partnership of the multitree-based mesh. We resort to a heuristic of fast response time and low delay in high churn environment. The scheduling algorithm concerns the deadline, bandwidth of the nods and the average segment fetch delay, so it can pull more data in the same interval. It reduces the delay and accommodates node dynamics. Through analysis and simulation, we show that MTreeTV supports fast channel switching (<8s), it can provide high playback continuity (>98%) with little control overload (<2%).4,Video broadcast imposes stringent real-time performance requirements in terms of bandwidth and latency, in view of the spread model of video-segment, we explore how the streaming rate of video, the bandwidth of nodes and servers etc. can influence MTreeTV's performance. The total bandwidth contribution of overlay nodes is a direct measure of system capacity. To reduce the setup delay, we must get the system capacity to get max as fast as we can. We analyze (qualified) system-capacity in a discrete-time manner, and also simulate the tradeoff of these key parameters. And therefore provide a rigorous basis for the design of P2P video broadcast systems.5,To answer the above question of the placement of the streaming servers, we transform the placement problem to the set cover problem and solve it using 0-1 programming, linear programming (LP) relaxation and greedy heuristics. The super-optimality of the 0-1 programming problem provides a lower bound to the set cover problem. Using simulation, we show that the rounding technique of LP approaches the lower bound very closely. Meanwhile, the greedy heuristic also provides good performance in all instances with significantly less computation complexity. We also evaluate the effect of the server parameters. By programming servers and optimizing nodes, MTreeTV can support fast channel switching (<4s) and satisfy the requirements of quality of experience guarantees of IPTV.