Researches On The Implementation Of High-Performance Video Services On IP Networks

With the development of Broad Band and video coding/decoding technology, (IP)Network-based video service is becoming a more and more important service effecting on people's life and work. (IP)Network-based video services mainly include video live broadcast and video on demand. Like the traditional radio or television broadcast service,(IP)Network-based video live broadcast provides single-source multi-receivers service, which can effectively overlay network users distributing extensively, while (IP)Network-based video on demand provides the video service that allows users to choose the video program and access time on their own initiative, which breaks through the service model that users can only receive passively as in the traditional way. IP networks contain the networks constructed with TCP/IP protocols, and whose data transmission method can be unicast or multicast . Unicast is the basic data transmission method on IP networks. When using unicast to realize the single-source multi-receivers service, each data packet may traverse the same link multiple times, and this decreases the data transferring efficiency. On the other hand, the multicast router can distribute the data packets and keep each data packet traverses the same link only one times on multicast networks. So using multicast can reduce a great deal of consumption of bandwidth for the type service. For the sake of discussion without losing its generality, our researches on realization technologies for video services are based on the Internet that only supports unicast and any IP networks that can support multicast. These researches contain two parts: 1)To realize highly efficient video live broadcast service on the Internet; 2)To realize high-performance video on demand service on IP multicast networks.Since video live broadcast service is a typical single-source multi-receivers service, IP multicast transmission technology provides an ideal solution for it. But multicast network is difficult to be deployed all over the Internet at present. In order to realize highly efficient video live broadcast service on the current Internet, application level multicast technology with similar high efficiency that provided by multicast technology is proposed. This technology utilizes user nodes that join multicast groups to realize the data transmission instead of multicast routers on multicast networks. It can reduce a great deal of consumption of concurrent streams from severs and trunk bandwidth of video live broadcast service on the Internet. Therefore it becomes a research hotspot. But without utilizing topology information of physical network effectively, most existed application level multicast algorithms can't build high-performance logic application level multicast networks. In addition, both a large amount of application of NAT equipments using to solve the problem of IP address shortage and deployment of firewalls for network security might interdict the direct connection between users, which is also an important problem that needs to be solved in application level multicast. Aiming at these problems, this paper has done a lot of researches and proposes a new physical topology oriented application level multicast solution of HHAM and a symmetric NAT traversal algorithm of PS-STUN.The characters of users accessing in Network-based video on demand service are different from those in video live broadcast: Different users may access to the same video at different time in Network-based video on demand service. So, the multicast channels can't be used in video on demand service like to do it in video live broadcast. With the segmenting video and scheduling these segmentations in multi-multicast channels, Static stream scheduling algorithms can make use of multicast channels to realize highly efficient transmission for hot videos. The main performance in these algorithms is demonstrated by the relationship between the number of channels (network bandwidth) assigned to broadcast a video and the client waiting time. It's higher performance of the algorithm that can bring fewer client wait time with the given number of channels (network bandwidth) for a video. However, to a given algorithm, the more channels (network bandwidth) assigned to broadcast a video can bring fewer client waiting time. For improving the performance, most of the existing static broadcasting algorithms use the channels assigned to each individual video statically. It would be impossible to increase or decrease the number of channels if it has being broadcasted. In order to meet most clients' requirement for minimal waiting time in whole system, VOD service system usually assigns more bandwidth to hotter video. But with the hot degree on a video changing dynamically, the static number of assigned channels is not optimal, and the performance of the whole system will be degraded. And there is another problem with this kind of algorithms that they are exigent with user resources. To solve these problems, this paper proposes some high-performance stream scheduling algorithms including improving stream scheduling algorithms that meet limit of users' resources and adaptive stream scheduling algorithms for CBR coding video and VBR coding video respectively. Through theoretical analysis and performance simulation testing, algorithms proposed in this paper is of high performance prior to current algorithms, which can effectively economize consumption of network bandwidth and concurrent streams from servers in Network-based video on demand service system.In total, this paper has done researches on some key technologies of high-performance Network-based video service and proposes some practical solutions on existing problems. These achievements are described as follows:1. This paper proposes a new physical topology oriented application level multicast solution -HHAM. By effectively utilizing physical topology information to build a high-performance application level overlay network, HHAM aims at making users receive and play live broadcast video continuously and steadily on the Internet. It is characterized by following: less transmission delay between data source and every user node; economizing bandwidth of trunk network; steadily receiving video flows at user terminals; meeting large-scale access demands of users distributing extensively.2.To solve the problem of NAT, this paper analyzes shortages of STUN algorithm deal with symmetric NAT and proposes a new symmetric NAT traversal algorithm PS-STUN. According to the dynamic and statistical characteristics of symmetric NAT, PS-STUN do symmetric NAT traversal by using forecasting technology based on measurement and scanning methods. Proved by testing in real network environment, this algorithm can improve effectively success rate of connecting directly between users in different private networks and increase transmission efficiency of the whole system consequently.3.As to realizing high-performance video on demand service, this paper completely analyzes and summarizes static stream scheduling algorithms proper for CBR coding video and proposes an adaptive stream-scheduling algorithm. The algorithm proposed in this paper excels existing algorithms of this kind in performance via formal theory analysis and contrast in performance test. In addition, an improved stream scheduling algorithm was proposed, which can apply to clients with different limited resource.4.Aiming at different characters between VBR and CBR coding video, this paper proposes a new scalable stream scheduling algorithm applying to VBR coding video. Based on videos used in simulation testing on network transmission, we do simulation testing on performance. These experimental results prove that the stream-scheduling algorithm proposed in this paper is high-performance. It has highly practical value for realization of video on demand service.Via formal calculation and simulation testing of these algorithms proposed in this paper, the results show their validity of solving related problems. These achievements are practically valued in realizing high-performance Network-based video service.