Research Of Application Layer Multicast Mechanism For Mobile Streaming Media In Heterogeneous Network

With the development of mobile network, the demand of mobile video streaming is rising up in the recent years. The scenarios of Real-Time multimedia streaming is shifting from broadband network to narrow-band network, and from PC clients to mobile clients, in which situation the heterogeneous between network and clients becomes more and more obviously.Currently, most architectures of mobile streaming server still use the strategy of direct connect between server and clients. As a result of that, the performance of the server will become the bottleneck of the whole system once the number of clients is rising up. Traditionally we propose to use a P2P network in order to share the pressure of streaming server. It works well in the PC clients'scenarios, but not the mobile ones because of the limitation of upload bandwidth.Therefore there is still no a comprehensive solution for the heterogeneous between network and clients. Furthermore, the duplication of data for PC and mobile clients in different will increase the pressure of the physical layer in the network.In this paper, we propose a new algorithm of ALM (application layer multicast) in single source of real-time streaming which face to the heterogeneous network. The core concept of this algorithm describes as follow: we utilize the PC clients to build a P2P overlay network by organizing them into a hierarchy of bounded-size clusters. And the mobile clients can get the real-time streaming service just connecting into this overlay network directly and simply. The PC clients can release the stress from the server side effectively and the mobile clients can also share the benefits of P2P overlay network, like the low end-to-end delay and low pressure of physical transport layer. It can come to a conclusion that the algorithm proposed in my paper can be an effectively solution for heterogeneous network.Compare to other ALM algorithm, the salient features of my work include:1) We use the PC clients to construct a P2P overlay network in order to release the pressure from server side. In other hand, we use this overlay network to provide streaming service for mobile clients that the narrow-band network peers can also share the benefits from P2P. 2) Reference to the concept of Zigzag and NICE algorithm, we organize the clients into a hierarchy of bounded-size clusters. As a result of that, we can control the impact of failure of peers on the ALM tree locally. And the overhead for both PC and mobile clients are in a constant value.3) As the process of hierarchy of clusters take place, we can organize the peer with enough effective bandwidth and minimal of network distance close to the top of ALM tree. This will lead to an improvement of performance for the ALM tree, opposed to Zigzag and NICE.