Multicast communication in grid computing network for real-time streaming media data

Liu, Xiaodong,Dai, Qionghai,Lin, Chuang

2006-01-01

Abstract:This paper presents a three-level group-tree Application Layer Multicast(ALM) overlay network with transcoding capability on special grid node to support real-time stream data distributing. The Grid is a rapidly developing computing infrastructure that normally deals with large cooperative problems on geographically distributed resources. But grid computing is traditionally suited to computing-intensive tasks like astronomy, medical and biological. Recently, delivering realtime streams data gathering from sensors, health monitors and audio/video devices across the Internet is becoming more and more popular, so we incorporate the peer-to-peer technique and multicast technique into grid to support these time-sensitive applications. Peer-to peer network has emerged as a popular alternative to traditional client/server architecture, its distribution nature avoid single point failure and bandwidth bottleneck to make our system more reliable and scalable. In today's heterogeneous environment, end-users use different devices such as desktop, laptops, PDAs, smart phones and others. Especially increasing number of wireless and cellular users is connected to the Internet. In this environment, sending customers real-time stream media data forces the sender to supply the data in different formats to be properly accommodated by the end users' devices. So, we proposed a solution of transcoding data on some high performance grid node. An algorithm constructing minimum cost spanning tree is presented in this paper. We analysed our algorithm in contrast to other methods and simulation results show our algorithm solved the problem of multiple data sources with lower cost than other methods.