Adaptive Resource Management Protocol for Prioritized Delay-Bounded Multiple-Stream Multicast

• Main Authors: Liu Chun-kun, 劉醇錕
• Other Authors: Longsong Lin
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/66020661667276746477

碩士 === 國立雲林科技大學 === 電子與資訊工程技術研究所 === 86 === This paper proposes an integrated resource management protocol that comprises resource reservation/allocation and admission control mechanisms coupled with multicast routing algorithm to provide quality-of-service(QoS) guaranteed multicast service. The multicast service is design to deliver as many as poss-ible streams of media offered by a source to a group of receivers and guarant-ees to fulfill each receiver's QoS requirement, which is specified in terms of the priorities of the preferred streams and their bounds on the tolerable del-ay. The resource management protocol does not presume the existence of a mult-icast tree but establishes the multicast tree "on-the-fly" based on the recei-vers' specifications. It is our goal to maximize the efficiency gain over the receivers' requests as well as the network utilization by admitting more requ-ests into the multicasting. We characterize the flow in the network in terms of a variable fluid model in which a flow sees a diverse amount of bandwidth on the links of the path from source to receiver. The essential idea of the adaptive bandwidth reservation/allocation protocol is to allow diverse bandwi-dth to be allocated for the same stream by considering the capacity and the traffic condition of individual link segment. This adaptive bandwidth reserva- tion/allocation policy is capable of accommodating more streams in each link while satisfying the end-to-end delay bound for each stream. We compare our that our protocol is able to scale best as the number of network nodes and of-fered streams increase. In addition, in case that the delay bound tends to re-lax and the capacity of the network is insufficient to convey all the streams offered by the source, our algorithm can admit the streawith larger priori-ties so that the overall utilization and efficiency are leveraged.