首页 | 本学科首页   官方微博 | 高级检索  
     

DLMCC:一种动态层次组播拥塞控制机制
引用本文:任立勇,卢显良,吴劲,周旭.DLMCC:一种动态层次组播拥塞控制机制[J].计算机科学,2004,31(1):49-52.
作者姓名:任立勇  卢显良  吴劲  周旭
作者单位:电子科技大学计算机学院,成都,610054
基金项目:国防预研基金(51406070201DZ0211),电子科技大学青年科技基金(YF020803)
摘    要:由于Internet本身固有的异构性,累积层次组播受到广泛关注,但现有的层次组播大多是粗粒度和静态的,不能适应动态变化的网络环境。为解决这一问题,提出了一种动态层次组播拥塞控制机制DLMCC。在该机制中:①给出了一种反馈聚集算法,动态确定层次组播的层次数和层次速率,从而有效地提高了网络的带宽利用率;②采用了一种逐级分组对探测带宽方法,可准确快速测量本地带宽,同时可保证粗拉度的TCP友好;③所有组播层次的数据在同一个组播组中传输,彻底消除了接收者加入新的层次和离开最高层次时需执行复杂和耗时的IGMP操作,以及多个组播组导致的新的异构问题。仿真实验结果表明DLMCC是层次动态变化、TCP友好和可扩展的。

关 键 词:DLMCC  层次组播  拥塞控制  异构性  反馈聚集算法  粗拉度  通信协议

DLMCC:A Congestion Control Mechanism for Dynamic Layered Multicast
REN Li-Yong LU Xian-Liang WU Jin ZHOU Xu.DLMCC:A Congestion Control Mechanism for Dynamic Layered Multicast[J].Computer Science,2004,31(1):49-52.
Authors:REN Li-Yong LU Xian-Liang WU Jin ZHOU Xu
Abstract:More and more attentions have been paid on layered multicast for Internet intrinsic heterogeneity, but most existing layered multicast protocols cannot adapt to dynamic network conditions because their layers are coarse granularity and static. In this paper, a new congestion control mechanism for dynamic layered multicast (DLMCC)is presented. To implement this mechanism: first, a novel feedback aggregating algorithm is put forward. It can dynamically determine the number of layers and the rate of each layer, and can efficiently improve bandwidth utilization of network. Second, a bandwidth measurement method based on hierarchical packet-pair is introduced, which can infer local bandwidth quickly and accurately. Third, because the data of all the layers is transferred in only one group. the intricate and time-consuming IGMP operations, caused by receiver joining a new layer or leaving the topmost subscribed layer, are thoroughly eliminated. And this mechanism also avoids other problems resulted from multiple groups. Simulation results show that DLMCC is adaptive, TCP friendly and scalable.
Keywords:Dynamic layered multicast  Congestion control  TCP friendly  Aggregate algorithm
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《计算机科学》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号