适合GEO卫星网络的分层组播拥塞控制算法

Layered Multicast Congestion Control Algorithm for GEO Satellite Networks

针对GEO卫星网络高误码率、长延时的特点,提出了一种适合GEO卫星分层组播的拥塞控制算法,SLM-CC。采用系统瓶颈处的队列信息作为调整层的依据,能够及时对网络状况做出反应,解决了卫星网络长延时特性带来的IGMP离开延时大的问题。通过周期性的比较组播数据流和TCP流的平均速率,动态调整组播层,满足了TCP友好性;针对不同分层体制的组播数据流,提出一种基于累积速率的层优先级调整策略,保证了多个不同分层体制的组播流之间的带宽公平性。另外,SLMCC设置一个比较系数,避免了卫星TCP性能下降导致组播数据流吞吐量的同步下降。仿真结果证明了SLMCC在GEO卫星网络环境中的有效性。

A new layered multicast congestion control algorithm SLMCC is proposed for GEO satellite networks in this paper, aiming at the network characteristics of high BER and long propagation delays. In SLMCC, the queue information at system bottlenecks is used in adjusting the multicast layers in time, which eliminates the influence of the GEO long propagation delays. The average rate of multicast flows and of TCP flows passing through the bottlenecks are measured and compared for TCP-friendliness, and a novel rate-based layer priority modification policy is designed for bandwidth fairness among heterogeneous multicast sessions. In order to avoid the unnecessary decrease of multicast rate along with the bad-performance TCP flows in high BER satellite channels, a comparison coefficient α is adopted in SLMCC. Simulation results show that SLMCC significantly improves the performance of layered multicast in GEO satellite environments.