摘要
在区别服务网络内为使传输控制协议 (TCP)连接在网络拥塞时得到同目标速率相一致的吞吐量 ,需要分析影响 TCP连接吞吐量的主要因素 .基于流模型假定 ,推导了一个确信服务 TCP连接在边缘路由器采用漏桶标记算法和核心路由器采用 In和 Out随机提前检测算法的吞吐量模型 .TCP连接吞吐量是一个漏桶参数、确信服务 TCP连接目标速率、TCP连接的端到端时延和数据包丢失率的函数 。
For a transfer control protocol (TCP) flow in differentiated service network to get a throughput consistent with target rate in the case of congestion, it needs to identify the key factors with the effect on TCP throughput. Based on a fluid model, this paper presented a throughput model of TCP flow when edge routers use token bucket tagged algorithm in conjunction with that core routers use Random early detection with In and Out algorithm. The throughput of a TCP flow is a function of parameters with token bucket, target rate, end-to-end delay and packet loss rate. The model then is validated by the simulation results.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2001年第6期870-873,共4页
Journal of Shanghai Jiaotong University
