摘要
为了应对实际的业务应用需求,在一定程度上提高网络带宽利用率,在一般实时数据转发系统(RTDFS)的基础上,针对该类系统,提出了优化TCP链路和转发缓冲区取值的方法。主要根据转发系统的网络链路状况,以及实时数据到达转发系统的特征,利用排队论的相关理论建立相应的数学模型,并利用影响TCP传输吞吐量的主要参数,计算出系统运行服务器中最佳的转发缓冲区大小和并行的线程数量。实际系统的实现结果表明,标称带宽为2 M,实时数据记录端到实时数据转发端的往返时延约为2 ms,实时数据转发端到集控中心服务器的往返时延约为154 ms的情况下,转发端的缓冲区无限大、多线程个数为4时,系统运行带宽利用率能达到90%以上,并且验证了该数学模型的理论计算。
In order to cope with the actual requirements of business applications and improving utilization of network bandwidth to a certain extent.This paper provided a method to optimize the TCP links and the data forwarding buffer on the basis of the general real-time data forwarding system（RTDFS）.The method was mainly based on the network link status and the characteristics of real-time data arrived at forwarding system, utilized the relevant theories of queuing theory to build the corresponding mathematical model, and made use of main parameters which affected TCP transfer throughput, then figured out the best data forwarding buffer size and the number of parallel threads in communication server. The results of actual system implementation show：when the nominal bandwidth is 2M, the round-trip delay between real-time data recording and real-time data forwarding end is about 2ms, the round-trip delay between real-time data forwarding and control center is about 154ms,the bandwidth utilization can be above 90% when buffer is infinite and the number of multi-thread is 4, which verifies the theory computation of mathematic model.
出处
《电子设计工程》
2010年第4期80-82,共3页
Electronic Design Engineering
关键词
实时数据转发
排队模型
缓冲区
多线程
real-time data forwarding
queuing model
buffer
multi-thread
作者简介
李贞（1985-），女，浙江舟山人，硕士研究生。研究方向：嵌入式系统与宽带通信。
