一种基于分组丢失率测量的差错容忍式拥塞控制算法  

An Error Tolerated Congestion Control Algorithm Based on Packet Loss Rate Measurement in Deep Space Communication Network

作　　者：严军荣[1] 卢玉龙[1] 潘鹏[1,2] 

机构地区：[1]杭州电子科技大学通信工程学院,杭州310018 [2]南京邮电大学通信与信息工程学院,南京210003

出　　处：《电信科学》2015年第11期39-43,共5页Telecommunications Science

基　　金：浙江省自然科学基金青年基金资助项目(No.LQ13F010013);第56批中国博士后科学基金面上基金资助项目(No.2014M561693);国家自然科学基金青年科学基金资助项目(No.61401130)~~

摘　　要：空间通信的TCP大多数是基于Vegas算法,该算法需要对往返时延进行较为精确的测量,这在具有极长且可变时延的信道特征的深空通信环境中很难实现。提出一种基于分组丢失率测量的差错容忍式拥塞控制算法,该算法采用数据块的形式发送数据,依据历史数据设定差错容忍度,利用分组丢失率测量值进行拥塞状态判断及发送窗口大小调整,从而使用较小的开销达到较高的传输效率。最后,利用数学建模方法,证明基于分组丢失率测量的差错容忍式拥塞控制算法的吞吐量比传统TCP的Tahoe算法提高34c,比Reno算法提高22%。Congestion control algorithms of TCP protocol previously used in space communications are mostly based on TCP Vegas algorithm. They always require accurately measurement of RTY, but it＇s very difficult to achieve in deep space communication networks which have very long and variable delay channel characteristics. An errors tolerated congestion control algorithm is proposed based on packet loss rate measurement. The algorithm sends data in the form of data blocks and gives an errors tolerated degree based on the history data. It measures the congestion state based on the current packet loss rate measurement, so it can achieve a smaller overhead and higher transmission efficiency. In addition, the usage of mathematical modeling methods simplifies the networks and simulates the real network traffic. The mentioned algorithm is compared with the classical congestion control algorithm. Analysis and simulation prove that thethroughput of improved algorithm is 34 percent better than Tahoe and 22 percent better than Reno, so it gets a better performance.

关 键 词：拥塞控制 数据块 分组丢失率 差错容忍 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]