• •    

一种基于马尔可夫相遇时间间隔预测的拥塞控制策略CCSMP

杨永健,王恩   

  1. 吉林大学计算机科学与技术学院
  • 收稿日期:2012-11-26 修回日期:2013-05-07 发布日期:2013-06-20
  • 通讯作者: 杨永健

A congestion control strategy based on Markov meeting time span prediction model

  • Received:2012-11-26 Revised:2013-05-07 Published:2013-06-20

摘要: 为了解决容迟网络中由于资源受限所带来的投递率较低的问题,通常采用基于多副本的路由机制来提高投递成功率,随之而来的问题是由于副本数过多而导致节点缓存超负荷,进而发生了网络拥塞。本文提出一种基于马尔可夫相遇时间间隔预测的拥塞控制策略CCSMP(the congestion control strategy based on Markov meeting time span prediction model),主要是通过规定节点缓存的排队方式和丢弃机制,将预测得到的较早与目的节点相遇的报文排于队首,尽可能丢弃效用值较低的报文,进而解决由于节点缓存有限而带来的拥塞问题。通过在ONE环境下进行仿真,与Drop-Front( DF) 和Drop-Oldest( DO)两种拥塞控制策略对比,实验结果表明文中提出的拥塞控制策略提高了报文投递率,减小了平均网络时延,并且在一定程度上减少了网络负载比率和丢包率。

关键词: 计算机科学技术基础学科, 拥塞, 马尔可夫相遇时间间隔, 排队方式, 丢弃机制

Abstract: In order to solve the problem of the low delivery ratio caused by the limited resources in DTN, routing mechanism based on multiple copies is usually used to improve the delivery ratio, the following problem is the overload of nodes buffer caused by the excessive numbers of copies, then network congestion happens. This paper proposes the congestion control strategy based on Markov meeting time span prediction model CCSMP, it mainly specified the line method and discarding mechanism of nodes buffer, the predicted messages meeting earlier with destination node are lined in the first, and discard the messages with lower utility values as far as possible, then solve the congestion problem caused by the limited node buffer. Through the simulation in ONE environment, compared with the two kinds of congestion control strategy Drop-Front( DF) and Drop-Oldest( DO), the experimental results show that the congestion control strategy proposed by this paper improves the message delivery ratio, and reduces the average network delay, and to a certain extend reduces the overhead ratio and packet dropped rate.

Key words: Basic disciplines of computer science and technology, congestion, Markov meeting time span , line method , discarding mechanism

中图分类号: 

  • TP393
[1] 刘洲洲, 彭寒. 基于节点可靠度的无线传感器网络拓扑控制算法[J]. 吉林大学学报(工学版), 2018, 48(2): 571-577.
[2] 杨永健, 王恩, 杜占玮. 基于马尔可夫相遇时间间隔预测的拥塞控制策略[J]. 吉林大学学报(工学版), 2014, 44(01): 149-157.
[3] 李鑫, 李京春, 林冉, 王少杰, 王文杰, 刘驰. 基于社交网络的实时共享服务高效拥塞管理机制[J]. 吉林大学学报(工学版), 2012, 42(增刊1): 255-260.
[4] 韩永, 姚念民, 蔡绍滨, 韩启龙. 拥塞感知的iSCSI虚拟存储系统[J]. 吉林大学学报(工学版), 2012, 42(增刊1): 368-372.
[5] 王萍,陈虹,卢晓晖. 基于预测控制的时滞网络主动队列管理[J]. 吉林大学学报(工学版), 2011, 41(4): 1089-1095.
[6] 夏海轮, 曾志民, 丁炜. Ad-hoc网络中一种新的自适应退避算法[J]. 吉林大学学报(工学版), 2009, 39(05): 1369-1373.
[7] 肖萍萍,田彦涛,杨晓萍 . 基于分组丢失的高带宽流鉴别算法[J]. 吉林大学学报(工学版), 2006, 36(05): 793-0798.
[8] 杨晓萍,史帅,陈虹. 一种改进的TCP拥塞控制算法[J]. 吉林大学学报(工学版), 2006, 36(03): 433-0437.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!