项目批准号：61773108

项目名称：大数据环境下的TCP网络拥塞控制算法研究     英文名称  Algorithms of TCP network congestion control in big data environments

项目起止年月：2018年01月至 2021年12月

中文摘要：

随着云计算、社交媒体等新兴信息技术和应用模式的快速发展，促使全球数据量急剧增加，推动人类社会迈入大数据时代。因此，基于传输控制协议(TCP)的通信网络得到飞速的发展和广泛的应用。为此，本项目针对TCP网络的主动队列管理方案，设计拥塞控制算法。首先，根据实际网络的随机特点，引入马尔科夫过程建立TCP网络的数学模型，相比较于现有的模型，所得到的模型更加准确的描述了系统的变化特性；其次，考虑大数据环境下UDP流以及TCP网络的饱和等因素，设计非线性拥塞控制器实现网络的拥塞避免；再次，提出基于事件的触发策略，以减少控制器的传输数据，进而可改善网络的拥塞情况；最后，本项目提出一种全局拥塞控制策略，使拥塞节点可利用与其相连接的其他节点的链路容量来间接增加自身的链路容量，进而实现拥塞控制。本项目拟开展的研究工作完全从TCP网络实际应用中存在的问题出发，将会为TCP网络拥塞控制的应用做出有益的尝试和探索。

Abstract：

With the rapid development of Cloud Computing and Internet of Things, the data has increased sharply in the word. This urges our society into the age of Big Data, so, the communication networks based on TCP have developed rapidly and been applied widely. Therefore, this project focuses on the congestion control algorithm for the TCP network based on the Active Queue Management (AQM). Firstly, based on the stochastic character of the real network, the Markov process is used to build the mathematic model of the TCP network. Compared with the existing model, the new model will describe the change accurately. Secondly,  considering the User Datagram Protocol (UDP), the saturation of input and other factors in the network, the nonlinear controller is proposed to avoid congestion. Thirdly, the control algorithm based on event-triggered will be proposed in order to reduce the transferred data and then the congestion can be reduced. Finally, this project presents a global congestion control algorithm. When the congestion happens, the router could cooperate with other connected routers which can  expand the capacity indirectly so as to distribute more and more data from the congested router. The researches to be developed are based on the practical problems, undoubtedly, the project is a beneficial attempt and exploration of the application for the congestion control of TCP networks.

项目简介     项目团队     内容展示     基本问题

  

Home | Researches | Education | Awards | Teaching | Interactions | Publications | Links

Copyright © 2021. 
All rights reserved. Reproduction without permission is strictly prohibited.