文档介绍：摘要
TCP的拥塞控制由“慢启动(Slow start)”和“拥塞避免(Congestion avoidance)”组成,后来TCP Reno版本中又针对性的加入了“快速重传(Fast retransmit)”、“快速恢复(Fast Recovery)”算法,再后来在TCP NewReno中又对“快速恢复”算法进行了改进,近些年又出现了选择性应答(SACK)算法,还有其他方面的大大小小的改进,成为网络研究的一个热点。
TCP的拥塞控制主要原理依赖于一个拥塞窗口(cwnd)来控制,窗口值的大小就代表能够发送出去的但还没有收到ACK的最大数据报文段,显然窗口越大那么数据发送的速度也就越快,但是也就越可能使得网络出现拥塞,如果窗口值为1,那么就简化为一个停等协议,每发送一个数据,都要等到对方的确认才能发送第二个数据包,显然数据传输效率低下。TCP拥塞控制算法就是要在这两者之间权衡,选取最好的cwnd值,从而使得网络吞吐量最大化且不产生拥塞。[1]
计算机网络已经逐渐发展成为现代生活中必不可少的一个部分,本论文的主要研究目的就是针对计算机网络中的TCP协议中的Reno拥塞控制,对其实行可视化,使其变得更加容易理解,让我们对计算机网络有更深刻的认识。
关键词:慢启动;拥塞避免;快速重传;快速恢复;可视化
Abstract
TCP congestion control from " Slow start" and " Congestion avoidance" composed of TCP Reno versions later then targeted to join the " Fast retransmit", " Fast Recovery"algorithm, and then later again in the TCP NewReno in the" fast recovery "algorithm has been improved in recent years has emerged selective acknowledgement algorithm, as well as other large and small improvements e a research work.
The main principle of TCP congestion control relies on a congestion window (cwnd) to control the window size value represents the ability to send out but not yet received the maximum data packet ACK Duan, clear window, so the greater the speed of data sent the faster, but also more likely to make work congestion occurs, if the window is 1, then reduced to a stop such agreement, each sending a data, must wait for confirmation of the other party can send a second packet, the data clearly transmission efficiency is low. TCP congestion control algorithm is to balance between these two, choose the best cwnd value, allowing work to maximize throughput and does not create congestion.
work has gradually e an essential of modern life, a part of the main research objective of this thesis is work for the TCP Reno congestion control protocol in its implementation of visualization to make it more easy to understand , let work has a better understanding.
Key words: Slow start; Congestion avoidance; Fast retransmit;Fast R