| With the development of computer and communication technology, people's demand for the Internet has been increasing in an indescribable way, so the more and the better control mechanisms which could smooth the network play a very important part .The congestion control is one of the cruxes, that is how to avoid or eliminate the congestion in the network, so as to make the network basically run in the best and the low-grade congest state.Congestion in the network stems from the disequilibrium about the source and the flux in the network, it wouldn't be eliminated because the improvement of the network's process ability, to this day, the congestion hasn't had a perfect solution scheme all the while. Facing to the multifarious environment of the network, it has a certain difficulty not only in the design about the algorithm of the congestion control, but also in the function comparison of those protocols which needs a centralized norm. According to the position of the algorithm achieve, they fall into two algorithms: Source algorithm and Link algorithm. Source algorithm was performed at the mainframe and the end machines, its effect is to regulate the sending rate according to the feedback information; Link algorithm was performed at the network equipments, such as the router and the switch, its effect is to examine the emergence of the network congestion, and bring the feedback information. The crux is how to bring the feedback information and how to response to them.TCP protocol is the most widely used with source algorithm, and also the main popular transmission protocol, including four epochs, that is, the slow start, congestion avoidance, fast transmission and fast discovery, its core algorithm of congestion avoidance used a mechanism called AIMD (additive increase multiplicative decrease). TCP protocol has come through several improved versions. Every step makes possible higher bandwidth utilization rate .From RFC1122 to TCP Tahoe and to the most widely used TCP Reno, each modification has made a great enhancement in its performance. But at this HWAN period, the development of network technology such as optical communication technology leads to the appearance of many high-speed networks with bandwidth larger than 1Gbps, or even 10Gbps.Widely adopted as a data transfer protocol in current Internet, TCP performs badly in high-speed networks because of its congestion control algorithm. It is more and more obvious about its localizations, especially its response too slow at the large windows, and decrease too fast when the congestion happens. As a result, some new congestion control algorithms have been proposed.First, this paper analyzes the reason that leads to congestion and the current survey of the research on congestion control, points that congestion is closely relatived with the design mechanism of TCP/IP network. It then introduces the existing TCP source algorithm, the IP link algorithm associated with it. Also, this paper analyzes every phase of the TCP source algorithm deeply .In the end, the paper takes a proposal on the congestion avoidance of the TCP source algorithm. It introduces a proportional item (q) into the algorithm of this phase to make the transmission kept in the congestion avoidance as likely as possible, thereby reduces the times that congestion happens and improves the throughput. To compare the performance of algorithms easily, the leading tool of network simulation in TCP/IP research realm, NS, is introduced in this paper to make experiment result more objective and true.In this paper, we firstly gave an overview of the related work, and made then some contribution work as follows:1. Read the reference literatures, comprehend the basic congestion control and the current situation, analyze and summarize those new protocols.2. Use the simulation tool-NS-2, combine the theory, and put forward a new improvement algorithm QTCP. It enhances the cwnd , especially improved the quickly in the slow start, do better in the high-speed networks. |
PDF Full Text Request