Research On Congestion Control For Satellite Networks

In today's information age, Internet in people's daily life play an increasing role and its development speed is very fast, so the ground communication has been unable to meet the coverage, communication speed and data bandwidth demand of modern communication. With Internet and the development of space technology supports, satellite networks has been develop rapidly for its high-speed, timely, reliable and wide-area coverage without grapes.However, there are many problems have to be solved when satellite links directly access to the Internet, because of the differences between satellite channel and ground channel. Some characteristics of satellite networks limit the development of the Internet core technology, the TCP performance. Over satellite network, the existing TCP congestion control mechanism is not so efficient as on the ground channel. Many improved or new congestion control algorithm has been proposed at home and abroad, promoted the use of congestion control research, and made a good run effect. However, some of these programs only have made for a particular problem, for other problems are not addressed; some programs in these processes required relatively large network overhead, it is a waste of resources; some programs in addition to hardware, and other than the TCP protocol level or routing protocols have additional requirements, it is difficult to achieve. This thesis expounds the cause of congestions, analyzes the existing TCP performance improvement programs for satellite networks, and studies the congestion control scheme for satellite networks, discusses the improvement program of end to end protocols and the schemes based on Middleware, and proposed the research focuses. This thesis based on end to end improved protocol, have been studied and improved, the main research results are as follows:1. In TCP-Peach+ programs,NIL segment lose seriously in the high error rate environment, like satellite networks, cwnd is difficult to restore to the initial state in the moments later Quick Recovery, and with the cwnd decreased , the data transfer rate is also reducing. This paper proposed an improved Quick Recovery mechanism, named Modified Quick Recovery,to restrain the lower of cwnd and stop the deterioration of TCP-Peach+ performance.2. TCP-Peach+ set the ssthresh to half of cwnd in the congestion avoidance phase, and it is a waste of existing bandwidth resources. This paper proposed a new scheme, named TCP-nPeach, based on Bandwidth Estimation. This algorithm use ABSE Mechanism to get Bandwidth Estimation by observing the arrival of ACK. TCP-nPeach set cwnd and ssthresh by the estimated bandwidth when receive three ACK. Simulation results shows that the new algorithm has a very good improvement in Goodput and Fairness.