Congestion Control Strategy Based-on Active Network

In 21st century, Internet is getting rapid development. With the increase of various new service requests and the improvement of network technologies, the traditional network architecture can not meet the tremendous needs of users. Various emerging business and applications make network communication traffic significantly increase, which increment far greater than the increase of network bandwidth, and lead to serious network congestion.Firstly, this thesis analyzes the cause of network congestion. Then, this thesis reviews and summarizes the fundamental of TCP/IP congestion control and its typical techniques, and discusses the troubles of TCP/IP congestion control; include self-similarity, efficiency, fairness and algorithms improvement, etc. Whereafter, we compare the congestion control of TCP and IP, and analyze their characteristics, advantages and disadvantages, provide helps for future research.Secondly, this thesis gives an in-depth study of current design ideas of active networks architectures. Active networks not only have the basic capability of forward packets toward their destination as of traditional packet network, but also have the capability to support dynamic programming. Based on this study, the design ideas of active networks are introduced into the research of congestion control algorithms. In purpose of solving their problems of traditional network, we present a new congestion control algorithms based on active network - CCAAN.This algorithm can not only precisely detect the network congestion and timely notify the sources, but also provide more effectively congestion control according to the characteristic of data flows.Finally, the algorithm is evaluated in a simulated environment. Its results show that CCAAN can improve users' average throughput obviously, and the time delays of calculation are also very short.