A Network Congestion Control Algorithm Based on History Connections and Its Performance Analysis

It is well known that the end to end congestion control mechanism embedded in TCP has important contribution to the success of the Internet It ensures the stability of network, and prevents the congestion from collapsing The slow start is a vital phase in TCP congestion control, but its performance is usually not good for the major applications on Internet, such as the World Wide Web In order to solve the problem, an improved algorithm is proposed, which makes use of recent history information to initialize the parameters of new connections, and paces the packets by using the token technique during the first RTT The performance of the algorithm has been tested and evaluated on NS simulator The simulation results demonstrate that the proposed algorithm performs better in average transmitting time than traditional TCP employing the large initial window, especially for short connections, and it could also coexist with TCP flows friendly