Research And Improvement On Congestion Control Scheme Over Multimedia Real-time Applications

The rapid development of the Internet provides a significant platform for the development of the multimedia real-time applications. The multimedia real-time applications as a new method of the information transform are getting more and more attention of the researchers.TCP protocol provides reliable data transform for the Internet, its congestion control scheme based on AIMD algorithm makes it a crucial guarantee for the steady operation of the Internet. The multimedia real-time applications have a high requirement for the performance of real-time and steadiness. The window-based sending rate adjusting scheme brings a superior jitter of the TCP sending rate. The retransmission scheme of TCP results that the data can not get to the receivers in time, that is to say, TCP has a bad performance of real-time. So TCP can not satisfy multimedia real-time applications the requirement for the performance of real-time and steady. General multimedia real-time applications use UDP as transport protocol which solves the problem of the drawback of superior delay and jitter and suits good for the multimedia real-time applications. But UDP protocol lacks congestion control scheme. When there is congestion it will not only occupy lots of TCP bandwidth but also bring more serious congestion. So it is required to devise certain protocols which support congestion control scheme on the network layer in order to satisfy multimedia real-time applications the requirement for congestion control and makes multimedia real-time applications share bandwidth with TCP flows friendly. A common multicast TCP-friendly real-time applications congestion control scheme is TFMCC protocol. TFMCC adopts a TCP long-term throughput equation to adjust the sending rate and the sending rate of which has a good performance of steadiness. To reduce the delay and jitter of TFMCC further more is necessary for multimedia real-time applications.TFMCC protocol only uses packet loss ratio as the congestion symbol. Before packet loss occurs the source can not forecast congestion. It can be informed congestion by the time the packet loss occurs. Before the packet loss occurs, TFMCC protocol can not forecast congestion and the network congestion accumulates in a long time which will result in a high loss ratio. TFMCC protocol does not take delay and jitter into account when it achieves congestion signal. It has no special module and scheme to calculate delay and jitter and can not control network delay and jitter efficiently. The sending rate of TFMCC has a great jitter with the variation of the delay and jitter. In order to solve the problems pointed out above, we presented an improved algorithm TFMCC-BDJ (Improved TFMCC protocol based on end-to-end Delay and Jitter). This algorithm uses end-to-end delay jitter as the network congestion signal before the packet loss occurs and adds to the source an efficient control of the performance of real-time and steadiness. So this algorithm is suitable for the multimedia real-time applications. TFMCC protocol only uses network property to adjust sending rate and does not take specific applications into account. In order to solve this problem, we presented an improved algorithm TFMCC-BMR (Improved TFMCC protocol based on Minimal Rate). This algorithm uses no feedback timer to judge congestion and judge crowd via comparing current sending rate and minimal rate and then adjust the sending rate according to the degree of the crowd. Finally, we used the network simulation software NS to verify the performance of TFMCC-BDJ and TFMCC-BMR. The result of the simulation proved TFMCC-BDJ brings remarkable performance ameliorate to the average end-to-end delay, jitter and packet loss ratio and can discover network congestion in time on the premise of preserving TCP-Friendliness with TCP flows. The improved TFMCC-BMR can ensure the effective operation of the applications on the premise of overall TCP Friendliness.