Research On The Prediction Algorithm Of The Multimedia Avaliable Bandwidth Range

With the increasing of multimedia on the Internet, multimedia not only produces mess flows, but also can not response to the network congestion. Moreover, it can try its best to use up the available bandwidth. When the media streaming based on UDP and the TCP flows share the bandwidth, UDP flows will behave aggressively and oppcuy the resources unfairly. Seriously, it leads to the collapse of the network congestion or the starvation of TCP traffics. A fair bandwidth allocation mechanism must be needed in order to insure enough bandwidth for the traditional TCP applications and to prevent the multimedia robbing bandwidth.Media streaming may rob the available bandwidth without control and causes unfair transmission between media streaming and TCP flows. Predicting the variation of the available bandwidth may balance the usage of bandwidth. Previous woks focused on predicting the average of the bandwidth, largely ignoring available bandwidth is a variety. New bandwidth prediction is proposed based on analyzing the importance of predicting the range of the end-to-end available bandwidth: using Gaussian model to describe the available bandwidth and then obtaining the corresponding available bandwidth variation through two percentiles of the available bandwidth distribution. Multimedia adjusts sending rate according to the available bandwidth range.According to the prediction algorithm, a fair transmission system which ensures the fair allocation is realized. It can predict the range at the receiver and adjust sending rate at the sender. The predictive information is sent to the sender through Real-time Transfer Protocol. The range can be predicted accurately under different network load. The fair allocation between multimedia and TCP flows and the requirements of quality of services are realized.