Schemes For Handling Self-Similarity Traffic Based On The AQM Algorithm

It is a great discovery in the area of performance evaluation of high speed network since 1990, i.e., that the real traffic has self-similarity. The consequence of the discovery is that the conventional Markov models are not applicable to network traffic modeling. New models and tools have to be developed. Self-Similarity presents that the traffic in all the time scales (or in a wide range of time scales) is statistical similarity. The burst exists without a finite length. We cannot get rid of them by a long time scales. So the cluster traffic will not decrease and will become more burst. It presented that the traffic with self-similarity must do a bad affect on the performance of the high-speed network.There is a very useful Scheme for handling the traffic, which is"Active Queue Management". The domain strategy is dropping the Surplus packets before the route is in a traffic jam. In that way, the node can do something to deal with the traffic jam. It can avoid a lot of unnecessary loss (a lot of waste of bandwidth). It can raise the utilization of network with the above.RED Algorithm is the most important Algorithm in the AQM family. It is used in a wide range. The domain strategy is to monitor the average length of the queue in the post of route. When the traffic jam is coming, it tells the route to decrease the speed to avoid the traffic jam. RED Algorithm bases on FIFO and only drops the packet that has already arrived route, so it is very easy to realize. But there is few members of RED Algorithm can get rid of the bad affect of self-similar traffic.The innovation of my article is to develop a better algorithm, which bases on the shared buffer management schemes. The algorithm can get rid of the bad affect of self-similar traffic. The domain strategy is that setting the maximum and minimum, which is up to the data that is in a fit range recent. The maximum and minimum is fit to the fluid of the self-similar traffic. First, I analyze the characteristic and the mathematic model of the self-similarity; And, I present several character of self- -similar traffic. Then I write a strategy for deal with the self-similarity. The last, I introduce a software Ns-2 and show the numerical results.