A Study Of Traffic Matrix Estimation Algorithm Of IP Network

Nowadays,as the Internet develops rapidly,the scale of the network is expanding quickly,the complexity of it is increasing promptly and the appearance of the new services and applications.So it requires high quality of network measuring.Managing such a massive network becomes an extremely challenging.It is necessary to get into the inner property of the specific network to design,control and manage it successfully.Traffic Matrix is one of the important parameters to identify the performance of the network..But it is hard to measure it directly as the Internet is becoming massive,distributed and heterogeneous.As the Traffic Matrix estimation is an important index of network measuring,it attracts more attention too.It is very useful to infer Traffic Matrix from link measurements and routing information,especially for the task of capacity planning, traffic engineering and network reliability analysis.It is also very important 1 for IP network management.But,estimation techniques based on partial information are used to populate Traffic Matrices because amassing sufficient data from direct measurements to populate a Traffic Matrix is typically prohibitively expensive.The term Network Tomography for the TM estimation problem when the partial data comes from repeated measurements of the traffic flowing along directed links of the network..Network tomography is based on end-end measurements in order to gain the information indirectly in the network.This paper analyzes the estimation algorithm of Traffic Matrix based on Network Tomography.First Introduce the Network Tomography,then Make some research on the existing algorithm about Traffic Matrix estimation,a new algorithm is called Gaussian Mixture Model method which estimates Traffic matrix based on Network Tomography,existing traffic matrix and Normal distribution is proposed.This method is mainly use GMM to describe the OD flow,and then use EM algorithm to estimate the parameters of this model.Basing on the information entropy,we propose Maximum Entropy to deal with the TM through the key mathematical techniques being the matrix transformations and Lagrange multipliers method,an expression is derived for calculating the TMs from the link measurements and the routing matrix. Finally use the new mathematical theory to simulate the algorithm.In accordance with the existing theory,experimental basis and backbone network topology analyze and compare the estimation algorithm of TM,then give the further research plan.Through both theoretical analysis and simulation result,it is shown that the GMM method achieve better performance and closer to the actual value of network.