Research On The Methods Of Traffic Matrix Estimation In IP Routing Networks

With the rapid development of the Internet,technologies such as 4G,WIFI,and the Internet of Things have been applied to the Internet.The number of mobile Internet and IP network terminals has grown exponentially.The scale of the network has grown,and the number of businesses in the network,such as online chat,video games,TV movies has also grown.All of these have led to an increasingly complex network structure,and also make the data types transmitted in the network under the current environment more diversified,which has led to more and more difficulties in the monitoring,management,and maintenance of the network.At the same time,the non-core services in the network also consume too many network resources which affects the normal operation of the network.In this environment,the traffic matrix estimation has extremely important significance.It is not only an important support for many network technologies,but also an important factor in network management.The traffic matrix,as a role that can observe the traffic of the whole network,reflects the composition of all the traffic in the entire network and describes the composition of the traffic in each link.The most important point in the traffic matrix estimation is how to overcome the ill-conditioned nature of the traffic matrix.That is,the number of rows in the traffic matrix is much smaller than the number of columns.The actual form in the network is that the number of OD pairs is much larger than the number of links.It is also a key point of this study.This paper firstly introduces the background,current situation,purpose and significance of the research.Next,the concept of the traffic matrix,the acquisition method,and the difficulty of acquisition are given.Then the traffic matrix is mathematically modeled to obtain the relationship among the traffic matrix,the routing matrix,and the link traffic.Then the traffic matrix estimation is described in detail,include the development of the method and the advantages and disadvantages of each algorithm.Then this paper establishes a new optimization model,and based on this model,a new method of generating a priori information is given.Then the matrix of generalized inverse and {1} inverse are used to calculate the traffic matrix.Finally,the traffic matrix is obtained,and is compared with the traffic matrices obtained under Poisson distribution and Gaussian distribution respectively.The results show that the estimated value obtained by our method is closer to the actual traffic matrix and the estimation effect is better.