Search Results

1. I. Juva, Traffic Matrix Estimation, Licentiate thesis, Networking Laboratory, Helsinki University of Technology, 2005 (bib)

   Abstract: In a communication network, the traffic has a source where that particular traffic flow is originating from and a destination where it is terminating at. Each origin-destination combination constitutes an origin destination (OD) pair. The knowledge of the amount of traffic of each such OD pair in the network is represented by the \emphtraffic matrix. The traffic matrix is a required input in many network management and traffic engineering tasks, where in many cases the knowledge on the traffic volumes are assumed to be known. However, in reality, they are seldom readily obtainable in networks, as only the link count measurements and routing information is available. Solving the OD counts from these is an underconstrained problem. Thus, it is not solvable, unless some extra information is brought into the situation. This thesis gives a comprehensive overview of the estimation methods proposed in the literature. These are divided into a few main groups based on the nature of the extra information each approach uses. The methods based on the gravity model assume that the traffic between two nodes is proportional to the product of the total traffic volumes of the nodes. In the Maximum likelihood methods the sample covariance of link counts is used. In the Bayesian methods there is an assumption about a prior distribution for the estimate. The thesis describes each proposed method and reviews the comparative studies made to evaluate the performance of the methods We propose a novel method for traffic matrix estimation: The Quick method based on link covariances, which yields an analytical expression for the estimate and is thus computationally light-weight. The accuracy of the method is compared with that of other methods using second moment estimates by simulation under synthetic traffic scenarios.