A comparison of some bottleneck-link detection methods for network tomography

Network tomography offers a useful method to identify internal problems in a network using data which can be obtained at the network's edge. Provided the topology of the network is known then it is possible to recover from the edge measurements some properties of internal links of the network which may not be accessible for any number of reasons - cost, ownership, physical location etc. In this paper we introduce two new estimation algorithms based on the Pearson type-1 distribution and compare these with existing estimator and detector algorithms used to find the bottleneck link in a wired network, that is, the link experiencing the highest delay.