Bayesian Network Tomography and Inference

Abstract

The aim of this technical report is to give a short overview of known techniques for network tomography (introduced in the paper of Vardi (1996)), extended by a Bayesian approach originating Tebaldi and West (1998). Since the studies of A.K. Erlang (1878-1929) on telephone networks in the last millennium, lots of needs are seen in todays applications of networks and network tomography, so for instance networks are a critical component of the information structure supporting finance, commerce and even civil and national defence. An attack on a network can be performed as an intrusion in the network or as sending a lot of fault information and disturbing the network flow. Such attacks can be detected by modelling the traffic flows in a network, by counting the source destination packets and even by measuring counts over time and by drawing a comparison with this 'time series' for instance.