Using Bayesian networks on network tomography

Network tomography aims to obtain link-level performance characteristics, such as loss rate and average delay on each link, by end-to-end measurement. The obtained information can help us to understand the dynamic nature of networks. A number of methods have been proposed in recent years, which can be divided into two classes: multicast-based and unicast-based. In this paper, we propose an approach in the multicast class that uses the Bayesian network to carry out statistical inference. Simulations based on the network simulator 2 (ns2) were conducted, which shows our approach produced almost identical result as that produced by the maximum likelihood estimator previous proposed