Skip to Main Content
Network tomography aims to obtain network characteristics by end-to-end measurements. Most works carried out in recent years focused on the methods and methodologies to identify some of the characteristics, such as loss rate, delay distribution, etc. which show long-term network behaviours. We in this paper turn our attention to link-level temporal correlation, in particular the transition probability of each link, and introduce a new method that is simple and fast to identify link-level transition probability. In addition, we will show that the information embedded in the transition probability is richer in some aspects than some of those long-term characteristics since those long-term characteristics can be derived from the temporal ones. The proposed method is tested in simulations, the results show the loss rates obtained by the proposed method are comparable to those obtained by the traditional maximum likelihood estimate (MLE).