Maximum Likelihood Estimation for Multiple-Source Loss Tomography with Network Coding

Loss tomography aims at inferring the loss rate of links in a network from end-to-end measurements. Previous work has developed optimal maximum likelihood estimators (MLEs) for link loss rates in a single-source multicast tree. However, only sub-optimal algorithms have been developed for multiple-source loss tomography. In this paper, we revisit multiple-source loss tomography in tree networks with multicast and network coding capabilities, and we provide, for the first time, low-complexity MLEs for the link loss rates. We also derive the rate of convergence of the estimators.