Robust Loss Inference in the Presence of Noisy Measurements

This paper addresses the problem of inferring link loss rates based on network performance tomography in noisy network systems. Since the network tomography emerged, all existing tomography-based methods are limited to the basic condition that both network topologies and end-to-end routes must be absolutely accurate, which in most cases is impractical, especially for large-scale heterogeneous networks. To overcome the impracticability of tomography-based methods, we propose a robust tomography-based loss inference method capable of accurately inferring all link loss rates even when the network system may change dynamically. The new method first measures the end-to-end loss rates of selected paths to reduce the probing cost, and then calculates an upper bound for the loss rate of each link using the measurement results. Finally, it finds all the link loss rates that most closely conform to the measurement results within their upper bounds. Compared with two traditional loss inference methods (with and without path selection, respectively), the results strongly confirm the promising performance of our proposed approach.