Machine Learning Network Tomography with partial topology knowledge and dynamic routing

Networks are always progressing to support the evolving and diverse applications and the needs for improved capacity, latency and security. To this end, monitoring is key to ensuring the uninterrupted network operation and the QoS of the applications. Network Tomography uses a subset of monitoring information (corresponding to partial view of the network state) to estimate wide-sense network performance, including unmonitored parameters. In this paper, we present a novel Machine Learning (ML) formulation for Network Tomography. The proposed formulation accounts for realistic scenarios where: i) the existence of certain links of the network is not known (e.g., due to security reasons), ii) the routing is dynamic (non-deterministic), i.e., for the same origin-destination node pair, a different route may be selected depending on the state of certain links. Our simulations indicate that our proposal has better estimation accuracy compared to traditional algebraic or other ML approaches that cannot or do not take into account these two assumptions.