Network failure assessment based on graph signal processing

Network fault analysis is done to assess network degradation and pursue network planning. A real-time assessment of the extent of degradation is usually provided corresponding to the forwarding capacity of the failed routing node or link. However, the network operates in a diminished capacity and reroutes services to cope up with the faults. This introduces congestion on certain flows, or latency on services and functions, which are many hops away from the faults. The spread of such failures is referred to as the blast radius. To assess network degradation, existing methods use correlation of flows, link utilization and routing schemes to services and network functions, which are time-consuming. In this work, we propose a new method to assess the spread of such failures, which follows the progression of flows with network topology. This new method is based on sequential estimation and graph signal processing. The method provides an estimate of the blast-radius and eventual flow distribution in the network. We simulate one or multiple faults in different network topologies to verify our method. The simulation results show that our method provides an effective way to assess network degradation due to node or link failures.