| 17 |
How likely do severe infrastructure failures occur?Exact computation of network's scale-wise unreliability 
|
|---|
For network infrastructures such as telecommunication and power, to avoid severe outages, a high-reliability design where the probability of occurrence of large-scale outages is significantly low is in demand. We propose an algorithm that precisely computes the probability of occurrence of outages for every outage scale when network components fail occasionally. To improve efficiency, we use data structures called decision diagrams, enabling us to avoid repetitively performing equivalent computations. As a result, the proposed method successfully computes the probability of occurrence of every outage scale precisely for real-world network topologies with around 100 nodes. The proposed method contributes to checking whether a designed network meets a severe reliability requirement needed for modern network infrastructures. In the future, we want to automatically design network infrastructures such that large-scale outages are less likely to occur by extending the proposed method.
[1] K. Nakamura, T. Inoue, M. Nishino, N. Yasuda, S. Minato, “Exact and efficient network reliability evaluation per outage scale,” in Proc. The 2023 IEEE International Conference on Communications (ICC), 2023.
Kengo Nakamura
Linguistic Intelligence Research Group, Innovative Communication Laboratory