<p>The reliability of networks exposed to large disasters has grasped the research community's attention and has become a critical concern in network studies during the last decade. Looking at the damages caused by recent disasters, with earthquakes top the list of those showing more significant impacts on communication networks, and simultaneously, the least predictable events. This study uses the Probabilistic Seismic Hazard Analysis method to estimate the network element state after an earthquake. The approach considers a seismic source model and ground prediction equations to assess the intensity measure for each element according to its location. In the simulation, nodes fail according to the building's fragility curves. Similarly, links fail according to a failure rate estimated based on the intensity measure and the cable's characteristics. We use the source-terminal, and the diameter constrained reliability metrics. The approach goes beyond the graph representation of the network accounting for the seismic risk in the geographical region where the network is embedded. Incorporating the terrain characteristics and the component's robustness into the network performance analysis at an a ordable computational cost.</p>

<p>The research community’s attention has been attracted to the reliability of networks exposed to large-scale disasters and this has become a critical concern in network studies during the last decade. Earthquakes are high on the list of those showing the most significant impacts on communication networks, and simultaneously, they are the least predictable events.  This study uses the Probabilistic Seismic Hazard Analysis method to estimate the network element state after an earthquake. The approach considers a seismic source model and ground prediction equations to assess the intensity measure for each element according to its location. In the simulation, nodes fail according to the building's fragility curves. Similarly, links fail according to a failure rate depending on the intensity measure and the cable's characteristics. We use the source-terminal, and the diameter constrained reliability metrics. The approach goes beyond the graph representation of the network and incorporates the terrain characteristics and the component's robustness into the network performance analysis at an affordable computational cost.  We study the method on a network in a seismic region with almost 9000 km of optical fiber. We observed that for source-terminal that are less than 500 km apart the improvements are marginals while for those that are more than 1000 km apart, reliability improves near a 30\% in the enhanced designs. We also showed that these results depend heavily on the robustness/fragility of the infrastructure, showing that performance measures based only the network topology are not enough to evaluate new designs.</p>

<p>The reliability of networks exposed to large disasters has grasped the research community's attention and has become a critical concern in network studies during the last decade. Looking at the damages caused by recent disasters, with earthquakes top the list of those showing more significant impacts on communication networks, and simultaneously, the least predictable events. This study uses the Probabilistic Seismic Hazard Analysis method to estimate the network element state after an earthquake. The approach considers a seismic source model and ground prediction equations to assess the intensity measure for each element according to its location. In the simulation, nodes fail according to the building's fragility curves. Similarly, links fail according to a failure rate estimated based on the intensity measure and the cable's characteristics. We use the source-terminal, and the diameter constrained reliability metrics. The approach goes beyond the graph representation of the network accounting for the seismic risk in the geographical region where the network is embedded. Incorporating the terrain characteristics and the component's robustness into the network performance analysis at an a ordable computational cost.</p>