Impact of Facility Factors on Robustness of Communication Networks under Natural Disasters

Communication networks are critical social infrastructures, and network failure due to natural disasters can impact severely our daily lives. The research community has extensively studied how to assess the robustness of networks against natural disasters. Although communication networks are protected by facilities such as buildings and underground conduits, conventional studies have not considered networks in robustness assessments. Thus, this study investigated the impact of underground conduits on network robustness during earthquakes. An actual conduit dataset was considered, including tens of thousands of conduits with their attributes (structure, material, length, and age) and damage status from past earthquake inspections. We employed a conduit damage prediction technique, a machine learning method developed in infrastructure engineering, and evaluated the robustness of regional communication networks with and without conduit attributes. The evaluation results revealed that the estimated outage scales could differ by a maximum factor of seven, depending on the conduit attributes. Additionally, a what-if analysis for conduit upgrades was conducted, which involved recognizing conduit attributes. These findings support the significance of facility factors in network assessment during natural disasters, and open up a new interdisciplinary research field between infrastructure engineering and telecommunication networks.