Critical region identification and geodiverse routing protocol under massive challenges

Abstract

Regionally-correlated failures or attacks pose a great challenge to the normal network communication for physical backbone networks. When the same intensity of challenges occur at different physical locations, the damage to the network connectivity varies greatly. In this paper, we propose a critical region identification model and demonstrate its effectiveness in finding critical regions for fiber-level networks under regionally-correlated failures or attacks. We apply the model on several real-world backbone networks to demonstrate its efficiency using both unweighted and weighted topologies. Furthermore, the identified critical region result is used to improve the routing performance using GeoDivRP, a resilient routing protocol with geodiversity considered.