Paper detail

Robustness of Complex Networks against Attacks Guided by Damage

Extensive researches have been dedicated to investigating the performance of real networks and synthetic networks against random failures or intentional attack guided by degree (degree attack). Degree is one of straightforward measures to characterize the vitality of a vertex in maintaining the integrity of the network but not the only one. Damage, the decrease of the largest component size that was caused by the removal of a vertex, intuitively is a more destructive guide for intentional attack on networks since the network functionality is usually measured by the largest component size. However, it is surprising to find that little is known about behaviors of real networks or synthetic networks against intentional attack guided by damage (damage attack), in which adversaries always choose the vertex with the largest damage to attack. In this article, we dedicate our efforts to understanding damage attack and behaviors of real networks as well as synthetic networks against this attack. To this end, existing attacking models, statistical properties of damage in complex networks are first revisited. Then, we present the empirical analysis results about behaviors of complex networks against damage attack with the comparisons to degree attack. It is surprising to find a cross-point for diverse networks before which damage attack is more destructive than degree attack. Further investigation shows that the existence of cross-point can be attributed to the fact that: degree attack tends produce networks with more heterogenous damage distribution than damage attack. Results in this article strongly suggest that damage attack is one of most destructive attacks and deserves our research efforts.Our understandings about damage attack may also shed light on efficient solutions to protect real networks against damage attack.