Multi-node attack strategy of complex networks due to cascading breakdown. (January 2018)

• Record Type: Journal Article
• Authors: Chaoqi, Fu; Ying, Wang; Xiaoyang, Wang; Yangjun, Gao
• Is Part Of: Chaos, solitons and fractals. Volume 106(2018)
• Page Start: 61
• Page End: 66
• Publication Date: 2018-01
• Subjects: Cascading failure -- Multi-node attack -- Average avalanche scale -- Tolerance coefficient; Chaotic behavior in systems -- Periodicals; Solitons -- Periodicals; Fractals -- Periodicals; Chaotic behavior in systems; Fractals; Solitons; Periodicals; 003.7
• DOI: 10.1016/j.chaos.2017.11.009
• Languages: English
• ISSNs: 0960-0779

Abstract

Highlights: The effect of multi-node attack strategies under different conditions is studied. As with single node attacks, when θ = 1, network has the strongest robustness under multi-node attacks. If the distance among attacked nodes was 2, the redistribution load could be better used. Three kinds of conditional attack strategies have different advantages under different conditions. Abstract: Studying attack strategy of complex networks is the basis of investigating network characteristics such as robustness, invulnerability, and network security. Knowing means of attack can help us take more effective measures to ensure network security. Presently, most research conclusions focus on a single vertex being attacked, and the choice of a set of attack nodes is also limited to a complete understanding of network information. In this paper, considering the effect of cascading failure, we focus on the multi-node attack strategy. Our results showed that the distance between attack targets has a great effect on the attacking effect. Taking both the average avalanche scale and maximum destruction size into account, when the distance between attack targets was 2, the network suffered the most serious damage. If the information about the network was unclear, we presented 3 kinds of conditional attack strategies. Under the condition of different tolerance coefficients and different degrees of known information, each strategy had its own unique advantages. In conclusion, the research in this paper supports the easy and quick selection of attack targets under the condition of incomplete information.