Academia.eduAcademia.edu

Outline

Title
Abstract
FAQs
Introduction
Conclusions
References
All Topics
Computer Science
Algorithms and Data Structures

Error and attack tolerance of complex networks

Profile image of Vito LatoraVito Latora

2004, Physica A: Statistical …

https://doi.org/10.1016/J.PHYSA.2004.04.031
visibility

…

description

7 pages

link

1 file

Abstract
sparkles

AI

The paper explores the error and attack tolerance of complex networks, highlighting the differences between scale-free networks and random networks. It demonstrates that scale-free networks, while robust to random errors, are highly vulnerable to targeted attacks on critical nodes. A dynamical model is presented to illustrate the cascading failures that can result from the malfunction of a single component, emphasizing the importance of protecting scale-free networks from attacks given their real-world prevalence.

FAQs

sparkles

AI

What key property differentiates scale-free networks from random networks?add

Scale-free networks exhibit a power-law degree distribution P(k) ∼ k^(-α) with α between 2 and 3, contrasting with the Poissonian distribution of random networks.

How does node removal affect the efficiency of scale-free networks?add

The study reveals that removing 15% of nodes through targeted attacks reduces the efficiency of scale-free networks to about 0.33, with total destruction at 35% removal.

Which attack strategies are most impactful on network robustness?add

Degree-based attacks lead to rapid efficiency decline in scale-free networks, while in Erdős-Rényi graphs, differences in attack impact are less pronounced.

What models explain cascading failures in real-world networks?add

A dynamical model demonstrates that when critical nodes are removed, cascading effects can lead to network-wide failures, prevailing in systems like the Internet.

How does load redistribution influence network performance after node failure?add

Load redistribution causes congestion in unaffected nodes, which can result in communication delays, further propagating degradation across the network.

Related papers

Resilience Notions for Scale-free Networks
Gunes Ercal

Procedia Computer Science, 2013

Much of traditional graph theoretic analysis of networks had focused on regular or near-regular network models such as random Erdos-Renyi models, where the degree distribution is either the same for every node or highly concentrated about the mean. As such, much beautiful theoretical machinery exists to analyze various properties including analysis of network resilience via eigenvalues of the adjacency matrix representing the network. However, it has been recently observed that real world networks tend to be scale-free, which usually implies a high variance and power-law degree distribution. This poses a problem in applying existing theoretical machinery to some problems, particularly that of the resilience of networks to node attacks. In this work we examine networks in which the greatest discrepancy arises in attempting to apply previous resilience notions, and we tailor a new mathematical notion of resilience that works for scale-free networks in the presence of node attacks.

View PDFchevron_right
A quantitative method for determining the robustness of complex networks
Hongrun Wu

• This paper proposes a quantitative method for determining the robustness. • This method can deal with both node and edge attacks and assure the similar results.

View PDFchevron_right
Optimization of Robustness of Complex Networks
Eugene Stanley

Networks with a given degree distribution may be very resilient to one type of failure or attack but not to another. The goal of this work is to determine network design guidelines which maximize the robustness of networks to both random failure and intentional attack while keeping the cost of the network (which we take to be the average number of links per node) constant. We find optimal parameters for: (i) scale free networks having degree distributions with a single power-law regime, (ii) networks having degree distributions with two power-law regimes, and (iii) networks described by degree distributions containing two peaks. Of these various kinds of distributions we find that the optimal network design is one in which all but one of the nodes have the same degree, k1 (close to the average number of links per node), and one node is of very large degree, k2 ∼ N 2/3 , where N is the number of nodes in the network.

View PDFchevron_right
An Experimental Evaluation of Robustness of Networks
Hyoungshick Kim

IEEE Systems Journal, 2013

Models of conflict in networks provide insights into applications from epidemiology to guerilla warfare. Barabási, Albert and Jeong modelled selective attacks on networks in which an attacker targets high-order nodes to destroy connectivity; Nagaraja and Anderson extended this to iterated attacks where the attacker and defender take turns to remove and rebuild nodes and edges according to given strategies. We extend the iterative model by introducing the cost required to perform network operations. This gives much finer granularity than previous models, whether we are interested in network resilience against random failures or intentional attacks. We empirically study how to design more effective attacks and/or defenses through intensive simulation on several well-known network topologies including the three real-world networks. In particular, an effective defence against many attacks is to add new links connecting low-centrality nodes to maintain the overall balance of network centrality.

View PDFchevron_right
Robustness of complex networks to cascading failures induced by Poisson fluctuating loads
Xijin Tang

Physica D: Nonlinear Phenomena, 2019

Considering the fact that overload failures in real-world functional networks are usually caused by extreme values of temporally fluctuating loads that exceed the allowable range, we study the robustness of complex networks against cascading overload failures in such situation. In our model, at each cascading step the fluctuating load on a network is described by the total number of walkers, which is a random number that subjects to Poisson distribution. And a node fails when its overload probability is beyond its risk tolerance. Relative size of the giant

View PDFchevron_right
Multiscale vulnerability of complex networks
julio flores

Chaos, 2007

We present a novel approach to quantify the vulnerability of a complex network, i.e., the capacity of a graph to maintain its functional performance under random damages or malicious attacks. The proposed measure represents a multiscale evaluation of vulnerability, and makes use of combined powers of the links' betweenness. We show that the proposed approach is able to properly describe some cases for which earlier measures of vulnerability fail. The relevant applications of our method for technological network design are outlined.

View PDFchevron_right
Effect of random failures on traffic in complex networks
JORDI DUCH

2007

We study the effect of random removal of nodes in networks on the maximum capability to deliver information in communication processes. Measuring the changes on the onset of congestion, we observe different behaviors depending on the network structure, governed by the distribution of the algorithmic betweenness (number of paths traversing a node given a communication protocol) of the nodes, and particularly by the node with the highest betweenness. We also compare the robustness of networks from a topological and dynamical point of view. We find that for certain values of traffic load, after suffering a random failure, the network can be physically connected but the nodes are unable to communicate due congestion. These results highlight the necessity to include dynamical considerations in studies about resilience of complex networks.

View PDFchevron_right
"Characterizing behavior of Complex networks against perturbations and generation of Pseudo-random networks", Accepted in IEEE conference INDICON 2014.
Sk Jahid Ahmed, Bitan BANERJEE

Vulnerability of a real-world complex network against unwanted attacks and random link failures is an issue of immense concern. A small attack or failure of the network, has the potential to trigger a global cascading breakdown, thereby raising questions with regard to the possible strategies to combat such a mishap. Many works have been published lately, that deals mainly with the revival of a complex network after an attack or failure. In this paper, we propose to build the network architecture in an efficient manner, so that the network can withstand attacks or link failures up to some certain pre-specified limit. We introduce a novel approach to enhance the robustness of a network from the prevention point of view, that is prior to an attack or failure. Simulation results reveal that with a slight increase in the number of driver nodes, from that obtained using the existing maximum matching algorithm, enhances the stability of the network up to a large number of link failures. We also observe that, the sparse and inhomogeneous networks are difficult to control and are less robust, compared to dense and homogeneous networks.

View PDFchevron_right
Robustness envelopes of networks
Stojan Trajanovski

Journal of Complex Networks, 2013

We study the robustness of networks under node removal, considering random node failure, as well as targeted node attacks based on network centrality measures. Whilst both of these have been studied in the literature, existing approaches tend to study random failure in terms of average-case behavior, giving no idea of how badly network performance can degrade purely by chance. Instead of considering average network performance under random failure, we compute approximate network performance probability density functions as functions of the fraction of nodes removed. We find that targeted attacks based on centrality measures give a good indication of the worst-case behavior of a network. We show that many centrality measures produce similar targeted attacks and that a combination of degree centrality and eigenvector centrality may be enough to evaluate worst-case behavior of networks. Finally, we study the robustness envelope and targeted attack responses of networks that are rewired to have high-and lowdegree assortativities, discovering that moderate assortativity increases confer more robustness against targeted attacks whilst moderate decreases confer more robustness against random uniform attacks.

View PDFchevron_right
Optimization of Robustness and Connectivity in Complex Networks
Hiroki Sayama

Physical Review Letters, 2003

Scale-free networks rely on a relatively small number of highly connected nodes to achieve a high degree of interconnectivity and robustness to random failure, but suffer from a high sensitivity to directed attack. In this paper we describe a parametrized family of networks and analyze their connectivity and sensitivity, identifying a network that has an interconnectedness closer to that of a scale-free network, a robustness to attack closer to that of an exponential network, and a resistance to failure better than that of either of those networks.

View PDFchevron_right

References (20)

  1. S.N. Dorogovtesev, J.F.F. Mendes, Evolution of Networks, Oxford University Press, Oxford, 2003.
  2. S.H. Strogatz, Exploring complex networks, Nature 10 (2001) 268.
  3. R. Albert, H. Jeong, A.-L. Barabà asi, Nature 401 (1999) 130.
  4. A.-L. Barabà asi, R. Albert, Science 286 (1999) 509.
  5. M. Faloutsos, P. Faloutsos, C. Faloutsos, Comput. Comm. Rev. 29 (1999) 251.
  6. W. Li, X. Cai, cond-mat/0309236.
  7. H. Jeong, B. Tombor, R. Albert, Z.N. Oltvai, A.-L. Barabà asi, Nature 407 (2000) 651.
  8. H. Jeong, S.P. Mason, A.-L. Barabà asi, Z.N. Oltvai, Nature 411 (2001) 41.
  9. P. Erd os, A. RÃ enyi, Publ. Math. 6 (1959) 290.
  10. R. Albert, A.-L. Barabà asi, Rev. Mod. Phys. 74 (2002) 47.
  11. S. Wasserman, K. Faust, Social Networks Analysis, Cambridge University Press, Cambridge, 1994.
  12. R. Albert, H. Jeong, A.-L. Barabà asi, Nature 406 (2000) 378;
  13. R. Albert, H. Jeong, A.-L. Barabà asi, Correction, Nature 409 (2001) 542.
  14. P. Crucitti, V. Latora, M. Marchiori, A. Rapisarda, Physica A 320 (2003) 622.
  15. P. Holme, B.J. Kim, Phys. Rev. E 65 (2002) 066109.
  16. A.E. Motter, Y. Lai, Phys. Rev. E 66 (2002) 065102(R).
  17. Y. Moreno, R. Pastor-Satorras, A. VÃ asquez, A. Vespignani, Europhys. Lett. 62 (2003) 292.
  18. P. Crucitti, V. Latora, M. Marchiori, cond-mat/0309141 and Phys. Rev. E 69 (2004) 045104 (R).
  19. V. Latora, M. Marchiori, Phys. Rev. Lett. 87 (2001) 198701.
  20. V. Latora, M. Marchiori, Europ. Phys. J. B 32 (2003) 249.

Related papers

Vulnerability of complex networks
Mario Biey

Communications in Nonlinear Science and Numerical Simulation, 2011

We consider normalized average edge betweenness of a network as a metric of network vulnerability. We suggest that normalized average edge betweenness together with is relative difference when certain number of nodes and/or edges are removed from the network is a measure of network vulnerability, called vulnerability index. Vulnerability index is calculated for four synthetic networks: Erdős-Rényi (ER) random networks, Barabási-Albert (BA) model of scale-free networks, Watts-Strogatz (WS) model of small-world networks, and geometric random networks. Real-world networks for which vulnerability index is calculated include: two human brain networks, three urban networks, one collaboration network, and two power grid networks. We find that WS model of small-world networks and biological networks (human brain networks) are the most robust networks among all networks studied in the paper. interaction networks, neural networks, ecological networks) and etc. A central issue in the analysis of complex networks is the assessment of their robustness and vulnerability. Different approaches to address network robustness and vulnerability have recently been proposed by research community. The first approach is related to structural robustness : how different classes of network topologies are affected by the removal of a finite number of links and/or nodes. It was concluded that the more heterogeneous a network is in terms of, e.g., degree distribution, the more robust it is to random failures, while, at the same time, it appears more vulnerable to deliberate attacks on highly connected nodes. The second approach concerns dynamical robustness . For networks supporting the flow of a physical quantity, the removal of a node or link will cause the flow to redistribute with the risk that some other nodes or links may be overloaded and failure prone. Hence, a triggering event can cause a whole sequence of failures due to overload, and may even threaten the global stability of the network. Such behavior is termed cascading failure.

View PDFchevron_right
Resilience of Complex Networks to Random Breakdown
Eugene Stanley

Using Monte Carlo simulations we calculate f c , the fraction of nodes that are randomly removed before global connectivity is lost, for networks with scale-free and bimodal degree distributions. Our results differ from the results predicted by an equation for f c proposed by Cohen et al. We discuss the reasons for this disagreement and clarify the domain for which the proposed equation is valid.

View PDFchevron_right
Characterising the robustness of complex networks
Caterina Scoglio

International Journal of Internet Technology and Secured Transactions, 2010

With increasingly ambitious initiatives such as GENI and FIND that seek to design future internets, it becomes imperative to define the characteristics of robust topologies, and build future networks optimized for robustness. This paper investigates the characteristics of network topologies that maintain a high level of throughput in spite of multiple attacks. To this end, we select network topologies belonging to the main network models and some real world networks. We consider three types of attacks: removal of random nodes, high degree nodes, and high betweenness nodes. We use elasticity as our robustness measure and, through our analysis, illustrate that different topologies can have different degrees of robustness. In particular, elasticity can fall as low as 0.8% of the upper bound based on the attack employed. This result substantiates the need for optimized network topology design. Furthermore, we implement a tradeoff function that combines elasticity under the three attack strategies and considers the cost of the network. Our extensive simulations show that, for a given network density, regular and semi-regular topologies can have higher degrees of robustness than heterogeneous topologies, and that link redundancy is a sufficient but not necessary condition for robustness.

View PDFchevron_right
Characterizing the Robustness of Complex Networks
Caterina Scoglio

arXiv (Cornell University), 2008

With increasingly ambitious initiatives such as GENI and FIND that seek to design future internets, it becomes imperative to define the characteristics of robust topologies, and build future networks optimized for robustness. This paper investigates the characteristics of network topologies that maintain a high level of throughput in spite of multiple attacks. To this end, we select network topologies belonging to the main network models and some real world networks. We consider three types of attacks: removal of random nodes, high degree nodes, and high betweenness nodes. We use elasticity as our robustness measure and, through our analysis, illustrate that different topologies can have different degrees of robustness. In particular, elasticity can fall as low as 0.8% of the upper bound based on the attack employed. This result substantiates the need for optimized network topology design. Furthermore, we implement a tradeoff function that combines elasticity under the three attack strategies and considers the cost of the network. Our extensive simulations show that, for a given network density, regular and semi-regular topologies can have higher degrees of robustness than heterogeneous topologies, and that link redundancy is a sufficient but not necessary condition for robustness.

View PDFchevron_right
Comparison of Failures and Attacks on Random and Scale-Free Networks
Jean-Loup Guillaume

2004

It appeared recently that some statistical properties of complex networks like the Internet, the World Wide Web or Peer-to-Peer systems have an important influence on their resilience to failures and attacks. In particular, scale-free networks (i.e. networks with power-law degree distribution) seem much more robust than random networks in case of failures, while they are more sensitive to attacks. In this paper we deepen the study of the differences in the behavior of these two kinds of networks when facing failures or attacks. We moderate the general affirmation that scale-free networks are much more sensitive than random networks to attacks by showing that the number of links to remove in both cases is similar, and by showing that a slightly modified scenario for failures gives results similar to the ones for attacks. We also propose and analyze an efficient attack strategy against links.

View PDFchevron_right

Related topics

  • Pharmacology
  • Biochemistry
  • Bioinformatics
  • Evolutionary Biology
  • Genetics
  • Marine Biology
  • Neuroscience
  • Mathematical Physics
  • Environmental Science
  • Geophysics
  • Physics
  • Materials Science
  • Quantum Physics
  • Developmental Biology
  • Immunology
  • Climate Change
  • Molecular Biology
  • Structural Biology
  • Biotechnology
  • Systems Biology
  • Biology
  • Ecology
  • Drug Discovery
  • Nanotechnology
  • Intelligent Transport System
  • Astrophysics
  • Medicine
  • Scale-Free Networks
  • Scale free network
  • Community Networks

    • Cited by

    Functional essentiality from topology features in metabolic networks: A case study in yeast
    Lorenzo Farina

    FEBS Letters, 2005

    The relation between the position of mutations in Saccharomyces cerevisiae metabolic network and their lethality is the subject of this work. We represent the topology of the network by a directed graph: nodes are metabolites and arcs represent the reactions; a mutation corresponds to the removal of all the arcs referring to the deleted enzyme. Using publicly available knock-out data, we show that lethality corresponds to the lack of alternative paths in the perturbed network linking the nodes affected by the enzyme deletion. Such feature is at the basis of the recently recognized importance of ÔmarginalÕ arcs of metabolic networks.

    View PDFchevron_right
    Exploring and challenging the network of angiogenesis
    Raúl Montañez

    Scientific reports, 2011

    Angiogenesis is one of the hallmarks of cancer and, as such, one of the alternative general targets for anticancer therapy. Since angiogenesis is a complex process involving a high number of interconnected components, a network approach would be a convenient systemic way to analyse responses to directed drug attacks. Herein we show that, although the angiogenic network is easily broken by short combinations of directed attacks, it still remains essentially functional by keeping the global patterns and local efficiency essentially unaltered after these attacks. This is a clear sign of its high robustness and resilience and stresses the need of directed, combined attacks for an effective blockade of the process. The results of this theoretical study could be relevant for the design of new antiangiogenic therapies and the selection of their targets.

    View PDFchevron_right
    Failure mechanisms of load-sharing complex systems
    Vitali Volovoi

    Physical Review E, 2014

    We investigate the failure mechanisms of load sharing complex systems. The system is composed of multiple nodes or components whose failures are determined based on the interaction of their respective strengths and loads (or capacity and demand respectively) as well as the ability of a component to share its load with its neighbors when needed. We focus on two distinct mechanisms to model the interaction between components' strengths and loads. The failure mechanisms of these two models demonstrate temporal scaling phenomena, phase transitions and multiple dis-* In the last decade a significant body of research has accumulated in the study of complex systems, their structure and dynamics [1]. Static robustness in terms of node removals have been explored in [2]. But most real networks undergo dynamic failures where the failure of a single or multiple nodes might trigger cascades of failure through the network. Dynamical redistribution of flow have been considered in different real world networks: power grids [3], air transportation networks [4], and communication networks [5]. Many physical systems fail as their capacity or strength degrades over time under constant load or load increases over time as strength remains fixed. For example, loss of strength phenomena is observed in stress-rupture or creep rupture [6], tire wear, and level of fluid in a hydraulic system [7]; whereas load buildup is commonly considered in fiber-bundle models on complex systems [8, 9]. Failure occurs when component load is greater than its strength. Component failure due to overloading is a serious threat in networks: a single component failure and its subsequent load redistribution can trigger cascades of failures through the network, ultimately bringing down the entire system [3, 8]. On the other hand, many communication and transportation systems exhibit congestion phenomena as data or customer traffic density increases beyond certain thresholds. Congestion or jamming phenomena for critical values of traffic density has been demonstrated in models of transportation [10] and communication networks [11-13]. Traffic flow models for air transportation systems have been explored in [4, 14, 15]. Congestion in one part of the network has the effect of rerouting traffic to other parts of the network resulting in slowing down or clogging traffic in the entire system. In this work we explore two different models of interaction between component strength and load to understand the failure mechanisms of complex systems. The first one is a Loss of Strength (LOS) model where components loose strength over time following prescribed

    View PDFchevron_right
    Attack Robustness and Centrality of Complex Networks
    Bala Sundaram

    PLoS ONE, 2013

    Many complex systems can be described by networks, in which the constituent components are represented by vertices and the connections between the components are represented by edges between the corresponding vertices. A fundamental issue concerning complex networked systems is the robustness of the overall system to the failure of its constituent parts. Since the degree to which a networked system continues to function, as its component parts are degraded, typically depends on the integrity of the underlying network, the question of system robustness can be addressed by analyzing how the network structure changes as vertices are removed. Previous work has considered how the structure of complex networks change as vertices are removed uniformly at random, in decreasing order of their degree, or in decreasing order of their betweenness centrality. Here we extend these studies by investigating the effect on network structure of targeting vertices for removal based on a wider range of non-local measures of potential importance than simply degree or betweenness. We consider the effect of such targeted vertex removal on model networks with different degree distributions, clustering coefficients and assortativity coefficients, and for a variety of empirical networks.

    View PDFchevron_right
    Structural Tendencies — Effects of Adaptive Evolution of Complex (Chaotic) Systems
    Andrzej Gecow

    International Journal of Modern Physics C

    We describe systems using Kauffman and similar networks. They are directed functioning networks consisting of finite number of nodes with finite number of discrete states evaluated in synchronous mode of discrete time. In this paper we introduce the notion and phenomenon of "structural tendencies". Along the way we expand Kauffman networks, which were a synonym of Boolean networks, to more than two signal variants and we find a phenomenon during network growth which we interpret as "complexity threshold". For simulation we define a simplified algorithm which allows us to omit the problem of periodic attractors. We estimate that living and human designed systems are chaotic (in Kauffman sense) which can be named — complex. Such systems grow in adaptive evolution. These two simple assumptions lead to certain statistical effects, i.e., structural tendencies observed in classic biology but still not explained and not investigated on theoretical way. For example, term...

    View PDFchevron_right
    580 California St., Suite 400
    San Francisco, CA, 94104
    © 2025 Academia. All rights reserved