Information Theory Perspective on Network Robustness
Autor: | Panos M. Pardalos, Tiago A. Schieber, Alejandro C. Frery, Martín Gómez Ravetti, Osvaldo A. Rosso, Laura C. Carpi |
---|---|
Jazyk: | angličtina |
Rok vydání: | 2014 |
Předmět: |
FOS: Computer and information sciences
Physics - Physics and Society Ciencias Físicas FOS: Physical sciences General Physics and Astronomy Physics and Society (physics.soc-ph) Network theory Information theory computer.software_genre Network topology 01 natural sciences 010305 fluids & plasmas Network simulation Robustness (computer science) 0103 physical sciences NETWORK ROBUSTNESS 010306 general physics Condensed Matter - Statistical Mechanics COMPLEX NETWORKS Social and Information Networks (cs.SI) Physics Statistical Mechanics (cond-mat.stat-mech) Interdependent networks Computer Science - Social and Information Networks INFORMATION THEORY Complex network Network formation Astronomía Data mining computer CIENCIAS NATURALES Y EXACTAS |
Popis: | A crucial challenge in network theory is the study of the robustness of a network after facing a sequence of failures. In this work, we propose a dynamical definition of network's robustness based on Information Theory, that considers measurements of the structural changes caused by failures of the network's components. Failures are defined here, as a temporal process defined in a sequence. The robustness of the network is then evaluated by measuring dissimilarities between topologies after each time step of the sequence, providing a dynamical information about the topological damage. We thoroughly analyze the efficiency of the method in capturing small perturbations by considering both, the degree and distance distributions. We found the network's distance distribution more consistent in capturing network structural deviations, as better reflects the consequences of the failures. Theoretical examples and real networks are used to study the performance of this methodology. 5 pages, 2 figures, submitted |
Databáze: | OpenAIRE |
Externí odkaz: |