Propriedades dinâmicas em redes de Kleinberg

• Ano de defesa: 2015
• Autor(a) principal: Silva, Samuel Morais da
• Orientador(a): Araújo, Ascânio Dias
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Redes de Kleinberg; Sistemas complexos; Propriedades de transporte; Criticalidade; Modelo small world; Física computacional
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/13864

A great number of systems defined as complex consist of interconnected parts or individual components performing a network or graph. Communication between the parts is essential for their existence so that it is necessary a better understanding of their ability to communicate depending on the amount of information that transits. The dynamics of package transport in these systems and the emergence of congestion are problems of high scientific and economic interest. In this work we investigate the dynamical properties of transport of packages (informations) between sources and previously defined destinations, considering different models of spatially embbeded networks such as lattice and Kleinberg. More precisely, we study a second-order continuous phase transition from a phase of free transport to a congestion phase, when the packages are accumulated in certain regions of the network. By means of a Finite Size Scaling, we describe this phase transition characterizing its critical exponents. For 1D and 2D lattice networks, we observe that the critical parameter $p_c$ scales with exponents approximately $-1$ and $-0.5$ with respect to the system size. In the case of Kleinberg newtorks where shortcuts between two nodes $i$ and $j$ are added to the network according to a probability distibution given by $P(r_ {ij}) sim r_{ij}^{-alpha}$, we show that the best scenario occurs when $alpha = d$, where $d$ is the dimention of the topology structure. In this regime, package traffic were shown to be more resilient to the increase of number of packages in the network. The confirmation of our result is obtained not only from direct measure of order parameter, that is, the ratio between undelivered and generated packets, but is also supported by our analysis of finite size.