Abstract

The study of a majority of complex networks started initially from the desire to understand various natural systems, ranging from communication networks to ecologic chains or networks (Costa et al. 2003). This chapter lists the significant results in the research of complex networks, focusing on the large-scale traffic networks. Beyond the fields from where data was extracted, special emphasis will be placed on the three robust indicators for network topology: average path length, clustering coefficient, and degree distribution. To model a distributed network environment like the Internet, it is necessary to integrate the data collected from multiple points in a network in order to get a complete picture of net-work-wide view of the traffic. Knowledge of dynamic characteristics is essential to network management (e.g., detection of failures/congestion, provisioning, and traffic engineering like QoS routing or server selections). However, given the enormity of the huge scale and restrictions due to access rights, it is expensive (sometime impossible) to measure such characteristics directly. To solve this problem, a host of methods and tools to infer the unobservable network performance characteristics are used in large-scale networking environment. A model where inference based on self-similarity and fractal behavior is best represented is the scale-free network, and so this model will be largely discussed in this chapter.