ABSTRACT
In the last twenty years, the Internet has experienced tremendous growth, which has transformed it from a small-scale research network to the largest and most complex artificially deployed system. The Internet possesses similar
of Complex
structural properties to the ones characterizing many other complex systems pervading science: a plethora of often heterogeneous subsystems (sources and routers) performing complex functions, interconnected by heterogeneous links (wired, wireless, satellite links) often incorporating complex dynamics themselves. There are several factors contributing to the immense complexity of the system: the large scale and size as a result of its exponential growth; the fragmented nature of the underlying infrastructure; the hierarchical organization; the extreme heterogeneity as a result of the diverse network technologies and communication services that are accommodated; the distributed management of the available resources; the complex structures that arise in the implementation of the various functionalities of the layered protocols, and so on [1]. Many of the complex network functions that drive the current Internet have been developed using engineering intuition, heuristics, and ad hoc nonlinear techniques, with the objective of making the system resilient to failures and robust to changing environments. The problem with this approach is that very little is known about why these methods work and very little explanation can be given when they fail. Given the lack of a coherent and unified theory of complex systems, these methods do not have analytically proven performance properties and can thus prove to be ineffective as the system evolves over time. When such vulnerabilities do show up, designers usually resort to even more complex network functions to solve the problem, thus contributing to a spiral of increasing complexity [1]. These observations highlight the necessity to develop a new theoretical framework to help explain the complex and unpredictable behaviors of the Internet and offer alternative network protocols that are provably effective and robust. Such a framework can serve as a starting point to develop a unified theory for complex systems, useful in explaining how the interaction between the individual components of such systems allows the emergence of a global behavior that would not be anticipated from the behavior of components in isolation.
