Measurement, analysis, and modeling of multi-layered communication networks

Empirical analysis and mathematical modeling have been the foundation for a great deal of network research and have resulted in significant improvements to Internet systems, protocols and practices. Recent progress applying a constantly expanding set of sophisticated statistical tools and mathematical techniques suggests the emergence of a new type of measurement-based networking research that could benefit from a more principled approach for representing, analyzing, and modeling data collected from tomorrow's Internet or other multi-layered communication networks. To this end, we discuss a structured approach to Internet data representation and modeling that provides a framework for systematically applying tools from statistical analysis, signal processing or machine learning, and techniques from mathematical modeling. By respecting the multi-scale (in time, space, and layers) nature of the data, the proposed approach offers critical insights into a number of challenging network research problems. We illustrate the approach with two concrete examples and compare it with alternatives that are based on recent advances in the theory of complex networks