Data-Driven Intra-Autonomous Systems Graph Generator
Caio Vinicius Dadauto, Nelson L. S. da Fonseca, Ricardo da Silva Torres · IEEE Transactions on Network and Service Management · 2024
Accurate modeling of realistic network topologies is essential for evaluating novel Internet solutions. Numerous investigations have used topologies generated by graph generators employing scale-free-based models. Although scale-free networks accurately encode node degree distribution, they overlook crucial graph properties, such as betweenness, clustering, and assortativity. The limitations of existing generators pose challenges for evaluating network mechanisms and protocols, such as routing. This paper introduces a novel deep learning-based generator of synthetic graphs representing intra-autonomous on the Internet, named Deep-Generative Graphs for the Internet (DGGI). It also presents a massive new dataset of real intra-AS graphs extracted from the project Internet Topology Data Kit (ITDK), called Internet Graphs (IGraphs). DGGI creates synthetic graphs that accurately reproduce the properties of centrality, clustering, assortativity, and node degree. DGGI overperforms existing Internet topology generators. On average, DGGI improves the Maximum Mean Discrepancy (MMD) metric by 84.4%, 95.1%, 97.9%, and 94.7% for assortativity, betweenness, clustering, and node degree, respectively.