Performance Rescaling of Complex Networks
Abstract: Recent progress in network topology modeling [1], [2] has shown that it is possible to create smaller-scale replicas of large complex networks, like the Internet, while simultaneously preserving several important topological properties. However, the constructed replicas do not include notions of capacities and latencies, and the fundamental question of whether smaller networks can reproduce the performance of larger networks remains unanswered. We address this question in this letter, and show that it is possible to predict the performance of larger networks from smaller replicas, as long as the right link capacities and propagation delays are assigned to the replica's links. Our procedure is inspired by techniques introduced in [2] and combines a time-downscaling argument from [3]. We show that significant computational savings can be achieved when simulating smaller-scale replicas with TCP and UDP traffic, with simulation times being reduced by up to two orders of magnitude.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.