Reproduce “Super-resolution on network telemetry time series” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Fengchen Gong, Divya Raghunathan, Aarti Gupta, Maria Apostolaki

Zoom2Net is a software framework that infers fine-grained network monitoring data from multiple correlated coarse-grained time series, enhancing existing infrastructures without requiring hardware upgrades. It improves accuracy and performance across diverse network management tasks by integrating domain knowledge to address data ambiguities and ensure consistency with operational principles.

Key insights

  • Ambiguity inherent in single coarse-grained measurements can be significantly reduced by leveraging correlations among multiple distinct, yet related, coarse-grained telemetry signals.
  • For network telemetry, the goal should be to generate functionally equivalent fine-grained outputs that are plausible and consistent with measurements, rather than precise point-to-point reconstructions, to mitigate issues like 'coarse-grained collisions'.
  • Explicitly incorporating domain knowledge, such as physical laws and operational principles, through loss function regularization and post-processing, is crucial for ensuring the imputed data's accuracy and adherence to real-world network behavior.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.