Skip to content

About

Reproducible candidate-link sentinel for safe OSPF reconfiguration under network-digital-twin drift, with Containerlab/FRRouting experiments and holdout validation.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

When the Network Twin Drifts: A Candidate-Link Sentinel for Safe OSPF Reconfiguration

Reproducibility workspace for the sole-authored manuscript by Mohammed Alabdullah.

  • ORCID: 0009-0001-0826-4380
  • Correspondence: m.w@sudo-gate.com
  • Affiliation: Department of Network Engineering, School of Computer Engineering, Iran University of Science and Technology (IUST), Tehran, Iran
  • Release: v1.0.0

Submission-ready manuscript

The canonical manuscript source in paper/main.tex and paper/sections/ is synchronized with the finalized IEEE Transactions on Network and Service Management (TNSM) submission version after the final independent review pass.

  • Initial submission target: 9 pages in IEEE two-column format.
  • paper/manuscript.pdf is rebuilt from the canonical source by GitHub Actions and is required to remain 9 pages.
  • The finalized manuscript uses SLO terminology consistently, qualifies the holdout reporting, strengthens the transferable-design-pattern framing, uses the corrected February 2026 IETF draft revisions, and contains the minimized truthful AI-use acknowledgment.
  • Experimental measurements, frozen-policy parameters, validation/holdout outcomes, and scientific results are unchanged.

Research question

The study asks whether a network digital twin can approve an OSPF link-cost change even when hidden operational latency on the activated link makes the resulting path violate a 50 ms service-level objective.

Main empirical result

The frozen candidate-link sentinel accepted 110 of 186 validation candidates, including 109 safe candidates, with one unsafe deployment. On the 33-sample holdout, it accepted all 19 safe candidates and rejected all 14 unsafe candidates. The holdout result is empirical; the one-sided score covered only 20 of 33 realized RTTs and does not establish nominal conformal coverage under distribution shift.

Repository layout

  • paper/: finalized manuscript source, generated PDF, references, and figures.
  • paper/sections/: section-level LaTeX source used by the canonical manuscript.
  • lab/: Containerlab/FRRouting configurations, scripts, schedules, and frozen result records.
  • processed-data/: analysis-ready Phase 3--5 datasets.
  • results/: Phase 3 summaries.
  • figures/: manuscript figures.
  • docs/: design, provenance, correction, and traceability records.
  • tools/verify_release.py: read-only verification.

Integrity constraints

Do not rerun, overwrite, or tune against the Phase 5 holdout. Do not modify lab/phase5/results/frozen_policy.json. See lab/phase5/results/frozen_hash_resolution.json.

Quick verification

python3 tools/verify_release.py

The verifier writes no experimental result files.

Public release artifacts

  • Reproducibility artifact: ndt-ospf-artifact-v1.0.0.zip
  • License: MIT for software and artifact files

Archived release

About

Reproducible candidate-link sentinel for safe OSPF reconfiguration under network-digital-twin drift, with Containerlab/FRRouting experiments and holdout validation.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages