• Introduces traversal concept generalizing paths with gap transitions for incomplete networks. • Treats planned connections as first-class transitions, enabling realistic infrastructure reasoning. • Parametric framework uses multi-dimensional criteria to prune traversal search efficiently. • Scales to large graphs via candidate filtering, avoiding combinatorial explosion. • Evaluated on datacenter circuit design and optical telecom routing scenarios. • Demonstrates conditional feasibility, non-scalarizable trade-offs, and policy calibration beyond classical methods.
Article Summaries:
- Summary
Researchers introduce “traversal,” a generalized path concept that blends existing network links with planned, missing connections-representing feasible but not yet built links. The new parametric framework treats these planned transitions as first‑class elements, enabling scalable analysis of large graphs through efficient candidate filtering. It evaluates traversals using multi‑dimensional cost criteria, allowing the algorithm to prune infeasible routes early. Applied to datacenter circuit design and optical telecommunication routing, the method demonstrates conditional feasibility, uncovers trade‑offs that cannot be captured by single‑metric optimization, and offers policy‑calibration tools beyond traditional path‑finding models.
Sources: