Simone Angelini

Fondazione "Ugo Bordoni"

Unclaimed · AI-structured profileOpenAlex

Learning-to-Optimize for AI-native 6G: GNN surrogates supervised by MILP for NR-V2X wireless networks

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

No output data is available yet, so this axis is unscored. It would measure publication volume and productivity over time, such as total paper count and output trajectory; the current snapshot contains only 2 cataloged arXiv papers.

Impact—

Impact data is missing, so no score is estimated. This axis would measure citation-based influence — citation counts, h-index and citation trajectory — none of which were provided for this researcher in the current snapshot.

Focus0/100

Focus scores 0, the lowest concentration reading available. The raw field distribution splits output exactly evenly — 0.5 to 'AI-native 6G network architecture / network optimization' and 0.5 to 'wireless networking / AI for communications' — across only two cataloged papers. With such a small sample, the zero reflects an even two-field split rather than confirmed scattered interests; the two fields are closely related, so this reading should be treated as a small-sample artifact until more papers accumulate.

Collaboration32/100

The collaboration score is 32, derived from a small pool of 4 distinct coauthors across the cataloged output. All four frequent collaborators — Giambattista Amati, Federica Mangiatordi, Emiliano Pallotti and Pierpaolo Salvo — point to a stable, institution-bound group around Fondazione Ugo Bordoni. Angelini sits in the last author position on both 2026 papers without being corresponding author, a pattern consistent with a supervisory role inside a closed team rather than wide external networking; the data shows no cross-institution coauthorship.

Activity—

Activity data is insufficient for scoring. This axis would measure publishing rhythm and recency, such as papers per recent year and gaps in output; with only two 2026 papers cataloged, no temporal pattern can yet be established.

Reproducibility38/100

Reproducibility scores 38 (reproMean), averaged over 2 scored papers (reproPapers). Evolve rates the NR-V2X relay-optimisation paper medium: a credible MILP-supervised GINE simulation study claiming ~100x speed-up, but with no baselines or error bars. The architectural-layer paper rates low because its results are imported from the authors' prior papers rather than independently re-derived. The mean therefore reflects early, partially documented results that are not yet straightforwardly reproducible by third parties.

Research summary

Simone Angelini is a researcher at Fondazione Ugo Bordoni whose cataloged work centres on learning-to-optimize (LTO) methods for AI-native wireless networks. Two 2026 arXiv papers anchor this profile. In 'AI-Driven Real-Time Relay Optimisation in Smart Urban NR-V2X Networks via Learning-to-Optimise Graph Neural Networks', a GINE graph neural network supervised by MILP solutions is reported to deliver near-optimal real-time relay selection with roughly a 100x speed-up; Evolve's review finds the simulation study credible but notes the absence of baselines and error bars. A companion paper, 'Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks', argues that LTO deserves status as a first-class architectural layer of AI-native 6G networks, illustrated through an NR-V2X GNN case whose results are imported from the authors' prior papers. Angelini holds the last author position on both papers and is not the corresponding author on either, a pattern conventionally associated with a senior or supervisory role in a compact four-person coauthor group — Giambattista Amati, Federica Mangiatordi, Emiliano Pallotti and Pierpaolo Salvo. Bibliometric coverage remains narrow: radar scores exist only for collaboration (32), focus (0) and reproducibility (38), while output, impact and activity lack sufficient data; reproducibility readiness across the two papers is rated low and medium.

Collaboration pattern

Collaboration is compact and closed: 4 distinct coauthors, all recurring — Giambattista Amati, Federica Mangiatordi, Emiliano Pallotti and Pierpaolo Salvo — around Fondazione Ugo Bordoni. Angelini takes the last author position on both 2026 papers without corresponding authorship, a pattern consistent with a supervisory or senior-contributor role. No external or cross-institution coauthorship appears in the cataloged data, and the collaboration score of 32 mirrors this small, stable team.

Research keywords

Learning-to-OptimizeAI-native networks6G architectureNR-V2Xrelay selectiongraph neural networksGINEMILPsurrogate modelsAI for communications

Highlights

Near-optimal real-time NR-V2X relay selection via MILP-supervised GNNs
A GINE graph neural network trained on MILP solutions is reported to select relays in smart urban NR-V2X networks in real time with near-optimal quality and a ~100x speed-up; Evolve's review deems the simulation study credible but flags missing baselines and error bars.
Learning-to-Optimize proposed as an architectural layer of AI-native networks
The paper argues that learning-to-optimize should be treated as a first-class architectural layer in AI-native 6G network design, demonstrated through an NR-V2X GNN case study; Evolve notes the results are imported from the authors' prior papers rather than newly derived.
A recurring MILP-to-GNN methodological pipeline
Both cataloged papers share one recipe — exact MILP optimization as supervision, graph neural network surrogates as fast inference, and NR-V2X as the testbed — indicating a deliberate, coherent research programme rather than isolated outputs.

Timeline

2026Publishes 'AI-Driven Real-Time Relay Optimisation in Smart Urban NR-V2X Networks via Learning-to-Optimise Graph Neural Networks' on arXiv, in the last author position.
2026Publishes 'Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks', framing LTO as an architectural layer of AI-native 6G networks.
2026Both outputs land within a single year, splitting evenly between AI-native 6G architecture and wireless AI networking, alongside frequent coauthors Giambattista Amati, Federica Mangiatordi, Emiliano Pallotti and Pierpaolo Salvo.

Papers in the Evolve catalog

Frequent co-authors

Giambattista Amati×2 Federica Mangiatordi×2 Emiliano Pallotti×2 Pierpaolo Salvo×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Radar scores and this profile rest on a limited bibliometric snapshot covering only 2 cataloged arXiv papers.
Output, impact and activity axes have no underlying data: publication counts, citation metrics and per-year activity were not provided.
No ORCID or country is available; the profile is name-resolved from public bibliographic data, and 'Simone Angelini' is a fairly common Italian name, so identity should be verified before reuse.
The reproducibility mean (38) is computed from only 2 scored papers, and readiness ratings come from Evolve's reviews of arXiv preprints, not peer-reviewed versions.