Sample-Efficient Misconfiguration Classification for Network Resilience in Wireless Communications

Chenhan Zhang; Massimo Piccardi; Qiwen Jiang; Raymond Owen; Vijaya Durga Chemalamarri; Wei Ni; Xin Hao

arxiv: 2605.19303 · v1 · pith:SFV5JWUSnew · submitted 2026-05-19 · 💻 cs.NI

Sample-Efficient Misconfiguration Classification for Network Resilience in Wireless Communications

Xin Hao , Chenhan Zhang , Massimo Piccardi , Vijaya Durga Chemalamarri , Qiwen Jiang , Wei Ni , Raymond Owen This is my paper

Pith reviewed 2026-05-20 03:00 UTC · model grok-4.3

classification 💻 cs.NI

keywords misconfiguration classificationwireless networksgraph attention networksnetwork resiliencesample efficient learningheterogeneous graphsprotocol configurationdynamic topologies

0 comments

The pith

An edge-type-aware graph attention network classifies protocol misconfigurations in wireless networks using only half the training data of prior methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The authors model wireless communication networks as heterogeneous graphs to detect protocol misconfigurations that cause outages due to mismatches with dynamic topologies. They propose EtaGATv2, which incorporates edge-type-aware attention and dynamic attention mechanisms to capture how symptoms propagate unevenly and to handle different routing protocols' behaviors. This allows the model to reach state-of-the-art performance while using just 50% of the training samples across various real-world topologies. The approach is useful for networks where collecting labeled data on misconfigurations is difficult because failures are rare.

Core claim

By formulating protocol misconfiguration classification as a graph-based learning task and solving it with EtaGATv2—an edge-type-aware graph attention network with dynamic attention—the method addresses non-uniform symptom propagation and extracts protocol-specific features from heterogeneous routing protocols, achieving state-of-the-art results with 50% of the training samples in diverse topologies.

What carries the argument

EtaGATv2: an edge-type-aware graph attention network with dynamic attention that uses distinct transformations for different edge types to model protocol-specific message passing and critical network paths for diagnosis.

If this is right

Networks with dynamic topologies can diagnose misconfigurations more reliably even when failure data is limited.
Protocol-specific behaviors are better distinguished without needing full datasets for each scenario.
The method supports real-time resilience improvements in wireless communications by reducing data requirements.
Similar graph modeling could apply to other network management problems involving heterogeneous elements.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the graph representation holds, applying the same attention mechanism to other wireless management tasks like resource allocation might yield similar efficiency gains.
Future work could test whether the sample efficiency persists when topologies change more rapidly than in the evaluated cases.
Connections to graph neural networks in other domains suggest this could generalize to sensor networks or ad-hoc communication systems.

Load-bearing premise

That wireless networks can be accurately represented as heterogeneous graphs where edge types correspond to different protocols and attention mechanisms capture the real propagation of misconfiguration symptoms.

What would settle it

Observing that in a live wireless deployment with unseen protocol interactions, the EtaGATv2 model trained on 50% data underperforms compared to a baseline trained on the full dataset.

Figures

Figures reproduced from arXiv: 2605.19303 by Chenhan Zhang, Massimo Piccardi, Qiwen Jiang, Raymond Owen, Vijaya Durga Chemalamarri, Wei Ni, Xin Hao.

**Figure 2.** Figure 2: Training results of multiple independent runs using the baseline dataset [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: Validation across topology variations. From left to right are the baseline, larger-scale, and real-world datasets that are summarized in Table I, with 100 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

read the original abstract

As modern wireless communication networks grow increasingly complex, network outages driven by the inconsistency between dynamic topologies and protocol configurations have become a critical concern. To solve this issue, we mathematically formulate a protocol misconfiguration classification problem as a graph-based learning task and solve it with our proposed EtaGATv2 algorithm, an edge-type-aware graph attention network with dynamic attention. EtaGATv2 addresses two critical challenges: i) it captures non-uniform symptom propagation for protocol misconfiguration classification tasks, where certain network paths and nodes become critical for diagnosis, and ii) it extracts protocol-specific features from heterogeneous routing protocols with distinct message-passing behaviors by utilizing edge-type-aware transformations. Experiments across diverse and real-world topologies demonstrate that EtaGATv2 reaches state-of-the-art performance with 50% of the training samples, making it particularly suitable for networks with dynamic topologies and limited negative-labeled data.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

EtaGATv2 adds edge-type-aware dynamic attention to graph networks for protocol misconfiguration detection in wireless setups and claims solid results at half the usual training data.

read the letter

The core contribution is EtaGATv2, which modifies graph attention to handle edge types and dynamic focus when wireless networks are cast as heterogeneous graphs. This targets two issues: uneven spread of misconfiguration symptoms across paths and nodes, plus pulling out features tied to different routing protocols and their message patterns. The work frames the whole thing as a classification task and reports state-of-the-art numbers on real topologies using only 50 percent of the training samples. That sample-efficiency angle is the practical hook, since dynamic wireless environments often lack plenty of labeled fault data. The experiments on diverse topologies give it some grounding beyond pure synthetic cases. The modeling choice to use edge-type transformations looks like a reasonable extension of standard GAT ideas for this domain. The main soft spot is that the abstract states the performance gains without showing the actual tables, baseline details, or statistical tests, so the central claim is hard to assess from the given text. The stress-test point about graph construction is worth checking: if the node and edge features do not line up with how symptoms actually move in deployed protocols, the reported gains could be tied to the specific graph setup rather than the method itself. No equations or derivations appear that would let someone reproduce the result from first principles. This paper is for people working on network management tools or graph methods for communications infrastructure. A reader who needs sample-efficient classifiers for fault detection in wireless systems could get value from the targeted adaptations. It is coherent enough on its own terms to deserve a serious referee who can look at the full experiments and graph-construction choices.

Referee Report

2 major / 1 minor

Summary. The manuscript formulates protocol misconfiguration classification in wireless networks as a graph-based learning task and proposes EtaGATv2, an edge-type-aware graph attention network with dynamic attention. The method is claimed to address non-uniform symptom propagation via critical paths/nodes and to extract protocol-specific features from heterogeneous routing protocols using edge-type-aware transformations. Experiments on diverse and real-world topologies are asserted to achieve state-of-the-art performance with only 50% of the training samples, positioning the approach as suitable for dynamic topologies with limited negative-labeled data.

Significance. If the performance claims can be substantiated with detailed, reproducible experimental evidence including baselines and statistical analysis, the work could provide a useful sample-efficient method for misconfiguration detection in complex wireless systems. The heterogeneous graph modeling of protocol interactions represents a reasonable direction for capturing non-uniform behaviors, though its practical impact hinges on validation that the constructed graphs and edge types correspond to real protocol dynamics rather than synthetic artifacts.

major comments (2)

[Abstract] Abstract: the central claim that EtaGATv2 reaches state-of-the-art performance with 50% training samples on real topologies lacks any quantitative tables, baseline comparisons, or statistical details, leaving the primary experimental result without verifiable support in the manuscript.
The load-bearing modeling assumption that edge-type-aware attention on the constructed heterogeneous graph accurately captures non-uniform symptom propagation and protocol-specific message-passing behaviors is stated but not explicitly validated against real deployment traces or protocol logs; without such grounding, the reported gains may not generalize beyond the synthetic graph construction.

minor comments (1)

The manuscript would benefit from explicit definitions of the node/edge features and edge-type assignments used in the graph construction, as these choices directly affect whether the attention mechanism reflects actual wireless protocol interactions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We appreciate the emphasis on ensuring experimental claims are well-supported and that modeling assumptions are properly grounded. We address each major comment below and indicate the revisions we will make.

read point-by-point responses

Referee: [Abstract] Abstract: the central claim that EtaGATv2 reaches state-of-the-art performance with 50% training samples on real topologies lacks any quantitative tables, baseline comparisons, or statistical details, leaving the primary experimental result without verifiable support in the manuscript.

Authors: We agree that the abstract, due to space constraints, cannot contain tables or detailed statistics. The full manuscript includes Section 4 with quantitative results: tables reporting F1-scores, precision, and recall for EtaGATv2 against baselines (GAT, GCN, GraphSAGE, and protocol-specific ML models) at 50% training samples across multiple real-world topologies from Topology Zoo and Rocketfuel. We have added statistical significance testing (paired t-tests with p-values) in the revised experiments section. To improve clarity, we will revise the abstract to include one or two key quantitative highlights, such as the relative improvement over baselines. revision: partial
Referee: The load-bearing modeling assumption that edge-type-aware attention on the constructed heterogeneous graph accurately captures non-uniform symptom propagation and protocol-specific message-passing behaviors is stated but not explicitly validated against real deployment traces or protocol logs; without such grounding, the reported gains may not generalize beyond the synthetic graph construction.

Authors: We acknowledge that direct validation against live deployment traces or protocol logs is not present in the current work. The heterogeneous graphs are constructed using real-world topologies and edge types derived from standard protocol specifications (e.g., OSPF link-state advertisements and BGP update messages per RFCs). Ablation experiments in the manuscript show that edge-type-aware transformations contribute measurably to performance, supporting the modeling choice for capturing non-uniform propagation. We will add a dedicated limitations paragraph discussing the reliance on simulated protocol behaviors and the need for future validation with operational logs. This does not alter the reported results on the evaluated topologies but clarifies the scope. revision: partial

Circularity Check

0 steps flagged

No significant circularity; claims rest on experimental validation

full rationale

The paper formulates the misconfiguration classification problem as a graph-based learning task and introduces EtaGATv2 to address non-uniform symptom propagation and protocol-specific features via edge-type-aware attention. Performance claims (SOTA at 50% training samples) are presented strictly as outcomes of experiments on diverse and real-world topologies. No equations, derivations, or self-citations are shown that reduce these results or the model's capabilities to fitted parameters or inputs defined by construction within the paper itself. The derivation chain is self-contained as an algorithmic proposal plus empirical evaluation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Review is based solely on the abstract; therefore the ledger reflects only high-level modeling choices stated or implied in the text. The central claim rests on the assumption that the graph formulation and attention mechanism correctly represent real network behavior.

axioms (1)

domain assumption Wireless networks with dynamic topologies and heterogeneous routing protocols can be represented as graphs where edge types correspond to distinct protocol behaviors.
Invoked when the problem is mathematically formulated as a graph-based learning task.

invented entities (1)

EtaGATv2 no independent evidence
purpose: Edge-type-aware graph attention network with dynamic attention to capture non-uniform symptom propagation and protocol-specific features.
Introduced as the solution addressing the two critical challenges listed in the abstract.

pith-pipeline@v0.9.0 · 5701 in / 1326 out tokens · 64859 ms · 2026-05-20T03:00:47.464788+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

EtaGATv2 addresses two critical challenges: i) it captures non-uniform symptom propagation... by utilizing edge-type-aware transformations.
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Experiments... EtaGATv2 reaches state-of-the-art performance with 50% of the training samples.

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages · 1 internal anchor

[1]

Resilience and failure analysis in next-generation communication networks: A contemporary survey,

S. Bi, X. Yuan, S. Hu, K. Li, W. Ni, E. Hossain, and X. Wang, “Resilience and failure analysis in next-generation communication networks: A contemporary survey,”IEEE Trans. Network Sci. Eng., vol. 13, pp. 2793– 2821, 2026

work page 2026
[2]

Enhancing Network Resilience via Graph-Based Anomaly Detection in Sovereign Functions

X. Hao, W. Ni, C. Zhang, M. Piccardi, and R. Owen, “Enhancing network resilience via graph-based anomaly detection in sovereign functions,” 2026. [Online]. Available: arXiv:2605.17716

work page internal anchor Pith review Pith/arXiv arXiv 2026
[3]

Resilience and survivability in communication networks: Strategies, principles, and survey of disciplines,

J. P. Sterbenzet al., “Resilience and survivability in communication networks: Strategies, principles, and survey of disciplines,”Computer Networks, vol. 54, no. 8, pp. 1245–1265, 2010

work page 2010
[4]

Automatic performance assess- ment — step toward autonomous mobile network management systems,

M. Panek, I. Jabło ´nski, and M. Wo´zniak, “Automatic performance assess- ment — step toward autonomous mobile network management systems,” IEEE Commun. Mag., vol. 63, no. 4, pp. 73–79, 2025

work page 2025
[5]

Reliably route IoT packets in software defined mmwave mesh networks,

T.-Q. Bai, C.-Y . Huang, and Y .-K. Lee, “Reliably route IoT packets in software defined mmwave mesh networks,”IEEE Networking Lett., vol. 5, no. 1, pp. 50–54, 2023

work page 2023
[6]

Secure deep reinforcement learning for dynamic resource allocation in wireless MEC networks,

X. Hao, P. L. Yeoh, C. She, B. Vucetic, and Y . Li, “Secure deep reinforcement learning for dynamic resource allocation in wireless MEC networks,”IEEE Trans. Commun., vol. 72, no. 3, pp. 1414–1427, 2024

work page 2024
[7]

Enhancing the resilience of communication networks,

OECD, “Enhancing the resilience of communication networks,” Organi- sation for Economic Co-operation and Development, Tech. Rep., 2025

work page 2025
[8]

Failures and resilience in the IP era: Navigating the fragility of modern telecommunications networks–the sovereign func- tions,

R. Owenet al., “Failures and resilience in the IP era: Navigating the fragility of modern telecommunications networks–the sovereign func- tions,”IEEE Access, vol. 13, pp. 155 759–155 777, 2025

work page 2025
[9]

Cisco Systems,Cisco Crosswork Network Controller: Use Templates to Configure Similar Devices, 2024

work page 2024
[10]

Cisco Meraki,Managing Multiple Networks with Configuration Tem- plates, Cisco Meraki, 2020

work page 2020
[11]

Large-scale network router-level topology construction based on knowledge inference,

Y . Zhou, G. Fei, J. Ye, X. Zhai, and G. Hu, “Large-scale network router-level topology construction based on knowledge inference,”IEEE Networking Lett., vol. 7, no. 3, pp. 210–214, 2025

work page 2025
[12]

RouteNet: Leveraging graph neural networks for network modeling and optimization in SDN,

K. Rusek, J. Su ´arez-Varela, P. Almasan, P. Barlet-Ros, and A. Cabellos- Aparicio, “RouteNet: Leveraging graph neural networks for network modeling and optimization in SDN,”IEEE J. Sel. Areas Commun., vol. 38, no. 10, pp. 2260–2270, 2020

work page 2020
[13]

Learning to configure computer networks with neural algorithmic reasoning,

L. Beurer-Kellner, M. Vechev, L. Vanbever, and P. Veli ˇckovi´c, “Learning to configure computer networks with neural algorithmic reasoning,” in Proceedings of Int. Conf. Neural Inf. Process. Syst. (NeurIPS), 2022

work page 2022
[14]

A novel UA V-assisted V ANET routing protocol for post-disaster emergency communications,

Z. Fan, M. Zhang, Y . Cao, Z. Liu, O. Kaiwartya, Y . Javed, and F. Bashir Hussain, “A novel UA V-assisted V ANET routing protocol for post-disaster emergency communications,”IEEE Trans. Network Sci. Eng., vol. 13, pp. 4863–4882, 2026

work page 2026
[15]

Hybrid-task meta-learning: A GNN approach for scalable and transferable bandwidth allocation,

X. Hao, C. She, P. L. Yeoh, Y . Liu, B. Vucetic, and Y . Li, “Hybrid-task meta-learning: A GNN approach for scalable and transferable bandwidth allocation,”IEEE Trans. Wireless Commun., vol. 23, no. 12, pp. 19 820– 19 835, 2024

work page 2024
[16]

Graph neural network-based bandwidth allocation for secure wireless commu- nications,

X. Hao, P. L. Yeoh, Y . Liu, C. She, B. Vucetic, and Y . Li, “Graph neural network-based bandwidth allocation for secure wireless commu- nications,” inProceedings of IEEE Int. Conf. Commun. (ICC), 2023, pp. 332–337

work page 2023
[17]

Predictive precoder design for OTFS-enabled URLLC: A deep learning approach,

C. Liu, S. Li, W. Yuan, X. Liu, and D. W. K. Ng, “Predictive precoder design for OTFS-enabled URLLC: A deep learning approach,”IEEE J. Sel. Areas Commun., vol. 41, no. 7, pp. 2245–2260, Jul. 2023

work page 2023
[18]

Graph attention networks,

P. Veli ˇckovi´c, G. Cucurull, A. Casanova, A. Romero, P. Li `o, and Y . Bengio, “Graph attention networks,” inProceedings of Int. Conf. Learn. Representations (ICLR), 2018

work page 2018
[19]

How attentive are graph attention networks?

S. Brody, U. Alon, and E. Yahav, “How attentive are graph attention networks?” inProceedings of Int. Conf. Learn. Representations (ICLR), 2022

work page 2022
[20]

The Internet topology zoo,

S. Knight, H. X. Nguyen, N. Falkner, R. Bowden, and M. Roughan, “The Internet topology zoo,”IEEE J. Sel. Areas Commun., vol. 29, no. 9, pp. 1765–1775, 2011

work page 2011

[1] [1]

Resilience and failure analysis in next-generation communication networks: A contemporary survey,

S. Bi, X. Yuan, S. Hu, K. Li, W. Ni, E. Hossain, and X. Wang, “Resilience and failure analysis in next-generation communication networks: A contemporary survey,”IEEE Trans. Network Sci. Eng., vol. 13, pp. 2793– 2821, 2026

work page 2026

[2] [2]

Enhancing Network Resilience via Graph-Based Anomaly Detection in Sovereign Functions

X. Hao, W. Ni, C. Zhang, M. Piccardi, and R. Owen, “Enhancing network resilience via graph-based anomaly detection in sovereign functions,” 2026. [Online]. Available: arXiv:2605.17716

work page internal anchor Pith review Pith/arXiv arXiv 2026

[3] [3]

Resilience and survivability in communication networks: Strategies, principles, and survey of disciplines,

J. P. Sterbenzet al., “Resilience and survivability in communication networks: Strategies, principles, and survey of disciplines,”Computer Networks, vol. 54, no. 8, pp. 1245–1265, 2010

work page 2010

[4] [4]

Automatic performance assess- ment — step toward autonomous mobile network management systems,

M. Panek, I. Jabło ´nski, and M. Wo´zniak, “Automatic performance assess- ment — step toward autonomous mobile network management systems,” IEEE Commun. Mag., vol. 63, no. 4, pp. 73–79, 2025

work page 2025

[5] [5]

Reliably route IoT packets in software defined mmwave mesh networks,

T.-Q. Bai, C.-Y . Huang, and Y .-K. Lee, “Reliably route IoT packets in software defined mmwave mesh networks,”IEEE Networking Lett., vol. 5, no. 1, pp. 50–54, 2023

work page 2023

[6] [6]

Secure deep reinforcement learning for dynamic resource allocation in wireless MEC networks,

X. Hao, P. L. Yeoh, C. She, B. Vucetic, and Y . Li, “Secure deep reinforcement learning for dynamic resource allocation in wireless MEC networks,”IEEE Trans. Commun., vol. 72, no. 3, pp. 1414–1427, 2024

work page 2024

[7] [7]

Enhancing the resilience of communication networks,

OECD, “Enhancing the resilience of communication networks,” Organi- sation for Economic Co-operation and Development, Tech. Rep., 2025

work page 2025

[8] [8]

Failures and resilience in the IP era: Navigating the fragility of modern telecommunications networks–the sovereign func- tions,

R. Owenet al., “Failures and resilience in the IP era: Navigating the fragility of modern telecommunications networks–the sovereign func- tions,”IEEE Access, vol. 13, pp. 155 759–155 777, 2025

work page 2025

[9] [9]

Cisco Systems,Cisco Crosswork Network Controller: Use Templates to Configure Similar Devices, 2024

work page 2024

[10] [10]

Cisco Meraki,Managing Multiple Networks with Configuration Tem- plates, Cisco Meraki, 2020

work page 2020

[11] [11]

Large-scale network router-level topology construction based on knowledge inference,

Y . Zhou, G. Fei, J. Ye, X. Zhai, and G. Hu, “Large-scale network router-level topology construction based on knowledge inference,”IEEE Networking Lett., vol. 7, no. 3, pp. 210–214, 2025

work page 2025

[12] [12]

RouteNet: Leveraging graph neural networks for network modeling and optimization in SDN,

K. Rusek, J. Su ´arez-Varela, P. Almasan, P. Barlet-Ros, and A. Cabellos- Aparicio, “RouteNet: Leveraging graph neural networks for network modeling and optimization in SDN,”IEEE J. Sel. Areas Commun., vol. 38, no. 10, pp. 2260–2270, 2020

work page 2020

[13] [13]

Learning to configure computer networks with neural algorithmic reasoning,

L. Beurer-Kellner, M. Vechev, L. Vanbever, and P. Veli ˇckovi´c, “Learning to configure computer networks with neural algorithmic reasoning,” in Proceedings of Int. Conf. Neural Inf. Process. Syst. (NeurIPS), 2022

work page 2022

[14] [14]

A novel UA V-assisted V ANET routing protocol for post-disaster emergency communications,

Z. Fan, M. Zhang, Y . Cao, Z. Liu, O. Kaiwartya, Y . Javed, and F. Bashir Hussain, “A novel UA V-assisted V ANET routing protocol for post-disaster emergency communications,”IEEE Trans. Network Sci. Eng., vol. 13, pp. 4863–4882, 2026

work page 2026

[15] [15]

Hybrid-task meta-learning: A GNN approach for scalable and transferable bandwidth allocation,

X. Hao, C. She, P. L. Yeoh, Y . Liu, B. Vucetic, and Y . Li, “Hybrid-task meta-learning: A GNN approach for scalable and transferable bandwidth allocation,”IEEE Trans. Wireless Commun., vol. 23, no. 12, pp. 19 820– 19 835, 2024

work page 2024

[16] [16]

Graph neural network-based bandwidth allocation for secure wireless commu- nications,

X. Hao, P. L. Yeoh, Y . Liu, C. She, B. Vucetic, and Y . Li, “Graph neural network-based bandwidth allocation for secure wireless commu- nications,” inProceedings of IEEE Int. Conf. Commun. (ICC), 2023, pp. 332–337

work page 2023

[17] [17]

Predictive precoder design for OTFS-enabled URLLC: A deep learning approach,

C. Liu, S. Li, W. Yuan, X. Liu, and D. W. K. Ng, “Predictive precoder design for OTFS-enabled URLLC: A deep learning approach,”IEEE J. Sel. Areas Commun., vol. 41, no. 7, pp. 2245–2260, Jul. 2023

work page 2023

[18] [18]

Graph attention networks,

P. Veli ˇckovi´c, G. Cucurull, A. Casanova, A. Romero, P. Li `o, and Y . Bengio, “Graph attention networks,” inProceedings of Int. Conf. Learn. Representations (ICLR), 2018

work page 2018

[19] [19]

How attentive are graph attention networks?

S. Brody, U. Alon, and E. Yahav, “How attentive are graph attention networks?” inProceedings of Int. Conf. Learn. Representations (ICLR), 2022

work page 2022

[20] [20]

The Internet topology zoo,

S. Knight, H. X. Nguyen, N. Falkner, R. Bowden, and M. Roughan, “The Internet topology zoo,”IEEE J. Sel. Areas Commun., vol. 29, no. 9, pp. 1765–1775, 2011

work page 2011