pith. machine review for the scientific record. sign in

arxiv: 2605.00169 · v1 · submitted 2026-04-30 · 💻 cs.NI · cs.DC· cs.LG

Network Digital Untwinning: Towards Backward Optimization of Digital Twins

Zifan Zhang , Dianwei Chen , Anjun Gao , Manhua Wang , Mingzhe Chen , Minghong Fang , Xianfeng Yang , Yuchen Liu This is my paper

Pith reviewed 2026-05-09 20:24 UTC · model grok-4.3

classification 💻 cs.NI cs.DCcs.LG
keywords network digital twinsdata removalmodel integrityrollback mechanismprivacy compliancenetwork managementGaussian noisetwin optimization
0
0 comments X

The pith

A framework lets network operators remove specific digital twin contributions without rebuilding the model from scratch.

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

The paper develops a backward optimization method for network digital twins so that selected data can be erased when devices go offline or regulations demand it. Instead of discarding the whole twin and starting over, the approach locates the affected parts through metrics of geography, data patterns, and network attributes, then rolls back to an earlier checkpoint, adds controlled noise, and remaps the model. Two versions handle one removal at a time or many removals together by grouping similar twins. If the method works, operators gain a practical way to stay compliant and private while keeping the twin accurate and efficient for ongoing network management.

Core claim

The paper presents a network digital untwinning framework with two parts: Single Request Untwinning identifies a target twin and its influence using connectivity metrics, then removes that influence through an optimal rollback checkpoint plus injected Gaussian noise and a remapping step; Parallel Request Untwinning clusters similar twins and runs a coordinated schedule for multiple removals at once. Theoretical analysis shows the resulting model remains indistinguishable from one built from scratch without the removed data, and experiments on real traffic traces confirm that accuracy and speed are preserved.

What carries the argument

The untwinning process, which identifies propagating influence via connectivity metrics and then applies a rollback checkpoint augmented with Gaussian noise followed by remapping to excise the target contribution.

If this is right

  • Device deactivation or network changes can be reflected in the twin without a full rebuild.
  • Regulatory data-removal requests can be satisfied while the twin continues to support management tasks.
  • Multiple simultaneous removals become feasible through clustering and joint rollback scheduling.
  • Overall computational cost drops because only the affected portion is recalculated.

Where Pith is reading between the lines

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

  • The same rollback-plus-noise pattern could be tested on digital twins of other systems such as power grids or transportation networks.
  • If the indistinguishability property holds at scale, it opens the door to continuous, on-the-fly updates rather than periodic full reconstructions.
  • Operators might combine the method with streaming data pipelines so that untwinning triggers automatically when privacy thresholds are crossed.

Load-bearing premise

That connectivity metrics based on geographical proximity, data distribution, and network attributes can correctly locate the exact target twin and all parts it influences.

What would settle it

Running the untwinning procedure on a known traffic dataset and finding that the resulting model gives measurably different predictions or error rates than a fresh twin built from the same data minus the removed contributions.

Figures

Figures reproduced from arXiv: 2605.00169 by Anjun Gao, Dianwei Chen, Manhua Wang, Minghong Fang, Mingzhe Chen, Xianfeng Yang, Yuchen Liu, Zifan Zhang.

Figure 1
Figure 1. Figure 1: The workflow of SRU. We use neural network symbols to represent NDT models, but can be extended to any models. sensory data, as shown in Step 1 of [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: PRU consists of four main steps: (1) Parallel forward twinning; (2) Context-aware scheduling to determine cluster￾specific rollbacks; (3) Cluster-level perturbation for obfuscation; (4) Staggered re-twinning until global convergence. simultaneously, as shown in [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Testbed configurations on I-15 in Salt Lake, Utah. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison between untwinning target NDT only and [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: PED under diverse request settings. improvements with PRU. These performance gains can be attributed to its effective network clustering of NDTs and the coordinated rollback mechanism, which carefully determines a per-cluster rollback checkpoint to excise the contributions of all target NDTs and their propagating influences. Especially in cases where multiple untwinning requests occur within the same clust… view at source ↗
read the original abstract

Network digital twins (NDTs) are transforming network management by offering precise virtual replicas of physical network systems. However, their reliance on diverse and sensitive data introduces significant challenges related to data management, regulatory compliance, and user privacy. In scenarios where selective data removal is necessary, such as device deactivation, network reconfiguration, or regulatory compliance, traditional approaches often fall short of preserving the integrity of the twin model. To address this gap, we introduce a network digital untwinning framework that enables the targeted removal of deprecated NDT contributions while maintaining model integrity. Our approach comprises two complementary components: Single Request Untwinning (\algO) and Parallel Request Untwinning (\algM) mechanisms. \algO leverages connectivity metrics based on geographical proximity, data distribution, and network-level attributes to identify and remove the target NDT along with its propagating influence. This is achieved through an optimally selected rollback checkpoint augmented with injected Gaussian noise, followed by a precise remapping phase. \algM extends this mechanism to efficiently handle multiple removal requests by clustering NDTs with similar attributes and performing a coordinated rollback and untwinning schedule. We provide theoretical guarantees on model indistinguishability from scratch-built twins, and validate the framework through extensive experiments on real-world traffic data, demonstrating its effectiveness and operational efficiency.

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.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes a network digital untwinning framework for selectively removing deprecated contributions from network digital twins (NDTs) while preserving overall model integrity. It defines two mechanisms: Single Request Untwinning (O), which identifies target NDTs via connectivity metrics (geographical proximity, data distribution, network attributes), rolls back to an optimal checkpoint, injects Gaussian noise, and performs remapping; and Parallel Request Untwinning (M), which clusters similar NDTs for coordinated rollback and untwinning. The paper claims theoretical guarantees that the resulting model is indistinguishable from a scratch-built twin and validates the approach via experiments on real-world traffic data showing effectiveness and operational efficiency.

Significance. If the indistinguishability guarantees can be formally established and the experiments quantitatively confirm that utility is preserved with negligible distortion relative to baselines, the framework would address a practical gap in privacy-compliant NDT management for networks. The combination of rollback-plus-noise with clustering for parallel handling extends standard differential-privacy techniques to digital-twin maintenance and could influence operational practices in dynamic network environments.

major comments (3)
  1. [Abstract] Abstract: the central claim of 'theoretical guarantees on model indistinguishability from scratch-built twins' is load-bearing yet unsupported by any equation, theorem statement, or proof sketch; the rollback-checkpoint + Gaussian-noise + remapping procedure must be shown formally to bound a suitable distance (e.g., total variation or KL divergence) to a scratch-built model, otherwise the guarantee reduces to an unverified assertion.
  2. [Framework description] Methodology description: the assumption that connectivity metrics based on geographical proximity, data distribution, and network attributes reliably isolate the target NDT and its propagating influence is not accompanied by any robustness analysis or counter-example; if these metrics fail to capture influence accurately, both the single and parallel mechanisms lose their correctness foundation.
  3. [Evaluation] Experimental validation: the abstract states 'extensive experiments on real-world traffic data' but supplies no dataset size, baseline comparisons (full retraining, naive deletion), quantitative metrics (e.g., prediction error, latency overhead), or statistical tests; without these, the claims of 'effectiveness and operational efficiency' cannot be assessed as evidence for the indistinguishability result.
minor comments (2)
  1. [Abstract] Notation: the symbols O and M for the two mechanisms are introduced without an explicit definition table or first-use expansion; a short notation section would improve readability.
  2. [Single Request Untwinning] The phrase 'optimally selected rollback checkpoint' is used without stating the optimization criterion or algorithm; a brief description of the selection procedure would clarify the method.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their insightful comments, which have helped us identify areas for improvement in our manuscript. We will make major revisions to formalize the theoretical guarantees, provide robustness analysis for the connectivity metrics, and enhance the experimental validation with detailed quantitative information. Our point-by-point responses are as follows.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim of 'theoretical guarantees on model indistinguishability from scratch-built twins' is load-bearing yet unsupported by any equation, theorem statement, or proof sketch; the rollback-checkpoint + Gaussian-noise + remapping procedure must be shown formally to bound a suitable distance (e.g., total variation or KL divergence) to a scratch-built model, otherwise the guarantee reduces to an unverified assertion.

    Authors: We acknowledge that the abstract asserts theoretical guarantees on indistinguishability, yet the current manuscript lacks explicit theorem statements, equations, or proof sketches formalizing the bound. This is a substantive gap. In the revised manuscript we will insert a dedicated 'Theoretical Analysis' section that defines indistinguishability via total variation distance (or KL divergence) between the untwinning output and a scratch-built model. The section will prove that the optimal-checkpoint rollback combined with calibrated Gaussian noise and remapping yields a bounded distance, leveraging the post-processing property of the Gaussian mechanism and the fact that rollback restores a state free of the deprecated contribution. revision: yes

  2. Referee: [Framework description] Methodology description: the assumption that connectivity metrics based on geographical proximity, data distribution, and network attributes reliably isolate the target NDT and its propagating influence is not accompanied by any robustness analysis or counter-example; if these metrics fail to capture influence accurately, both the single and parallel mechanisms lose their correctness foundation.

    Authors: The connectivity metrics follow standard network-science practice for influence propagation. Nevertheless, the referee correctly notes the absence of explicit robustness analysis or counter-examples. We will add a new subsection that examines metric sensitivity, presents counter-examples (e.g., virtualized overlays where geography decouples from data influence), and quantifies degradation in untwinning accuracy under metric error. We will also outline a fallback using learned influence graphs when the base metrics prove insufficient. revision: yes

  3. Referee: [Evaluation] Experimental validation: the abstract states 'extensive experiments on real-world traffic data' but supplies no dataset size, baseline comparisons (full retraining, naive deletion), quantitative metrics (e.g., prediction error, latency overhead), or statistical tests; without these, the claims of 'effectiveness and operational efficiency' cannot be assessed as evidence for the indistinguishability result.

    Authors: We agree that the abstract and evaluation section would be strengthened by explicit quantitative details. In the revision we will update the abstract to report dataset characteristics, the two baselines (full retraining and naive deletion), the concrete metrics (prediction error and latency overhead), and the statistical tests employed. The evaluation section will be expanded with additional tables, figures, and analysis that directly link these measurements to the indistinguishability claim. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's central claims rest on a proposed untwinning framework using connectivity metrics, rollback checkpoints, Gaussian noise injection, and remapping to achieve model indistinguishability from scratch-built twins. No equations, derivations, or self-referential definitions are present in the abstract or described process that reduce the theoretical guarantees or experimental validations to fitted parameters or prior self-citations by construction. The approach aligns with standard privacy-preserving techniques without evident self-definition, fitted-input-as-prediction, or ansatz smuggling. The derivation chain appears self-contained against external benchmarks like real-world traffic data experiments.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 2 invented entities

The central claim depends on newly introduced mechanisms whose effectiveness is asserted without independent evidence or detailed derivation visible in the abstract.

free parameters (1)
  • Gaussian noise variance
    Injected during rollback to mask influence; specific value or selection rule not stated in abstract
axioms (1)
  • domain assumption Connectivity metrics based on geographical proximity, data distribution, and network-level attributes accurately capture propagating influence of a target NDT
    Used to select rollback checkpoint and determine removal scope
invented entities (2)
  • Single Request Untwinning (O) mechanism no independent evidence
    purpose: Targeted removal of one deprecated NDT contribution via rollback and remapping
    Newly defined component of the framework
  • Parallel Request Untwinning (M) mechanism no independent evidence
    purpose: Coordinated removal of multiple similar NDT contributions via clustering and scheduled rollback
    Newly defined component of the framework

pith-pipeline@v0.9.0 · 5550 in / 1425 out tokens · 48283 ms · 2026-05-09T20:24:28.659376+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 2 canonical work pages · 1 internal anchor

  1. [1]

    Digital twin networks: A survey,

    Y . Wu, K. Zhang, and Y . Zhang, “Digital twin networks: A survey,” in IEEE IoT-J, 2021

    2021

  2. [2]

    Synergizing ai and digital twins for next-generation network optimization, forecasting, and security,

    Z. Zhang, M. Fang, D. Chen, X. Yang, and Y . Liu, “Synergizing ai and digital twins for next-generation network optimization, forecasting, and security,” inIEEE WCM, 2025

    2025

  3. [3]

    Resource allocation based on digital twin-enabled federated learning framework in heterogeneous cellular network,

    Y . He, M. Yang, Z. He, and M. Guizani, “Resource allocation based on digital twin-enabled federated learning framework in heterogeneous cellular network,”IEEE TVT, 2022

    2022

  4. [4]

    Digital twin for intelligent context-aware iot healthcare systems,

    H. Elayan, M. Aloqaily, and M. Guizani, “Digital twin for intelligent context-aware iot healthcare systems,” inIEEE IoT-J, 2021

    2021

  5. [5]

    Smart mobility digital twin based automated vehicle navigation system: A proof of concept,

    K. Wang, Z. Li, K. Nonomura, T. Yu, K. Sakaguchi, O. Hashash, and W. Saad, “Smart mobility digital twin based automated vehicle navigation system: A proof of concept,” inIEEE TIV, 2024

    2024

  6. [6]

    Towards intelligent transportation with pedestrians and vehicles in-the-loop: A surveillance video-assisted federated digital twin framework,

    X. Li, J. Wei, H. Wang, L. Dong, R. Chen, C. Yi, J. Cai, D. Niyato, and X. Shen, “Towards intelligent transportation with pedestrians and vehicles in-the-loop: A surveillance video-assisted federated digital twin framework,”IEEE Network, 2025

    2025

  7. [7]

    VaN3Twin: the Multi-Technology V2X Digital Twin with Ray-Tracing in the Loop

    R. Pegurri, D. Gasco, F. Linsalata, M. Rapelli, E. Moro, F. Raviglione, and C. Casetti, “Van3twin: the multi-technology v2x digital twin with ray-tracing in the loop,”arXiv preprint arXiv:2505.14184, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  8. [8]

    Digital twin of wireless systems: Overview, taxonomy, challenges, and opportunities,

    L. U. Khan, Z. Han, W. Saad, E. Hossain, M. Guizani, and C. S. Hong, “Digital twin of wireless systems: Overview, taxonomy, challenges, and opportunities,” inIEEE ComST, 2022

    2022

  9. [9]

    Wireless network digital twin for 6g: Generative ai as a key enabler,

    Z. Tao, W. Xu, Y . Huang, X. Wang, and X. You, “Wireless network digital twin for 6g: Generative ai as a key enabler,” inIEEE WCM, 2024

    2024

  10. [10]

    Digital twin- assisted data-driven optimization for reliable edge caching in wireless networks,

    Z. Zhang, Y . Liu, Z. Peng, M. Chen, D. Xu, and S. Cui, “Digital twin- assisted data-driven optimization for reliable edge caching in wireless networks,” inIEEE JSAC, 2024

    2024

  11. [11]

    Colosseum as a digital twin: Bridging real- world experimentation and wireless network emulation,

    D. Villa, M. Tehrani-Moayyed, C. P. Robinson, L. Bonati, P. Johari, M. Polese, and T. Melodia, “Colosseum as a digital twin: Bridging real- world experimentation and wireless network emulation,” inIEEE TMC, 2024

    2024

  12. [12]

    Two-phase authentication for secure vehicular digital twin communications,

    X. Zhang, C. Lai, G. Li, and D. Zheng, “Two-phase authentication for secure vehicular digital twin communications,”Computer Networks, 2025

    2025

  13. [13]

    Secure and flexible data sharing with dual privacy protection in vehicular digital twin networks,

    C. Wang, Y . Ming, H. Liu, J. Feng, and N. Zhang, “Secure and flexible data sharing with dual privacy protection in vehicular digital twin networks,”IEEE TITS, 2024

    2024

  14. [14]

    Securing vehicle-to-digital twin communications in the internet of vehicles,

    S. Jabeen Siddiqi, A. H. Alobaidi, M. Ahmad Jan, and M. Tariq, “Securing vehicle-to-digital twin communications in the internet of vehicles,”ACM TOMM, 2025

    2025

  15. [15]

    Rsaka- vdt: Designing reliable and provably secure authenticated key agreement scheme for vehicular digital twin networks,

    K. Wang, J. Dong, S. Wang, Z. Yuan, L. Sha, and F. Xiao, “Rsaka- vdt: Designing reliable and provably secure authenticated key agreement scheme for vehicular digital twin networks,”IEEE TVT, 2025

    2025

  16. [16]

    Secr: A secure and efficient charging reservation scheme based on digital twin in vehicular network,

    G. Li, T. H. Luan, J. Zheng, C. Lai, K. Zhang, and S. Yu, “Secr: A secure and efficient charging reservation scheme based on digital twin in vehicular network,”IEEE IoT-J, 2024

    2024

  17. [17]

    Digital network twins for next-generation wireless: Creation, optimization, and challenges,

    Z. Zhang, Z. Peng, H. Yu, M. Chen, and Y . Liu, “Digital network twins for next-generation wireless: Creation, optimization, and challenges,” IEEE network, 2025

    2025

  18. [18]

    Securing dis- tributed network digital twin systems against model poisoning attacks,

    Z. Zhang, M. Fang, M. Chen, G. Li, X. Lin, and Y . Liu, “Securing dis- tributed network digital twin systems against model poisoning attacks,” IEEE IoT-J, 2024

    2024

  19. [19]

    Digital twin network – requirements and architecture,

    “Digital twin network – requirements and architecture,” International Telecommunication Union, Y . 3090 Recommendations, 2024

    2024

  20. [20]

    Digital twin network - capability levels and evaluation methods,

    “Digital twin network - capability levels and evaluation methods,” International Telecomm. Union, Y . 3091 Recommendations, 2024

    2024

  21. [21]

    Sequential informed federated unlearning: Efficient and provable client unlearning in feder- ated optimization,

    Y . Fraboni, R. Vidal, L. Kameni, and M. Lorenzi, “Sequential informed federated unlearning: Efficient and provable client unlearning in feder- ated optimization,” inAISTATS, 2024

    2024

  22. [22]

    Towards efficient and certified recovery from poisoning attacks in federated learning,

    Y . Jiang, J. Shen, Z. Liu, C. W. Tan, and K.-Y . Lam, “Towards efficient and certified recovery from poisoning attacks in federated learning,” in arXiv preprint arXiv:2401.08216, 2024

    work page arXiv 2024

  23. [23]

    Fedrecovery: Differentially private machine unlearning for federated learning frame- works,

    L. Zhang, T. Zhu, H. Zhang, P. Xiong, and W. Zhou, “Fedrecovery: Differentially private machine unlearning for federated learning frame- works,” inIEEE TIFS, 2023

    2023

  24. [24]

    Fedme2: Memory evaluation & erase promoting federated unlearning in dtmn,

    H. Xia, S. Xu, J. Pei, R. Zhang, Z. Yu, W. Zou, L. Wang, and C. Liu, “Fedme2: Memory evaluation & erase promoting federated unlearning in dtmn,” inIEEE JSAC, 2023

    2023

  25. [25]

    Federaser: Enabling efficient client-level data removal from federated learning models,

    G. Liu, X. Ma, Y . Yang, C. Wang, and J. Liu, “Federaser: Enabling efficient client-level data removal from federated learning models,” in IEEE IWQoS, 2021

    2021

  26. [26]

    Mapping wireless networks into digital reality through joint vertical and horizontal learning,

    Z. Zhang, M. Chen, Z. Yang, and Y . Liu, “Mapping wireless networks into digital reality through joint vertical and horizontal learning,” in IFIP/IEEE Networking, 2024

    2024

  27. [27]

    Distributed digital twin migration in multi-tier computing systems,

    Z. Chen, W. Yi, A. Nallanathan, and J. A. Chambers, “Distributed digital twin migration in multi-tier computing systems,”IEEE JSTSP, 2024

    2024

  28. [28]

    6g digital twin networks: From theory to practice,

    X. Lin, L. Kundu, C. Dick, E. Obiodu, T. Mostak, and M. Flaxman, “6g digital twin networks: From theory to practice,”IEEE CommMag, 2023

    2023

  29. [29]

    A comprehensive survey on digital twin for future networks and emerging internet of things industry,

    A. Hakiri, A. Gokhale, S. B. Yahia, and N. Mellouli, “A comprehensive survey on digital twin for future networks and emerging internet of things industry,” inComputer Networks, 2024

    2024

  30. [30]

    Digital-twin-enabled federated learning and cnn-based channel estimation for urban vehicular channels,

    C. Ding and I. W.-H. Ho, “Digital-twin-enabled federated learning and cnn-based channel estimation for urban vehicular channels,”IEEE IoT-J, 2025

    2025

  31. [31]

    Mobility digital twin: Concept, architecture, case study, and future challenges,

    Z. Wang, R. Gupta, K. Han, H. Wang, A. Ganlath, N. Ammar, and P. Tiwari, “Mobility digital twin: Concept, architecture, case study, and future challenges,”IEEE IoT-J, 2022

    2022

  32. [32]

    Digital twin empowered content caching in social-aware vehicular edge networks,

    K. Zhang, J. Cao, S. Maharjan, and Y . Zhang, “Digital twin empowered content caching in social-aware vehicular edge networks,”IEEE TCSS, 2021

    2021

  33. [33]

    Communication-Efficient Learning of Deep Networks from Decentralized Data,

    B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. A. y. Arcas, “Communication-Efficient Learning of Deep Networks from Decentralized Data,” inAISTATS, 2017

    2017

  34. [34]

    The algorithmic foundations of differential privacy,

    C. Dwork, A. Rothet al., “The algorithmic foundations of differential privacy,” inFound Trends Theor Comput Sci, 2014

    2014

  35. [35]

    Traffic data,

    Utah Department of Transportation, “Traffic data,” https://www.udot.utah.gov/connect/business/traffic-data/, 2025, accessed: March 20th, 2025

    2025

  36. [36]

    Net- work digital twin: Context, enabling technologies, and opportunities,

    P. Almasan, M. Ferriol-Galm ´es, J. Paillisse, J. Su ´arez-Varela, D. Perino, D. L´opez, A. A. P. Perales, P. Harvey, L. Ciavaglia, L. Wonget al., “Net- work digital twin: Context, enabling technologies, and opportunities,” in IEEE CommMag, 2022

    2022

  37. [37]

    Digital twin for 5G and beyond,

    H. X. Nguyen, R. Trestian, D. To, and M. Tatipamula, “Digital twin for 5G and beyond,” inIEEE CommMag, 2021

    2021

  38. [38]

    Research on integrated sensing, communication resource allocation and digital twin placement based on digital twin in iov,

    L. Tang, A. Wang, B. Xia, Y . Tang, and Q. Chen, “Research on integrated sensing, communication resource allocation and digital twin placement based on digital twin in iov,”IEEE IoT-J, 2025

    2025

  39. [39]

    O-ran-based digital twin function virtualization for sustainable iov service response: An asynchronous hierarchical reinforcement learning approach,

    Y . Tao, J. Wu, Q. Pan, A. K. Bashir, and M. Omar, “O-ran-based digital twin function virtualization for sustainable iov service response: An asynchronous hierarchical reinforcement learning approach,”IEEE TGCN, 2024

    2024

  40. [40]

    Intelligent reflecting surface and network slicing assisted vehicle digital twin update,

    L. Li, L. Tang, Y . Wang, T. Liu, and Q. Chen, “Intelligent reflecting surface and network slicing assisted vehicle digital twin update,”IEEE TITS, 2024

    2024

  41. [41]

    Energy- efficient federated learning training optimization for digital twin driven 6g air-ground integrated vehicular networks,

    C. Tan, P. Yu, Z. Qu, L. Zhang, W. Li, X. Qiu, and S. Guo, “Energy- efficient federated learning training optimization for digital twin driven 6g air-ground integrated vehicular networks,”IEEE TITS, 2025

    2025

  42. [42]

    A graph neural network-based digital twin for network slicing management,

    H. Wang, Y . Wu, G. Min, and W. Miao, “A graph neural network-based digital twin for network slicing management,” inIEEE TII, 2020

    2020

  43. [43]

    Connecting the twins: A review on digital twin technology & its networking requirements,

    M. Mashaly, “Connecting the twins: A review on digital twin technology & its networking requirements,” inProcedia Computer Science, 2021

    2021

  44. [44]

    Intelligent digital twin-based software-defined vehicular networks,

    L. Zhao, G. Han, Z. Li, and L. Shu, “Intelligent digital twin-based software-defined vehicular networks,” inIEEE Network, 2020

    2020

  45. [45]

    Joint vehicle connection and beamforming optimization in digital twin assisted inte- grated sensing and communication vehicular networks,

    W. Ding, Zhang, M. Chen, Y . Liu, and M. Shikh-Bahaei, “Joint vehicle connection and beamforming optimization in digital twin assisted inte- grated sensing and communication vehicular networks,” inIEEE IoT-J, 2024

    2024

  46. [46]

    Optimizing synchronization delay for digital twin over wireless networks,

    Yang, Zhaohui and Chen, Mingzhe and Liu, Yuchen and Zhang, Zhaoyang, “Optimizing synchronization delay for digital twin over wireless networks,” inIEEE ICASSP, 2024

    2024

  47. [47]

    Towards making systems forget with machine unlearning,

    Y . Cao and J. Yang, “Towards making systems forget with machine unlearning,” inIEEE Security & Privacy, 2015

    2015

  48. [48]

    Machine unlearning,

    L. Bourtoule, V . Chandrasekaran, C. A. Choquette-Choo, H. Jia, A. Travers, B. Zhang, D. Lie, and N. Papernot, “Machine unlearning,” inIEEE Security & Privacy, 2021

    2021

  49. [49]

    Federated unlearn- ing: How to efficiently erase a client in fl?

    A. Halimi, S. Kadhe, A. Rawat, and N. Baracaldo, “Federated unlearn- ing: How to efficiently erase a client in fl?” inICML, 2022

    2022

  50. [50]

    Toward efficient and robust federated unlearning in iot networks,

    Y . Yuan, B. Wang, C. Zhang, Z. Xiong, C. Li, and L. Zhu, “Toward efficient and robust federated unlearning in iot networks,” inIEEE IoT- J, 2024

    2024

  51. [51]

    Dt-fu: Digital twin-driven federated unlearning for resilient vehicular networks in the 6g era,

    W. Daluwatta, S. Edirimannage, C. Elvitigala, I. Khalil, and M. Atiquz- zaman, “Dt-fu: Digital twin-driven federated unlearning for resilient vehicular networks in the 6g era,” inIEEE CommMag, 2024

    2024