Pith. sign in

REVIEW 5 major objections 7 minor 30 references

A Knowledge-Centric Communication For Autonomous Cislunar Networks

T0 review · 5 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Knowledge-centric digital twin keeps cislunar networks reasoning through outages by tracking uncertainty as 'knowledge entropy' and 'knowledge freshness.'

desk verdict A useful cislunar digital-twin application built on standard Bayesian fusion and AoI; the metrics are renamings and some claims overreach, but the framework, code, and honest Longjiang-2 checks merit a serious referee. read the letter →

arxiv 2608.04817 v1 pith:RRTMGGSD submitted 2026-08-05 eess.SP

Afan Ali , Daniel Benevides da Costa , Ali Arshad Nasir This is my paper · ORCID
classification eess.SP MSC 94A1593E1193E10
keywords knowledge-centriccommunicationdigitaltwincislunarnetworksknowledgeentropyfreshnessBayesianfusionageofinformationLongjiang-2
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that autonomous cislunar communication systems should reason over an evolving Operational Knowledge State, not just over the latest observation. Because light-speed delay, orbital geometry, and intermittent visibility make it impossible to ever fully reconcile with ground truth, the system must explicitly track what it knows, how uncertain it remains, and how stale that knowledge has become. The authors prove that fusing independent observation sources always reduces posterior uncertainty, and they show in a specifically constructed scenario that no single source can meet a mission's knowledge requirement while the fused state succeeds every time. They define two metrics, Knowledge Entropy and Knowledge Freshness, to quantify the state, and they demonstrate the framework's behavior against simulated cislunar links and against real flight data from the Longjiang-2 lunar micro-satellite.

What carries the argument

The Operational Knowledge State $K(t) = (G(t), \{b_{ij}(t)\})$ is the central object: the time-varying communication graph together with a Gaussian posterior belief $b_{ij}(t) \approx \mathcal{N}(\mu_{ij}(t), \Sigma_{ij}(t))$ for each active link. The two quantities defined over this belief, Knowledge Entropy, the differential entropy $H_{ij}(t) = \frac{1}{2}\log((2\pi e)^d |\Sigma_{ij}(t)|)$, and Knowledge Freshness, the age $\Delta_{ij}(t) = t - t_{\text{last}}$, are what allow the system to reason about what it does not know. The update mechanism is a standard Bayesian precision-weighted fusion, Eqs. (21)-(22), which takes the learned predictive variance $\Sigma_{r,ij}(t)$ as measurement noise; Propositions 1 and 2 show that this update guarantees fusion never increases uncertainty and that prediction between observations always increases it.

What would settle it

A concrete falsifier would be a cislunar or Earth-orbiting experiment where two independent observation sources are fused and the resulting posterior variance exceeds the minimum of the two individual variances, which would directly contradict Proposition 1. Alternatively, reproducing the Fusion Necessity scenario with real radio links and finding a case where the fused belief fails the mission entropy requirement while a single source meets it would refute Corollary 2.

Watch

Extended reading notes

Core claim

The paper claims that treating operational knowledge itself as the object of autonomous reasoning, rather than treating communication as the faithful delivery of symbols or the tracking of instantaneous state, enables better-calibrated decisions in cislunar networks. Concretely, it constructs a Knowledge-Centric Digital Twin that fuses delayed observations, a physics-based link model, a learned residual model, and mission objectives into a Gaussian belief per link, then defines Knowledge Entropy as the differential entropy of that belief and Knowledge Freshness as the time since the belief was last updated by a real observation. The central theoretical results are that Bayesian precision-weighted fusion never increases uncertainty (Fusion Dominance) and that uncertainty strictly grows between observations (Knowledge Aging); together these imply Fusion Necessity, meaning there are mission thresholds that only a fused belief can meet. The authors support the framework with simulations and with comparisons to published Longjiang-2 flight data, showing that the fusion benefit and the entropy response to a real anomaly appear in the flight data as the framework predicts.

Load-bearing premise

The load-bearing premise is that the learned predictive variance $\Sigma_r(h)$ is a valid measurement-noise covariance and that different observation sources are conditionally independent, so that the precision-weighted fusion weights are correct; the paper's own hop3 calibration ratio of 0.055 shows this can fail badly when the residual model is used outside its training distribution.

Editorial extensions

If this is right

  • Intermittent, long-delay cislunar links can keep producing hedged decisions during outages, because the belief explicitly tracks growing uncertainty rather than freezing at the last observation.
  • Fusing multiple independent observation sources becomes a necessity rather than an option: there are mission entropy requirements that no single asset can meet but that the fused state meets with certainty, as demonstrated in the Fusion Necessity scenario.
  • A learned residual model that transfers across orbital regimes, provided its predictive variance is capped at the stationary variance of the residual process, enables zero-shot deployment with calibration ratios close to 1.0, and only 15% target-domain data can fix the remaining mismatch.
  • The per-link independence of the belief update keeps the computational cost linear in the number of active links, making the framework scalable to hundreds of tracked links.
  • Real flight data from Longjiang-2 corroborate the core pattern: fusing the direct and moonbounce paths and fusing four ground stations both reduce RMSE relative to the best single source, and the knowledge-centric entropy responds to the documented oscillator-jump anomaly while the state-centric entropy stays flat.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The framework's value is most pronounced when there are long outages and mission thresholds that are tight enough that no single source suffices; in regimes with near-continuous visibility, the authors' own Table 2 shows the advantage over baselines nearly vanishes, so the practical payoff is greatest for far-side, polar, or relay-dependent cislunar links.
  • A natural testable extension is to replace the reduced-order NRHO approximation with a full CR3BP propagator or mission ephemerides and check whether the outage statistics, and hence the entropy and freshness dynamics, change in a way that affects the PMC conclusions.
  • A deeper open question the paper exposes is whether the learned predictive variance $\Sigma_r(h)$ remains well calibrated under distribution shift; the hop3 extrapolation failure (calibration ratio 0.055) shows that the framework's guarantees depend concretely on keeping that uncertainty estimate physically sensible, and a broader evaluation across more orbital regimes would test the robustness of
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 7 minor

Summary. The manuscript proposes a 'knowledge-centric' communication framework (KCDT) for cislunar networks, in which each link is represented by a Gaussian belief over SNR, Doppler, and delay, and an Operational Knowledge State aggregates delayed observations, physics models, learned residuals, and mission objectives. The paper defines Knowledge Entropy and Knowledge Freshness, proves a fusion-dominance result (Prop. 1), an aging result (Prop. 2), and a fusion-necessity corollary (Cor. 2), and evaluates the framework in two simulated orbital regimes plus comparisons with Longjiang-2 flight data. The core proof is a standard precision-weighted Bayesian fusion argument; the evaluation shows mixed results, with the framework's advantage over baselines appearing only when the residual uncertainty model is well calibrated.

Significance. If fully supported, the framework would be a useful conceptual step: it gives communication and digital-twin research a common object of reasoning, defines interpretable metrics over the fused belief, and provides a scalable per-link fusion structure. The credits here are real: the proofs are correct (though elementary), the code is publicly archived, the authors state several limitations explicitly, and the attempt to test against real flight data is commendable. However, the headline guarantee is an internal-covariance statement that does not automatically transfer to operational accuracy when the learned measurement covariance is miscalibrated, and the mission-success advantage is not demonstrated in the reported simulations. The paper is therefore promising but requires substantially tightened claims and a fairer decision-layer evaluation before it can be accepted.

major comments (5)
  1. [§4.3, Eqs. (21)–(22), Table 2] Proposition 1 proves a statement about the algorithm's internal covariance, not about operational estimation error. The precision-weighted update treats the learned variance Σ_{r,ij}(t) = g_θ(h_ij(t)) as the true measurement-noise covariance; if this quantity is miscalibrated, the fusion weights in Eq. (22) are wrong and the fused posterior is not the true posterior. The manuscript's own results demonstrate such miscalibration (Table 1, hop3 zero-shot calibration 0.055 for the unbounded architecture; Table 2, KCDT calibration 29.531 on Regime B/hop2). The abstract's statement that 'distributed knowledge fusion systematically improves operational knowledge by reducing uncertainty' should therefore be qualified: fusion reduces the internal belief covariance, and this translates into better-calibrated operational knowledge only when Σ_r is a calibrated predictive variance.
  2. [§2.1, Cor. 2, Fig. 4b] The headline 'there are regimes where no amount of improving a single observation source gets you to mission success' is an overstatement of Corollary 2. Corollary 2 only compares N individually insufficient sources against their fusion; it says nothing about improving one source's precision. In the S3 scenario, if source A's measurement variance were reduced sufficiently, Eq. (21) would allow source A alone to satisfy Σ_req. The supported claim is the weaker but still meaningful statement that, for fixed per-source configurations that individually fail, fusion can meet the requirement. Please rephrase the Results and Discussion accordingly.
  3. [§2.3, Table 2, §4.5.2] The decision-layer experiment does not evaluate the proposed architecture under a calibrated model. In Regime A, the PMC is identical for KCDT and the state-centric baseline (0.406 vs 0.406), and equal to the oracle, so no mission-success advantage is demonstrated there. In Regime B/hop2, KCDT has RMSE 2.464 dB versus 0.397 dB for state-centric and 0.529 dB for data-driven, and its calibration ratio of 29.531 is the worst in the table, because the analytic residual heuristic was intentionally mismatched. This heuristic is a different object from the learned residual model reported in Table 1, which achieves hop2 zero-shot calibration 1.403 and 0.993 after few-shot tuning. Since the actual proposed system includes the learned model, the comparison should either be repositioned as an explicit model-mismatch ablation or run with the learned model so that the headline decision-layer claim is evaluated on the real architecture.
  4. [§2.2, Table 1, §4.2] The introduction of the bounded predictive-variance head is presented as a 'fix' after the hop3 zero-shot calibration collapsed to 0.055. As reported, the cap at the stationary variance of the residual process is a free parameter whose value appears to have been chosen in response to a target-domain failure. The zero-shot transfer claim would be stronger if the cap were fixed using Regime A alone or derived from the residual process's physical stationarity before evaluating Regime B; as written, the reader cannot rule out post-hoc selection on the hop3 test result. Please state explicitly when and on what data the cap was determined, and report the sensitivity of the bounded calibration to the cap value.
  5. [§2.4, Fig. 8] The Longjiang-2 validation does not specify how the 'truth' for the RMSE comparisons in Fig. 8 is obtained. If the ground truth is derived from the same public flight data that supplies the observations, the comparison is partially circular and should be stated as a consistency check rather than an independent validation. Additionally, the moonbounce reflection is acknowledged to be a simplified single-point geometry that does not reproduce the reported Doppler near-cancellation; with these caveats, the reported RMSE reductions (1.259 vs 1.092; 1.031 vs 0.847) should be described as illustrative rather than as evidence that the framework improves accuracy on real data.
minor comments (7)
  1. [§2.2, Table 1] The text 'the same light touch improved hop3 from 0.46 to 0.980' appears to be a typo; Table 1 lists the bounded hop3 zero-shot calibration as 1.046, so the sentence should read 'from 1.046 to 0.980'.
  2. [§4.4.3, Eq. (31)] The definition of PMC over continuous time, Pr(∀t∈[0,T] ...), needs a formal statement of the underlying stochastic process and of how the probability is estimated from Monte Carlo runs; otherwise the reported PMC values depend on an unspecified temporal discretization.
  3. [§4.4.2, Eq. (30)] Knowledge Freshness as defined is exactly the age of the last observation; the claimed generalization to the 'temporal validity of the fused Operational Knowledge State' is not formalized. Please add a decay or validity function, or clarify explicitly where the definition goes beyond standard AoI.
  4. [§2.1, Fig. 4b] The S3 Fusion Necessity scenario is referenced in the Results but not specified in the Methods; please add the scenario definition, including how the two sources and the mission entropy requirement were generated.
  5. [§4.2, Eqs. (13)–(20)] Several equations are introduced with 'can be illustrated as' or 'can be defined as'; these should be replaced with standard mathematical phrasing such as 'is given by' or 'is defined as'.
  6. [Fig. 5(b)] The calibration axis in Fig. 5(b) extends to 40, compressing the baseline values near 1; a logarithmic scale or a broken axis would make the comparison readable.
  7. [§4.1, Regime B] The reduced-order NRHO approximation is acknowledged in the Discussion, but a brief quantitative validation against a full CR3BP orbit, for example one orbit's occultation times, would strengthen the claim that the approximation reproduces representative outage statistics.

Circularity Check

2 steps flagged · score 4.0 of 10

The Bayesian fusion proof is a standard, non-circular identity of the precision-weighted update, but the cross-regime 'zero-shot' claim is weakened by a post-hoc variance cap, and Knowledge Freshness is Age of Information under a new name.

  1. fitted input called prediction [Section 2.2, 'Operational knowledge transfers across communication regimes'; Table 1]
    "Moreover, hop3 was the outlier, the calibration ratio of the original, unbounded predictive-variance architecture dropped to 0.055. Digging into the traces showed why a handful of outages on hop3 ran as long as 79.7 h, well past the 17.6 h longest gap the model had encountered in training. So this was extrapolation failure, not a flaw in the residual model itself. The fix was simple, it required a cap on the predictive variance at the stationary variance of the residual process."

    The unbounded model's zero-shot hop3 calibration was 0.055; only after observing this target-hop failure did the authors introduce a predictive-variance cap set using the residual process's stationary variance, which is target-environment information. The capped model was then re-evaluated on the same hop and the improved calibration (1.046) was reported as evidence of cross-regime transfer. The final model is therefore not the originally trained Regime-A model: its architectural hyperparameter was selected in response to the very test point it is later used to validate, so the hop3 'zero-shot' prediction is partially forced by a post-hoc adjustment rather than being a pure out-of-distribution forecast.

  2. renaming known result [Section 4.4.2, 'Knowledge Freshness']
    "Knowledge Freshness, ∆ij(t), is the time since the belief was last updated by a real observation, which can be defined as ∆ij(t)=t−t last(i, j). It plays the same role AoI plays for a packet, but for a whole probabilistic belief rather than a single message."

    This definition is exactly the Age of Information of the belief-update stream: elapsed time since the most recent received update. Introducing it as a separate 'Knowledge Freshness' metric, and claiming in the Abstract that it generalizes AoI 'to the temporal validity of the fused Operational Knowledge State', is a rename of a known quantity rather than a derivation of a new temporal-validity concept. The metric contains no fusion or multi-source information beyond the timestamp of the last observation, so presenting it as a novel contribution is a relabeling rather than an organizational or theoretical advance.

full rationale

The central fusion mathematics is not circular. Equations (21)-(22) define a standard precision-weighted Bayesian update, and Proposition 1 is a valid algebraic consequence of summing positive-definite precisions; Proposition 2 follows directly from the assumed process-noise growth. These are identities of the update rule, not empirical predictions, and they do not depend on any self-citation. The Longjiang-2 comparisons use external published flight data [25] and are therefore independent evidence rather than self-support. The circularity that exists is in the evaluation framing. First, the zero-shot cross-regime claim for hop3 is weakened by a post-hoc architectural fix: after the unbounded uncertainty head produced calibration 0.055 on hop3, a predictive-variance cap was introduced and the same hop re-evaluated, so the reported 1.046 is not the prediction of a model fixed before seeing the target. Second, Knowledge Freshness (Eq. 30) is a direct rename of Age of Information, so presenting it as a new generalization is relabeling rather than derivation. The 'Fusion Necessity' scenario is an illustration of Corollary 2 with variances chosen to make the outcome hold; it is not an empirical discovery that single-source estimators can never succeed, but it is also an explicitly constructed example rather than a hidden circular step. I did not score the Longjiang-2 RMSE ground-truth concern as a circular step because the paper does not state how the RMSE truth was constructed, and flagging it would require speculation. Overall, the core fusion result remains self-contained and standard, while the evaluation protocol contains partial circularity, giving a score of 4.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper does not introduce new physical entities. The Operational Knowledge State is a mathematical data structure, and the two metrics are renamings of existing quantities. Free parameters are listed above; the most consequential is the post-hoc predictive variance cap that fixes the hop3 calibration failure.

free parameters (5)
  • Process noise covariance Q_proc = not specified
    Controls how quickly knowledge entropy grows during outages; hand-chosen in simulations.
  • Predictive variance cap (stationary variance of residual) = not specified numerically
    Introduced post hoc to fix hop3 calibration collapse from 0.055 to 1.046; a free parameter fit to the held-out hop3 data.
  • Correlation time tau_c of analytic residual heuristic = fit on Regime A
    Used in the Regime B/hop2 comparison; deliberately carried over from Regime A, leading to calibration 29.5.
  • Initial posterior covariance Sigma_0 = not specified
    Initial uncertainty of each link belief; required to start the filter.
  • Neural network weights theta (residual model) = learned from Regime A data
    Trained on Regime A communication data; part of the learned residual and uncertainty head. Not a closed-form parameter but is fitted to data.
assumptions (5)
  • domain assumption Posterior belief over link state is Gaussian
    Equation (16) approximates the posterior as Gaussian, standard in Bayesian filtering but not exact for nonlinear cislunar dynamics.
  • ad hoc to paper Observation sources are conditionally independent
    Proposition 1 assumes N independent assets; the moonbounce/direct path example likely violates this since both paths originate from the same transmitted signal.
  • domain assumption LoS visibility is determined solely by geometric sphere occlusion
    Eqs. (2)-(3) exclude other causes of link failure (e.g., interference, antenna pointing), simplifying the communication model.
  • ad hoc to paper Reduced-order NRHO approximation reproduces representative outage statistics
    Regime B uses a reduced-order Gateway orbit, not full CR3BP; the paper acknowledges exact timestamps are not mission-accurate.
  • domain assumption Learned residual and predictive variance are stationary across regimes after bounding
    The zero-shot transfer assumes the residual process in Regime B is drawn from the same family as Regime A, which hop2 violated until few-shot tuning.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Knowledge-Centric Communication For Autonomous Cislunar Networks." pith.science (2026). https://pith.science/paper/RRTMGGSD

@misc{pith2026260804817,
  author       = {Pith},
  title        = {Pith review of: A Knowledge-Centric Communication For Autonomous Cislunar Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RRTMGGSD}},
  note         = {Machine review of arXiv:2608.04817}
}
read the original abstract

Future lunar infrastructure requires communication as a persistent service rather than a mission specific capability. Communication, digital twin, and artificial intelligence (AI) research have advanced autonomy independently while largely assuming that system state can eventually be reconciled with ground truth. Cislunar communication violates this assumption because propagation delay is fundamentally limited by the speed of light, visibility is governed by orbital geometry, and autonomous decisions often precede confirming observations. Here we introduce a knowledge-centric communication framework in which a digital twin continuously integrates delayed observations, communication physics, learned models, uncertainty quantification, and mission objectives into an evolving Operational Knowledge State. To quantify what is known, how uncertain it remains, and how current it is, we define two quantities over the fused Bayesian belief maintained for each communication link: \emph{Knowledge Entropy}, the differential entropy of the posterior belief, and \emph{Knowledge Freshness}, which generalizes Age of Information (AoI) from a single information stream to the temporal validity of the fused Operational Knowledge State. We prove that distributed knowledge fusion systematically improves operational knowledge by reducing uncertainty and demonstrate its importance for mission success through simulations and comparisons with flight data from the Longjiang-2 lunar micro-satellite.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 25 canonical work pages

  1. [1]

    Technical report, NASA, Washington, DC, USA (2023)

    National Aeronautics and Space Administration: Moon to Mars architecture: Lunar communications, positioning, navigation, and timing (CPNT) architecture. Technical report, NASA, Washington, DC, USA (2023). https://www.nasa.gov

    work page 2023

  2. [3]

    In: 2017 IEEE Aerospace Conference, pp

    Schlesinger, A., Willman, B.M., Pitts, L., Davidson, S.R., Pohlchuck, W.A.: Delay/Disruption Tolerant Networking for the International Space Station (ISS). In: 2017 IEEE Aerospace Conference, pp. 1–14 (2017). https://doi. org/10.1109/AERO.2017.7943857 .https://ieeexplore.ieee.org/document/7943857 Accessed 2026-07-09

    work page arXiv 2017

  3. [4]

    IEEE Journal of Selected Topics in Quantum Electronics32(1: Advances in Free Space Laser), 1–13 (2026) https://doi.org/10

    Velasco, A.E., Meenehan, S.M., Wright, M.W., Alerstam, E., Allmaras, J.P., Andrews, K., Buehlman, W.C., Garkanian, V., Jones, C.M.G., Srinivasan, M.: Operational Results From the Deep Space Optical Communications (DSOC) Project Ground Laser Transmitter. IEEE Journal of Selected Topics in Quantum Electronics32(1: Advances in Free Space Laser), 1–13 (2026) ...

    work page arXiv 2026

  4. [5]

    NASA Technical Report (2026)

    Laboratory, N.J.P., Team, D.: Deep space optical communications: Operational results. NASA Technical Report (2026). https://www.nasa.gov

    work page 2026

  5. [6]

    IEEE Transactions on Aerospace and Electronic Systems62, 1582–1594 (2026) https://doi.org/10.1109/ TAES.2025.3633210

    C ¸ etin, S.G., D¨ onmez, B., Kurt, G.K.: Advancing Lunar Communication Through Interdomain Space Networks and Dynamic Orchestration. IEEE Transactions on Aerospace and Electronic Systems62, 1582–1594 (2026) https://doi.org/10.1109/ TAES.2025.3633210 . Accessed 2026-07-08

    work page arXiv 2026

  6. [7]

    IEEE Access8, 165570–165582 (2020) https://doi.org/10.1109/ACCESS

    Fernandez, L., Ruiz-De-Azua, J.A., Calveras, A., Camps, A.: Assessing LoRa for Satellite-to-Earth Communications Considering the Impact of Ionospheric Scintil- lation. IEEE Access8, 165570–165582 (2020) https://doi.org/10.1109/ACCESS. 2020.3022433 . Accessed 2026-07-08

    arXiv 2020

  7. [8]

    IEEE Transactions on Wireless Communications25, 11530–11545 (2026) https://doi.org/10.1109/TWC.2026.3659523

    Wei, Y., Chen, C., Yao, L., Zhang, H., Dong, F., Li, F., Wei, F., Jiang, G., Bao, F., Zhou, Y., Li, Z., Shen, C., Zhang, J., Chi, N., Shi, J.: A Semantic-Empowered Free-Space Optical Communication System With Turbulence-Resilient Vector Beams. IEEE Transactions on Wireless Communications25, 11530–11545 (2026) https://doi.org/10.1109/TWC.2026.3659523 . Acc...

    work page arXiv 2026

  8. [9]

    Bell System Technical Journal27(3), 379–423 (1948)

    Shannon, C.E.: A mathematical theory of communication. Bell System Technical Journal27(3), 379–423 (1948)

    1948

Show all 30 references
  1. [10]

    2731–2735 (2012)

    Kaul, S., Yates, R., Gruteser, M.: Real-time status: How often should one update? In: IEEE INFOCOM, pp. 2731–2735 (2012)

    2012

  2. [11]

    IEEE Journal on Selected Areas in Communications41(1), 5–41 (2023)

    G¨ und¨ uz, D., Qin, Z., Aguerri, I.E., Dhillon, H.S., Yang, Z., Yener, A., Wong, K.K., Chae, C.-B.: Beyond transmitting bits: Context, semantics, and task-oriented communications. IEEE Journal on Selected Areas in Communications41(1), 5–41 (2023)

    2023

  3. [12]

    arXiv preprint arXiv:2201.01389 (2021)

    Qin, Z., Tao, X., Lu, J., Li, G.Y.: Semantic communications: Principles and challenges. arXiv preprint arXiv:2201.01389 (2021)

    2021 arXiv

  4. [13]

    Nature573, 490–491 (2019) https: //doi.org/10.1038/d41586-019-02849-1

    Tao, F., Qi, Q.: Make more digital twins. Nature573, 490–491 (2019) https: //doi.org/10.1038/d41586-019-02849-1

    2019 doi

  5. [14]

    Nature Computational Science4, 166–168 (2024) https://doi.org/10.1038/ s43588-024-00609-4

    Willcox, K., Segundo, B.: The role of computational science in digital twins. Nature Computational Science4, 166–168 (2024) https://doi.org/10.1038/ s43588-024-00609-4

    2024

  6. [15]

    Nature Computational Science4(3), 178–183 (2024) https://doi.org/10.1038/ s43588-024-00613-8

    Ferrari, A., Willcox, K.: Digital twins in mechanical and aerospace engineering. Nature Computational Science4(3), 178–183 (2024) https://doi.org/10.1038/ s43588-024-00613-8

    2024

  7. [16]

    Nature Computational Science6(1), 6–10 (2026) https://doi.org/10.1038/s43588-025-00944-0

    San, O., Rasheed, A., Bozdemir, E., Deng, J.: The Evolution of Digital Twins from Reactive to Agentic Systems. Nature Computational Science6(1), 6–10 (2026) https://doi.org/10.1038/s43588-025-00944-0 . arXiv:2605.26066 [cs.CE]. Accessed 2026-07-08

    2026 arXiv

  8. [17]

    IEEE Transactions on Mobile Computing25(4), 4839–4856 (2026) https://doi.org/10.1109/TMC.2025

    Zhou, L., Leng, S., Liu, Y., Xiong, Z., Quek, T.Q.S.: Digital Twins for Low- Altitude UA V Networks–Cooperation and Learning. IEEE Transactions on Mobile Computing25(4), 4839–4856 (2026) https://doi.org/10.1109/TMC.2025. 3626747 . Accessed 2026-07-08

    2026 doi

  9. [18]

    Scientific Reports (2026) https: //doi.org/10.1038/s41598-026-58750-1

    Hamwi, A.A., Mittal, M.: FL-TWIN: a unified federated learning system for intrusion detection with digital twins modelling. Scientific Reports (2026) https: //doi.org/10.1038/s41598-026-58750-1 . Accessed 2026-07-08

    2026 doi

  10. [19]

    IEEE Open Journal of the Communications Society7, 3046–3059 (2026) https://doi.org/10

    Li, X., Wang, W., Liu, P., Zhou, T., Xia, H.: Exploring Generative AI for Miti- gating Digital Twin Deviation in Secure Vehicular Edge Computing. IEEE Open Journal of the Communications Society7, 3046–3059 (2026) https://doi.org/10. 1109/OJCOMS.2026.3668197 . Accessed 2026-07-08

    2026

  11. [20]

    IEEE Open Journal of the Communications Society7, 4259–4277 (2026) https://doi.org/10

    Parwez, M.S., Srivillibhutturu, S.T., Alshammari, A., Olufowobi, H.: Digital Twin–Guided AI Path Planning for Connectivity-Aware Mobility. IEEE Open Journal of the Communications Society7, 4259–4277 (2026) https://doi.org/10. 1109/OJCOMS.2026.3683886 . Accessed 2026-07-08

    2026

  12. [21]

    IEEE Open Journal of the Communications Society7, 153–168 (2026) https://doi.org/10

    Wang, L., Ma, X., Fan, J., Li, F., Li, Y.: Digital Twin Architecture Design and Testbed Deployment for Satellite-Terrestrial Integrated Networks. IEEE Open Journal of the Communications Society7, 153–168 (2026) https://doi.org/10. 1109/OJCOMS.2025.3638849 . Accessed 2026-07-08

    2026

  13. [22]

    IEEE Com- munications Surveys & Tutorials28, 1845–1878 (2026) https://doi.org/10.1109/ COMST.2025.3635582

    Duran, K., Verda Cakir, L., Yigit, Y., Huseynov, K., Ram Kusu, S., Ali Ert¨ urk, M., Canberk, B.: Toward Digital Twin-as-a-Service (DTaaS) Platforms: A Survey on Architecture, Design Requirements, and Performance Metrics. IEEE Com- munications Surveys & Tutorials28, 1845–1878 ...

    2026

  14. [23]

    IEEE Geoscience and Remote Sensing Magazine14(3), 392–441 (2026) https://doi.org/10.1109/MGRS.2026.3665978

    Barretta, D., Marini, L., Razzano, F., Guerrisi, G., Sandberg, S., Del Prete, R., Zancan, V., Mariotti, E., Furano, G., Manovi, L., Guo, J., Ullo, S.L., Long´ ep´ e, N., Meoni, G.: Toward Real-Time Earth System Intelligence Through Advance- ments in Onboard AI Processing: A re...

    2026

  15. [24]

    IEEE Communications Standards Magazine7(4), 76–83 (2023) https://doi.org/10.1109/MCOMSTD.0005.2200070

    Lin, X., Kundu, L., Dick, C., Velayutham, S.: Embracing ai in 5g-advanced toward 6g: A joint 3gpp and o-ran perspective. IEEE Communications Standards Magazine7(4), 76–83 (2023) https://doi.org/10.1109/MCOMSTD.0005.2200070

    2023

  16. [25]

    Nature Communica- tions11(1), 3425 (2020) https://doi.org/10.1038/s41467-020-17272-8

    Wei, M., Hu, C., Est´ evez, D., Tai, M., Zhao, Y., Huang, J., Bassa, C., Jan Dijkema, T., Cao, X., Wang, F.: Design and flight results of the VHF/UHF communication system of Longjiang lunar microsatellites. Nature Communica- tions11(1), 3425 (2020) https://doi.org/10.1038/s414...

    2020 doi

  17. [26]

    Microcosm Press, Hawthorne, CA (2013)

    Vallado, D.A.: Fundamentals of Astrodynamics and Applications, 4th edn. Microcosm Press, Hawthorne, CA (2013)

    2013

  18. [27]

    Proceedings of the IRE 34(5), 254–256 (1946) https://doi.org/10.1109/JRPROC.1946.234568

    Friis, H.T.: A Note on a Simple Transmission Formula. Proceedings of the IRE 34(5), 254–256 (1946) https://doi.org/10.1109/JRPROC.1946.234568 . Accessed 2026-07-19

    1946

  19. [28]

    Intelligent Robotics and Autonomous Agents

    Thrun, S., Burgard, W., Fox, D.: Probabilistic Robotics. Intelligent Robotics and Autonomous Agents. MIT Press, Cambridge, MA (2005)

    2005

  20. [29]

    barshalom: Adaptive Estimation and Maneuvering Targets, pp. 421–490. John Wiley & Sons, Ltd. Chap. 11. https://doi.org/10.1002/0471221279.ch11 . https: //onlinelibrary.wiley.com/doi/abs/10.1002/0471221279.ch11

    doi

  21. [30]

    Cambridge University Press, Cambridge; New York (2013)

    Horn, R.A., Johnson, C.R.: Matrix Analysis, 2nd edn. Cambridge University Press, Cambridge; New York (2013)

    2013

  22. [31]

    When observations stop, communication should not stop reasoning

    Bhatia, R.: Positive Definite Matrices. Princeton Series in Applied Mathematics. Princeton University Press, Princeton, NJ (2007) Author contributions Afan Ali and Daniel Benevides da Costa initiated and conceptualized the original draft. Afan Ali designed the methodology, inv...

    2007

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.