AI-Assisted ISAC Localization-as-a-Service for 6G UAV-IoT Networks

Ruhul Amin Khalil

REVIEW 4 major objections 4 minor 15 references

AI-Assisted ISAC Localization-as-a-Service for 6G UAV-IoT Networks

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A weighted utility score over link, sensing, and cost features selects a compact anchor–beam subset that approaches all-anchor localization accuracy while reducing pilot, energy, and beam-training overhead.

desk verdict Plausible incremental anchor–beam selection scheme, but the algorithm is not fully specified and the evaluation is too qualitative to confirm the claimed tradeoff. read the letter →

arxiv 2608.10571 v1 pith:3HMJGOJ4 submitted 2026-08-11 eess.SP

Ruhul Amin Khalil This is my paper · ORCID

classification eess.SP

keywords integratedsensingandcommunicationlocalization-as-a-serviceanchor-beamselectionUAV-IoTnetworksedgeintelligence6Ggradientboostingresourceallocation

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

The paper argues that reliable localization-as-a-service in 6G UAV-IoT networks does not require activating every available aerial and terrestrial anchor and beam. It proposes AIRS-LaaS, an edge-intelligence layer that scores each candidate anchor–beam pair by a weighted combination of SINR, sensing confidence, LoS probability, geometric contribution, mobility risk, and resource cost, then greedily activates a compact budget-limited subset. In simulation, this selection approaches the localization accuracy of all-anchor activation while reducing pilot, energy, and beam-training overhead, and it outperforms nearest-anchor, strongest-SINR, random, and FIM/CRLB-greedy baselines under LoS/NLoS and UAV-mobility conditions. The paper also maps the framework onto standardization-oriented KPIs, ISAC reporting, localization confidence, and AI model lifecycle management. If correct, the practical consequence is that network-native localization can be delivered with a small fraction of the sensing resources.

What carries the argument

The load-bearing object is the utility score $\eta_{a,m,k} = w_1 \tilde{\gamma}_{a,m,k} + w_2 \tilde{q}_{a,m,k} + w_3 \tilde{\rho}_{a,k} + w_4 \tilde{g}_{a,k} - w_5 \tilde{r}_{a,m,k} - w_6 \tilde{c}_{a,m}$, a weighted sum of six normalized features per anchor–beam pair. It is the ranking function that turns raw ISAC reports into an ordering, combined with a greedy budgeted selection loop that enforces a SINR floor and an incremental Fisher-information diversity check $\Delta_i = \log\det\bigl(J_{S_k\cup\{i\}} + \epsilon I\bigr) - \log\det\bigl(J_{S_k} + \epsilon I\bigr)$. The score carries the argument because it is what lets the framework balance communication reliability, sensing confidence, LoS likelihood, geometry, mobility risk, and cost without activating all links; the Fisher check prevents redundant geometry from consuming the budget.

What would settle it

Run the same selection rule with link features drawn from a channel model or field dataset different from the one used to train the weights, and compare the localization-outage versus activated-links curve against FIM/CRLB-greedy; if the advantage over that baseline vanishes or reverses, the claimed balanced tradeoff does not transfer.

Watch

Extended reading notes

Core claim

The central claim is that anchor–beam selection for ISAC localization should be treated as a joint resource-selection problem, not a pure geometry or pure link-quality problem. AIRS-LaaS computes a utility score $\eta_{a,m,k}$ for every candidate pair from normalized features—communication quality, sensing confidence, LoS probability, geometric contribution, mobility risk, and resource cost—with weights trained offline by a gradient-boosting regressor on simulated link observations. The selector removes links below an SINR/confidence threshold, sorts by score, and adds pairs under a pilot/energy budget while requiring a positive incremental Fisher-information gain to avoid geometric redundancy. The reported simulations show AIRS-LaaS operating in a low-localization-outage region with fewer activated links than full-anchor selection, communication outage close to the strongest-SINR baseline, and better robustness than FIM/CRLB-greedy as NLoS probability increases.

Load-bearing premise

The utility-score weights learned offline in the simulator are assumed to transfer to real deployment link-feature distributions; if the field statistics differ, the reported balance among localization, communication, and overhead may not hold.

Editorial extensions

If this is right

All-anchor activation is not required for reliable localization-as-a-service; a budget-limited subset selected by the utility score reaches a low-outage operating region with fewer activated links.
Joint scoring beats single-factor selection: AIRS-LaaS keeps communication outage close to strongest-SINR while also preserving localization robustness under NLoS, unlike nearest-anchor or strongest-SINR alone.
The greedy selection with an incremental Fisher-information check prevents geometrically redundant anchors from consuming the pilot/energy budget.
The edge-side complexity grows roughly as the number of candidates times the log of that number, plus the product of selected pairs and candidates, which is small enough for real-time selection in dense UAV-IoT deployments.
The framework supports standardization-style operation by exposing measurable indicators and a fallback to conventional selection when AI model confidence or feedback is missing.

Reading between the lines

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

The same feature-weighted scoring could be reused for other measurement-activation decisions in ISAC, such as which beams to train or which sensing waveforms to turn on, wherever the dominant cost is activation rather than estimation.
The ablation table suggests the six weights have separable effects; an online service could adapt them per quality-of-service class instead of fixing them offline, using outage and residual feedback to nudge weights.
A field-validated variant would likely need a small per-deployment calibration dataset or periodic retraining, since the current evaluation trains and tests on the same simulator; a cross-environment test would reveal how much of the advantage is environment-specific.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Short version: this is a plausible incremental systems paper, not a breakthrough. The genuinely new combination is a six-feature utility score (SINR, sensing confidence, LoS probability, geometry, mobility risk, resource cost) fused with an FIM-based diversity check inside a greedy selection loop. I agree with the reader that no cited prior work uses that exact rule. The paper is clearly organized, the system model is reasonable, and the ablation summary gives modest support for the multi-feature design.

The soft spots are real and they sit inside the algorithm, not just in the evaluation. Section V-A says the gradient-boosting regressor is trained with utility labels "derived from localization residuals, outage states, and resource cost," but no equation or step-by-step procedure defines how those components become a scalar label. Section IV-B presents an interpretable weighted sum η = w1γ̃ + w2q̃ + w3ρ̃ + w4g̃ − w5r̃ − w6c̃, but then says the reported results use a gradient-boosting regressor. It is never stated whether the regressor replaces the formula, predicts it, or calibrates the weights. So Algorithm 1 is under-specified; the stress-test note lands. That is a basic reproducibility problem, and it is more fundamental than the generalization concern.

The circularity issue is real but secondary. The labels come from the same simulator that produces the evaluation metrics, so the reported balanced tradeoff could be an in-sample artifact. I also agree that the figures are smooth curves with no numerical tables or error bars, which makes the magnitude of the claimed gains hard to judge. And since the FIM/CRLB-greedy baseline uses the same diversity criterion, the marginal benefit of the learned scoring layer needs a sharper quantitative comparison.

On the positive side, the paper does not oversell itself: it positions AIRS-LaaS as a framework, includes a fallback rule, and addresses standardization aspects without claiming to solve them.

Who is this for? Researchers working on ISAC resource selection or UAV-assisted localization, particularly those interested in 3GPP/ETSI alignment. It is not a formal contribution in the sense of a theorem or released artifacts, but it deserves a serious referee because the problem is relevant and the approach is a legitimate extension. I would send it to review with a clear request: give a precise definition of the utility labels and the scoring model, release or describe the Monte Carlo setup enough for reproduction, and add out-of-sample or at least error-barred numerical results.

Referee Report

4 major / 4 minor

Summary. The paper proposes AIRS-LaaS, an edge-intelligent anchor–beam selection framework for ISAC-based Localization-as-a-Service in 6G UAV-IoT networks. Candidate anchor–beam pairs are scored from LoS probability, SINR, sensing confidence, geometry, mobility risk, and resource cost; a greedy, budget-limited selection then activates a compact subset before localization. The manuscript presents a system model, an algorithm with complexity analysis, simulation comparisons against all-anchor, FIM/CRLB-greedy, strongest-SINR, nearest-anchor, and random selection, an ablation summary, and a discussion of standardization aspects. The central claim is that AIRS-LaaS approaches all-anchor localization accuracy while reducing overhead and maintaining communication reliability.

Significance. If the claims are supported, the paper addresses a practically relevant pre-localization resource-selection problem in UAV-IoT ISAC networks, and the proposed framework has a sensible design: it jointly considers link reliability, sensing quality, geometry, mobility, and cost, and it is positioned against several established anchor-selection baselines. The manuscript also gives a concrete per-target complexity bound, an ablation of the utility-score components, and a useful discussion of 3GPP/ETSI alignment, AI lifecycle management, and privacy-aware reporting. However, the evidence currently rests on qualitative simulation curves, and the core scoring rule is not specified precisely enough to be reproduced; these issues must be resolved before the central claim can be assessed.

major comments (4)

[§IV-B and §V-A] The utility scoring function that drives Algorithm 1 is not well-defined. Section IV-B presents an interpretable weighted-sum score η_{a,m,k}=w1 γ̃ + w2 q̃ + w3 ρ̃ + w4 g̃ - w5 r̃ - w6 c̃, but then states that the reported results use a lightweight gradient-boosting regressor; it is unclear whether the regressor replaces the weighted sum, predicts its output, or calibrates the weights. Section V-A says the regressor is trained with utility labels 'derived from localization residuals, outage states, and resource cost,' but gives no equation or procedure for constructing a scalar label for each anchor–beam pair, no normalization rule, and no explanation of how residuals and outage states are assigned to individual links (e.g., via counterfactual deletion). Because every selection decision and every reported tradeoff depends on η_{a,m,k}, the algorithm is currently not reproducible and the claimed advantage cannot be attributed to a concrete selection rule.
[§V-A and §V-C, Fig. 3] The training and evaluation setup creates a circularity risk. The gradient-boosting model is trained on Monte Carlo link observations with labels derived from localization residuals, outage states, and resource cost, and the same simulator then evaluates the method using localization outage (Fig. 3a, 3c, 3d) and communication outage (Fig. 3b), which are exactly the quantities appearing in the training labels. The LoS probability, sensing confidence, and mobility risk features are also generated by the same simulator. Under this setup, good test performance can reflect in-sample calibration rather than a generalizable selection rule. The authors should evaluate on a held-out scenario with different channel parameters, anchor geometries, or blockage maps, or use real measurement traces, and report the validation performance separately from the calibration performance.
[§V-C, Fig. 3, and Table I] The performance evidence is purely qualitative: the figures show curves without error bars, numerical tables, or a stated number of Monte Carlo trials, and statements such as 'AIRS-LaaS closely follows' the all-anchor reference or 'remains close to' the strongest-SINR baseline cannot be quantified from the plots. The ablation in Table I is also qualitative ('Higher communication outage', 'Poorer spatial diversity') with no measured magnitudes. Since the central claim is a balanced localization–communication–overhead tradeoff, the authors should report numerical outage values, the number of activated links, overhead/cost values, and confidence intervals or percentile spreads across independent trials.
[§IV-C and Algorithm 1] The geometric diversity check is not fully specified. Algorithm 1 line 10 requires checking that 'geometry diversity is improved', and Section IV-C defines the incremental Fisher-information gain Δ_i = log det(J_{S_k∪{i}} + εI) - log det(J_{S_k} + εI), but the manuscript never defines the localization Fisher information matrix J for the selected set, the measurement model (range, angle, or hybrid), or how the candidate pair's measurements enter Δ_i. This matters because the claimed difference from the FIM/CRLB-greedy baseline is that AIRS-LaaS accounts for link reliability and cost in addition to geometry; without an implementable definition of J and Δ_i, the algorithm cannot be reimplemented and the comparison cannot be reproduced.

minor comments (4)

[Throughout] The text contains many spacing errors in 'UA V' that should be unified to 'UAV'.
[§V-A] The paper says the 28 GHz link model is given 'for reproducibility', but several simulation parameters are missing, including the noise figure, antenna gains, beam codebook size, number of beams per anchor, the model for LoS probability as a function of distance and altitude, and the number of Monte Carlo trials per figure.
[§V-C, Fig. 3] Figure 3 does not include error bars or shaded confidence regions; adding them would help distinguish real differences from random fluctuations, especially for the curves that are described as 'close' to each other.
[Table I] The ablation table would be more informative if it reported numerical changes in localization outage, communication outage, and overhead when each term is removed, rather than only qualitative dominant impacts.

Circularity Check

1 steps flagged · score 6.0 of 10

The claimed balanced tradeoff is partially an in-sample fit: the AI scorer is trained on labels derived from outage/residual/cost and then evaluated on the same outage/residual/cost metrics.

fitted input called prediction [Section V-A (Simulation Scenario and Parameters) and Section V-C/Fig. 3 (Localization Accuracy and Communication Reliability)]
"For the reported results, the scoring block is implemented using a gradient-boosting regressor with 100 shallow trees, maximum depth 3, learning rate 0.05, and squared-error loss. Features are min–max normalized, and the model is trained offline using an 80/20 training/validation split from Monte Carlo link observations, with utility labels derived from localization residuals, outage states, and resource cost."

The utility score that drives Algorithm 1's ranking is learned from training labels 'derived from localization residuals, outage states, and resource cost.' The paper's headline results are then reported as localization outage (Fig. 3a,c,d, with outage defined as e_k > tau_L), communication outage (Fig. 3b, with outage defined as gamma < gamma_min), and number of activated pairs/overhead (Fig. 3d). These are the same outcome quantities from which the labels were constructed. Thus the reported 'balanced tradeoff' shows how well the fitted scorer does on the objectives it was trained to optimize, not an independent validation of the selection framework.

full rationale

The central claim of AIRS-LaaS is that it reaches a low-outage operating region with fewer activated links than full-anchor selection. That claim is evaluated with localization and communication outage curves. The same outage/residual/cost quantities are explicitly used to construct the training labels of the gradient-boosting scorer. Consequently, the reported advantage over baselines is in part a re-statement of the fitted objective: the selector was trained to prefer links whose labels indicate low residuals/outage and low cost, and the evaluation then measures those same quantities. This is the fitted-input-called-prediction pattern, and it is load-bearing because no external benchmark or independent scenario is used to validate the learned weights. I do not find other circularity: the self-citations [12] and [13] are not load-bearing (they only motivate energy-efficient anchor activation and supply an optional localization estimator), no uniqueness theorem is imported from the authors, and the feature weighting plus diversity check has independent algorithmic content. The lack of an explicit equation for the utility labels is a reproducibility problem, but the circularity is already visible in the described label construction. Overall score 6 reflects partial circularity: the framework itself is not purely definitional, but the paper's headline performance claim reduces substantially to the training objective.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim relies on six calibrated weights, GBM hyperparameters, and operational thresholds that are not reported, together with channel-model and feature-availability assumptions that are not independently validated. No new physical entities are introduced.

free parameters (3)

Utility-score weights w1..w6 = not reported
Trained or calibrated offline from simulated localization residuals, outage states, and resource cost in Section V-A; no values or sensitivity analysis are given.
Gradient-boosting hyperparameters = 100 trees, max depth 3, learning rate 0.05
Chosen for the reported scoring block in Section V-A without sensitivity analysis; Section IV-B also describes a weighted-sum scoring layer, so the actual fitted model is ambiguous.
Resource budget C_max and SINR threshold gamma_min = not reported
Operational thresholds that determine which links are selected; their values affect the claimed tradeoff but are not reported in the text.

assumptions (4)

standard math FIM-based incremental gain Delta_i = log det(J + eps I) - log det(J + eps I) measures the localization information contributed by a candidate link.
Uses the standard Fisher information and Cramer-Rao lower bound relation for range-angle measurements; it is invoked in Section IV-C without derivation.
domain assumption The 28 GHz LoS/NLoS path-loss and shadowing model with exponents 2.1 and 3.5 and NLoS range bias in [3,10] m describes the simulated network.
This is a simulation input from Section V-A; the claimed results hold only under this channel model.
domain assumption The network can supply accurate features: SINR, sensing confidence, LoS probability, geometric contribution, mobility risk, and resource cost for every candidate pair.
The algorithm's utility score consumes these features in Section IV-B; no model is given for how sensing confidence and LoS probability are obtained, and errors in them would change selection.
ad hoc to paper Utility labels derived from localization residuals, outage states, and resource cost in the same simulator provide a training signal that generalizes to online deployment.
Section V-A states the model is trained offline with such labels; the evaluation on the same metrics assumes this transfer, which is not tested.

how reviews work

0 comments

Cite this review

Pith. "Pith review of AI-Assisted ISAC Localization-as-a-Service for 6G UAV-IoT Networks." pith.science (2026). https://pith.science/paper/3HMJGOJ4

@misc{pith2026260810571,
  author       = {Pith},
  title        = {Pith review of: AI-Assisted ISAC Localization-as-a-Service for 6G UAV-IoT Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3HMJGOJ4}},
  note         = {Machine review of arXiv:2608.10571}
}

read the original abstract

Integrated sensing and communication (ISAC) can enable sixth-generation (6G) unmanned aerial vehicle-assisted Internet of Things (UAV-IoT) networks to provide reliable Localization-as-a-Service (LaaS), but activating all aerial/terrestrial anchors and beams increases pilot overhead, energy use, and beam-training delay. This article proposes artificial intelligence (AI)-assisted ISAC resource selection for LaaS (AIRS-LaaS), an edge-intelligent framework that ranks candidate anchor--beam pairs using line-of-sight (LoS) likelihood, signal-to-interference-plus-noise ratio (SINR), sensing confidence, geometry, mobility risk, and resource cost. A lightweight selector then activates only a compact subset before localization. Simulations compare AIRS-LaaS with all-anchor, Fisher information matrix/Cram\'er--Rao lower bound (FIM/CRLB)-greedy, strongest-SINR, nearest-anchor, and random schemes under LoS/non-line-of-sight (NLoS) conditions and UAV mobility. Results show a balanced localization--communication--overhead tradeoff, while the discussion highlights standard-driven key performance indicators (KPIs), ISAC reporting, localization confidence, fallback operation, AI model management, and privacy-aware data exchange.

Figures

Figures reproduced from arXiv: 2608.10571 by the authors.

**Figure 2.** Robustness evaluation under UAV mobility and dense candidate sets. For practical 3GPP/ETSI integration, AIRS-LaaS relies on measurable indicators such as SINR, beam quality, ToA/AoA/Doppler, sensing confidence, localization uncertainty, anchor/beam capability, and AI model status [6]. These indicators support a compact signaling flow that includes capability exposure, periodic or event-triggered ISAC reporting, ed… view at source ↗

**Figure 3.** AIRS-LaaS performance comparison with benchmark anchor–beam selection schemes. selection, which sequentially selects the anchor–beam pair with the largest Fisher-information gain, or equivalently the largest reduction in CRLB-oriented localization uncertainty. Unlike AIRS-LaaS, this baseline mainly captures localization geometry and does not jointly account for sensing confidence, LoS probability, mobility risk, com… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 13 canonical work pages

[1]

Framework and Overall Objectives of the Future Development of IMT for 2030 and Beyond,

ITU-R, “Framework and Overall Objectives of the Future Development of IMT for 2030 and Beyond,” International Telecommunication Union, Radiocommunication Sector, Recommendation ITU-R M.2160-0, Nov

work page
[2]

Integrated Sensing and Communications for IoT: Synergies with Key 6G Tech- nology Enablers,

A. Kaushik, R. Singh, M. Li, H. Luo, S. Dayarathna, R. Senanayake, X. An, R. A. Stirling-Gallacher, W. Shin, and M. D. Renzo, “Integrated Sensing and Communications for IoT: Synergies with Key 6G Tech- nology Enablers,”IEEE Internet of Things Magazine, vol. 7, no. 5, pp. 136–143, Sep. 2024

work page 2024
[3]

RAN Rel-19 Status and a Look Beyond,

3GPP, “RAN Rel-19 Status and a Look Beyond,” 3GPP Technology Update, Apr. 2025, by W. Chen, 3GPP TSG RAN Chair. [Online]. Available: https://www.3gpp.org/technologies/ran-rel-19

work page 2025
[4]

Isac Channel Models from ETSI and 3GPP,

M. Heggo, A. Shojaeifard, A. Mourad, C. Jiang, R. Liu, and J. Liu, “Isac Channel Models from ETSI and 3GPP,” inIEEE 36th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), 2025, pp. 1–6

work page 2025
[5]

Integrated Sensing and Communications (ISAC); Use Cases and Deployment Scenarios,

ETSI, “Integrated Sensing and Communications (ISAC); Use Cases and Deployment Scenarios,” European Telecommunications Standards Institute, Group Report ETSI GR ISC 001 V1.1.1, Mar. 2025. [Online]. Available: https://www.etsi.org/deliver/etsigr/ISC/001099/001/ 01.01.0160/grISC001v010101p.pdf

work page 2025
[6]

Overview of AI/ML Related Work in 3GPP,

3GPP, “Overview of AI/ML Related Work in 3GPP,” 3GPP News and Events, 2025. [Online]. Available: https://www.3gpp.org/news-events/ 3gpp-news/ai-ml-2025

work page 2025
[7]

An Overview of the 3GPP Study on Artificial Intelligence for 5G New Radio,

X. Lin, “An Overview of the 3GPP Study on Artificial Intelligence for 5G New Radio,” 2023. [Online]. Available: https://arxiv.org/abs/ 2308.05315

arXiv 2023
[8]

AI/ML Life Cycle Management for Interoperable AI Native RAN,

C.-H. Huang, C.-K. Wen, and G. Y . Li, “AI/ML Life Cycle Management for Interoperable AI Native RAN,” 2025. [Online]. Available: https://arxiv.org/abs/2507.18538

arXiv 2025

Show all 15 references

[9]

Efficient UA V Hovering, Resource Allocation, and Trajectory Design for ISAC With Limited Backhaul Capacity,

A. Khalili, A. Rezaei, D. Xu, F. Dressler, and R. Schober, “Efficient UA V Hovering, Resource Allocation, and Trajectory Design for ISAC With Limited Backhaul Capacity,”IEEE Transactions on Wireless Communications, vol. 23, no. 11, pp. 17 635–17 650, 2024

2024
[10]

Generative AI-Aided QoE Maxi- mization for RIS-Assisted Digital Twin Interaction,

J. Chen, Y . Li, C. Yi, and S. Gong, “Generative AI-Aided QoE Maxi- mization for RIS-Assisted Digital Twin Interaction,” inIEEE INFOCOM 2025 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS), 2025, pp. 1–6

2025
[11]

From Data Mirror to Smart Copilot: A Survey on NextG Semantic Communication for Propelling Digital Twin World Into Cognitive Stage,

F. Zhu, J. Chen, J. Wen, Y . Yang, C. Yi, Y . Tie, P. Zhang, J. Cai, D. Niyato, and M. Guizani, “From Data Mirror to Smart Copilot: A Survey on NextG Semantic Communication for Propelling Digital Twin World Into Cognitive Stage,”IEEE Communications Surveys & Tutorials, vol. 28...

2026
[12]

Energy- Efficient Anchor Activation Protocol for Non-Cooperative Localization of Industrial Internet of Things (IIoT),

R. A. Khalil, N. Saeed, M. Almutiry, and A. H. Alenezi, “Energy- Efficient Anchor Activation Protocol for Non-Cooperative Localization of Industrial Internet of Things (IIoT),”ICT Express, vol. 9, no. 5, pp. 815–820, 2023

2023
[13]

Robust Multi-Target Localization in ISAC Systems: Leveraging Multidimensional Scaling,

R. A. Khalil and N. Saeed, “Robust Multi-Target Localization in ISAC Systems: Leveraging Multidimensional Scaling,”IEEE Open Journal of the Communications Society, vol. 5, pp. 3678–3689, 2024

2024
[14]

UA V Assisted Integrated Sensing and Communications for Internet of Things: 3D Trajectory Optimization and Resource Allocation,

Z. Liu, X. Liu, Y . Liu, V . C. M. Leung, and T. S. Durrani, “UA V Assisted Integrated Sensing and Communications for Internet of Things: 3D Trajectory Optimization and Resource Allocation,”IEEE Transactions on Wireless Communications, vol. 23, no. 8, pp. 8654–8667, Aug. 2024

2024
[2023]

Available: https://www.itu.int/rec/R-REC-M.2160-0- 202311-I/en

[Online]. Available: https://www.itu.int/rec/R-REC-M.2160-0- 202311-I/en

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Framework and Overall Objectives of the Future Development of IMT for 2030 and Beyond,

ITU-R, “Framework and Overall Objectives of the Future Development of IMT for 2030 and Beyond,” International Telecommunication Union, Radiocommunication Sector, Recommendation ITU-R M.2160-0, Nov

work page

[2] [2]

Integrated Sensing and Communications for IoT: Synergies with Key 6G Tech- nology Enablers,

A. Kaushik, R. Singh, M. Li, H. Luo, S. Dayarathna, R. Senanayake, X. An, R. A. Stirling-Gallacher, W. Shin, and M. D. Renzo, “Integrated Sensing and Communications for IoT: Synergies with Key 6G Tech- nology Enablers,”IEEE Internet of Things Magazine, vol. 7, no. 5, pp. 136–143, Sep. 2024

work page 2024

[3] [3]

RAN Rel-19 Status and a Look Beyond,

3GPP, “RAN Rel-19 Status and a Look Beyond,” 3GPP Technology Update, Apr. 2025, by W. Chen, 3GPP TSG RAN Chair. [Online]. Available: https://www.3gpp.org/technologies/ran-rel-19

work page 2025

[4] [4]

Isac Channel Models from ETSI and 3GPP,

M. Heggo, A. Shojaeifard, A. Mourad, C. Jiang, R. Liu, and J. Liu, “Isac Channel Models from ETSI and 3GPP,” inIEEE 36th International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), 2025, pp. 1–6

work page 2025

[5] [5]

Integrated Sensing and Communications (ISAC); Use Cases and Deployment Scenarios,

ETSI, “Integrated Sensing and Communications (ISAC); Use Cases and Deployment Scenarios,” European Telecommunications Standards Institute, Group Report ETSI GR ISC 001 V1.1.1, Mar. 2025. [Online]. Available: https://www.etsi.org/deliver/etsigr/ISC/001099/001/ 01.01.0160/grISC001v010101p.pdf

work page 2025

[6] [6]

Overview of AI/ML Related Work in 3GPP,

3GPP, “Overview of AI/ML Related Work in 3GPP,” 3GPP News and Events, 2025. [Online]. Available: https://www.3gpp.org/news-events/ 3gpp-news/ai-ml-2025

work page 2025

[7] [7]

An Overview of the 3GPP Study on Artificial Intelligence for 5G New Radio,

X. Lin, “An Overview of the 3GPP Study on Artificial Intelligence for 5G New Radio,” 2023. [Online]. Available: https://arxiv.org/abs/ 2308.05315

arXiv 2023

[8] [8]

AI/ML Life Cycle Management for Interoperable AI Native RAN,

C.-H. Huang, C.-K. Wen, and G. Y . Li, “AI/ML Life Cycle Management for Interoperable AI Native RAN,” 2025. [Online]. Available: https://arxiv.org/abs/2507.18538

arXiv 2025