arxiv: 2605.07163 · v1 · submitted 2026-05-08 · 📡 eess.SP

Recognition: 2 theorem links

· Lean Theorem

Towards Intelligent Low-Altitude Wireless Network Deployment: Differentiable Channel Knowledge Map Construction and Trajectory Design

Le Zhao , Zesong Fei , Wenge Shi , Xinyi Wang , Jingxuan Huang , Jihao Luo , Yong Zeng

Authors on Pith no claims yet

Pith reviewed 2026-05-11 02:06 UTC · model grok-4.3

classification 📡 eess.SP

keywords channel knowledge mapdifferentiable CKMUAV trajectory optimizationlow-altitude wireless networksconditional neural networksjoint resource allocationcKANCNN feature encoding

0 comments

The pith

A neural network builds differentiable channel knowledge maps from continuous UAV locations and environmental features to jointly optimize power, bandwidth, and trajectories in low-altitude networks.

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

The paper develops a method to build channel knowledge maps that can be differentiated with respect to continuous UAV locations by using a convolutional neural network to extract environmental features and then a conditional multilayer perceptron or Kolmogorov-Arnold network to predict channel gains. This allows the maps to be used directly in optimization problems for joint power, bandwidth, and trajectory design in multi-UAV low-altitude wireless networks. The approach achieves higher prediction accuracy than feature-less methods and delivers higher minimum throughput than traditional statistical channel models. Sympathetic readers would care because it moves toward data-driven, location-aware planning that could make UAV network deployment more efficient and reliable.

Core claim

We propose a location-oriented CKM construction method that directly maps continuous spatial coordinates to channel gain. In particular, a shared convolutional neural network is employed to encode high-level environmental features from conditional inputs. These features are then sampled based on location information to form a fused regressor-conditional multilayer perceptron or conditional Kolmogorov-Arnold network for channel gain prediction. We further propose a joint power, bandwidth, and trajectory optimization method for multi-UAV systems, with the constructed differentiable CKM employed to evaluate the communication performance, solved via alternating optimization and successive convex

What carries the argument

The location-oriented differentiable CKM, built by encoding environmental features with a shared CNN and predicting gains via c-MLP or cKAN from sampled location features, which provides differentiable channel information for optimization.

If this is right

Enables location-aware differentiability of the CKM for continuous UAV positions.
Achieves higher accuracy in channel prediction than methods without environmental features.
The CKM-JPBTO method yields significantly higher minimum throughput than statistical channel model-based approaches.
Supports efficient deployment of low-altitude wireless networks with multiple UAVs.

Where Pith is reading between the lines

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

This framework could allow real-time updates to UAV paths based on live environmental data without retraining.
The use of conditional networks might generalize to predicting other wireless metrics like interference levels.
In practice, it could lower the cost of network planning by relying more on learned models than exhaustive simulations.
Extensions might include handling moving obstacles or time-varying channels in the CKM construction.

Load-bearing premise

The neural network-based CKM accurately predicts channel gains for any continuous location using the encoded environmental features, and the alternating optimization reliably finds a high-quality solution to the non-convex joint optimization problem.

What would settle it

Measuring actual channel gains at several arbitrary continuous UAV locations in a real or simulated environment and comparing them to the CKM predictions, or running the JPBTO and comparing the achieved minimum throughput against a statistical model baseline.

Figures

Figures reproduced from arXiv: 2605.07163 by Jihao Luo, Jingxuan Huang, Le Zhao, Wenge Shi, Xinyi Wang, Yong Zeng, Zesong Fei.

**Figure 2.** Figure 2: The illustration of the proposed model-data dual-accelerated CKM construction architecture. The CNN is employed to encode the multi-channel input [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: Visualization of building distribution in simulation. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

**Figure 4.** Figure 4: Visual comparison of CKM predictions using: (a) MLP [ [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗

**Figure 5.** Figure 5: NMSE of reconstructed CKM with different channel gain measure ˜ [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗

**Figure 6.** Figure 6: Convergence of CKM-JPBTO with Pmax = 10 W, Bmax = 10 MHz, , N = 50, and T = 100 s, by |Ds |/|D| ˜ = 2% sampled cKAN. Table IV. EMPIRICAL LATENCY COMPARISON Operation Stage Method Latency (s) CKM Inference RadioUNet [20] 0.058 cKAN (Ours) 0.035 Gradient Computation Backpropagation 0.028 Optimization (Per Iteration) GWO-JPBTO 0.888 CKM-JPBTO 0.293 suffer from limited accuracy as they fail to capture complex … view at source ↗

**Figure 9.** Figure 9: The trajectories, power, bandwidth, and communication rate optimized [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗

**Figure 10.** Figure 10: Minimum achievable rate under different transmit power [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗

**Figure 11.** Figure 11: Minimum achievable rate under different bandwidth [PITH_FULL_IMAGE:figures/full_fig_p012_11.png] view at source ↗

**Figure 12.** Figure 12: Minimum achievable rate under different timeslot number [PITH_FULL_IMAGE:figures/full_fig_p012_12.png] view at source ↗

read the original abstract

Channel knowledge map (CKM) has emerged as a promising technique to leverage prior propagation knowledge in low-altitude wireless networks (LAWNs), yet state-of-the-art grid-based CKM construction methods struggle to support efficient LAWN deployment due to their lack of differentiability with respect to continuous locations of unmanned aerial vehicles (UAVs). To overcome this limitation, we propose a differentiable CKM-triggered trajectory optimization framework for LAWNs. Firstly, we propose a location-oriented CKM construction method that directly maps continuous spatial coordinates to channel gain. In particular, a shared convolutional neural network (CNN) is employed to encode high-level environmental features from conditional inputs. These features are then sampled based on location information to form a fused regressor-conditional multilayer perceptron (c-MLP) or conditional Kolmogorov-Arnold network (cKAN)-for channel gain prediction. We further propose a joint power, bandwidth, and trajectory optimization (JPBTO) method for multi-UAV systems, with the constructed differentiable CKM employed to evaluate the communication performance. The formulated non-convex problem is solved via alternating optimization and successive convex approximation. Numerical results show that the proposed framework enables location-aware differentiability of the CKM, while achieving higher accuracy than the methods without environmental features. Furthermore, the proposed CKM-JPBTO achieves a significantly higher minimum throughput than the conventional statistical channel model-based JPBTO.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note private letter to a colleague

The paper builds a differentiable CKM with a shared CNN for environmental features feeding into conditional MLP or KAN regressors, then folds that into joint power-bandwidth-trajectory optimization for multi-UAV systems, but the numerical claims rest on limited validation details.

read the letter

The new piece here is the location-oriented CKM that directly takes continuous coordinates and stays differentiable. A shared CNN encodes environmental features from conditional inputs, those features get sampled at the UAV position, and a c-MLP or cKAN turns them into channel-gain predictions. That map then supplies the performance metric inside an alternating optimization with successive convex approximation for the joint power, bandwidth, and trajectory problem in multi-UAV LAWNs. The architecture is a clear step past grid-based CKM methods that lose differentiability at non-grid points, and the integration into JPBTO is straightforward and technically coherent. Numerical results are reported to show higher accuracy than feature-free baselines and a clear lift in minimum throughput over statistical channel models. That combination is the useful part for anyone already working on UAV trajectory design or location-aware channel modeling. The main limitation is the thin experimental backing. The abstract states accuracy and throughput gains but supplies no training-set size, test-location distribution, error bars, ablation on the CNN versus regressor choice, or checks on how the model behaves outside the training support. Without those, the generalization claim for arbitrary continuous locations stays unverified, and the reported throughput improvement is conditional on the CKM staying accurate wherever the SCA solver evaluates it. The optimization itself is standard SCA, so any gains inherit the same dependence on the learned map. This work is aimed at the wireless-communications community focused on low-altitude networks and differentiable channel models. A reader already familiar with CKM or UAV optimization would pick up the architecture and the joint formulation quickly. It is coherent enough on its own terms to merit a serious referee, even though the experiments need more detail before the performance claims can be taken as settled.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes a differentiable channel knowledge map (CKM) construction technique for low-altitude wireless networks that directly maps continuous UAV spatial coordinates to channel gains. A shared CNN encodes high-level environmental features from conditional inputs; these are location-sampled and fed to a fused c-MLP or cKAN regressor. The resulting differentiable CKM is embedded in a joint power-bandwidth-trajectory optimization (JPBTO) problem for multi-UAV systems, which is solved by alternating optimization combined with successive convex approximation. Numerical results are reported to show higher prediction accuracy than methods lacking environmental features and significantly higher minimum throughput than conventional statistical-channel-model-based JPBTO.

Significance. If the generalization and optimization claims hold, the work would provide a practical route to gradient-based trajectory design in LAWNs that exploits location-specific propagation knowledge, potentially improving spectral efficiency over purely statistical models. The explicit construction of a location-aware differentiable CKM and the exploration of cKAN regressors are technically interesting contributions that could influence future environment-aware network planning.

major comments (3)

[Numerical results] Numerical results section: the abstract states higher accuracy and significantly higher minimum throughput, yet no information is supplied on training-set size, spatial distribution of training versus test locations, validation metrics (RMSE/MAE) with error bars, or ablation studies isolating the contribution of the CNN-encoded environmental features versus the choice of c-MLP versus cKAN. Without these, the central performance claims remain only partially supported.
[Proposed CKM construction and JPBTO] CKM construction and JPBTO sections: the headline performance gains rest on the neural CKM supplying accurate channel gains (and their gradients) at arbitrary continuous locations queried by the trajectory optimizer. No out-of-distribution tests, sensitivity analysis to distribution shift, or verification that prediction error remains bounded along optimized trajectories are provided; any reported throughput improvement is therefore conditional on unverified generalization.
[JPBTO formulation and solution] Optimization section: the non-convex JPBTO is addressed by alternating optimization and successive convex approximation, but the manuscript supplies neither convergence analysis nor results from multiple random initializations to establish that the obtained solutions are reliably high-quality rather than sensitive to initialization or approximation error.

minor comments (2)

[CKM construction] Notation for the fused regressor-conditional MLP (c-MLP) and cKAN could be clarified with an explicit diagram or pseudocode showing how location sampling interfaces with the CNN feature map.
[Numerical results] Figure captions and axis labels in the numerical results should explicitly state the number of Monte-Carlo runs and the precise definition of 'minimum throughput' (e.g., per-user or system-wide).

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We address each major point below and have revised the manuscript to incorporate additional details, analyses, and clarifications that strengthen the support for our claims.

read point-by-point responses

Referee: [Numerical results] Numerical results section: the abstract states higher accuracy and significantly higher minimum throughput, yet no information is supplied on training-set size, spatial distribution of training versus test locations, validation metrics (RMSE/MAE) with error bars, or ablation studies isolating the contribution of the CNN-encoded environmental features versus the choice of c-MLP versus cKAN. Without these, the central performance claims remain only partially supported.

Authors: We agree that these details are essential for rigorously supporting the performance claims. In the revised manuscript, we have expanded the Numerical Results section to specify the training set size (10,000 samples generated via ray-tracing), the spatial distribution (uniform random sampling over the 1 km × 1 km area with an 80/20 train/test split), and validation metrics (RMSE and MAE reported with error bars from five independent runs). We have also added ablation studies that isolate the CNN environmental encoder's contribution and compare c-MLP versus cKAN regressors, confirming their respective impacts on accuracy. revision: yes
Referee: [Proposed CKM construction and JPBTO] CKM construction and JPBTO sections: the headline performance gains rest on the neural CKM supplying accurate channel gains (and their gradients) at arbitrary continuous locations queried by the trajectory optimizer. No out-of-distribution tests, sensitivity analysis to distribution shift, or verification that prediction error remains bounded along optimized trajectories are provided; any reported throughput improvement is therefore conditional on unverified generalization.

Authors: We acknowledge that explicit verification of generalization is important for the differentiable CKM. The revised manuscript now includes out-of-distribution tests on unseen location sets with altered environmental parameters. We have added sensitivity analysis to distribution shifts (e.g., changes in building density and height) and verified that prediction error remains bounded along the optimized trajectories by evaluating CKM outputs at each iteration of the JPBTO solver. These additions confirm that the reported throughput gains are not conditional on unverified assumptions. revision: yes
Referee: [JPBTO formulation and solution] Optimization section: the non-convex JPBTO is addressed by alternating optimization and successive convex approximation, but the manuscript supplies neither convergence analysis nor results from multiple random initializations to establish that the obtained solutions are reliably high-quality rather than sensitive to initialization or approximation error.

Authors: We thank the referee for highlighting this gap. The revised manuscript includes a convergence analysis of the alternating optimization procedure, showing that the objective stabilizes within approximately 25 iterations on average across scenarios. We have also added results from 10 independent runs with different random initializations of the UAV trajectories and resource allocations, reporting the mean and standard deviation of the achieved minimum throughput to demonstrate that the solutions are robust and not highly sensitive to initialization or approximation errors. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The paper constructs a differentiable CKM via a CNN encoder for environmental features followed by location-sampled c-MLP or cKAN regressors trained on data, then feeds the resulting differentiable map into a standard alternating optimization + SCA solver for the JPBTO problem. All performance claims (accuracy gains, throughput improvements) rest on numerical simulations against external baselines rather than any quantity defined by construction from fitted parameters or prior self-citations. No self-definitional loops, fitted inputs renamed as predictions, or load-bearing self-citation chains appear in the derivation.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the assumption that neural networks can learn accurate differentiable mappings from location and environmental features to channel gains, and that the non-convex optimization can be effectively handled by alternating optimization and successive convex approximation.

free parameters (1)

Neural network weights
Parameters of the shared CNN, c-MLP, and cKAN that are fitted during training to enable channel gain prediction.

axioms (2)

domain assumption A shared CNN can extract high-level environmental features from conditional inputs that are useful for channel prediction across locations.
Invoked in the location-oriented CKM construction method.
domain assumption The formulated non-convex JPBTO problem can be solved to a good local optimum via alternating optimization and successive convex approximation.
Used to obtain the reported performance gains.

pith-pipeline@v0.9.0 · 5573 in / 1518 out tokens · 35671 ms · 2026-05-11T02:06:47.704849+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

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

location-oriented CKM construction method that directly maps continuous spatial coordinates to channel gain... fused regressor-conditional multilayer perceptron (c-MLP) or conditional Kolmogorov-Arnold network (cKAN)
IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

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

gradient of the CKM with respect to q... chain rule... B-spline basis functions

What do these tags mean?

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

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages · 1 internal anchor

[1]

Dif- ferentiable channel knowledge map reconstruction via kolmogorov- arnold networks,

L. Zhao, S. Zhao, X. Wang, J. Huang, Z. Fei, and Y . Zeng, “Dif- ferentiable channel knowledge map reconstruction via kolmogorov- arnold networks,” inIEEE Int. Conf. Wireless Commun. Signal Process. (WCSP), Chongqing, China, 2025, Accepted

work page 2025
[2]

Accessing from the sky: A tutorial on UA V communications for 5G and beyond,

Y . Zeng, Q. Wu, and R. Zhang, “Accessing from the sky: A tutorial on UA V communications for 5G and beyond,”Proceedings of the IEEE, vol. 107, no. 12, pp. 2327–2375, Dec. 2019

work page 2019
[3]

Air-ground integrated sensing and communications: Opportunities and challenges,

Z. Fei, X. Wang, N. Wu, J. Huang, and J. A. Zhang, “Air-ground integrated sensing and communications: Opportunities and challenges,” IEEE Communications Magazine, vol. 61, no. 5, pp. 55–61, 2023

work page 2023
[4]

Joint optimization of trajectory, propulsion, and thrust powers for covert UA V-on-UA V video tracking and surveillance,

S. Hu, W. Ni, X. Wang, A. Jamalipour, and D. Ta, “Joint optimization of trajectory, propulsion, and thrust powers for covert UA V-on-UA V video tracking and surveillance,”IEEE Trans. Inf. F orensics Secur ., vol. 16, pp. 1959–1972, 2021

work page 1959
[5]

Energy management and trajectory optimization for UA V-enabled legitimate monitoring systems,

S. Hu, Q. Wu, and X. Wang, “Energy management and trajectory optimization for UA V-enabled legitimate monitoring systems,”IEEE Trans. Wireless Commun., vol. 20, no. 1, pp. 142–155, Jan. 2021

work page 2021
[6]

Unmanned aerial vehicle-aided intelligent transportation systems: Vision, challenges, and opportunities,

A. Telikani, A. Sarkar, B. Du, F. Santoso, J. Shen, J. Yan, J. Yong, and E. Yap, “Unmanned aerial vehicle-aided intelligent transportation systems: Vision, challenges, and opportunities,”IEEE Commun. Surveys Tuts., pp. 1–1, 2025

work page 2025
[8]

A novel 3D UA V channel model for A2G communication environments using AoD and AoA estimation algorithms,

H. Jiang, Z. Zhang, C.-X. Wang, J. Zhang, J. Dang, L. Wu, and H. Zhang, “A novel 3D UA V channel model for A2G communication environments using AoD and AoA estimation algorithms,”IEEE Trans. on Commun., vol. 68, no. 11, pp. 7232–7246, 2020

work page 2020
[9]

Energy-efficient UA V communication with trajectory optimization,

Y . Zeng and R. Zhang, “Energy-efficient UA V communication with trajectory optimization,”IEEE Trans. on Wireless Commun., vol. 16, no. 6, pp. 3747–3760, 2017

work page 2017
[10]

Joint trajectory and communication design for multi-UA V enabled wireless networks,

Q. Wu, Y . Zeng, and R. Zhang, “Joint trajectory and communication design for multi-UA V enabled wireless networks,”IEEE Trans. on Wireless Commun., vol. 17, no. 3, pp. 2109–2121, 2018

work page 2018
[11]

Hybrid offline-online design for UA V-enabled data harvesting in probabilistic LoS channels,

C. You and R. Zhang, “Hybrid offline-online design for UA V-enabled data harvesting in probabilistic LoS channels,”IEEE Trans. on Wireless Commun., vol. 19, no. 6, pp. 3753–3768, 2020

work page 2020
[12]

Ultra reliable UA V communication using altitude and cooperation diversity,

M. M. Azari, F. Rosas, K.-C. Chen, and S. Pollin, “Ultra reliable UA V communication using altitude and cooperation diversity,”IEEE Trans. Commun., vol. 66, no. 1, pp. 330–344, Jan. 2018

work page 2018
[13]

Multi-objective deployment optimization of uavs for energy-efficient wireless coverage,

X. Zhu, L. Zhai, N. Li, Y . Li, and F. Yang, “Multi-objective deployment optimization of uavs for energy-efficient wireless coverage,”IEEE Trans. on Commun., vol. 72, no. 6, pp. 3587–3601, 2024

work page 2024
[14]

Grey wolf optimizer,

S. Mirjalili, S. M. Mirjalili, and A. Lewis, “Grey wolf optimizer,” Advances in Engineering Software, vol. 69, pp. 46–61, 2014

work page 2014
[15]

Toward environment-aware 6G communications via channel knowledge map,

Y . Zeng and X. Xu, “Toward environment-aware 6G communications via channel knowledge map,”IEEE Wireless Commun., vol. 28, no. 3, pp. 84–91, June 2021

work page 2021
[16]

A tutorial on environment-aware communications via channel knowledge map for 6G,

Y . Zeng, J. Chen, J. Xu, D. Wu, X. Xu, S. Jin, X. Gao, D. Gesbert, S. Cui, and R. Zhang, “A tutorial on environment-aware communications via channel knowledge map for 6G,”IEEE Commun. Surveys Tuts., vol. 26, no. 3, pp. 1478–1519, Feb. 2024

work page 2024
[17]

An adaptablek-nearest neighbors algorithm for MMSE image interpolation,

K. S. Ni and T. Q. Nguyen, “An adaptablek-nearest neighbors algorithm for MMSE image interpolation,”IEEE Trans. Image Process., vol. 18, no. 9, pp. 1976–1987, May 2009

work page 1976
[18]

Fixed rank kriging for cellular coverage analysis,

H. Braham, S. B. Jemaa, G. Fort, E. Moulines, and B. Sayrac, “Fixed rank kriging for cellular coverage analysis,”IEEE Trans. V eh. Technol., vol. 66, no. 5, pp. 4212–4222, Aug. 2017

work page 2017
[19]

How much data is needed for channel knowledge map construction?

X. Xu and Y . Zeng, “How much data is needed for channel knowledge map construction?”IEEE Trans. Wireless Commun., vol. 23, no. 10, pp. 13 011–13 021, May 2024

work page 2024
[20]

RadioUNet: Fast radio map estimation with convolutional neural networks,

R. Levie, c. Yapar, G. Kutyniok, and G. Caire, “RadioUNet: Fast radio map estimation with convolutional neural networks,”IEEE Trans. Wireless Commun., vol. 20, no. 6, pp. 4001–4015, Feb. 2021

work page 2021
[21]

RadioGAT: A joint model-based and data-driven framework for multi-band radiomap reconstruction via graph attention networks,

X. Li, S. Zhang, H. Li, X. Li, L. Xu, H. Xu, H. Mei, G. Zhu, N. Qi, and M. Xiao, “RadioGAT: A joint model-based and data-driven framework for multi-band radiomap reconstruction via graph attention networks,” IEEE Trans. Wireless Commun., vol. 23, no. 11, pp. 17 777–17 792, Sep. 2024

work page 2024
[22]

Fast and accurate cooperative radio map estimation enabled by GAN,

Z. Zhang, G. Zhu, J. Chen, and S. Cui, “Fast and accurate cooperative radio map estimation enabled by GAN,” inIEEE Int. Conf. Commun. Workshops (ICC Workshops), 2024, pp. 1641–1646

work page 2024
[23]

CKMDiff: A generative diffusion model for CKM construction via inverse problems with learned priors,

S. Fu, Y . Zeng, Z. Wu, D. Wu, S. Jin, C.-X. Wang, and X. Gao, “CKMDiff: A generative diffusion model for CKM construction via inverse problems with learned priors,” 2025. [Online]. Available: https://arxiv.org/abs/2504.17323

work page arXiv 2025
[24]

RadioDiff: An effective generative diffusion model for sampling-free dynamic radio map construction,

X. Wang, K. Tao, N. Cheng, Z. Yin, Z. Li, Y . Zhang, and X. Shen, “RadioDiff: An effective generative diffusion model for sampling-free dynamic radio map construction,”IEEE Trans. Cogn. Commun. Netw., 2024

work page 2024
[25]

3D-RadioDiff: An altitude-conditioned diffusion model for 3D radio map construction,

L. Zhao, Z. Fei, X. Wang, J. Luo, and Z. Zheng, “3D-RadioDiff: An altitude-conditioned diffusion model for 3D radio map construction,” IEEE Wireless Commun. Lett., vol. 14, no. 7, pp. 1969–1973, April 2025

work page 1969
[26]

RadioDiff-3D: A 3D× 3D radio map dataset and generative diffusion based benchmark for 6g environment-aware communication,

X. Wang, Q. Zhang, N. Cheng, J. Chen, Z. Zhang, Z. Li, S. Cui, and X. Shen, “RadioDiff-3D: A 3D× 3D radio map dataset and generative diffusion based benchmark for 6g environment-aware communication,” IEEE Trans. on Netw. Sci. and Eng., pp. 1–18, 2025, doi:10.1109/TNSE. 2025.3590545

work page doi:10.1109/tnse 2025
[27]

Aerial base station placement via propagation radio maps,

D. Romero, P. Q. Viet, and R. Shrestha, “Aerial base station placement via propagation radio maps,”IEEE Trans. on Commun., vol. 72, no. 9, pp. 5349–5364, 2024

work page 2024
[28]

Radio map-based 3D path planning for cellular- connected UA V,

S. Zhang and R. Zhang, “Radio map-based 3D path planning for cellular- connected UA V,”IEEE Trans. Wireless Commun., vol. 20, no. 3, pp. 1975–1989, Mar. 2021

work page 1975
[29]

Aerial video streaming over 3D cellular networks: An environment and channel knowledge map approach,

C. Zhan, H. Hu, Z. Liu, J. Wang, N. Cheng, and S. Mao, “Aerial video streaming over 3D cellular networks: An environment and channel knowledge map approach,”IEEE Trans. Wireless Commun., vol. 23, no. 2, pp. 1432–1446, Feb. 2024

work page 2024
[30]

Energy minimization for cellular-connected UA V: from optimization to deep reinforcement learning,

C. Zhan and Y . Zeng, “Energy minimization for cellular-connected UA V: from optimization to deep reinforcement learning,”IEEE Trans. Wireless Commun., vol. 21, no. 7, pp. 5541–5555, Jul. 2022

work page 2022
[31]

Simultaneous navigation and radio mapping for cellular-connected UA V with deep reinforcement learning,

Y . Zeng, X. Xu, S. Jin, and R. Zhang, “Simultaneous navigation and radio mapping for cellular-connected UA V with deep reinforcement learning,”IEEE Trans. on Wireless Commun., vol. 20, no. 7, pp. 4205– 4220, 2021

work page 2021
[32]

Channel Knowledge Map for Environment-Aware Communications: EM Algorithm for Map Construc- tion,

K. Li, P. Li, Y . Zeng, and J. Xu, “Channel Knowledge Map for Environment-Aware Communications: EM Algorithm for Map Construc- tion,” inIEEE Wireless Commun. Netw. Conf. (WCNC), Austin, TX, USA, 2022, pp. 1659–1664

work page 2022
[33]

Learning radio maps for UA V- aided wireless networks: A segmented regression approach,

J. Chen, U. Yatnalli, and D. Gesbert, “Learning radio maps for UA V- aided wireless networks: A segmented regression approach,” inIEEE Int. Conf. Commun. (ICC), Paris, France, 2017, pp. 1–6

work page 2017
[34]

Indoor radio map construction and localization with deep Gaussian processes,

X. Wang, X. Wang, S. Mao, J. Zhang, S. Periaswamy, and J. Patton, “Indoor radio map construction and localization with deep Gaussian processes,”IEEE Internet Things J., vol. 7, no. 11, pp. 11 238–11 249, Nov. 2020

work page 2020
[35]

Joint trajectory and transmit power design for cellular-connected UA Vs via differentiable channel knowledge map,

Y . Li, X. Wang, Z. Zheng, J. Guo, and Z. Fei, “Joint trajectory and transmit power design for cellular-connected UA Vs via differentiable channel knowledge map,”IEEE Trans. V eh. Technol., pp. 1–16, May 2025, doi:10.1109/TVT.2025.3567741

work page doi:10.1109/tvt.2025.3567741 2025
[36]

KAN: Kolmogorov-Arnold Networks

Z. Liu, Y . Wang, S. Vaidya, F. Ruehle, J. Halverson, M. Solja ˇci´c, T. Y . Hou, and M. Tegmark, “KAN: Kolmogorov-arnold networks,” 2025. [Online]. Available: https://arxiv.org/abs/2404.19756

work page internal anchor Pith review arXiv 2025
[37]

A survey of efficient ray-tracing techniques for mobile radio propagation analysis,

T. Imai, “A survey of efficient ray-tracing techniques for mobile radio propagation analysis,”IEICE Trans. on Commun., vol. 100, no. 5, pp. 666–679, 2017

work page 2017
[38]

Wireless InSite®,

REMCOM, Inc., “Wireless InSite®,” https://www.remcom.com/ wireless-insite-em-propagation-software

work page