pith. sign in

arxiv: 2603.28280 · v2 · pith:Q57ZXU7Wnew · submitted 2026-03-30 · 📡 eess.SP

Multimodal-NF: A Wireless Dataset for Near-Field Low-Altitude Sensing and Communications

Mengyuan Li , Qianfan Lu , Jiachen Tian , Hongjun Hu , Yu Han , Xiao Li , Chao-Kai Wen , Shi Jin This is my paper

Pith reviewed 2026-05-22 10:42 UTC · model grok-4.3

classification 📡 eess.SP
keywords near-field CSImultimodal datasetlow-altitude sensingXL-MIMOwireless communicationsspatial semanticsoverhead reduction6G networks
0
0 comments X

The pith

Multimodal sensor data reduces the search space for near-field wireless tasks in low-altitude settings.

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

The paper presents Multimodal-NF, a dataset that pairs high-fidelity near-field channel state information with RGB images, LiDAR point clouds, and GPS data. It establishes that these multimodal priors supply spatial semantics which shrink the near-field search space and cut overhead in sensing and communication. A sympathetic reader would care because current datasets focus on 2D far-field scenarios and fail to support the 3D low-altitude XL-MIMO systems expected in 6G networks. The work includes a generation framework and validates the approach with case studies on beam selection and line-of-sight classification.

Core claim

This letter introduces the Multimodal-NF dataset and generation framework for upper midband operation. It synchronizes near-field CSI with precise labels such as Top-5 beam indices and LoS/NLoS status alongside RGB, LiDAR, and GPS modalities. The central claim is that multimodal priors provide spatial semantics that reduce the near-field search space and thereby lower the overhead of wireless sensing and communication tasks.

What carries the argument

Multimodal priors that furnish spatial semantics to constrain the near-field search space.

If this is right

  • Lower overhead in acquiring channel state information for extremely large MIMO arrays.
  • Improved accuracy in selecting optimal beams using visual and depth cues.
  • More efficient detection of line-of-sight versus non-line-of-sight conditions.
  • Support for environment-aware operations in low-altitude unmanned aerial vehicle scenarios.

Where Pith is reading between the lines

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

  • Future systems could fuse these modalities in real time to adapt beamforming dynamically without exhaustive search.
  • The dataset generator might be extended to other frequency bands or incorporate additional sensors like radar.
  • Integration with machine learning models could further automate the overhead reduction process.

Load-bearing premise

The synthetic near-field CSI and its exact alignment with the sensory data faithfully represent real low-altitude environments and yield spatial semantics that actually cut computational or signaling overhead.

What would settle it

An experiment that applies the dataset to train a model for beam index prediction and shows no reduction in search overhead or accuracy compared to a baseline using only wireless data.

Figures

Figures reproduced from arXiv: 2603.28280 by Chao-Kai Wen, Hongjun Hu, Jiachen Tian, Mengyuan Li, Qianfan Lu, Shi Jin, Xiao Li, Yu Han.

Figure 1
Figure 1. Figure 1: Illustration of the low-altitude XL-MIMO system. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Example visualizations of (a) the LAE scene with [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Visualizations of (a) the Cartesian-domain channel [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Spatial-temporal variation of beam indices in (a) [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: System achievable rate comparison of the LLM [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
read the original abstract

Environment-aware 6G wireless networks demand the deep integration of multimodal and wireless data. However, most existing datasets are confined to 2D terrestrial far-field scenarios, lacking the 3D spatial context and near-field characteristics crucial for low-altitude extremely large-scale multiple-input multiple-output (XL-MIMO) systems. To bridge this gap, this letter introduces Multimodal-NF, a large-scale dataset and specialized generation framework. Operating in the upper midband, it synchronizes high-fidelity near-field channel state information (CSI) and precise wireless labels (e.g., Top-5 beam indices, LoS/NLoS) with comprehensive sensory modalities (RGB images, LiDAR point clouds, and GPS). Crucially, these multimodal priors provide spatial semantics that help reduce the near-field search space and thereby lower the overhead of wireless sensing and communication tasks. Finally, we validate the dataset through representative case studies, demonstrating its utility and effectiveness. The open-source generator and dataset are available at https://lmyxxn.github.io/6GXLMIMODatasets/.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 2 minor

Summary. The manuscript introduces Multimodal-NF, a large-scale dataset and generation framework for upper-midband near-field low-altitude XL-MIMO systems. It synchronizes simulated high-fidelity near-field CSI (including labels such as Top-5 beam indices and LoS/NLoS) with RGB images, LiDAR point clouds, and GPS data. The central claim is that these multimodal priors supply spatial semantics that reduce the near-field search space and thereby lower overhead for wireless sensing and communication tasks, with utility shown through representative case studies. The generator and dataset are released open-source.

Significance. If the generated CSI and labels accurately reproduce real low-altitude propagation, the dataset would address a clear gap in existing 2D far-field terrestrial collections and support research on environment-aware 6G systems. The open-source release of both generator and data is a concrete strength that enables reproducibility and follow-on work.

major comments (2)
  1. [Case Studies] Case Studies section: the claim that multimodal priors reduce near-field search space and overhead is presented as validated by representative case studies, yet no quantitative metrics (e.g., beam-search reduction factor, overhead savings, or comparison against unimodal baselines) are reported; without these numbers the central utility argument remains unverified.
  2. [Dataset Generation] Dataset Generation / Simulation Framework: the high-fidelity near-field CSI is produced by simulation; the manuscript provides no side-by-side comparison against over-the-air measurements collected in low-altitude conditions, leaving open whether spherical-wave effects, multipath structure, and dynamic scattering are faithfully reproduced—directly load-bearing for the reliability of the supplied spatial semantics.
minor comments (2)
  1. [Abstract] Abstract and introduction: the carrier frequency or exact upper-midband range used for the XL-MIMO simulations is not stated, which affects reproducibility of the near-field regime.
  2. [Dataset Description] Dataset description: the total number of synchronized samples, number of distinct scenarios, and precise temporal/spatial alignment procedure between CSI and sensory modalities should be tabulated for clarity.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments, which highlight important aspects for strengthening the manuscript. We respond to each major comment below and indicate planned revisions.

read point-by-point responses

  1. Referee: [Case Studies] Case Studies section: the claim that multimodal priors reduce near-field search space and overhead is presented as validated by representative case studies, yet no quantitative metrics (e.g., beam-search reduction factor, overhead savings, or comparison against unimodal baselines) are reported; without these numbers the central utility argument remains unverified.

    Authors: We agree that the case studies as currently presented are primarily illustrative and do not supply the quantitative metrics needed to fully verify the overhead-reduction claim. In the revised manuscript we will add explicit quantitative evaluations, including beam-search reduction factors, overhead savings percentages, and direct comparisons against unimodal baselines, using the released dataset. revision: yes

  2. Referee: [Dataset Generation] Dataset Generation / Simulation Framework: the high-fidelity near-field CSI is produced by simulation; the manuscript provides no side-by-side comparison against over-the-air measurements collected in low-altitude conditions, leaving open whether spherical-wave effects, multipath structure, and dynamic scattering are faithfully reproduced—directly load-bearing for the reliability of the supplied spatial semantics.

    Authors: The dataset is generated through a controlled simulation framework that incorporates spherical-wave near-field models and ray-tracing for low-altitude environments to ensure reproducibility and coverage of diverse conditions. We will expand the manuscript with additional details on the underlying propagation models, parameter settings, and theoretical validation steps. A direct empirical side-by-side comparison with over-the-air measurements, however, lies outside the present scope. revision: partial

standing simulated objections not resolved
  • Direct side-by-side comparison against over-the-air measurements in low-altitude XL-MIMO settings, which would require new experimental campaigns beyond the resources of this work.

Circularity Check

0 steps flagged

No circularity: dataset paper with no derivation chain or fitted predictions

full rationale

The manuscript introduces Multimodal-NF as a generated dataset and open-source framework that synchronizes simulated near-field CSI with RGB, LiDAR, and GPS modalities. No equations, first-principles derivations, parameter fitting, or predictions appear in the provided text. The claim that multimodal priors reduce near-field search space is presented as a motivating utility statement and is illustrated via case studies rather than derived from any internal model or self-referential input. Because the work contains no load-bearing mathematical steps that could reduce to their own definitions or fitted values, the circularity score is zero and the derivation (such as it is) is self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based solely on the abstract, the paper does not explicitly list or rely on additional free parameters, axioms, or invented entities beyond standard wireless channel modeling assumptions.

pith-pipeline@v0.9.0 · 5740 in / 1133 out tokens · 65785 ms · 2026-05-22T10:42:43.693788+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

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

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

18 extracted references · 18 canonical work pages · 1 internal anchor

  1. [1]

    Low-altitude UA V position prediction- assisted near-field adaptive beamwidth control for XL- MIMO systems,

    Y. Wu et al., “Low-altitude UA V position prediction- assisted near-field adaptive beamwidth control for XL- MIMO systems,” IEEE Internet Things J., vol. 13, no. 4, pp. 5477–5490, Feb. 2026

    work page 2026

  2. [2]

    Integrated sensing and communication for low altitude economy: Opportunities and challenges,

    Y. Jiang et al., “Integrated sensing and communication for low altitude economy: Opportunities and challenges,” IEEE Commun. Mag., vol. 63, no. 12, pp. 72–78, 2025

    work page 2025

  3. [3]

    New midband for 6G: Several con- siderations from the channel propagation characteristics perspective,

    J. Zhang et al., “New midband for 6G: Several con- siderations from the channel propagation characteristics perspective,” IEEE Commun. Mag., vol. 63, no. 1, pp. 175–180, Jan. 2025

    work page 2025

  4. [4]

    Mid-band extra large-scale MIMO sys- tem: Channel modeling and performance analysis,

    J. Tian et al., “Mid-band extra large-scale MIMO sys- tem: Channel modeling and performance analysis,” IEEE Trans. Commun., vol. 73, no. 2, pp. 1025–1041, Feb. 2025

    work page 2025

  5. [5]

    ComAI: The convergence of communica- tion and artificial intelligence,

    P. Zhang et al., “ComAI: The convergence of communica- tion and artificial intelligence,” IEEE Commun. Surv. & Tut., vol. 28, pp. 2163–2197, 2026

    work page 2026

  6. [6]

    DeepSense 6G: A large-scale real- world multi-modal sensing and communication dataset,

    A. Alkhateeb et al., “DeepSense 6G: A large-scale real- world multi-modal sensing and communication dataset,” IEEE Commun. Mag., vol. 61, no. 9, pp. 122–128, Sept. 2023

    work page 2023

  7. [7]

    DeepMIMO: A generic deep learning dataset for millimeter wave and massive MIMO applications,

    ——, “DeepMIMO: A generic deep learning dataset for millimeter wave and massive MIMO applications,” in Proc. Inf. Theory Appl. Workshop (ITA), 2019, pp. 1– 8. 5

    work page 2019

  8. [8]

    BUPTCMCC-6G-DataAI+: A generative channel dataset for 6G AI air-interface research,

    L. Yu et al., “BUPTCMCC-6G-DataAI+: A generative channel dataset for 6G AI air-interface research,” Sci. China Inf. Sci., vol. 68, no. 9, p. 197301, Sep. 2025, doi: 10.1007/s11432-024-4445-0

    work page doi:10.1007/s11432-024-4445-0 2025

  9. [9]

    The LuViRA dataset: Synchronized vision, radio, and audio sensors for indoor localization,

    O. Yaman et al., “The LuViRA dataset: Synchronized vision, radio, and audio sensors for indoor localization,” in Proc. IEEE Int. Conf. Robot. Autom. (ICRA), 2024, pp. 11 920–11 926

    work page 2024

  10. [10]

    5G MIMO data for machine learning: Application to beam-selection using deep learning,

    A. Klautau et al., “5G MIMO data for machine learning: Application to beam-selection using deep learning,” in Proc. Inf. Theory Appl. Workshop (ITA), 2018, pp. 1– 9

    work page 2018

  11. [11]

    CA VIAR: Co-simulation of 6G commu- nications, 3-d scenarios, and AI for digital twins,

    J. Borges et al., “CA VIAR: Co-simulation of 6G commu- nications, 3-d scenarios, and AI for digital twins,” IEEE Internet Things J., vol. 11, no. 19, pp. 31 287–31 300, 2024

    work page 2024

  12. [12]

    Multimodal-Wireless: A large-scale dataset for sensing and communication,

    T. Mao et al., “Multimodal-Wireless: A large-scale dataset for sensing and communication,” arXiv preprint arXiv:2511.03220, 2025

    work page arXiv 2025

  13. [13]

    Effects of building materials and structures on ra- diowave propagation above about 100 MHz,

    ITU-R, “Effects of building materials and structures on ra- diowave propagation above about 100 MHz,” International Telecommunication Union (ITU), Geneva, Switzerland, Recommendation ITU-R P.2040-3, Aug. 2023. [Online]. A vailable:https://www.itu.int/rec/R-REC-P.2040/en

    work page 2040

  14. [14]

    Sionna RT: Differentiable ray tracing for radio propagation modeling,

    J. Hoydis et al., “Sionna RT: Differentiable ray tracing for radio propagation modeling,” in Proc. IEEE Globecom Workshops (GC Wkshps), Kuala Lumpur, Malaysia, Dec. 2023, pp. 317–321

    work page 2023

  15. [15]

    Open3D: A Modern Library for 3D Data Processing

    Q.-Y. Zhou, J. Park, and V. Koltun, “Open3D: A modern library for 3D data processing,” arXiv:1801.09847, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  16. [16]

    Keypoint detection empowered near-field user localization and channel reconstruction,

    M. Li et al., “Keypoint detection empowered near-field user localization and channel reconstruction,” IEEE Trans. Wireless Commun., vol. 24, no. 7, pp. 5664–5677, Jul. 2025

    work page 2025

  17. [17]

    Structure-aware multimodal LLM framework for trustworthy near-field beam prediction,

    ——, “Structure-aware multimodal LLM framework for trustworthy near-field beam prediction,” arXiv preprint arXiv:2603.16143, 2026. [Online]. A vailable: https://arxiv.org/abs/2603.16143

    work page arXiv 2026

  18. [18]

    Two-stage hierarchical beam training for near-field communications,

    C. Wu et al., “Two-stage hierarchical beam training for near-field communications,” IEEE Trans. Veh. Technol., vol. 73, no. 2, pp. 2032–2044, Feb. 2024. 6

    work page 2032