Pith. sign in

REVIEW 3 major objections 6 minor 39 references

GPS-Aided Deep Learning for Beam Prediction and Tracking in UAV mmWave Communication

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read GPS-only deep learning predicts drone beams three steps ahead.

desk verdict Useful GPS-only beam prediction extension, but the adjusted-splitting protocol leaks trajectory-level temporal information into the test set, so the headline accuracy and power-loss numbers are not demonstrated. read the letter →

arxiv 2505.17530 v2 pith:HINR5FCL submitted 2025-05-23 eess.SP cs.ITcs.LGmath.IT

classification eess.SPcs.ITcs.LGmath.IT
keywords millimeter-wavecommunicationbeampredictiontrackingUAVcommunicationsGPS-aideddeeplearningSense6GGRUencoder-decodertrainingoverhead
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 tries to show that a small neural network using only GPS position data can replace the expensive beam-sweeping step in millimeter-wave links between a drone and a base station. The model is asked to output both the current best beam and the best beams one, two, and three time steps ahead, at 60 GHz on a real-world drone dataset. The authors report Top-1 accuracy above 70% and average power loss below 0.6 dB at every prediction step, with overhead savings of about 93% because only 2–3 of 32 codebook beams need to be trained. If the claims hold, a drone's GPS receiver alone could provide lightweight beam management that keeps working when cameras or other sensors fail.

What carries the argument

The load-bearing object is the combined input feature $O[t] = \{g_{\mathrm{UE,norm}}[t], u_{\mathrm{UE-BS}}[t]\}$: the drone's latitude and longitude min-max normalized to $[0,1]$, joined with the unit vector from the base station to the drone computed through an ECEF coordinate transformation. Sequences of $W=8$ such feature vectors feed a 1D convolutional feature extractor, then a GRU encoder-decoder whose final encoder hidden state seeds the decoder, and finally a two-layer classifier with softmax over the 32 codebook beams; the decoder emits $V+1=4$ beam indices, covering the current and three future steps. The adjusted splitting algorithm is a second mechanism: it splits raw data into chunks, scores candidate chunk sizes by label-distribution similarity, then further rebalances each label group to the 65/15/20 train/validation/test ratio, which the paper argues is what makes the GPS features learnable.

What would settle it

Repeat the experiment holding out entire drone trips (each sequence index $q$ in only one partition) and compare Top-1 accuracy and mean power loss against the paper's numbers; if the trip-disjoint results fall well below 70% accuracy or push power loss above 0.6 dB, the central claim fails for truly unseen trajectories.

Watch

Extended reading notes

Core claim

The central discovery, stated on the paper's own terms, is that sequential GPS data—normalized drone latitude and longitude plus the unit vector pointing from the base station to the drone—carries enough information to predict the current optimal beam and three future beams simultaneously in a 60 GHz UAV scenario. This is achieved by a 1D-CNN plus GRU encoder-decoder that maps an 8-step position window to a sequence of beam indices from a 32-beam codebook. A data-splitting procedure called adjusted splitting, which rebalances labels so train, validation, and test sets mirror the original label distribution, is presented as a necessary ingredient: without it, accuracy drops by roughly 31–34 percentage points and mean power loss rises by 48–70%. The authors report that the combined position-plus-unit-vector input outperforms either feature alone, improving Top-1 accuracy by 8–14 percentage points and roughly halving power loss.

Load-bearing premise

The load-bearing premise is that the adjusted splitting yields a test set that fairly represents unseen drone flights rather than fragments of flights that already appeared in training.

Editorial extensions

If this is right

  • A drone link could maintain its beam without sweeping the full 32-beam codebook, training only 2–3 beams per reliability target and saving roughly 93% of training overhead.
  • The same model covers both current-beam prediction and future-beam tracking, so no separate tracking network is required.
  • With an 8-step observation window and 3-step horizon, the system has roughly three time slots of lookahead, enough to react to the drone's motion if the position data arrives in time.
  • Performance degrades as codebook size grows (63–64% accuracy at 64 beams) and as speed increases, so the approach is best suited to moderate speeds and 32-beam or smaller codebooks.
  • Medium flight heights (40–80 m) are the hardest height range, so future tuning should target that regime before deployment.

Reading between the lines

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

  • Because the adjusted splitting does not force samples from a single drone trip to stay in one partition, the reported accuracies may partly reflect the model recognizing specific trips; the natural test is a trip-disjoint split, which the paper does not report.
  • The unit-vector component expresses position relative to the base station, suggesting the model may transfer across base stations only after recalibrating that reference; absolute coordinates alone would not generalize.
  • If the GPS-only result survives a trip-disjoint test, then fusion with vision or radar should push accuracy higher where GPS is sparse, and the same architecture could extend to indoor drones using equivalent localization such as UWB or SLAM.
  • The preprocessing choice of dataset-wide min-max normalization means the model assumes the test set lies within the training coordinate range; deployment outside the trained area would need online normalization or recalibration.
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.

Referee Report

3 major / 6 minor

Summary. The manuscript proposes a GPS-aided deep learning model for simultaneous current and future beam prediction in UAV mmWave communication. The model uses a CNN-GRU encoder-decoder with a classifier to map a window of W=8 GPS-derived features (normalized UE coordinates and the UE-BS unit vector) to V+1=4 beam indices. The authors introduce an "adjusted splitting" protocol intended to balance label distributions across training, validation, and test sets, and report Top-1 accuracy above 70%, mean power loss below 0.6 dB, and about 93% training overhead savings on DeepSense6G Scenario 23. The paper also analyzes performance across UAV heights, speeds, and codebook sizes.

Significance. If the quantitative claims were valid, the work would offer a lightweight, GPS-only alternative to full codebook sweeping for UAV links, and the real-world evaluation on DeepSense6G would be a practical strength. The paper gives a detailed architecture, a reproducible description of training hyperparameters, and extensive diagnostic experiments by height, speed, codebook size, and overhead/reliability. However, the validity of all headline numbers rests on the data-splitting protocol and the normalization procedure, and the manuscript does not establish that the test set is a genuinely held-out set.

major comments (3)
  1. [III-A, Algorithm 1, lines 11–15 and 22–25] The adjusted splitting procedure does not constrain the sequence index q to a single partition. In the chunk loop, each chunk is split internally by ratios (lines 11–15), so a drone trip that crosses a chunk boundary can contribute frames to different partitions. In the label-based regrouping step (lines 22–25), samples for each beam label are drawn from all three already-populated partitions and split again by ratio, which can freely interleave frames from the same q across Dtrain, Dval, and Dtest. Since beam indices along a single trajectory are highly autocorrelated and adjacent frames are near-duplicates, the test set can contain GPS traces and beam sequences that are essentially copied from training. The integrity requirement in Section III-B only ensures that each constructed sample's input and output share the same q and are consecutive; it does not prevent this cross-partition leakage. Consequently, the "Top-1 accuracy exceeding 70%" and "average power loss below 0.6 dB" claims in the abstract and Table III are not validated as generalization results.
  2. [II-C1, Eqs. (6)–(7), and III-A] Min-max normalization is applied to the entire raw data set before splitting, so the extrema used in equations (6) and (7) are computed over samples that later enter the test set. This gives the model access to test-set distribution information through the input feature g_UE,norm. To make the evaluation clean, the normalization parameters should be estimated from the training split only and then applied unchanged to the validation and test splits.
  3. [IV-A, Figure 5] The reported 31–34 percentage point improvement of adjusted splitting over sequential splitting is confounded. Section III-A states that under sequential splitting the training set is empty for beam 0 while the test set contains many such samples, whereas the adjusted protocol guarantees balanced label proportions by construction. Therefore the gap shown in Figure 5 reflects not only the splitting strategy but also a change in label availability and a temporal-leakage artifact. The paper's claim that adjusted splitting "enhances model performance" requires a comparison in which both protocols preserve sequence-index integrity and both are evaluated on truly held-out trips.
minor comments (6)
  1. [Table II] The row labeled "Dtest raw 2,209 (20%) 2,050 (19.36%)" appears to report the development test set, not the raw test set; the label should likely be "Dtest dev" to be consistent with the preceding rows.
  2. [Eq. (4)] Equation (4) is incomplete: the sentence "Using the definition of the optimal beam index provided in equation" is followed directly by an unnumbered arg max expression with no operator or verb. The objective should be written as a complete mathematical statement.
  3. [II-C1] The subsection numbering contains a duplicated hierarchy: "a) 1) Min-max Normalization." One of the two markers should be removed.
  4. [Eq. (17)] The symbol K is already used for the raw data set size in Section II-C1 and for the chunk/sample counts elsewhere; reusing it for the number of test samples in the power-loss formula creates ambiguity. Use a distinct symbol such as N_test.
  5. [Abstract and IV-E] The abstract says "95% beam prediction accuracy guarantees," while Section IV-E and Figure 8(a) use the term "Guaranteed Reliability." Align the terminology to avoid confusing accuracy with reliability/outage probability.
  6. [Figure 5(b)] The numeric labels "0.30 0.28 0.27 0.57" are placed in a way that makes it unclear which bar they correspond to; adjust the labeling for unambiguous reading.

Circularity Check

1 steps flagged · score 5.0 of 10

Adjusted splitting can place the same drone trajectory in both train and test, so the headline 70% / 0.6 dB claims are not validated as independent predictions.

  1. other [Section III-A2 (Algorithm 1, lines 22–25), Section III-B, and Section IV-A (Figure 5) / Table III]
    "foreach label b ∈ {1, ..., M} do Db ← Get samples for label b from Dtrain_raw, Dval_raw, Dtest_raw; Split Db into Dtrain_b, Dval_b, Dtest_b using given ratios; Update Dtrain_raw, Dval_raw, Dtest_raw with new label-based splits; ... The primary goal is to ensure that the label distribution of each split data set reflects the label distribution of the original data set Draw and follows the desired splitting percentage."

    The test set is not an independent held-out set. Algorithm 1 pools samples by beam label across all three partitions and re-splits each label group by the same ratios, so the label distribution of the test set is forced, by construction, to match the training label distribution. Section III-B only requires that a sample's input and output sequences share the same sequence index q and be consecutive; it never requires all samples of a trip to remain in one partition. Since beam indices along a trajectory are strongly autocorrelated, the same drone trip can contribute near-duplicate GPS/beam fragments to both training and test.

full rationale

The core derivation is not circular: the model maps preprocessed GPS sequences to beam indices, and the labels b* are obtained from measured received power via Eq. (3), not from the model or from the split algorithm. The self-citation [2] is only a publication notice and is not load-bearing. However, the evaluation protocol is self-referential. Algorithm 1's label-based regrouping (lines 22–25) rebuilds train/val/test by re-splitting each label group, so the test set's label distribution is forced to match the training label distribution, while Section III-B's integrity constraint only ties a sample's input/output to the same q and never keeps all samples of a trip in one split. Because beam labels along a trajectory are autocorrelated, the same trip can appear in both training and test, inflating the reported generalization numbers. The claim that adjusted splitting 'outperforms sequential methods by 31–34%' is therefore not evidence of better generalization; it is largely a consequence of constructing the test set to resemble the training label distribution. The Top-1 >70% and <0.6 dB numbers in Table III inherit this problem. This is a data-leakage / circular-evaluation concern, not a self-citation or ansatz-smuggling issue.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central performance claims depend on a set of chosen hyperparameters and on domain assumptions about mmWave propagation and the representativeness of the DeepSense 6G dataset. No new physical entities are introduced; the main nonstandard item is the adjusted splitting protocol, which functions as an evaluation assumption rather than a physical postulate.

free parameters (8)
  • Observation window size W = 8
    Chosen by the authors without sensitivity analysis; the input sequence length directly affects how much trajectory context the model sees.
  • Prediction horizon V = 3
    The model predicts current beam plus three future steps; this choice is not justified and directly defines the claimed 'future beam' results.
  • CNN kernel, stride, padding = (3,1,1)
    Feature extraction hyperparameters chosen by the authors; no ablation shows their effect on accuracy.
  • GRU hidden size = 128
    Encoder and decoder hidden dimension; set by hand and not varied.
  • Training hyperparameters = 20 epochs, batch size 8, learning rate 5e-4, weight decay 0, LR reduction at epochs 12 and 18
    These are standard choices but unverified against alternatives; the reported single-run performance could depend on them.
  • Split ratios = 65% train, 15% validation, 20% test
    Arbitrary split percentages used in both sequential and adjusted splitting.
  • Chunk size percentages Pchunks = not specified
    Algorithm 1 takes Pchunks as input, but the paper never reports what chunk sizes were tried, making the adjusted split irreproducible.
  • Min-max normalization bounds = global min and max over full dataset before splitting
    The normalization statistics are computed over the entire dataset, including test samples, which leaks test information into preprocessing.
assumptions (5)
  • standard math WGS-84 ECEF coordinate transformation formulas (Eq. 8 to 11) are correct and applicable to the GPS data.
    Used to compute the UE-BS unit vector; this is a well-established geodetic transformation.
  • domain assumption mmWave links in the scenario follow a dominant single path and exhibit uplink/downlink reciprocity in received power.
    Stated in Section III to justify deriving optimal beams from received power and applying them bidirectionally.
  • domain assumption GPS position history plus the UE-BS unit vector is sufficient to predict the optimal beam index in this environment.
    This is the modeling premise behind the entire input design in Section II-C; it ignores blockage and multipath effects not captured by position.
  • domain assumption DeepSense 6G Scenario 23 is representative of UAV mmWave communications and its beam labels are accurate.
    The paper evaluates only this scenario and does not test on other environments, so generalization is assumed.
  • ad hoc to paper Adjusted splitting produces a valid held-out test set.
    Algorithm 1 is introduced by the authors, but it does not enforce that samples with the same sequence index q remain in one partition, risking temporal leakage.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GPS-Aided Deep Learning for Beam Prediction and Tracking in UAV mmWave Communication." pith.science (2026). https://pith.science/paper/HINR5FCL

@misc{pith2026250517530,
  author       = {Pith},
  title        = {Pith review of: GPS-Aided Deep Learning for Beam Prediction and Tracking in UAV mmWave Communication},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HINR5FCL}},
  note         = {Machine review of arXiv:2505.17530}
}
read the original abstract

Millimeter-wave (mmWave) communication enables high data rates for cellular-connected Unmanned Aerial Vehicles (UAVs). However, a robust beam management remains challenging due to significant path loss and the dynamic mobility of UAVs, which can destabilize the UAV-base station (BS) link. This research presents a GPS-aided deep learning (DL) model that simultaneously predicts current and future optimal beams for UAV mmWave communications, maintaining a Top-1 prediction accuracy exceeding 70% and an average power loss below 0.6 dB across all prediction steps. These outcomes stem from a proposed data set splitting method ensuring balanced label distribution, paired with a GPS preprocessing technique that extracts key positional features, and a DL architecture that maps sequential position data to beam index predictions. The model reduces overhead by approximately 93% (requiring the training of 2 ~ 3 beams instead of 32 beams) with 95% beam prediction accuracy guarantees, and ensures 94% to 96% of predictions exhibit mean power loss not exceeding 1 dB.

Figures

Figures reproduced from arXiv: 2505.17530 by the authors.

Figure 1
Figure 1. System Architecture represents the observation window size. Predictions with step v are denoted as [t + v], where v ∈ {0, . . . , V } and V is the maximum prediction step. F = {fm}M m=1 denotes the beam codebook with M beam vectors. uUE−BS represents the unit vector between UE and BS positions. II. SYSTEM MODEL AND PROPOSED SOLUTION This section commences by presenting the underlying sys￾tem model that forms the fou… view at source ↗
Figure 2
Figure 2. GPS data preprocessing pipeline for the proposed beam [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Neural network architecture used for beam prediction [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Impact of data set splitting methods on label distribution across train, validation, and test sets [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Impact of data set splitting methods on (a) model [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Model Performance on Various UAV Height TABLE IV: Model Performance on Various UAV Height with Equal Number of Samples Height Category Number of Samples Sampling Rounds Current Prediction Top1 Acc (%) Future Prediction 1 Top1 Acc (%) Future Prediction 2 Top1 Acc (%) Fu…
Figure 7
Figure 7. Figure 7: Model Performance on Various UAV Speed 80 85 90 95 99 Guaranteed Reliability [1 - Outage Probability] (%) 89 90 91 92 93 94 95 Overhead Savings (%) (a) Overhead Savings Current Beam Top1 Prediction Future Beam-1 Top1 Prediction Future Beam-2 Top1 Prediction Future Beam…
Figure 8
Figure 8. Figure 8: Model overhead savings and mean power loss reliability [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Model Performance on Various Beam Codebook Size [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 37 canonical work pages

  1. [29]

    Sensing-Aided 6G Drone Communications: Real-World Datasets and Demonstration,

    G. Charan and A. Alkhateeb, “Sensing-Aided 6G Drone Communications: Real-World Datasets and Demonstration,” Dec

  2. [1]

    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

  3. [2]

    Gps-aided deep learning for beam prediction and tracking in uav mmwave communication,

    V . A. Nugroho and B. M. Lee, “Gps-aided deep learning for beam prediction and tracking in uav mmwave communication,” IEEE Access, vol. 13, pp. 117 065–117 077, 2025

  4. [3]

    A Survey on Millimeter-Wave Beamforming Enabled UA V Communi- cations and Networking,

    Z. Xiao, L. Zhu, Y . Liu, P. Yi, R. Zhang, X.-G. Xia, and R. Schober, “A Survey on Millimeter-Wave Beamforming Enabled UA V Communi- cations and Networking,” IEEE Communications Surveys & Tutorials , vol. 24, no. 1, pp. 557–610, 2022

  5. [4]

    What Will the Future of UA V Cellular Communications Be? A Flight From 5G to 6G,

    G. Geraci, A. Garcia-Rodriguez, M. M. Azari, A. Lozano, M. Mezzav- illa, S. Chatzinotas, Y . Chen, S. Rangan, and M. D. Renzo, “What Will the Future of UA V Cellular Communications Be? A Flight From 5G to 6G,” IEEE Communications Surveys & Tutorials , vol. 24, no. 3, pp. 1304–1335, 2022

  6. [5]

    A Survey of Beam Management for mmWave and THz Communications Towards 6G,

    Q. Xue, C. Ji, S. Ma, J. Guo, Y . Xu, Q. Chen, and W. Zhang, “A Survey of Beam Management for mmWave and THz Communications Towards 6G,” IEEE Communications Surveys & Tutorials , pp. 1–1, 2024

  7. [6]

    Research Chal- lenges and Opportunities of UA V Millimeter-Wave Communications,

    C. Zhang, W. Zhang, W. Wang, L. Yang, and W. Zhang, “Research Chal- lenges and Opportunities of UA V Millimeter-Wave Communications,” IEEE Wireless Communications, vol. 26, no. 1, pp. 58–62, Feb. 2019

  8. [7]

    A Tutorial on Beam Management for 3GPP NR at mmWave Frequencies,

    M. Giordani, M. Polese, A. Roy, D. Castor, and M. Zorzi, “A Tutorial on Beam Management for 3GPP NR at mmWave Frequencies,” IEEE Communications Surveys & Tutorials, vol. 21, no. 1, pp. 173–196, 2019

Show all 39 references
  1. [8]

    Beam training and tracking in mmWave communication: A survey,

    W. Yi, W. Zhiqing, and F. Zhiyong, “Beam training and tracking in mmWave communication: A survey,” China Communications , vol. 21, no. 6, pp. 1–22, Jun. 2024

  2. [9]

    Beam Training Based on Dynamic Hierarchical Codebook for Millimeter Wave Massive MIMO,

    K. Chen and C. Qi, “Beam Training Based on Dynamic Hierarchical Codebook for Millimeter Wave Massive MIMO,”IEEE Communications Letters, vol. 23, no. 1, pp. 132–135, Jan. 2019

  3. [10]

    Hierarchical Codebook- Based Multiuser Beam Training for Millimeter Wave Massive MIMO,

    C. Qi, K. Chen, O. A. Dobre, and G. Y . Li, “Hierarchical Codebook- Based Multiuser Beam Training for Millimeter Wave Massive MIMO,” IEEE Transactions on Wireless Communications , vol. 19, no. 12, pp. 8142–8152, Dec. 2020

  4. [11]

    Channel Es- timation and Hybrid Precoding for Millimeter Wave Cellular Systems,

    A. Alkhateeb, O. El Ayach, G. Leus, and R. W. Heath, “Channel Es- timation and Hybrid Precoding for Millimeter Wave Cellular Systems,” IEEE Journal of Selected Topics in Signal Processing , vol. 8, no. 5, pp. 831–846, Oct. 2014

  5. [12]

    Two-stage compressed sensing for millimeter wave channel estimation,

    Y . Han and J. Lee, “Two-stage compressed sensing for millimeter wave channel estimation,” in 2016 IEEE International Symposium on Information Theory (ISIT) , Jul. 2016, pp. 860–864

  6. [13]

    Deep Learning Coordinated Beamforming for Highly-Mobile Millimeter Wave Systems,

    A. Alkhateeb, S. Alex, P. Varkey, Y . Li, Q. Qu, and D. Tujkovic, “Deep Learning Coordinated Beamforming for Highly-Mobile Millimeter Wave Systems,” IEEE Access, vol. 6, pp. 37 328–37 348, 2018

  7. [14]

    Multimodal Fusion Assisted Mmwave Beam Training in Dual-Model Networks,

    J. Liu, X. Li, T. Fan, S. Lv, and M. Shi, “Multimodal Fusion Assisted Mmwave Beam Training in Dual-Model Networks,” IEEE Transactions on Vehicular Technology, vol. 73, no. 1, pp. 995–1011, Jan. 2024. This work has been published in the IEEE Access with DOI: 10.1109/ACCESS.2025...

  8. [15]

    Robust Beam Tracking with Extended Kalman Filtering for Mobile Millimeter Wave Communications,

    X. Xin and Y . Yang, “Robust Beam Tracking with Extended Kalman Filtering for Mobile Millimeter Wave Communications,” in 2019 Com- puting, Communications and IoT Applications (ComComAp) , Oct. 2019, pp. 172–177

  9. [16]

    Adaptive Beam Tracking With the Unscented Kalman Filter for Millimeter Wave Communication,

    S. G. Larew and D. J. Love, “Adaptive Beam Tracking With the Unscented Kalman Filter for Millimeter Wave Communication,” IEEE Signal Processing Letters , vol. 26, no. 11, pp. 1658–1662, Nov. 2019

  10. [17]

    Position Prediction Based Fast Beam Tracking Scheme for Multi-User UA V-mmWave Communications,

    Y . Ke, H. Gao, W. Xu, L. Li, L. Guo, and Z. Feng, “Position Prediction Based Fast Beam Tracking Scheme for Multi-User UA V-mmWave Communications,” in ICC 2019 - 2019 IEEE International Conference on Communications (ICC) , May 2019, pp. 1–7

  11. [18]

    Learning-Based Predictive Beamforming for UA V Communications With Jittering,

    W. Yuan, C. Liu, F. Liu, S. Li, and D. W. K. Ng, “Learning-Based Predictive Beamforming for UA V Communications With Jittering,”IEEE Wireless Communications Letters , vol. 9, no. 11, pp. 1970–1974, Nov. 2020

  12. [19]

    Location-Aware Predictive Beamforming for UA V Communications: A Deep Learning Approach,

    C. Liu, W. Yuan, Z. Wei, X. Liu, and D. W. K. Ng, “Location-Aware Predictive Beamforming for UA V Communications: A Deep Learning Approach,” IEEE Wireless Communications Letters , vol. 10, no. 3, pp. 668–672, Mar. 2021

  13. [20]

    Position-Aided Beam Prediction in the Real World: How Useful GPS Locations Actually are?

    J. Morais, A. Bchboodi, H. Pezeshki, and A. Alkhateeb, “Position-Aided Beam Prediction in the Real World: How Useful GPS Locations Actually are?” in ICC 2023 - IEEE International Conference on Communications, May 2023, pp. 1824–1829

  14. [21]

    LiDAR Aided Future Beam Prediction in Real-World Millimeter Wave V2I Communications,

    S. Jiang, G. Charan, and A. Alkhateeb, “LiDAR Aided Future Beam Prediction in Real-World Millimeter Wave V2I Communications,” IEEE Wireless Communications Letters, vol. 12, no. 2, pp. 212–216, Feb. 2023

  15. [22]

    Vision-Position Multi-Modal Beam Prediction Using Real Millimeter Wave Datasets,

    G. Charan, T. Osman, A. Hredzak, N. Thawdar, and A. Alkhateeb, “Vision-Position Multi-Modal Beam Prediction Using Real Millimeter Wave Datasets,” in 2022 IEEE Wireless Communications and Network- ing Conference (WCNC) . Austin, TX, USA: IEEE, Apr. 2022, pp. 2727–2731

  16. [23]

    Millimeter Wave V2V Beam Tracking using Radar: Algorithms and Real-World Demonstration,

    H. Luo, U. Demirhan, and A. Alkhateeb, “Millimeter Wave V2V Beam Tracking using Radar: Algorithms and Real-World Demonstration,” in 2023 31st European Signal Processing Conference (EUSIPCO) , Sep. 2023, pp. 740–744

  17. [24]

    Computer Vision Aided Beam Tracking in A Real-World Millimeter Wave Deployment,

    S. Jiang and A. Alkhateeb, “Computer Vision Aided Beam Tracking in A Real-World Millimeter Wave Deployment,” in 2022 IEEE Globecom Workshops (GC Wkshps), Dec. 2022, pp. 142–147

  18. [25]

    Towards Real-World 6G Drone Communication: Position and Camera Aided Beam Prediction,

    G. Charan, A. Hredzak, C. Stoddard, B. Berrey, M. Seth, H. Nunez, and A. Alkhateeb, “Towards Real-World 6G Drone Communication: Position and Camera Aided Beam Prediction,” in GLOBECOM 2022 - 2022 IEEE Global Communications Conference , Dec. 2022, pp. 2951–2956

  19. [26]

    Millimeter Wave Drones with Cameras: Computer Vision Aided Wireless Beam Prediction,

    G. Charan, A. Hredzak, and A. Alkhateeb, “Millimeter Wave Drones with Cameras: Computer Vision Aided Wireless Beam Prediction,” in 2023 IEEE International Conference on Communications Workshops (ICC Workshops), May 2023, pp. 1896–1901

  20. [27]

    Vision-Assisted Beam Prediction for Real World 6G Drone Communication,

    I. Ahmad, A. R. Khan, R. N. B. Rais, A. Zoha, M. A. Imran, and S. Hussain, “Vision-Assisted Beam Prediction for Real World 6G Drone Communication,” in 2023 IEEE 34th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC) , Sep. 2023, pp. 1–7

  21. [28]

    Vision-assisted Beam Pre- diction for UA V-enabled Millimeter-Wave Communications using SE- ResNet50,

    Z. Zarei, F. D. Tilahun, and C. G. Kang, “Vision-assisted Beam Pre- diction for UA V-enabled Millimeter-Wave Communications using SE- ResNet50,” in 2023 14th International Conference on Information and Communication Technology Convergence (ICTC), Oct. 2023, pp. 1659– 1661

  22. [30]

    DeepSense 6G: A Large-Scale Real- World Multi-Modal Sensing and Communication Dataset,

    A. Alkhateeb, G. Charan, T. Osman, A. Hredzak, J. Morais, U. Demirhan, and N. Srinivas, “DeepSense 6G: A Large-Scale Real- World Multi-Modal Sensing and Communication Dataset,” IEEE Com- munications Magazine, vol. 61, no. 9, pp. 122–128, Sep. 2023

  23. [31]

    Characterizing Driving Styles with Deep Learning,

    W. Dong, J. Li, R. Yao, C. Li, T. Yuan, and L. Wang, “Characterizing Driving Styles with Deep Learning,” Oct. 2016. [Online]. Available: http://arxiv.org/abs/1607.03611

  24. [32]

    A note on computation of Geodetic coordinates from geocentric (Cartesian) coordinates,

    M. K. Paul, “A note on computation of Geodetic coordinates from geocentric (Cartesian) coordinates,” Bulletin géodésique, vol. 108, no. 1, pp. 135–139, Jun. 1973

  25. [33]

    Conversion of Earth-centered Earth-fixed coordinates to geode- tic coordinates,

    J. Zhu, “Conversion of Earth-centered Earth-fixed coordinates to geode- tic coordinates,” IEEE Transactions on Aerospace and Electronic Sys- tems, vol. 30, no. 3, pp. 957–961, Jul. 1994

  26. [34]

    State-of-the-Art in 1D Convolutional Neural Networks: A Survey,

    A. O. Ige and M. Sibiya, “State-of-the-Art in 1D Convolutional Neural Networks: A Survey,” IEEE Access, vol. 12, pp. 144 082–144 105, 2024

  27. [35]

    Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation,

    K. Cho, B. Van Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y . Bengio, “Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation,” inProceed- ings of the 2014 Conference on Empirical Methods in Natural Language Proce...

  28. [36]

    Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling,

    J. Chung, C. Gulcehre, K. Cho, and Y . Bengio, “Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling,” Dec

  29. [37]

    Multi-Modal Sensing-Aided Beam Prediction using Poolformer for UA V Communications,

    Y . Yeo and J. Kim, “Multi-Modal Sensing-Aided Beam Prediction using Poolformer for UA V Communications,” in 2024 Fifteenth International Conference on Ubiquitous and Future Networks (ICUFN), Jul. 2024, pp. 202–204

  30. [2014]

    Available: http://arxiv.org/abs/1412.3555

    [Online]. Available: http://arxiv.org/abs/1412.3555

  31. [2024]

    Available: http://arxiv.org/abs/2412.04734

    [Online]. Available: http://arxiv.org/abs/2412.04734

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.