Pith. sign in

REVIEW 4 major objections 8 minor 30 references

A large Sionna-simulated multi-antenna IQ dataset lets researchers train and compare jammer classification and direction finding under controlled indoor multipath.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-12 02:40 UTC pith:XOHZIZ5N

load-bearing objection Solid public multi-antenna IQ dataset with motion and OAT baselines; fills a real GNSS-interference data gap without overclaiming sim-to-real transfer. the 4 major comments →

arxiv 2607.03411 v1 pith:XOHZIZ5N submitted 2026-07-03 eess.SP cs.AI

The S-ICDF Dataset: Sionna-Simulated Dynamic Interference Characterization and Direction Finding

Christian Wielenberg , Lucas Heublein , Jonathan Ott , Alexander Mattick , Nisha L. Raichur , Jonas Pirkl , Lukas Schelenz , Tobias Feigl
show 3 more authors
George Yammine Christopher Mutschler Felix Ott
This is my paper
classification eess.SP cs.AI
keywords Sionnainterference monitoringdirection findingGNSS jammingIQ datasetMUSICESPRITray tracing
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Intentional radio jamming is illegal to collect at scale in the real world, and ground truth is hard to pin down once multipath, hardware, and motion mix in. This paper therefore builds S-ICDF: a large indoor interference dataset generated with Sionna ray tracing so that every waveform, SNR, array geometry, and multipath setting is known exactly. The receiver is a moving multi-antenna array that records raw complex IQ samples while a single interference source transmits one of 102 configurations. The authors then run classical direction-finding algorithms (MUSIC, ESPRIT, CAPON) and a modern ML model on the same data, showing how accuracy depends on waveform type, noise level, antenna spacing, and reflection depth. The public release is meant as a shared testbed so that characterization and localization methods can be iterated without needing illegal field campaigns.

Core claim

S-ICDF shows that a controlled, one-parameter-at-a-time Sionna ray-traced industrial hall can generate fully labeled multi-antenna IQ trajectories across 102 interference configurations, and that both classical direction finders and an XceptionTime model trained on raw IQ can recover class labels and angles with high accuracy under default settings while revealing clear sensitivity to SNR, array spacing, and multipath parameters.

What carries the argument

The S-ICDF generation pipeline: a fixed jammer, a moving 2x2 or 8x1 array, Sionna ray tracing with selectable reflection depth and refraction, and one-at-a-time sweeps over waveform, bandwidth, SNR, antenna spacing, and gain pattern, yielding 216,089 labeled IQ snapshots per configuration for characterization and direction finding.

Load-bearing premise

The claim that this metallic-hall ray-tracing setup with fixed reflection depths and material choices is realistic enough that the performance rankings and sensitivity trends will transfer to real indoor GNSS interference.

What would settle it

Record a real indoor multi-antenna campaign with known jammer locations and waveforms that match a subset of S-ICDF configurations, then check whether the same classical and ML methods keep the same accuracy ordering and the same sensitivity to spacing, SNR, and multipath as reported on the simulated set.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Researchers can train and compare jammer classifiers and direction finders on fully labeled array IQ without illegal field jamming.
  • One-parameter sweeps make it possible to isolate which design choices (spacing, array layout, reflection depth) most hurt or help classical DF.
  • The public baseline numbers (class accuracy, azimuth/elevation MSE) give a fixed reference for new ML architectures on the same split.
  • Synthetic-aperture motion of the array becomes a standard ingredient for time-series direction-finding benchmarks.

Where Pith is reading between the lines

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

  • If the simulation-to-reality gap is small, the same pipeline could later generate the large balanced corpora needed for foundation models of RF interference.
  • The strong dependence on inter-element spacing near half-wavelength suggests future releases should densify the spacing grid around the ambiguity region.
  • Because only a single interferer is present, multi-source or spoofing-plus-jamming extensions would be a natural next controlled experiment.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 8 minor

Summary. The manuscript introduces S-ICDF, a publicly released large-scale indoor multi-antenna IQ dataset for GNSS/wireless interference characterization and direction finding, generated with Sionna ray tracing in a metallic industrial-hall scene. Using a one-at-a-time parameter schedule, it covers many interference waveforms (chirp, noise, frequency-hopper, multitone, modulated, pulsed), bandwidths, SNR, array layouts/spacings/gain patterns, reflection depth, and refraction, with a moving 2×2 or 8×1 array providing time-series IQ and ground-truth geometry. The authors report classical DF baselines (MUSIC, ESPRIT, CAPON) and a multi-task XceptionTime model for classification/characterization and DoA/range, with headline numbers such as ~99.89% class accuracy and ESPRIT ~0.78° azimuth MSE under default settings, and release the data via GitLab.

Significance. If the resource is as described, S-ICDF is a useful contribution for a field where intentional jamming is often illegal and real measurements are confounded. Controlled multipath, array IQ, motion, and labeled DoA/range fill gaps relative to existing spoofing/jamming corpora that lack phase-coherent arrays or trajectory labels. Shipping classical and ML baselines plus a public link supports reproducible method development and sensitivity analysis. The work does not claim validated outdoor hardware transfer; its value is as a controlled simulation benchmark, which is appropriate and timely for ML-oriented interference monitoring research.

major comments (4)
  1. Sec. III-C and Sec. IV (ML split): The 80/20 train–test partition is described only as a split of trajectory samples (216,089 points per configuration; later 440,821/110,172 pooled). Adjacent samples on a continuous path are strongly correlated in geometry and multipath. Without an explicit trajectory- or segment-held-out protocol (or temporal gap), the reported XceptionTime DF/characterization numbers may be inflated by leakage. Please state the split rule precisely and, if random-by-sample, re-evaluate with trajectory-held-out folds so the ML baselines are interpretable as generalization rather than near-neighbor interpolation.
  2. Sec. IV(a) classical DF: Baseline claims (MUSIC/ESPRIT/CAPON azimuth/elevation MSE in Figs. 7–8 and Table III) lack load-bearing implementation details: snapshot/covariance window length, assumed number of sources vs. multipath components, subspace dimension selection, and whether elevation is jointly estimated on the stated 1° grid. In a dense multipath hall these choices dominate error. Without them, the ranking “ESPRIT most robust” and the SNR knee near −8 dB cannot be reproduced or fairly compared by users of the dataset.
  3. Abstract vs. Sec. I contributions vs. Sec. III-C: The paper alternately states “102 interference configurations,” “110 distinct interference parameterizations,” and “102 signal files.” Table II and the OAT design make the intended cardinality of the release load-bearing for the “large-scale / 102 configs” claim. Please reconcile the counts, define what constitutes one configuration, and align abstract, contributions, and dataset documentation.
  4. Table II vs. Table III default settings: Table II lists refraction default True, while Table III’s default line states “Refraction = False” (and reports a separate True row). Default reflection depth, spacing, and refraction define the reference operating point for all OAT curves. Inconsistent defaults undermine attribution of performance changes and the “default ~0.8° azimuth” baseline. Fix the table and restate which defaults were used for every reported figure.
minor comments (8)
  1. Table III, antenna distance 0.095 m row: elevation for MUSIC is written “7,45” (comma); should be a decimal consistent with other entries.
  2. Fig. 4 caption: “signal sharacteristics” → “characteristics.”
  3. Sec. I: “multi-patch” appears where “multi-element” or “multi-antenna” is meant; same phrasing recurs for the 2×2 array.
  4. Fig. 6 right panel axes are labeled 0–100 “Classes” for a fine-grained characterization task; clarify number of characterization labels and class taxonomy (the left panel is 6-way).
  5. Sec. III-B / Fig. 1: material properties, carrier wavelength consistency with 1.57542 GHz, and whether path loss and antenna element patterns are jointly applied should be stated briefly for reproducibility of the ray-traced channels.
  6. Related work (Table I): a short explicit column or sentence on which prior sets provide motion + phase-coherent arrays would sharpen the claimed gap versus Heublein et al. [9].
  7. Eq. (1): loss weights λ3=λ4=λ5=0.3 are free parameters; a one-sentence sensitivity note (or fixed seed/config in the repo) would help users reproduce the multi-task trade-off.
  8. GitLab URL in abstract uses a space (“darcy gnss”); ensure the published link matches the live repository path.

Circularity Check

0 steps flagged

No significant circularity: empirical dataset release and held-out baselines, not a self-derived prediction.

full rationale

S-ICDF is a simulation-and-benchmark resource paper. Its load-bearing claims are (1) generation of a public multi-antenna IQ dataset under controlled OAT parameter sweeps in Sionna RT and (2) empirical performance numbers of standard external algorithms (MUSIC/ESPRIT/CAPON) plus an XceptionTime model trained and evaluated on an 80/20 split of that same simulated data. Classification accuracy (99.89 %), characterization accuracy (71.99 %), and DF MSEs (e.g., ESPRIT 0.78° azimuth under defaults) are measurements on held-out trajectories, not quantities forced by construction from fitted inputs or self-defined identities. Self-citations to the authors’ prior real-world jammer recordings and ML papers appear only in related-work context and do not underwrite the numerical results or the dataset-generation pipeline. There is no uniqueness theorem, no ansatz smuggled via self-citation, and no renaming of a known empirical pattern as a first-principles derivation. The work is therefore self-contained against its own stated contribution; any sim-to-real transfer limitation is a validity concern, not circularity.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The central claim is the existence and utility of a simulated dataset plus empirical baselines. It rests on standard wireless-propagation and array-processing assumptions plus a handful of simulation defaults chosen by the authors; no new physical entities are postulated. Free parameters are the OAT defaults and training hyper-parameters that define the released corpus and the reported numbers.

free parameters (4)
  • default reflection depth = 5
    Fixed at 5 (varied only in OAT to 2 or 7); directly controls multipath richness and therefore DF error floors reported in Table III.
  • default antenna spacing = 0.09 m
    0.09 m chosen as default; authors note spatial ambiguities appear at 0.095 m and 0.2 m, so the reported “best” performance depends on this hand-chosen value.
  • XceptionTime loss weights λ3=λ4=λ5=0.3 = 0.3
    Hand-set multi-task weights that trade classification accuracy against angle/distance regression; affect the published ML DF numbers.
  • SNR evaluation grid and default = default 20 dB
    Default 20 dB and steps of 2 dB from −20 to 20 dB define the threshold behavior claimed around −8 dB.
axioms (4)
  • domain assumption Sionna RT with the chosen metallic hall geometry and material model produces physically grounded multipath channels representative of indoor industrial GNSS interference.
    Invoked throughout Sec. III-B and the claim of a “challenging yet realistic testbed”; no real-world validation of the channel statistics is provided.
  • ad hoc to paper One-at-a-time parameter variation with all other parameters at defaults isolates causal effects on DF performance.
    Explicit design choice in Sec. III-A/C; interactions among parameters are therefore unmeasured.
  • domain assumption Standard narrowband array manifold assumptions underlying MUSIC, ESPRIT and Capon remain applicable to the 100 MHz bandwidth, 1024-sample snapshots used.
    Classical baselines in Sec. IV-a rely on these textbook conditions.
  • domain assumption 80/20 random split of the continuous trajectory yields an unbiased test of generalization for both classical and ML methods.
    Used for all reported accuracies and MSEs; temporal correlation along the trajectory is not discussed as a possible leakage source.

pith-pipeline@v1.1.0-grok45 · 16466 in / 3106 out tokens · 29908 ms · 2026-07-12T02:40:11.346118+00:00 · methodology

0 comments
read the original abstract

Jamming and spoofing threaten wireless and satellite navigation by disrupting or manipulating radio frequency (RF) signals, undermining availability, integrity, and trust. Robust interference monitoring (i.e., detection, classification, characterization, and direction finding) is therefore essential to identify and localize anomalous signals. While machine learning (ML) promises improved performance in complex environments, its development and validation depend on large-scale datasets that capture realistic signal and channel variability. Collecting such data in the real world is difficult because intentional jamming is illegal and ground-truth attribution is confounded by propagation, hardware, and environmental effects. To address this gap, we create and publish S-ICDF, a large-scale indoor interference dataset generated with Sionna, a GPU-accelerated simulation library for physical-layer wireless communications. S-ICDF covers 102 interference configurations, including diverse antenna array patterns, bandwidths, and simulation settings such as noise level and reflection depth. We further provide baseline results by benchmarking S-ICDF with classical estimation and direction finding (DF) methods (MUSIC, ESPRIT, and CAPON) and with modern ML approaches. The dataset is publicly available at: https://gitlab.cc-asp.fraunhofer.de/darcy_gnss/sicdf_dataset

Figures

Figures reproduced from arXiv: 2607.03411 by Alexander Mattick, Christian Wielenberg, Christopher Mutschler, Felix Ott, George Yammine, Jonas Pirkl, Jonathan Ott, Lucas Heublein, Lukas Schelenz, Nisha L. Raichur, Tobias Feigl.

Figure 1
Figure 1. Figure 1: Sionna simulation with ray tracing. Contributions. In the fol￾lowing, we summarize our main contributions: (1) In Sionna, we model an indoor industrial environment and employ ray tracing to sim￾ulate propagation from an interference source and the resulting signals received by an antenna array (refer to [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Pipeline from parameter definition and Sionna simulation to IQ extraction, MUSIC/ML training, and evaluation. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Trajectory of the antenna. b) Parameter Simulation Methodology: The dataset is generated according to a one-at￾a-time (OAT) parameter-variation strategy. In each experiment, only a single parameter of interest is varied over its predefined range, while all remaining parameters are fixed at their respective default values. This design enables a clear and unambiguous attribution of performance changes to the… view at source ↗
Figure 4
Figure 4. Figure 4: Histograms of signal sharacteristics. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 Frequency [MHz] (a) Noise. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 Frequency [MHz] (b) Chirp. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 Frequency [MHz] (c) Hopper. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 Frequency [MHz] (d) Modul. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 Frequency [MHz] (e) Multit. 0 2 4 6 8 10 Time [μs] -50 -25 0 25 50 … view at source ↗
Figure 7
Figure 7. Figure 7: Results of the azimuth (dots) and elevation (cross) orientation errors (in [PITH_FULL_IMAGE:figures/full_fig_p005_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Evaluation results of the azimuth orientation error (in [PITH_FULL_IMAGE:figures/full_fig_p005_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Evaluation of the azimuth prediction error. [PITH_FULL_IMAGE:figures/full_fig_p006_9.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

30 extracted references · 2 linked inside Pith

  1. [1]

    Jammer Classification in GNSS Bands via Machine Learning Algorithms,

    R. M. Ferre, A. de la Fuente, and E. S. Lohan, “Jammer Classification in GNSS Bands via Machine Learning Algorithms,” inMDPI Sensors, vol. 19(22), Nov. 2019

    2019

  2. [2]

    GNSS Jamming Classification via CNN, Transfer Learning & the Novel Concatenation of Signal Repre- sentations,

    C. J. Swinney and J. C. Woods, “GNSS Jamming Classification via CNN, Transfer Learning & the Novel Concatenation of Signal Repre- sentations,” inCyberSA, Dublin, Ireland, Jun. 2021

    2021

  3. [3]

    Jamming Attacks and Anti-Jamming Strate- gies in Wireless Networks: A Comprehensive Survey,

    H. Pirayesha and H. Zeng, “Jamming Attacks and Anti-Jamming Strate- gies in Wireless Networks: A Comprehensive Survey,” inCOMST, vol. 24(2), Mar. 2022, pp. 767–809

    2022

  4. [4]

    A Navigation Signals Monitoring, Analysis and Recording Tool: Application to Real-Time Interference Detection and Classification,

    I. E. Mehr, A. Minetto, and F. Dovis, “A Navigation Signals Monitoring, Analysis and Recording Tool: Application to Real-Time Interference Detection and Classification,” inION GNSS+, Sep. 2023

    2023

  5. [5]

    Calibration of RFI Detection Levels in a Low-Cost GNSS Monitor,

    N. R. S. Miguel, Y .-H. Chen, S. Lo, T. Walter, and D. Akos, “Calibration of RFI Detection Levels in a Low-Cost GNSS Monitor,” inIEEE/ION PLANS, Monterey, CA, Apr. 2023. Acknowledgments.This work has been carried out within the PaiL project, funding code 50NP2506, sponsored by the German Federal Ministry for Transport (BMV) and supported by the German Spa...

    2023

  6. [6]

    Few-Shot Learning with Uncertainty-based Quadruplet Selection for Interference Classification in GNSS Data,

    F. Ott, L. Heublein, N. L. Raichur, T. Feigl, J. Hansen, A. R ¨ugamer, and C. Mutschler, “Few-Shot Learning with Uncertainty-based Quadruplet Selection for Interference Classification in GNSS Data,” inICL-GNSS, Antwerp, Belgium, Jun. 2024

    2024

  7. [7]

    Evaluating ML Robustness in GNSS Interference Classification, Characterization & Localization,

    L. Heublein, T. Feigl, T. Nowak, A. R ¨ugamer, C. Mutschler, and F. Ott, “Evaluating ML Robustness in GNSS Interference Classification, Characterization & Localization,” inICL-GNSS, Rome, Italy, Jun. 2025

    2025

  8. [8]

    Machine Learning- assisted GNSS Interference Monitoring Through Crowdsourcing,

    N. L. Raichur, T. Brieger, D. Jdidi, T. Feigl, J. R. van der Merwe, B. Ghimire, F. Ott, A. R ¨ugamer, and W. Felber, “Machine Learning- assisted GNSS Interference Monitoring Through Crowdsourcing,” in ION GNSS+, Denver, CO, Sep. 2022, pp. 1151–1175

    2022

  9. [9]

    Attention-Based Fusion of IQ and FFT Spectrograms with AoA Features for GNSS Jammer Localization,

    L. Heublein, C. Wielenberg, T. Nowak, T. Feigl, C. Mutschler, and F. Ott, “Attention-Based Fusion of IQ and FFT Spectrograms with AoA Features for GNSS Jammer Localization,” inRadarConf, Krakow, Poland, Oct. 2025

    2025

  10. [10]

    GNSS Interference Mitigation Method Based on Deep Learning,

    F. Chen, Z. Liu, L. Huang, Y . Xie, B. Ren, and Q. Zhou, “GNSS Interference Mitigation Method Based on Deep Learning,” inSec. Interdisciplinary Physics, Mar. 2025

    2025

  11. [11]

    Intentional GNSS Interference Detection and Characterization Algorithm Using AGC and Adaptive IIR Notch Filter,

    J. H. Yang, C. H. Kang, S. Y . Kim, and C. G. Park, “Intentional GNSS Interference Detection and Characterization Algorithm Using AGC and Adaptive IIR Notch Filter,” inIJASS, vol. 13(4), 2012, pp. 491–498

    2012

  12. [12]

    Characterizing Terrestrial GNSS Interference from Low Earth Orbit,

    M. J. Murrian, L. Narula, and T. E. Humphreys, “Characterizing Terrestrial GNSS Interference from Low Earth Orbit,” inION GNSS+, Miami, Florida, Sep. 2019, pp. 3239–3253

    2019

  13. [13]

    GNSS Spoofing, Jamming, and Mul- tipath Interference Classification Using a Maximum-Likelihood Multi- tap Multipath Estimator,

    J. N. Gross and T. E. Humphreys, “GNSS Spoofing, Jamming, and Mul- tipath Interference Classification Using a Maximum-Likelihood Multi- tap Multipath Estimator,” inION GNSS+, Jan. 2017

    2017

  14. [14]

    Bayesian Learning-driven Prototypical Contrastive Loss for Class-Incremental Learning,

    N. L. Raichur, L. Heublein, T. Feigl, A. R ¨ugamer, C. Mutschler, and F. Ott, “Bayesian Learning-driven Prototypical Contrastive Loss for Class-Incremental Learning,” inTMLR, Apr. 2025

    2025

  15. [15]

    Variational and Generative Models with Quantization for Disentanglement and Compressed Sensing of GNSS Spectrograms,

    L. Heublein, T. Feigl, A. R ¨ugamer, C. Mutschler, and F. Ott, “Variational and Generative Models with Quantization for Disentanglement and Compressed Sensing of GNSS Spectrograms,” inJ-ISPIN, Jan. 2026

    2026

  16. [16]

    Learning from Imbalanced Data,

    H. He and E. A. Garcia, “Learning from Imbalanced Data,” inTKDE, vol. 21(9), Jun. 2009, pp. 1263–1284

    2009

  17. [17]

    Jammer Enforcement,

    F. C. Commission, “Jammer Enforcement,” Apr. 2020. [Online]. Available: https://www.fcc.gov/general/jammer-enforcement

    2020

  18. [18]

    Eval- uation of NovAtel’s Jamming and Spoofing Detection and Mitigation Capabilities During Jammertest2024,

    A. Broumandan, A. Pirsiavash, I. Tremblay, and S. Kennedy, “Eval- uation of NovAtel’s Jamming and Spoofing Detection and Mitigation Capabilities During Jammertest2024,” inION GNSS+, Jan. 2025

    2025

  19. [19]

    The Texas Spoofing Test Battery: Toward a Standard for Evaluating GPS Signal Authentication Techniques,

    T. Humphreys, J. Bhatti, D. Shepard, and K. Wesson, “The Texas Spoofing Test Battery: Toward a Standard for Evaluating GPS Signal Authentication Techniques,” inION GNSS+, Sep. 2012

    2012

  20. [20]

    Raw IQ Dataset for GNSS GPS Jamming Signal Classification,

    C. J. Swinney and J. C. Woods, “Raw IQ Dataset for GNSS GPS Jamming Signal Classification,” inZenodo, Mar. 2021

    2021

  21. [21]

    Descriptor: Tuni2025 GNSS - Galileo and GPS Spoofing Datasets (TG-GGSD),

    S. Rahman, M. Z. H. Bhuiyan, J. Nurmi, and E. Lohan, “Descriptor: Tuni2025 GNSS - Galileo and GPS Spoofing Datasets (TG-GGSD),” in IEEE Data Descriptions, Oct. 2025

    2025

  22. [22]

    In-lab Validation of Jammer Detection and Direction Finding Algorithms for GNSS,

    R. M. Ferre, P. Richter, A. D. L. Fuente, and E. S. Lohan, “In-lab Validation of Jammer Detection and Direction Finding Algorithms for GNSS,” inICL-GNSS, Nuremberg, Germany, Jun. 2019

    2019

  23. [23]

    GATEMAN Project – Wide- Bandwidth, High-Precision GNSS and Jammer Raw Data,

    P. Richter, R. M. Ferre, and E.-S. Lohan, “GATEMAN Project – Wide- Bandwidth, High-Precision GNSS and Jammer Raw Data,” inZenodo, Apr. 2019

    2019

  24. [24]

    Validation of Evil WaveForms in a GNSS Simulator for GPS and Galileo Signals,

    D. G ´omez-Casco, P. Crosta, and M. Spangenberg, “Validation of Evil WaveForms in a GNSS Simulator for GPS and Galileo Signals,” in NAVITEC, Noordwijk, Netherlands, Apr. 2022

    2022

  25. [25]

    Oak Ridge Spoofing and Interference Test Battery (OAKBAT) - GPS,

    A. Albright, S. Powers, J. Bonior, and F. Combs, “Oak Ridge Spoofing and Interference Test Battery (OAKBAT) - GPS,” inU.S. Department of Energy, Sep. 2020

    2020

  26. [26]

    Sionna: An Open-Source Library for Next-Generation Physical Layer Research,

    J. Hoydis, S. Cammerer, F. A. Aoudia, A. Vem, N. Binder, G. Marcus, and A. Keller, “Sionna: An Open-Source Library for Next-Generation Physical Layer Research,” inarXiv:2203.11854, Mar. 2022

    Pith/arXiv arXiv 2022

  27. [27]

    XceptionTime: A Novel Deep Architecture based on Depthwise Separa- ble Convolutions for Hand Gesture Classification,

    E. Rahimian, S. Zabihi, S. F. Atashzar, A. Asif, and A. Mohammadi, “XceptionTime: A Novel Deep Architecture based on Depthwise Separa- ble Convolutions for Hand Gesture Classification,” inarXiv:1911.03803, Nov. 2019

    Pith/arXiv arXiv 1911

  28. [28]

    Multiple Emitter Location and Signal Parameter Estima- tion,

    R. Schmidt, “Multiple Emitter Location and Signal Parameter Estima- tion,” inTAP, vol. 43(3), Mar. 1986, pp. 276–280

    1986

  29. [29]

    ESPRIT – Estimation of Signal Parameters via Rotational Invariance Techniques,

    R. Roy and T. Kailath, “ESPRIT – Estimation of Signal Parameters via Rotational Invariance Techniques,” inTAES, Jul. 1989

    1989

  30. [30]

    A Generalized Capon Estimator for Localization of Multiple Spread Sources,

    A. Hassanien, S. Shahbazpanahi, and A. B. Gershman, “A Generalized Capon Estimator for Localization of Multiple Spread Sources,” inIEEE TSP, vol. 52(1), Jan. 2024

    2024