Advancing Single-Snapshot DOA Estimation with Siamese Neural Networks for Sparse Linear Arrays

Hongshan Liu; Ruxin Zheng; Shunqiao Sun; Yimin D. Zhang

REVIEW 3 major objections 5 minor 24 references

Advancing Single-Snapshot DOA Estimation with Siamese Neural Networks for Sparse Linear Arrays

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a Siamese neural network with a sparse augmentation layer improves single-snapshot direction-of-arrival estimation on sparse linear arrays, outperforming a compressive-sensing baseline and two ablated variants in…

desk verdict Smart training trick for single-snapshot DOA estimation, but the reported gains are measured on the same random-sparsity distribution used for training, so the generality claims outrun the evidence. read the letter →

arxiv 2501.07008 v1 pith:WZL4NPGO submitted 2025-01-13 eess.SP stat.ML

Ruxin Zheng , Shunqiao Sun , Hongshan Liu , Yimin D. Zhang This is my paper · ORCID

classification eess.SPstat.ML

keywords direction-of-arrivalestimationsingle-snapshotprocessingsparselineararraysSiameseneuralnetworkcontrastivelossaugmentationautomotiveradarmultilabelclassification

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

Single-snapshot direction-of-arrival (DOA) estimation is hard when a radar array has missing antennas, whether by design or by sensor failure. The paper claims that a Siamese neural network with a sparse augmentation layer solves this more accurately than traditional compressed sensing and than networks without the augmentation or the contrastive loss. The sparse augmentation layer randomly zeroes antenna elements during training so the encoder learns to extract DOA content that is unchanged across array geometries, while the contrastive loss pulls signals with identical target directions into nearby embeddings regardless of noise, reflectivity, or which antennas are missing. On simulated sparse linear arrays with 30% element dropout, the proposed network reports higher accuracy, precision, recall, and F1 than CS-OMP and both ablated baselines across 0-30 dB SNR. This matters for automotive radar, where only one snapshot may be available and antennas are often sparse.

What carries the argument

The load-bearing mechanism is the sparse augmentation (SA) layer paired with a contrastive loss inside a Siamese architecture. The SA layer randomly zeroes up to 30% of the antenna elements of each input with a binary mask, and supplies the active-element count $N_{\mathrm{SLA}}$ for normalization, so the encoder cannot memorize a fixed array geometry. A frequency embedding layer maps the masked signal to the continuous frequency domain via $g(y)=A^H(\theta)y/N_{\mathrm{SLA}}$, enabling convolutional feature extraction. The twin encoders are trained with the contrastive loss $S=\frac{1}{P}\sum_j\left[z_j g_j^2+(1-z_j)\max(0,m-g_j)^2\right]$, where $g_j=\|v_{j1}-v_{j2}\|$ is the Euclidean distance between paired embeddings; same-DOA pairs are pulled together and different-DOA pairs are pushed apart by margin $m$. This embedding is then fed to a binary-cross-entropy multilabel classifier.

What would settle it

Evaluate the trained model on test signals drawn from a fixed sparse array geometry, for example the 7-element SLA1 shown in the paper, rather than uniformly random 30% dropout masks, over the same 0-30 dB SNR range; if accuracy or F1 falls back to the level of CS-OMP or BaseNet1, the claimed robustness is specific to the random-mask training distribution.

Watch

Extended reading notes

Core claim

The paper's central claim is that a Siamese neural network trained with contrastive loss, after a sparse augmentation layer that randomly masks antenna elements, learns feature embeddings in which signals with the same directions-of-arrival but different sparse array geometries, reflection coefficients, and noise levels are pulled together, while signals with different DOAs are pushed apart. The resulting encoder, combined with a multilabel classification head, yields more accurate single-snapshot DOA estimates on random sparse linear arrays than the two ablated baselines and than compressive-sensing orthogonal matching pursuit, across SNR levels from 0 to 30 dB. The feature analysis shows that the proposed network's PCA-reduced embeddings cluster more tightly than the baselines' for both ULA and random SLA test signals.

Load-bearing premise

The load-bearing premise is that real sparse arrays and sensor failures behave like the random binary masks used in training, with the same 30% dropout, same number of targets, same 0-30 dB SNR range, and same 1-degree grid; under different conditions the reported gains may disappear.

Editorial extensions

If this is right

On simulated random sparse linear arrays with 30% element dropout, the proposed SNN reports higher accuracy, precision, recall, and F1 than CS-OMP and the two ablated baselines across 0-30 dB SNR.
Signals that share the same DOAs but differ in array geometry, reflection coefficients, and noise map to tightly clustered embeddings, which is what makes the classifier robust.
Because it operates on a single snapshot, the method applies where only one measurement is available, such as fast-moving automotive radar scenes.
The sparse augmentation layer alone accounts for a large part of the improvement (BaseNet2 beats BaseNet1), while the contrastive loss adds further robustness.
The design targets the exponential training-data growth that comes from refining the angle grid or increasing the target count, by learning pair similarities instead of enumerating label combinations.

Reading between the lines

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

A natural next test is to train the same architecture on a fixed, designed sparse array geometry rather than random masks, since automotive sparse arrays are often deliberately thinned; the paper does not report this regime.
Because the contrastive loss learns geometry-invariant embeddings, the method could in principle transfer to other single-snapshot inverse problems such as sparse channel estimation, although the paper only studies DOA.
The 1-degree grid and three-target limit mean the reported F1 values are tied to that resolution; extending to finer grids or more targets would show whether the claimed reduction in training-data demand holds.
A physical radar test with measured noise and correlated element failures would be a stronger check than simulated random dropout, which does not capture mutual coupling or failure clusters.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Here's the short version: this paper has a genuinely useful training trick—Siamese network plus contrastive loss plus random sparse-mask augmentation for single-snapshot DOA estimation on sparse linear arrays—and it demonstrates the trick works in a clean ablation on synthetic data. What it does not do is demonstrate the generalization its abstract claims, because the test distribution is the same as the training augmentation.

The novel piece is the combination, not the parts. The sparse augmentation layer is a domain-appropriate data augmentation; the frequency embedding layer is a sensible way to feed masked signals to a CNN; and the contrastive loss is a natural fit for making DOA embeddings invariant to which array elements are active. All are standard tools, but I haven't seen them put together this way in the cited deep-learning DOA literature. The paper is also well organized, the equations are consistent, and the code is linked. The PCA feature plots in Figure 4 are a nice touch and back up the intuition that the embeddings get tighter.

The soft spots are the evaluation. Training uses random masks with max sparsity 0.3; testing uses random masks with sparsity 0.3, same SNR range, same grid, same K<=3. That's in-distribution evaluation. The motivating scenarios—unintended antenna failures and deliberately sparse array geometries—involve fixed masks or structured patterns, not independent random masks per snapshot. None of that is tested. There are no error bars or repeated runs, and the only non-ablation baseline is CS-OMP; no prior deep learning DOA method is compared against. The claim in Section IV-B that the proposed SNN shows 'exceptional performance' is therefore overstated. It shows good performance on the exact distribution it was trained for.

This is not disqualifying. The method is plausible, the write-up is honest about the architecture, and the code gives you a basis for verification. The right fix is either to narrow the generality claims or to add out-of-distribution tests: fixed random masks, a few structured sparse geometries, sparsity levels other than 0.3, and repeated seeds with error bars. I'd bet the method still holds up reasonably outside its training distribution, but the paper doesn't show it.

I'd send this to review, not desk reject. It's a solid little engineering paper that needs a revision rather than a rejection. If you work on automotive radar DOA or deep learning for array processing, it's worth a skim; otherwise it's not essential reading.

Referee Report

3 major / 5 minor

Summary. The paper proposes a Siamese neural network (SNN) for single-snapshot DOA estimation in sparse linear arrays, combining a sparse augmentation (SA) layer that randomly zeroes antenna elements during training, a frequency embedding (FE) layer, and a contrastive loss that pulls together embeddings of signals sharing the same DOA labels but differing in reflection coefficients, SNR, and array masks. The method is trained on a 20-element ULA with up to three targets on a 1-degree grid and evaluated on 5,000 simulated test signals from random SLAs with sparsity 0.3, comparing accuracy, precision, recall, and F1 score against two ablations (BaseNet1 without SA and contrastive loss; BaseNet2 without contrastive loss) and CS-OMP. The paper also provides PCA-based feature visualization and reports superior performance across all metrics.

Significance. The code is publicly available and the ablation design is clear, which are strengths. The idea of using a Siamese architecture with contrastive learning to handle randomly sparsed arrays is sensible and could be relevant for automotive radar with sensor failures. However, the significance is conditional: the reported gains are measured entirely on the same random-mask distribution used for training augmentation, and the comparison set omits existing deep-learning DOA baselines. If the evaluation is extended to fixed masks, different sparsity levels, structured sparse geometries, and statistical significance testing, the work would be a useful contribution. As it stands, the paper demonstrates in-distribution performance gains over two ablations and CS-OMP, but does not yet support the broader generalization claims in the abstract and conclusion.

major comments (3)

[Section IV-B and III-B/III-C] The evaluation is in-distribution with respect to the training-time sparse augmentation. Test signals in Section IV-B are generated by zeroing six random positions of a 20-element ULA to achieve sparsity 0.3, which is exactly the operation of the SA layer during training with maximum sparsity 0.3, and the same SNR range (0–30 dB), 1-degree grid, and target count (K ≤ 3) are used. The reported gains over BaseNet2 and CS-OMP therefore demonstrate performance on the training distribution but do not substantiate the abstract's claim that the method is effective for sparse arrays and sensor failures in general. Real sensor failures and designed sparse geometries are fixed per deployment rather than independently re-randomized per signal, and they may have different sparsity levels or structured element positions. Please add evaluations with fixed random masks, several sparsity levels (e.g., 0.1, 0.2, 0.4, 0.5), and standard sparse-array geometries (e.g., nested or co-prime arrays), reporting per-configuration metrics.
[Section IV-B] The paper reports single runs for each model and SNR point without error bars, confidence intervals, or repeated-seed experiments. Because the test set includes random noise and random masks, the apparent superiority of the proposed method over BaseNet2 at low SNR, where the curves nearly overlap in Figures 5(c) and 5(d), may not be statistically significant. Please provide the mean and standard deviation across multiple test draws or bootstrap confidence intervals, and state the number of Monte Carlo runs used.
[Section III-C and IV-B] The comparison set is limited to the two ablated baselines and CS-OMP. Existing deep-learning DOA estimators cited in the introduction (e.g., Papageorgiou et al. [10], Fuchs et al. [11], Gall et al. [13,14]) are not used as baselines, so the claim of 'superior performance' relative to the state of the art in deep-learning-based DOA estimation is not directly supported. At minimum, include a recent deep-learning single-snapshot DOA method adapted to the same array size and grid, or explicitly restrict the claim to the ablations and CS-OMP.

minor comments (5)

[Abstract] The GitHub URL is printed as 'https://github.com/ruxinzh/SNNS SLA' but should be 'SNNS_SLA' (the underscore is missing in the rendered text).
[Eq. (8)] The contrastive loss expression is typeset ambiguously; it should read S = (1/P) Σ_j [ z_j g_j^2 + (1 − z_j) max(0, m − g_j)^2 ].
[Section IV-B] It would help to state how the detection threshold of 0.5 is applied to the sigmoid output and whether it was tuned; the contrastive loss margin m is also never reported, even though it is a free parameter.
[Figure 5] The accuracy subplot uses a y-axis range starting near 0.965 while the other subplots start near 0.55, making visual comparisons across metrics misleading; a common scale or explicit annotation of the ranges would improve clarity.
[Section III-D] Please state whether a validation set was used for early stopping and whether the 5,000 test signals are guaranteed disjoint from the training combinations; this would clarify the generalization claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the network is trained and evaluated with independent ground-truth DOA labels; the overlap between training augmentation and test-time sparse masks is an evaluation-domain limitation, not a circular derivation.

full rationale

The paper's central claim is that the proposed Siamese neural network with a sparse augmentation layer and contrastive loss improves single-snapshot DOA estimation on random sparse linear arrays. The derivation chain is supervised end-to-end by independent ground-truth labels: Eq. (9) constructs the multilabel target directly from the true DOA values on the fixed grid, and the contrastive loss in Eq. (8) uses binary pair labels z_j determined by whether two signals share identical DOAs. Neither label is defined in terms of the reported evaluation metrics (accuracy, precision, recall, F1), and no trained parameter is fitted to those metrics. The sparse augmentation layer is a training-time data transformation, and the evaluation in Section IV-B uses the same random-mask sparsity model (0.3) as training, but this is a distributional match rather than an equation-level circularity: the test masks are not constructed from the network outputs, and the reported gains are measured on simulated data rather than derived from the method's own assumptions. The paper does cite prior work by the same authors, but those citations concern automotive radar context and sparse array systems, not the load-bearing justification for the SNN architecture or the contrastive loss. No self-cited uniqueness theorem or ansatz is invoked to force the design. The genuine limitation is that the evaluation does not establish generalization to fixed per-deployment sensor-failure masks, different sparsity levels, or structured sparse array geometries; that concern belongs to correctness and external validity, not circularity. Therefore the derivation is self-contained and the circularity score is 0.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The ledger lists hand-chosen hyperparameters and simulation assumptions that the evaluation depends on. No new physical entities are introduced; the sparse augmentation layer and frequency embedding layer are algorithmic components, not invented physical objects.

free parameters (4)

contrastive loss margin m = not reported in text
Hyperparameter in Eq. (8) controlling the minimum separation for dissimilar pairs; no value or sensitivity analysis is provided.
SA layer maximum sparsity = 0.3
Chosen for both training augmentation and evaluation, controlling the number of zeroed elements; no other sparsity levels are tested.
detection threshold = 0.5
Applied to the classifier output in Section IV-B to decide whether a DOA label is present; no threshold sweep is reported.
loss combination weight = not reported in text
The total loss is a linear combination of contrastive and binary cross-entropy losses, but the weighting factor is not specified.

assumptions (5)

domain assumption Signals are narrowband, far-field, and embedded in additive white complex Gaussian noise.
Standard data model in Eqs. (1)-(3), Section II; not challenged in the evaluation.
domain assumption Source reflection coefficients have magnitude uniformly distributed in [0.5, 1], with at most three sources separated by at least 1 degree.
Data generation constraints in Section III-C; limits the scenario coverage of the reported results.
ad hoc to paper Sensor failures and array sparsity are modeled as independent random zeroing of ULA elements, with test-time sparsity equal to the training augmentation sparsity.
The SA layer in Section III-B and the evaluation in Section IV-B both use sparsity 0.3 random masks; structured failure patterns or other sparsity ratios are not tested.
domain assumption The array manifold over the fixed 1-degree grid is known exactly and is used in the frequency embedding layer.
Eq. (6) assumes perfect knowledge of array geometry and grid; calibration errors are not modeled.
domain assumption DOA estimation is treated as multilabel classification on a fixed grid of 121 angles.
Problem formulation in Section III-A; off-grid sources and continuous DOA estimation are not considered.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Advancing Single-Snapshot DOA Estimation with Siamese Neural Networks for Sparse Linear Arrays." pith.science (2026). https://pith.science/paper/WZL4NPGO

@misc{pith2026250107008,
  author       = {Pith},
  title        = {Pith review of: Advancing Single-Snapshot DOA Estimation with Siamese Neural Networks for Sparse Linear Arrays},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WZL4NPGO}},
  note         = {Machine review of arXiv:2501.07008}
}

read the original abstract

Single-snapshot signal processing in sparse linear arrays has become increasingly vital, particularly in dynamic environments like automotive radar systems, where only limited snapshots are available. These arrays are often utilized either to cut manufacturing costs or result from unintended antenna failures, leading to challenges such as high sidelobe levels and compromised accuracy in direction-of-arrival (DOA) estimation. Despite deep learning's success in tasks such as DOA estimation, the need for extensive training data to increase target numbers or improve angular resolution poses significant challenges. In response, this paper presents a novel Siamese neural network (SNN) featuring a sparse augmentation layer, which enhances signal feature embedding and DOA estimation accuracy in sparse arrays. We demonstrate the enhanced DOA estimation performance of our approach through detailed feature analysis and performance evaluation. The code for this study is available at https://github.com/ruxinzh/SNNS_SLA.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 21 canonical work pages

[10]

Deep networks for direction-of-arrival estimation in low SNR,

G. K. Papageorgiou, M. Sellathurai, and Y . C. Eldar, “Deep networks for direction-of-arrival estimation in low SNR,” IEEE Transactions on Signal Processing , vol. 69, pp. 3714– 3729, 2021

work page 2021
[11]

A machine learning perspective on automotive radar direction of arrival estimation,

J. Fuchs, M. Gardill, M. L ¨ubke, A. Dubey, and F. Lurz, “A machine learning perspective on automotive radar direction of arrival estimation,” IEEE Access , vol. 10, pp. 6775–6797, 2022

work page 2022
[1]

MIMO radar for advanced driver-assistance systems and autonomous driving: Advantages and challenges,

S. Sun, A. P. Petropulu, and H. V . Poor, “MIMO radar for advanced driver-assistance systems and autonomous driving: Advantages and challenges,” IEEE Signal Processing Maga- zine, vol. 37, no. 4, pp. 98–117, 2020

2020
[2]

Markel, Radar for Fully Autonomous Driving

M. Markel, Radar for Fully Autonomous Driving . Artech House, 2022

work page 2022
[3]

Deep-neural-network- enabled vehicle detection using high-resolution automotive radar imaging,

R. Zheng, S. Sun, H. Liu, and T. Wu, “Deep-neural-network- enabled vehicle detection using high-resolution automotive radar imaging,” IEEE Transactions on Aerospace and Elec- tronic Systems, vol. 59, no. 5, pp. 4815–4830, 2023

work page 2023
[4]

A sparse linear array approach in automotive radars using matrix completion,

S. Sun and A. P. Petropulu, “A sparse linear array approach in automotive radars using matrix completion,” in ICASSP 2020- 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2020, pp. 8614– 8618

work page 2020
[5]

4D automotive radar sensing for autonomous vehicles: A sparsity-oriented approach,

S. Sun and Y . D. Zhang, “4D automotive radar sensing for autonomous vehicles: A sparsity-oriented approach,” IEEE Journal of Selected Topics in Signal Processing , vol. 15, no. 4, pp. 879–891, 2021

2021
[6]

Automotive FMCW radar with difference co-chirps,

L. Xu, S. Sun, K. V . Mishra, and Y . D. Zhang, “Automotive FMCW radar with difference co-chirps,” IEEE Transactions on Aerospace and Electronic Systems, vol. 59, no. 6, pp. 8145– 8165, 2023

work page 2023

Show all 24 references

[7]

Time-sensitive and distance-tolerant deep learning-based vehicle detection using high-resolution radar bird’s-eye-view images,

R. Zheng, S. Sun, H. Liu, and T. Wu, “Time-sensitive and distance-tolerant deep learning-based vehicle detection using high-resolution radar bird’s-eye-view images,” in 2023 IEEE Radar Conference (RadarConf23) . IEEE, 2023, pp. 1–6

2023
[8]

4D automotive radar exploiting sparse array optimization and compressive sensing,

R. Zheng, S. Sun, W. Kuo, T. Abatzoglou, and M. Markel, “4D automotive radar exploiting sparse array optimization and compressive sensing,” in 2023 57th Asilomar Conference on Signals, Systems, and Computers . IEEE, 2023, pp. 01–05

2023
[9]

On the design of linear sparse arrays with beampattern shift invariant properties,

R. Lin, G. Wang, and J. Li, “On the design of linear sparse arrays with beampattern shift invariant properties,” Signal Processing, vol. 199, p. 108632, 2022

2022
[12]

Neural- network-based DOA estimation in the presence of non- Gaussian interference,

S. Feintuch, J. Tabrikian, I. Bilik, and H. Permuter, “Neural- network-based DOA estimation in the presence of non- Gaussian interference,” IEEE Transactions on Aerospace and Electronic Systems, vol. 60, no. 1, pp. 119–132, 2024

2024
[13]

Spectrum- based single-snapshot super-resolution direction-of-arrival es- timation using deep learning,

M. Gall, M. Gardill, T. Horn, and J. Fuchs, “Spectrum- based single-snapshot super-resolution direction-of-arrival es- timation using deep learning,” in 2020 German Microwave Conference (GeMiC). IEEE, 2020, pp. 184–187

2020
[14]

Learning representations for neural networks applied to spectrum- based direction-of-arrival estimation for automotive radar,

M. Gall, M. Gardill, J. Fuchs, and T. Horn, “Learning representations for neural networks applied to spectrum- based direction-of-arrival estimation for automotive radar,” in 2020 IEEE/MTT-S International Microwave Symposium (IMS). IEEE, 2020, pp. 1031–1034

2020
[15]

Guaranteed deep learning for reliable radar signal processing,

S. Khobahi, A. Mostajeran, M. Emadi, P. Wang, and M. Soltanalian, “Guaranteed deep learning for reliable radar signal processing,” in IEEE 12th Sensor Array and Multichan- nel Signal Processing Workshop (SAM) , 2022, pp. 291–295

2022
[16]

Automotive radar sensing with sparse linear arrays using one-bit Hankel matrix completion,

A. Eamaz, F. Yeganegi, Y . Hu, S. Sun, and M. Soltanalian, “Automotive radar sensing with sparse linear arrays using one-bit Hankel matrix completion,” in 2024 IEEE Radar Conference (RadarConf24). IEEE, 2024, pp. 1–6

2024
[17]

Robust direction of arrival estimation in the presence of array faults using snapshot diversity,

G. C. Lee, A. S. Rawat, and G. W. Wornell, “Robust direction of arrival estimation in the presence of array faults using snapshot diversity,” in 2019 IEEE Global Conference on Signal and Information Processing (GlobalSIP) . IEEE, 2019, pp. 1–5

2019
[18]

Element failure correction for a large monopulse phased array antenna with active amplitude weighting,

W. P. Keizer, “Element failure correction for a large monopulse phased array antenna with active amplitude weighting,” IEEE Transactions on Antennas and Propagation , vol. 55, no. 8, pp. 2211–2218, 2007

2007
[19]

Robustness of coarrays of sparse arrays to sensor failures,

C.-L. Liu and P. Vaidyanathan, “Robustness of coarrays of sparse arrays to sensor failures,” in 2018 IEEE Interna- tional Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2018, pp. 3231–3235

2018
[20]

Direction of arrival (DoA) estimation under array sensor failures using a minimal resource allocation neural network,

S. Vigneshwaran, N. Sundararajan, and P. Saratchandran, “Direction of arrival (DoA) estimation under array sensor failures using a minimal resource allocation neural network,” IEEE Transactions on Antennas and Propagation , vol. 55, no. 2, pp. 334–343, 2007

2007
[21]

base network

designed for single-snapshot DOA estimation in SLAs. SNNs, utilizing twin identical networks, are ideal for applications requiring similarity learning between data elements. Our approach employs SNNs to discern similarities between signals with identical DOAs but varying in re...

2025 arXiv
[22]

Signature verification using a ‘Siamese’ time delay neural network,

J. Bromley, I. Guyon, Y . LeCun, E. S ¨ackinger, and R. Shah, “Signature verification using a ‘Siamese’ time delay neural network,” Advances in Neural Information Processing Sys- tems, vol. 6, 1993

1993
[23]

A survey on image data augmentation for deep learning,

C. Shorten and T. M. Khoshgoftaar, “A survey on image data augmentation for deep learning,” Journal of Big Data , vol. 6, no. 1, pp. 1–48, 2019

2019
[24]

Signal recovery from random measurements via orthogonal matching pursuit,

J. A. Tropp and A. C. Gilbert, “Signal recovery from random measurements via orthogonal matching pursuit,” IEEE Trans- actions on Information Theory , vol. 53, no. 12, pp. 4655– 4666, 2007

2007

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [10]

Deep networks for direction-of-arrival estimation in low SNR,

G. K. Papageorgiou, M. Sellathurai, and Y . C. Eldar, “Deep networks for direction-of-arrival estimation in low SNR,” IEEE Transactions on Signal Processing , vol. 69, pp. 3714– 3729, 2021

work page 2021

[2] [11]

A machine learning perspective on automotive radar direction of arrival estimation,

J. Fuchs, M. Gardill, M. L ¨ubke, A. Dubey, and F. Lurz, “A machine learning perspective on automotive radar direction of arrival estimation,” IEEE Access , vol. 10, pp. 6775–6797, 2022

work page 2022

[3] [1]

MIMO radar for advanced driver-assistance systems and autonomous driving: Advantages and challenges,

S. Sun, A. P. Petropulu, and H. V . Poor, “MIMO radar for advanced driver-assistance systems and autonomous driving: Advantages and challenges,” IEEE Signal Processing Maga- zine, vol. 37, no. 4, pp. 98–117, 2020

2020

[4] [2]

Markel, Radar for Fully Autonomous Driving

M. Markel, Radar for Fully Autonomous Driving . Artech House, 2022

work page 2022

[5] [3]

Deep-neural-network- enabled vehicle detection using high-resolution automotive radar imaging,

R. Zheng, S. Sun, H. Liu, and T. Wu, “Deep-neural-network- enabled vehicle detection using high-resolution automotive radar imaging,” IEEE Transactions on Aerospace and Elec- tronic Systems, vol. 59, no. 5, pp. 4815–4830, 2023

work page 2023

[6] [4]

A sparse linear array approach in automotive radars using matrix completion,

S. Sun and A. P. Petropulu, “A sparse linear array approach in automotive radars using matrix completion,” in ICASSP 2020- 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2020, pp. 8614– 8618

work page 2020

[7] [5]

4D automotive radar sensing for autonomous vehicles: A sparsity-oriented approach,

S. Sun and Y . D. Zhang, “4D automotive radar sensing for autonomous vehicles: A sparsity-oriented approach,” IEEE Journal of Selected Topics in Signal Processing , vol. 15, no. 4, pp. 879–891, 2021

2021

[8] [6]

Automotive FMCW radar with difference co-chirps,

L. Xu, S. Sun, K. V . Mishra, and Y . D. Zhang, “Automotive FMCW radar with difference co-chirps,” IEEE Transactions on Aerospace and Electronic Systems, vol. 59, no. 6, pp. 8145– 8165, 2023

work page 2023