Pith. sign in

REVIEW 2 major objections 4 minor 60 references

Multi-target Range, Doppler and Angle estimation in MIMO-FMCW Radar with Limited Measurements

T0 review · 2 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

Pith's one-line read The paper shows that a MIMO-FMCW radar with a random sparse array and randomly selected chirps can match full-array detection, and proves recovery guarantees when Doppler and angle grids sit at characteristic-function zeros.

desk verdict Real contribution with a genuine theory-simulation mismatch: the high-resolution grids used in experiments violate the conditions of the recovery theorems, so the headline resolution claim is not actually proven. read the letter →

arxiv 2502.01147 v2 pith:43BVSTAN submitted 2025-02-03 eess.SP

classification eess.SP MSC 94A12
keywords MIMO-FMCWradarcompressivesensingsparselineararrayrandomchirpselectionrange-Doppler-angleestimationorthogonalmatchingpursuitrecoveryguaranteesmutualcoherence
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

The paper is trying to establish that a MIMO-FMCW radar does not need its full uniform array or its full chirp train to locate multiple targets in range, velocity, and angle. It proposes transmitting only a randomly chosen subset of chirps and placing a small number of transmitter and receiver elements randomly over the aperture, then solving a 2D sparse-recovery problem for Doppler and angle after range is estimated by DFT-focusing or OMP. The paper proves, under grid and distribution conditions, that the induced measurement matrix has low coherence and satisfies isotropy with high probability, yielding uniform and non-uniform recovery guarantees. If true, the practical payoff is that a 2x4 sparse array with 10 random chirps performs like a 4x8 uniform array with 32 chirps at realistic SNRs, with better grid-based resolution and lower runtime.

What carries the argument

The load-bearing object is the vectorized sensing matrix $D = B \otimes C$, built from the Doppler dictionary $B$ ($P \times G_D$, columns $\exp(j(4\pi T_c/\lambda)\rho_i \zeta_p)$) and the angular dictionary $C$ ($N_T N_R \times G_\theta$, columns $\exp(j(\pi A/\lambda)\sin\phi_j(\alpha_n+\beta_m))$). The analysis reduces recovery quality to the random variables $\Gamma_B$ and $\Gamma_C$, the normalized inner products of dictionary columns; coherence of $D$ is bounded by their maxima. Conditions (26)-(27) and (33) place the grid spacings at the zeros of the chirp and antenna-sum characteristic functions, making $\Gamma_B$ and $\Gamma_C$ asymptotically Gaussian, and that drives the coherence bounds and isotropy that feed Theorems 1-4.

What would settle it

Take the Section V setup and compute the mutual coherence of $D = B \otimes C$ on the fine grids (Doppler spacing 0.78 m/s, $\sin\theta$ spacing 0.0204) for one random 2x4 sparse array and $P=10$ chirps; if the coherence is far above the RIP-safe level $\Lambda/(2K-1)$ at $K=5$ while OMP still hits the reported detection rates, then the resolution claim rests on unpromised OMP behavior. A direct hit-rate comparison between the zero-spaced grids and the fine grids at the same SNR would settle whether the resolution advantage is real or a dictionary-coherence artifact.

Watch

Extended reading notes

Core claim

On the paper's own terms, target localization in a MIMO-FMCW radar is a 3D spectral estimation problem: the IF signal separates into a range beat frequency $\exp(j2\pi \Omega_R t)$, a Doppler phase $\exp(j2\pi \Omega_D \zeta_p)$, and an angular phase $\exp(j2\pi \Omega_\theta (\alpha_n+\beta_m))$. With random sparse chirps and a random sparse linear array, the joint Doppler-angle part becomes $Y = C Z B^T + W$, and after vectorization $y = (B \otimes C) z + w$. The central discovery is that the Kronecker dictionary $B \otimes C$ inherits controlled coherence and isotropy from the random chirp and antenna distributions, provided the Doppler and angular grids are uniform and spaced at the zeros of the characteristic functions of the chirp and antenna-sum distributions (conditions (26)-(27) and (33)). Under those conditions, Theorems 1-4 bound the number of chirps and antenna elements needed for exact $\ell^1$ recovery with high probability, and the simulations show detection performance close to full-measurement DFT and MUSIC with roughly half the antennas and less than one third of the chirps.

Load-bearing premise

The recovery guarantees apply only when the Doppler and angular grids are spaced exactly at the zeros of the characteristic functions of the random chirp and antenna-sum distributions; the much finer grids used in the paper's high-resolution simulations lie outside those conditions.

Editorial extensions

If this is right

  • A 2x4 random sparse array transmitting 10 random chirps per CPI can deliver hit rates comparable to a 4x8 uniform linear array transmitting 32 chirps for SNRs above roughly -20 dB.
  • Because the required chirp and antenna counts grow logarithmically with the number of grid points, finer Doppler and angular grids do not force a linear increase in hardware.
  • The non-uniform guarantee allows trading chirps against antennas: the product $P \cdot N_T N_R$ only needs to exceed a constant times $K \log^2(G_D G_\theta/\epsilon)$.
  • Range-OMP recovers three close targets (48.8, 50.0, 51.2 m) from one chirp and one array channel at -20 dB SNR, where coherent DFT detects only two.
  • 2D-OMP is the fastest CS solver among those tested, with a speedup factor of $N_T N_R P/(N_T N_R + P)$ over vectorized 1D-OMP.

Reading between the lines

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

  • The simulation's resolution claims use grids six to sixteen times finer than the theory's zero-spaced grids, so the practical resolution advantage is an empirical property of OMP rather than a consequence of Theorems 1-4.
  • If fine-grid dictionaries remain low-coherence at these measurement counts, the same design should support even sparser chirp sets or larger search sectors than the simulated 10 chirps and single sector.
  • The sparse-chirp design frees most of the CPI and the paper notes this can scan other angular sectors; a direct next test is to run the estimator sector-by-sector and measure detection loss against the single-sector curves.
  • The calibration-error study suggests the random sparse array is roughly as sensitive as MUSIC, which points toward calibration-aware dictionary design as the natural follow-up.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. This paper proposes a compressive-sensing (CS) framework for joint range, Doppler, and angle-of-arrival estimation in MIMO-FMCW radar that uses a random sparse linear array (2 transmitters, 4 receivers over a 12λ aperture) and transmits only P = 10 of P_max = 32 chirps per CPI. The IF signal is modeled as a separable mixture of three complex exponentials (Eq. (3)). Range is estimated either by DFT-focusing with binary integration or by a grid-based Range-OMP operating on a single chirp and channel; velocities and AOAs are then estimated jointly from the Kronecker dictionary D = B ⊗ C (Eqs. (19)-(20)) using vectorized OMP, BP, LASSO, or 2D-OMP. The authors derive uniform and non-uniform recovery guarantees (Theorems 1-4): coherence and isotropy conditions for this structured random matrix under grid-spacing conditions (26)-(27) and (33), sufficient measurement-count scaling laws, and a practical parameter recipe (Proposition 5) that maps chirp and antenna distributions to grid spacings at zeros of the characteristic functions. Simulations compare false-alarm and hit rates, RMSE, ROC, and runtime against classical-DFT and MUSIC with full ULA measurements, reporting parity in detection at moderate SNR with better grid-based RMSE at lower measurement counts, together with a calibration-error sensitivity study (Appendix F).

Significance. If the results hold, this is a meaningful step toward reducing hardware and computational cost in automotive MIMO-FMCW radar: matching a 4×8 ULA with 32 chirps using a 2×4 random SLA with 10 chirps at practical SNRs is a substantive claim, and the structured-random-matrix analysis of the Kronecker dictionary D = B ⊗ C goes beyond the pulsed-radar analysis of [32]. Strengths to credit explicitly: the proofs in Appendices A-E are detailed, Proposition 5 converts abstract characteristic-function conditions into concrete checkable radar parameters, the measurement-count scaling laws in Theorems 2 and 4 (P scaling as (K−1/2)^2 log(G_D/ε_1), N_T N_R as (K−1/2)^2 log^2(G_θ/ε_2), and P N_T N_R as K log^2(G_D G_θ/ε)) are falsifiable predictions, and the experimental section is unusually complete, including ROC curves, scaling with target number, runtime, and calibration-error analysis. The significance is currently qualified by the regime gap between the provable zero-spacing grids and the demonstrated fine-grid resolution (Major Comment 1); once that gap is closed or clearly labeled, the paper would be a solid contribution to the sparse-radar literature.

major comments (2)
  1. [Section IV-C / Section V-B] Section IV-C / Section V-B: the simulated fine grids violate the grid-spacing conditions on which Theorems 1-4 are built, so the paper's headline 'higher resolution' claim is not covered by its own guarantees. With Table III parameters (A = A_T + A_R = 12λ, P_max = 32, λ = 12.5 mm), Proposition 5 fixes the theorem-compliant Doppler spacing at λ/(2P_max T_c) = 4.88 m/s and the sin(θ) spacing at 2λ/A = 0.1667. The Section V-B simulations instead use 200 Doppler points over ±78 m/s (spacing 0.78 m/s) and 50 sin(θ) points over [−0.5, 0.5] (spacing 0.0204), about 6× and 8× finer. At these spacings conditions (26)-(27) and (33) fail: the adjacent-column normalized inner products of B and C are approximately 0.96 and 0.95 (the characteristic functions Ψ_p and Ψ_ξ evaluated at the fine-grid differences are far from their zeros), so the Rayleigh tail in Lemma 2, the coherence bound (28), and the isotropy argument behind Theorem 4 all break down in exactly the regime used for the RMSE comparisons in Fig. 4. This is an internal tension rather than a 'sufficient but not necessary' subtlety, because Remark 6 states that grid points 'are not free variables' once (26)-(27) are enforced, and Remark 3 concedes that OMP recovery degrades when grid points are too close. The same observation applies to Range-OMP: the 0.12 m range grid has adjacent steering-vector inner products essentially equal to 1, and no theorem covers it. To support the abstract's resolution claim, the authors should (i) run experiments that validate Theorems 1-4 in the regime they actually cover (e.g., spacings 4.88 m/s and 0.1667, varying P and N_T N_R), (ii) report the fine-grid dictionary coherence and either analyze the fine-grid regime or explicitly label those results as empirical findings outside the guarantees, and (iii) qualify the resolution claim in the abstract accordingly.
  2. [Section IV-B, Theorem 4] Section IV-B, Theorem 4: the proof is a one-sentence citation stating that non-uniform recovery 'can be guaranteed by generalizing [54, Theorem 2.1]', followed by three substitutions (M = P N_T N_R rows, sparsity K, dimension G_D G_θ). The hypotheses of [54, Theorem 2.1] (isotropic and bounded measurement rows, row-wise independence) are not verified for D = B ⊗ C. Section IV-B itself notes that the rows of C and hence of D are 'generally not independent'; in fact, rows of D sharing a chirp index ζ_p share the same B row, and rows of C sharing a transmitter position α_n are dependent. That dependence is exactly what makes the claimed generalization non-trivial, and the universal constants κ_5 ≤ 2.87 × 10^6 and κ_7 ≤ 23.513 need justification to transfer. As written, the non-uniform guarantee (34)-(35) is load-bearing but not established.
minor comments (4)
  1. [Section V-B] Section V-B: the hit-rate definition ('within the corresponding resolution from the true target parameters') uses a method-dependent tolerance whose numerical values are never reported. For MUSIC, Table I describes the resolution only qualitatively ('higher than classical-DFT; depends on the array aperture, number of chirps, SNR and search grid density'), so the Doppler and angular windows used for MUSIC in Figs. 5-7 are unclear. Please state the exact tolerance windows for every method and parameter, and ideally also report hit rates under one common fixed tolerance to check robustness of the detection-parity claim.
  2. [Appendix B, Lemma 2] Appendix B, Lemma 2: the lemma treats ζ_1..P as i.i.d. draws from P_p, but Section II specifies sampling 'without replacement' ('randomly drawn distinct integers'). The equality P(|Γ_B| > θ) = exp(−θ^2 P) is therefore not exact: the without-replacement variance carries a finite-population factor of about 1 − (P−1)/(P_max−1) ≈ 0.69 for the simulation parameters. Because this correction reduces the variance, the stated bound remains conservative, but the statement should be amended or the sampling model clarified, e.g., by invoking the CLT for sampling without replacement.
  3. [Section V-B] Section V-B: the claim that the 50-point sin(θ)-uniform grid over [−30°, 30°] gives 'an approximate angular resolution of 2°' is inconsistent with the stated grid; 50 points over sin(θ) in [−0.5, 0.5] gives spacing 1/49 ≈ 0.0204, which corresponds to about 1.17° near broadside and 1.35° at ±30°. Please correct the value or explain how the 2° figure is obtained.
  4. [Sections III-B.1 and III-C] Sections III-B.1 and III-C: several typos should be corrected: 'for thp-th chirp' should read 'the p-th chirp'; 'the choice of girds' should read 'the choice of grids'; and in Section V-A, 'Range-OMP considers onlyy 1,1,1 channel' should read 'only the y_{1,1,1} channel'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorems 1–4 are derived from explicit random-model assumptions via external standard CS results, and no fitted parameter is renamed as a prediction.

full rationale

The paper's derivation chain is self-contained. The IF signal model (3) is built from the FMCW mixing equations (2) under stated narrow-band approximations; the dictionaries B and C in (16)–(17) are defined directly from this model; and the recovery guarantees in Theorems 1–4 are proved from the characteristic-function conditions (26)–(27)/(33) using standard external results ([32], [53], [54]), not from a fit to the simulated data. The only self-citation is [1], a conference precursor explicitly described as lacking theoretical guarantees, and it is not load-bearing for any theorem or experiment. The resolution advantage reported in Section V is tied to user-selected grid spacings (0.78 m/s Doppler, about 0.0204 sin-theta spacing), which are finer than the zero-spacing grids required by the theorems; the paper itself notes in Remarks 6 and 7 that grid points are not free under (26)–(27)/(33). This is a genuine coverage gap between the theoretical conditions and the simulation regime, so it is a correctness risk for the 'higher resolution' portion of the central claim, but it is not an equation-level circularity: no predicted quantity is defined in terms of itself, and no fitted parameter is relabeled as a prediction. The empirical RMSE and hit-rate comparisons are experimental demonstrations rather than consequences derived from the theorems, and the statement that estimation accuracy is 'primarily determined by the resolution of the corresponding grids' is an acknowledged design choice, not a circular derivation.

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

The central result rests on the separable 3D exponential model, the sparsity assumption, and characteristic-function grid conditions. The user-defined grids and OMP termination sparsity drive the claimed resolution in simulations. No new physical entities are introduced.

free parameters (5)
  • Range-OMP grid spacing = 0.12 m over 1.2 m to 120 m
    User-defined grid in Section V-A. It determines the claimed range resolution and is not tied to any recovery guarantee; it is finer than the 0.6 m DFT range bin.
  • Doppler grid spacing = 0.78 m/s (200 points in -78 to 78 m/s)
    User-defined in Section V-B. It yields the claimed velocity resolution. The required theoretical spacing from Proposition 5 is lambda/(2 Pmax Tc) = 4.88 m/s, so the simulation operates outside the guaranteed regime.
  • Angular grid spacing = 50 points uniform in sin(theta) over -30 to 30 degrees
    User-defined in Section V-B, giving about 2 degree AOA resolution. The required theoretical sin(theta) spacing is 2 lambda/A = 0.333, so condition (27)/(33) is violated by the simulated grid.
  • OMP termination sparsity Kmax = 20
    Used to stop Range-OMP, 1D-OMP, and 2D-OMP in Section V-B. It is larger than the true target count K (5 in most runs), is chosen by hand, and affects detection performance without entering the theoretical guarantees.
  • LASSO regularization parameter = 0.6
    Chosen for the LASSO comparison in Section V-B. It affects the hit and false alarm tradeoff but is not part of the theoretical claims.
assumptions (4)
  • domain assumption The IF signal is a separable 3D mixture of complex exponentials in fast time, slow time, and array position, with all cross-phase terms absorbed into the target amplitude (Equation 3).
    Section II relies on tau_R >> tau_D and tau_R >> tau_theta and on (tau)^2 approximately (tau_R)^2. If the narrowband or far-field approximations fail, the model itself breaks.
  • domain assumption The target scene is sparse: at most K targets per detected range bin, and the discretization error from choosing a finite grid is negligible.
    CS recovery, Range-OMP, and 2D-OMP all depend on z being exactly sparse on a known grid. Off-grid targets are not analyzed.
  • domain assumption The random chirp and antenna distributions and the grids satisfy the characteristic-function zero conditions (26)-(27) and (33).
    Theorems 1-4 require these conditions. The simulations in Section V-B choose grids that do not satisfy them, so the guarantees do not apply to the demonstrated high-resolution setup.
  • ad hoc to paper Chirp indices sampled without replacement can be modeled with an i.i.d. central limit theorem in Lemma 2.
    Lemma 2 assumes an asymptotic Gaussian distribution for Gamma_B with variance 1/(2P), treating the P selected chirp indices as i.i.d. draws from P_p, while Section II states that the chirps are distinct and drawn without replacement. The resulting dependence is not modeled.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-target Range, Doppler and Angle estimation in MIMO-FMCW Radar with Limited Measurements." pith.science (2026). https://pith.science/paper/43BVSTAN

@misc{pith2026250201147,
  author       = {Pith},
  title        = {Pith review of: Multi-target Range, Doppler and Angle estimation in MIMO-FMCW Radar with Limited Measurements},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/43BVSTAN}},
  note         = {Machine review of arXiv:2502.01147}
}
read the original abstract

Multiple-input multiple-output (MIMO) radar offers several performance and flexibility advantages over traditional radar arrays. However, high angular and Doppler resolutions necessitate a large number of antenna elements and the transmission of numerous chirps, leading to increased hardware and computational complexity. While compressive sensing (CS) has recently been applied to pulsed-waveform radars with sparse measurements, its application to frequency-modulated continuous wave (FMCW) radar for target detection remains largely unexplored. In this paper, we propose a novel CS-based multi-target localization algorithm in the range, Doppler, and angular domains for MIMO-FMCW radar, where we jointly estimate targets' velocities and angles of arrival. To this end, we present a signal model for sparse-random and uniform linear arrays based on three-dimensional spectral estimation. For range estimation, we propose a discrete Fourier transform (DFT)-based focusing and orthogonal matching pursuit (OMP)-based techniques, each with distinct advantages, while two-dimensional CS is used for joint Doppler-angle estimation. Leveraging the properties of structured random matrices, we establish theoretical uniform and non-uniform recovery guarantees with high probability for the proposed framework. Our numerical experiments demonstrate that our methods achieve similar detection performance and higher resolution compared to conventional DFT and MUSIC with fewer transmitted chirps and antenna elements.

Figures

Figures reproduced from arXiv: 2502.01147 by the authors.

Figure 1
Figure 1. MIMO-FMCW radar setup with sparse chirps and antenna array: (a) random SLA (green and red triangles denote [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Graphical illustration of range estimation using (a) [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Range channels for DFT with (a) coherent integration, and (b) Channel-A ( [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: RMSE in (a) range, (b) velocity, and (c) AOA estimation at different SNRs for the OMP+Binary, OMP+Range-OMP, [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: (a) False-alarm, and (b) hit rates at different SNRs for [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Hit rate with varying number of targets K at SNR= 30dB, & -20dB, for OMP+Binary, classical-DFT and MUSIC. for SNRs above −20 dB, but with only half the number of antenna elements and less than half transmitted chirps. Note that the performance degradation at very low S…
Figure 8
Figure 8. Figure 8: (a) False alarm and (b) hit rates at different SNRs for [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: (a) False alarm, and (b) hit rates under varying [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]
Figure 10
Figure 10. Figure 10: RMSE in (a) range and (b) AOA estimation un [PITH_FULL_IMAGE:figures/full_fig_p018_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 53 canonical work pages

  1. [32]

    Spatial compressive sensing for MIMO radar,

    M. Rossi, A. M. Haimovich, and Y . C. Eldar, “Spatial compressive sensing for MIMO radar,”IEEE Transactions on Signal Processing, vol. 62, no. 2, pp. 419–430, 2013

  2. [1]

    Multi-target range and angle detection for mimo-fmcw radar with limited antennas,

    H. Singh and A. Chattopadhyay, “Multi-target range and angle detection for mimo-fmcw radar with limited antennas,” in2023 31st European Signal Processing Conference (EUSIPCO). IEEE, 2023, pp. 725–729

  3. [2]

    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 Magazine, vol. 37, no. 4, pp. 98– 117, 2020

  4. [3]

    Automotive radars: A review of signal processing techniques,

    S. M. Patole, M. Torlak, D. Wang, and M. Ali, “Automotive radars: A review of signal processing techniques,”IEEE Signal Processing Magazine, vol. 34, no. 2, pp. 22–35, 2017

  5. [4]

    Signal processing for FMCW SAR,

    A. Meta, P. Hoogeboom, and L. P. Ligthart, “Signal processing for FMCW SAR,”IEEE Transactions on Geoscience and Remote Sensing, vol. 45, no. 11, pp. 3519–3532, 2007

  6. [5]

    A new distortion cor- rection method for FMCW SAR real-time imaging,

    C. Gu, W. Chang, X. Li, G. Jia, and X. Luan, “A new distortion cor- rection method for FMCW SAR real-time imaging,”IEEE Geoscience and Remote Sensing Letters, vol. 14, no. 3, pp. 429–433, 2017

  7. [6]

    Radar sensor signal acquisition and multi- dimensional FFT processing for surveillance applications in transport systems,

    S. Saponara and B. Neri, “Radar sensor signal acquisition and multi- dimensional FFT processing for surveillance applications in transport systems,”IEEE Transactions on Instrumentation and Measurement, vol. 66, no. 4, pp. 604–615, 2017

  8. [7]

    A low-complexity FMCW surveillance radar algorithm using two random beat signals,

    B.-s. Kim, Y . Jin, S. Kim, and J. Lee, “A low-complexity FMCW surveillance radar algorithm using two random beat signals,”Sensors, vol. 19, no. 3, p. 608, 2019

Show all 60 references
  1. [8]

    Simultaneous monitoring of multiple people’s vital sign leveraging a single phased-MIMO radar,

    Z. Xu, C. Shi, T. Zhang, S. Li, Y . Yuan, C.-T. M. Wu, Y . Chen, and A. Petropulu, “Simultaneous monitoring of multiple people’s vital sign leveraging a single phased-MIMO radar,”IEEE Journal of Electromag- netics, RF and Microwaves in Medicine and Biology, vol. 6, no. 3, pp. ...

  2. [9]

    A 77-GHz FMCW MIMO radar based on an SiGe single- chip transceiver,

    R. Feger, C. Wagner, S. Schuster, S. Scheiblhofer, H. Jager, and A. Stelzer, “A 77-GHz FMCW MIMO radar based on an SiGe single- chip transceiver,”IEEE Transactions on Microwave theory and Tech- niques, vol. 57, no. 5, pp. 1020–1035, 2009

  3. [10]

    M. A. Richards,Fundamentals of radar signal processing. McGraw- Hill Education, 2014

  4. [11]

    MUSIC-based algorithm for range-azimuth FMCW radar data process- ing without estimating number of targets,

    G. O. Manokhin, Z. T. Erdyneev, A. A. Geltser, and E. A. Monastyrev, “MUSIC-based algorithm for range-azimuth FMCW radar data process- ing without estimating number of targets,” inMediterranean Microwave Symposium (MMS). IEEE, 2015, pp. 1–4

  5. [12]

    2D-MUSIC technique applied to a coherent FMCW MIMO radar,

    F. Belfiori, W. van Rossum, and P. Hoogeboom, “2D-MUSIC technique applied to a coherent FMCW MIMO radar,” inIET International Conference on Radar Systems (Radar 2012), 2012, pp. 1–6

  6. [13]

    Analysis of joint angle-frequency estimation using ESPRIT,

    A. N. Lemma, A.-J. Van Der Veen, and E. F. Deprettere, “Analysis of joint angle-frequency estimation using ESPRIT,”IEEE Transactions on Signal Processing, vol. 51, no. 5, pp. 1264–1283, 2003. 20

  7. [14]

    Joint DFT-ESPRIT estimation for TOA and DOA in vehicle FMCW radars,

    S. Kim, D. Oh, and J. Lee, “Joint DFT-ESPRIT estimation for TOA and DOA in vehicle FMCW radars,”IEEE Antennas and Wireless Propagation Letters, vol. 14, pp. 1710–1713, 2015

  8. [15]

    Eigenstructure-based algorithms for direc- tion finding with time-varying arrays,

    B. Friedlander and A. Zeira, “Eigenstructure-based algorithms for direc- tion finding with time-varying arrays,”IEEE Transactions on Aerospace and Electronic Systems, vol. 32, no. 2, pp. 689–701, 1996

  9. [16]

    Elad,Sparse and redundant representations: from theory to appli- cations in signal and image processing

    M. Elad,Sparse and redundant representations: from theory to appli- cations in signal and image processing. Springer, 2010, vol. 2, no. 1

  10. [17]

    Compressive radar imaging,

    R. Baraniuk and P. Steeghs, “Compressive radar imaging,” in2007 IEEE Radar conference. IEEE, 2007, pp. 128–133

  11. [18]

    Compressed sensing radar,

    M. Herman and T. Strohmer, “Compressed sensing radar,” in2008 IEEE Radar Conference. IEEE, 2008, pp. 1–6

  12. [19]

    High-resolution radar via compressed sensing,

    M. A. Herman and T. Strohmer, “High-resolution radar via compressed sensing,”IEEE Transactions on Signal Processing, vol. 57, no. 6, pp. 2275–2284, 2009

  13. [20]

    Orthogonal least squares methods and their application to non-linear system identification,

    S. Chen, S. A. Billings, and W. Luo, “Orthogonal least squares methods and their application to non-linear system identification,”International Journal of Control, vol. 50, no. 5, pp. 1873–1896, 1989

  14. [21]

    Sparsity and compressed sensing in radar imaging,

    L. C. Potter, E. Ertin, J. T. Parker, and M. Cetin, “Sparsity and compressed sensing in radar imaging,”Proceedings of the IEEE, vol. 98, no. 6, pp. 1006–1020, 2010

  15. [22]

    Compressed sensing technique for high- resolution radar imaging,

    Y .-S. Yoon and M. G. Amin, “Compressed sensing technique for high- resolution radar imaging,” inSignal Processing, Sensor Fusion, and Target Recognition XVII, vol. 6968. SPIE, 2008, pp. 506–515

  16. [23]

    Design and analysis of compressed sensing radar detectors,

    L. Anitori, A. Maleki, M. Otten, R. G. Baraniuk, and P. Hoogeboom, “Design and analysis of compressed sensing radar detectors,”IEEE Transactions on Signal Processing, vol. 61, no. 4, pp. 813–827, 2012

  17. [24]

    Spatial compressive sensing for direction-of-arrival estimation of multiple sources using dynamic sensor arrays,

    I. Bilik, “Spatial compressive sensing for direction-of-arrival estimation of multiple sources using dynamic sensor arrays,”IEEE Transactions on Aerospace and Electronic Systems, vol. 47, no. 3, pp. 1754–1769, 2011

  18. [25]

    Compressed-domain detection and estimation for colocated MIMO radar,

    E. Tohidi, A. Hariri, H. Behroozi, M. M. Nayebi, G. Leus, and A. P. Petropulu, “Compressed-domain detection and estimation for colocated MIMO radar,”IEEE Transactions on Aerospace and Electronic Systems, vol. 56, no. 6, pp. 4504–4518, 2020

  19. [26]

    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

  20. [27]

    Sub-Nyquist radar via Doppler focusing,

    O. Bar-Ilan and Y . C. Eldar, “Sub-Nyquist radar via Doppler focusing,” IEEE Transactions on Signal Processing, vol. 62, no. 7, pp. 1796–1811, 2014

  21. [28]

    SUMMeR: Sub-nyquist MIMO radar,

    D. Cohen, D. Cohen, Y . C. Eldar, and A. M. Haimovich, “SUMMeR: Sub-nyquist MIMO radar,”IEEE Transactions on Signal Processing, vol. 66, no. 16, pp. 4315–4330, 2018

  22. [29]

    Sub-Nyquist radar: Principles and prototypes,

    K. V . Mishra, Y . C. Eldar, A. Maio, and A. Haimovich, “Sub-Nyquist radar: Principles and prototypes,” inCompressed sensing in radar signal processing. Cambridge University Press, 2019, pp. 1–48

  23. [30]

    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,” inIEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2020, pp. 8614–8618

  24. [31]

    Sparse antenna array design for MIMO radar using softmax selection,

    K. Diamantaras, Z. Xu, and A. Petropulu, “Sparse antenna array design for MIMO radar using softmax selection,”arXiv preprint arXiv:2102.05092, 2021

  25. [33]

    MIMO radar using compressive sampling,

    Y . Yu, A. P. Petropulu, and H. V . Poor, “MIMO radar using compressive sampling,”IEEE Journal of Selected Topics in Signal Processing, vol. 4, no. 1, pp. 146–163, 2010

  26. [34]

    Three dimensional compressive sensing in MIMO radar,

    Y . Liu, Y . Zhang, J. Tang, N. Zhang, and W. Zhu, “Three dimensional compressive sensing in MIMO radar,” in2015 49th Asilomar Conference on Signals, Systems and Computers. IEEE, 2015, pp. 599–603

  27. [35]

    Sparse reconstruction of chirplets for automotive FMCW radar interference mitigation,

    A. Correas-Serrano and M. A. Gonzalez-Huici, “Sparse reconstruction of chirplets for automotive FMCW radar interference mitigation,” in2019 IEEE MTT-S International Conference on Microwaves for Intelligent Mobility (ICMIM), 2019, pp. 1–4

  28. [36]

    Spectrum sharing radar: Coex- istence via Xampling,

    D. Cohen, K. V . Mishra, and Y . C. Eldar, “Spectrum sharing radar: Coex- istence via Xampling,”IEEE Transactions on Aerospace and Electronic Systems, vol. 54, no. 3, pp. 1279–1296, 2017

  29. [37]

    FRaC: FMCW-based joint radar-communications system via index modula- tion,

    D. Ma, N. Shlezinger, T. Huang, Y . Liu, and Y . C. Eldar, “FRaC: FMCW-based joint radar-communications system via index modula- tion,”IEEE Journal of Selected Topics in Signal Processing, vol. 15, no. 6, pp. 1348–1364, 2021

  30. [38]

    Compressed sensing- based multitarget CFAR detection algorithm for FMCW radar,

    Z. Cao, J. Li, C. Song, Z. Xu, and X. Wang, “Compressed sensing- based multitarget CFAR detection algorithm for FMCW radar,”IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 11, pp. 9160–9172, 2021

  31. [39]

    On compressive sensing applied to radar,

    J. H. Ender, “On compressive sensing applied to radar,”Signal Process- ing, vol. 90, no. 5, pp. 1402–1414, 2010

  32. [40]

    De Wit, W

    J. De Wit, W. Van Rossum, and A. De Jong,Orthogonal waveforms for FMCW MIMO radar. IEEE, 2011

  33. [41]

    Nearly orthogonal waveforms for MIMO FMCW radar,

    G. Babur, O. A. Krasnov, A. Yarovoy, and P. Aubry, “Nearly orthogonal waveforms for MIMO FMCW radar,”IEEE Transactions on Aerospace and Electronic Systems, vol. 49, no. 3, pp. 1426–1437, 2013

  34. [42]

    Range Doppler detection for automotive FMCW radars,

    V . Winkler, “Range Doppler detection for automotive FMCW radars,” in2007 European Radar Conference. IEEE, 2007, pp. 166–169

  35. [43]

    Y . C. Eldar and G. Kutyniok,Compressed sensing: theory and applica- tions. Cambridge university press, 2012

  36. [44]

    Orthogonal matching pursuit: Recursive function approximation with applications to wavelet decomposition,

    Y . C. Pati, R. Rezaiifar, and P. S. Krishnaprasad, “Orthogonal matching pursuit: Recursive function approximation with applications to wavelet decomposition,” inProceedings of 27th Asilomar conference on Signals, Systems and Computers. IEEE, 1993, pp. 40–44

  37. [45]

    CoSaMP: Iterative signal recovery from in- complete and inaccurate samples,

    D. Needell and J. A. Tropp, “CoSaMP: Iterative signal recovery from in- complete and inaccurate samples,”Applied and computational harmonic analysis, vol. 26, no. 3, pp. 301–321, 2009

  38. [46]

    An introduction to compressive sampling,

    E. J. Cand `es and M. B. Wakin, “An introduction to compressive sampling,”IEEE Signal Processing magazine, vol. 25, no. 2, pp. 21–30, 2008

  39. [47]

    Regression shrinkage and selection via the lasso,

    R. Tibshirani, “Regression shrinkage and selection via the lasso,”Jour- nal of the Royal Statistical Society Series B: Statistical Methodology, vol. 58, no. 1, pp. 267–288, 1996

  40. [48]

    2D sparse signal recovery via 2D orthogonal matching pursuit,

    Y . hugel, J. Wu, and B. Huang, “2D sparse signal recovery via 2D orthogonal matching pursuit,”Science China Information Sciences, vol. 55, pp. 889–897, 2012

  41. [49]

    Fast OMP algorithm for 2D angle estimation in MIMO radar,

    Y . Liu, M. Wu, and S. Wu, “Fast OMP algorithm for 2D angle estimation in MIMO radar,”Electronics letters, vol. 46, no. 6, pp. 444–445, 2010

  42. [50]

    An algorithm for the machine calculation of complex Fourier series,

    J. W. Cooley and J. W. Tukey, “An algorithm for the machine calculation of complex Fourier series,”Mathematics of Computation, vol. 19, no. 90, pp. 297–301, 1965

  43. [51]

    Signal recovery from random mea- surements via orthogonal matching pursuit,

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

  44. [52]

    Foucart and H

    S. Foucart and H. Rauhut,An invitation to compressive sensing. Springer, 2013

  45. [53]

    Compressive sensing and structured random matrices,

    H. Rauhut, “Compressive sensing and structured random matrices,” Theoretical foundations and numerical methods for sparse recovery, vol. 9, no. 1, p. 92, 2010

  46. [54]

    Remote sensing via l1- minimization,

    M. H ¨ugel, H. Rauhut, and T. Strohmer, “Remote sensing via l1- minimization,”Foundations of Computational Mathematics, vol. 14, no. 1, pp. 115–150, 2014

  47. [55]

    A probabilistic and RIPless theory of compressed sensing,

    E. J. Candes and Y . Plan, “A probabilistic and RIPless theory of compressed sensing,”IEEE Transactions on Information Theory, vol. 57, no. 11, pp. 7235–7254, 2011

  48. [56]

    D. H. Johnson and D. E. Dudgeon,Array Signal Processing: Concepts and Techniques. Prentice-Hall, 1993

  49. [57]

    A mathematical theory of antenna arrays with randomly spaced elements,

    Y . Lo, “A mathematical theory of antenna arrays with randomly spaced elements,”IEEE Transactions on Antennas and Propagation, vol. 12, no. 3, pp. 257–268, 1964

  50. [58]

    TenDSuR: Tensor-based 4D sub-nyquist radar,

    S. Na, K. V . Mishra, Y . Liu, Y . C. Eldar, and X. Wang, “TenDSuR: Tensor-based 4D sub-nyquist radar,”IEEE Signal Processing Letters, vol. 26, no. 2, pp. 237–241, 2018

  51. [59]

    G. N. Watson,A treatise on the theory of Bessel functions. The University Press, 1922, vol. 2

  52. [60]

    On the Lambert W function,

    R. M. Corless, G. H. Gonnet, D. E. Hare, D. J. Jeffrey, and D. E. Knuth, “On the Lambert W function,”Advances in Computational mathematics, vol. 5, pp. 329–359, 1996

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.