SINR Maximizing Distributionally Robust Adaptive Beamforming

Kiarash Hassas Irani; Sergiy A. Vorobyov; Yongwei Huang

REVIEW 2 major objections 4 minor 1 cited by

SINR Maximizing Distributionally Robust Adaptive Beamforming

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that worst-case SINR beamforming under distributional uncertainty in both the INC matrix and steering vector reduces to a quadratic matrix inequality problem solvable by semidefinite relaxations.

desk verdict The central reformulation is invalid because Eq. (21) miscomputes the support function of a PSD-constrained Frobenius ball, so the paper's main claim does not hold as written. read the letter →

arxiv 2505.15493 v1 pith:G3EHHH7O submitted 2025-05-21 eess.SP

Kiarash Hassas Irani , Yongwei Huang , Sergiy A. Vorobyov This is my paper

classification eess.SP MSC 90C2290C2690C4694A12

keywords distributionallyrobustoptimizationadaptivebeamformingSINRmaximizationquadraticmatrixinequalitysemidefiniterelaxationrank-onerecoveryinterference-plus-noisecovariancemomentuncertainty

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes a robust adaptive beamforming design that treats both the interference-plus-noise covariance (INC) matrix and the desired-signal steering vector as random quantities with only partial distributional knowledge. It claims that the worst-case SINR maximization over two distributional ambiguity sets can be rewritten exactly as a quadratic matrix inequality (QMI) problem through strong duality and the S-procedure, and that a sequence of semidefinite (LMI) relaxations with a rank-one penalty converges to a rank-one solution, i.e., a usable beamforming vector. If correct, this gives a tractable beamformer whose output SINR is guarded against distributional mismatch in both the INC matrix and the steering vector, not just the steering vector alone. The paper supports the claim with simulations showing higher output SINR than two competing distributionally robust beamformers.

What carries the argument

The load-bearing object is the dual reformulation of the two inner distributionally robust problems. For the INC side, the support function of the set Z1 converts the worst-case expected power into an explicit convex term, such as rho2 times the Frobenius norm of ww^H + X when Z1 is a PSD Frobenius-norm ball. For the steering side, the semi-infinite constraint is converted by the S-procedure into the block matrix inequality (32), producing the QMI (35). Relaxing ww^H to a positive semidefinite matrix W yields the LMI (36), and Lemma IV.1, which states that tr(W) = ||W||_F for a nonzero PSD matrix W exactly when W has rank one, supplies the penalty term used in the iterative algorithm (39).

What would settle it

Compute the support function of the set Z1 = {R in H^N : R ≽ 0, ||R||_F ≤ 1} at the 2x2 matrix diag(1,-1). The paper's identity (21) predicts the value $\sqrt$(2), but the actual maximum is 1, attained at R = diag(1,0); checking this example settles whether the reformulated objective in (36) is exact.

Watch

Extended reading notes

Core claim

The paper's central claim is that problem (7), which minimizes over beamformers w the worst-case expected INC power over distributions in the set D1 while requiring the worst-case expected signal power over distributions in the set D2 to be at least one, can be equivalently reformulated as the QMI problem (35) and then solved through the LMI relaxation (36) together with Algorithm 1, which returns a rank-one solution W = ww^H. The reformulation uses strong duality for linear conic programs to replace the inner worst-case expectations by their duals, the S-procedure to encode the support-set constraints on the steering vector, and the trace-versus-Frobenius-norm equality to drive rank-one recovery.

Load-bearing premise

The whole derivation depends on the claim that the worst-case expected interference power over all positive semidefinite matrices whose Frobenius norm is bounded equals rho2 times the Frobenius norm of the beamformer's outer product; that equality fails for matrices that are not positive semidefinite, so the reformulated problem may not match the original worst case.

Editorial extensions

If this is right

A beamformer can be designed with distributional robustness on both the INC matrix and the steering vector, rather than only on the steering vector as in earlier distributionally robust beamformers.
The resulting optimization problem (36) is a convex LMI problem, so it can be handled by standard interior-point solvers.
Algorithm 1 produces a rank-one solution whose sequence of optimal values is non-increasing, so the stopping criterion is guaranteed to be met.
Alternative uncertainty sets for the steering vector and the INC matrix yield the QMI variants (59), (65), and (70), all solvable by the same rank-one recovery procedure.
Simulations indicate improved output SINR relative to the distributionally robust beamformers of references [21] and [17], and robustness to inaccurate presumed signal directions.
Note: the paper's own formulation (36), used in all simulations, relies on the support-function identity (21) that is not generally valid; see the weakest-assumption field below.

Reading between the lines

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

If the support-function identity (21) is not valid, then the objective actually minimized in (36) may be an upper bound on, rather than the exact value of, the worst-case expected INC power, so the reported SINR gains may not reflect the true worst-case performance of problem (7).
The rank-one penalty heuristic resembles a convex-concave procedure; a natural testable extension is to replace the Frobenius-ball support function with the exact PSD-ball support function, computed via an eigenvalue projection, and compare the resulting beamformers.
The same duality machinery should extend to other INC ambiguity sets, such as trace-bounded or spectral-norm-bounded sets where the support function is known in closed form; such variants would isolate the effect of the Frobenius-ball assumption on the reported SINR performance.

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.

Desk Editor's Note

read the letter

The headline is short: the paper's central reformulation rests on a false support-function identity, so the main claim doesn't hold as written. That said, the paper is not a throwaway. The joint distributional-uncertainty formulation for both the INC matrix and steering vector is genuinely new relative to the cited baselines [17] and [21], and the alternative uncertainty sets in Section V are useful extensions. The writing is mostly clear, and the simulation study is reasonably thorough, with sensitivity checks on the key radii.

Where it breaks: Eq. (21). The paper sets Z1 to the intersection of the Frobenius ball ||R||_F ≤ ρ2 with the PSD cone, and then claims by Cauchy-Schwarz that the support function is ρ2||ww^H + X||_F. That ignores the PSD constraint. For Hermitian M, the correct support function is ρ2 times the Frobenius norm of the positive-semidefinite part of M, not of M itself. A concrete example: M = diag(1,-1) gives support 1, not √2. Since X is free Hermitian and W is PSD, W+X is not guaranteed PSD, so the objective in (22) and the LMI problem (36) used in all simulations is not the worst-case expected INC power from (7). That is a load-bearing error, not a typo.

The convergence proof for Algorithm 1 has a separate gap: Eq. (51) asserts a proportionality between consecutive iterates that is never established from the stated premises. Even if equality in Cauchy-Schwarz holds asymptotically, the argument that β_{k+1}<1 is doing real work is hand-wavy. This is secondary to the Eq. (21) problem but would need fixing too.

On the citation front: the paper leans on the authors' own [31] for preliminary results, which is fine given the overlap is disclosed in footnote 1. Self-citation is not an issue here.

Who is this for? A reader working on robust beamforming might be interested in the problem formulation and the alternative uncertainty sets, but they should not trust the derived algorithm. The paper needs a corrected derivation—either changing Z1 to the unconstrained Hermitian ball, or computing the PSD-ball support function correctly and reworking the reformulation—plus a proper convergence analysis. As it stands, I would not send this to referees; I'd desk reject it. If the authors fix Eq. (21), the paper could become a solid contribution; the current version doesn't support its main claim.

Referee Report

2 major / 4 minor

Summary. The paper proposes a distributionally robust adaptive beamforming (RAB) formulation in which the interference-plus-noise covariance matrix and the steering vector are treated as random with moment-based uncertainty sets. The main claim is that the worst-case SINR maximization problem (7), with uncertainty sets D1 in (8) and D2 in (9), can be reformulated exactly as the quadratic matrix inequality problem (35) via strong duality and the S-procedure, and then relaxed to the LMI problem (36). An iterative rank-one recovery procedure (Algorithm 1) is proposed, together with a convergence analysis. Numerical simulations compare the resulting beamformer with two existing distributionally robust beamformers under various uncertainty sets and parameter settings.

Significance. If the central reformulation were correct, the paper would offer a genuinely new DRO-based beamforming framework that handles distributional uncertainty in both the INC matrix and the steering vector, with a tractable LMI-based solution method. The paper is clearly organized, the dual derivations are laid out step by step, and the simulation study covers several uncertainty-set variants. However, the central equivalence relies on a support-function identity in Eq. (21) that is false. Because this identity is used directly in the objective of problem (22) and then in (35), (36), (59), (65), and (70), the LMI problems solved in Algorithm 1 and in the simulations are not equivalent to the original DRO problem (7). The paper's main theoretical contribution is therefore not established, and the numerical results do not validate the claimed DRO reformulation. The convergence analysis of Algorithm 1 also has a separate gap. No code or machine-checked proofs are provided, so the analysis must stand on the analytic derivations, which contains a load-bearing error.

major comments (2)

[§III, Eq. (21)] The claimed support function identity is false. For Z1 = {R ∈ H^N : ‖R‖_F ≤ ρ2, R ≽ 0}, the support function is δ_{Z1}(M) = ρ2 · (∑_{λ_i>0} λ_i^2(M))^{1/2}, not ρ2‖M‖_F; the Cauchy–Schwarz maximizer R = ρ2 M/‖M‖_F is not PSD when M is indefinite. For M = diag(1,−1) and ρ2 = 1, the true support function value is 1, whereas Eq. (21) gives √2. Since X in problem (22) is an unrestricted Hermitian variable, W+X need not be PSD, so the equality cannot be rescued by assuming W+X ≽ 0. This error propagates into the objective of (22), (35), (36), (59), (65), and (70), so the LMI problem solved in Algorithm 1 and in all simulations is not an equivalent reformulation of the original DRO problem (7). The central claim of the paper is therefore not established.
[§IV-B, Eqs. (49)–(51)] The convergence analysis is incomplete. Equality in (49) for one pair (W^k, W^{k+1}) only yields W^{k+1} = β_k W^k for that pair; it does not imply that all subsequent iterates are proportional, nor does the monotone convergence of {v_k} force (49) to become equality at convergence. The asserted relation W^{k+2} = β_{k+1} W^{k+1} in Eq. (51) is therefore not derived from the stated premises, and the argument that the penalty term converges to zero is unsupported. Since Algorithm 1's termination guarantee rests on this reasoning, the rank-one convergence claim is not proven.

minor comments (4)

[§V-B, Eq. (70)] In the statement of problem (70), the symbol list contains a typographical double comma ('w, x∈C^N, , X, X′, Z≽0'); this should be cleaned up.
[§IV-A, p. 8] The claim that 'the number of iterations required for convergence is consistently small' is not supported by any experimental data; please provide iteration counts or a histogram across the simulated scenarios.
[§III, Eq. (16)] The support function δ_{Z1}(·) is used before it is defined; please define it explicitly at first use and clarify the relationship between the support set Z1 and the distributional set D1 in (8).
[§VI] The sensitivity analysis in Figs. 5 and 6 only sweeps ρ1 and γ1; the statement that the method is insensitive to parameters over a wide range is not demonstrated for ρ2, γ2, α, and η.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the DRO-to-QMI/LMI reformulation is a mathematical derivation with user-supplied uncertainty-set radii; the self-citation to [31] is a non-load-bearing pointer to a conference sketch.

full rationale

The paper's claimed derivation chain (7) -> (10)-(22) -> (28)-(35) -> (36) is a conic-duality and S-procedure reformulation. The support functions and dual programs are derived from the explicitly defined sets D1 and D2, and the parameters rho1, rho2, gamma1, gamma2, Delta, and alpha are chosen as user inputs in the numerical examples rather than recovered from the beamforming output, so no fitted quantity is being renamed as a prediction. The only self-reference is the footnote 'Some preliminary outcomes from this research work (covered in Subsection IV.A here) have been initially outlined in [31]' near the end of the Introduction; the rank-one algorithm and its convergence are then developed in Section IV with the paper's own Lemma IV.1, Proposition IV.2, and Algorithm 1, so the self-citation is not load-bearing. The concern raised about Eq. (21), namely that delta_{Z1}(ww^H+X)=rho2||ww^H+X||_F is not the correct support function when Z1 includes the PSD constraint R≽0, is a mathematical-correctness issue in the reformulation rather than an equivalence-by-construction or fit-by-construction circularity, and therefore does not increase the circularity score. Similarly, the asserted proportionality W^{k+2}=beta_{k+1}W^{k+1} behind Eq. (51) in Section IV.B is an omitted proof step, but it is a proof gap, not a circular reduction of the paper's claim to its own inputs. Overall, the derivation is self-contained in the sense relevant to circularity: its central claim depends on strong duality, support functions, and the S-procedure applied to the stated uncertainty sets, not on fitting or on the authors' prior results.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The central derivation rests on standard conic duality, the S-procedure, and the assumption that a0 and Sigma are known. The uncertainty-set radii are user-chosen simulation parameters, not fitted constants. No new physical or mathematical entities are postulated.

free parameters (7)

rho1 = 0.001||S0||_F for D1; 0.01||S0||_F in the N=32 scenario
Radius of the similarity constraint on the expected INC matrix in D1; hand-chosen for simulations.
rho2 = 1.1 tr(S0)
Upper bound on the support set Z1 for the INC matrix; used in the objective of (36).
gamma1 = 0.01||a0||
First-order moment bound for the steering vector distribution in D2; hand-chosen.
gamma2 = 0.1
Second-order moment slack for the steering vector distribution in D2; hand-chosen.
Delta = 0.1
Norm deviation parameter in the steering-vector support set Z2 = {a: (1-Delta)N <= ||a||^2 <= (1+Delta)N}.
alpha = 10^3
Penalty parameter in the iterative LMI problem (39), asserted to be sufficiently large to enforce a rank-one solution.
eta = 10^-6
Termination threshold in Algorithm 1.

assumptions (4)

standard math Strong duality holds for the distributional reformulations because the uncertainty sets are non-empty, as claimed via Prop. 3.4 of [34].
Invoked after Eq. (18) and later for D2 variants. The cited proposition requires regularity conditions beyond simple non-emptiness, and the paper does not verify Slater-type conditions for the PSD mean constraint and similarity constraint.
standard math The S-procedure is lossless for the two-quadratic support set Z2 defined by the annulus in (33).
Used in Eq. (34) and in Section V; the S-procedure requires an interior point and a constraint qualification, which are assumed without explicit verification.
domain assumption The true mean a0 and covariance Sigma of the random steering vector are known or reliably estimated.
Used to define D2 in (9) and throughout the dual reformulation; in practice a0 and Sigma are estimated from a presumed angular sector, so estimation error enters the model.
domain assumption Higher-order moments of the steering vector and INC distributions beyond those constrained do not materially affect the robust design.
Stated in Section II to justify truncating D2 to first- and second-order moments; this is a modeling choice that limits the family of distributions considered.

how reviews work

0 comments

Cite this review

Pith. "Pith review of SINR Maximizing Distributionally Robust Adaptive Beamforming." pith.science (2026). https://pith.science/paper/G3EHHH7O

@misc{pith2026250515493,
  author       = {Pith},
  title        = {Pith review of: SINR Maximizing Distributionally Robust Adaptive Beamforming},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G3EHHH7O}},
  note         = {Machine review of arXiv:2505.15493}
}

read the original abstract

This paper addresses the robust adaptive beamforming (RAB) problem via the worst-case signal-to-interference-plus-noise ratio (SINR) maximization over distributional uncertainty sets for the random interference-plus-noise covariance (INC) matrix and desired signal steering vector. Our study explores two distinct uncertainty sets for the INC matrix and three for the steering vector. The uncertainty sets of the INC matrix account for the support and the positive semidefinite (PSD) mean of the distribution, as well as a similarity constraint on the mean. The uncertainty sets for the steering vector consist of the constraints on the first- and second-order moments of its associated probability distribution. The RAB problem is formulated as the minimization of the worst-case expected value of the SINR denominator over any distribution within the uncertainty set of the INC matrix, subject to the condition that the expected value of the numerator is greater than or equal to one for every distribution within the uncertainty set of the steering vector. By leveraging the strong duality of linear conic programming, this RAB problem is reformulated as a quadratic matrix inequality problem. Subsequently, it is addressed by iteratively solving a sequence of linear matrix inequality relaxation problems, incorporating a penalty term for the rank-one PSD matrix constraint. We further analyze the convergence of the iterative algorithm. The proposed robust beamforming approach is validated through simulation examples, which illustrate improved performance in terms of the array output SINR.

Figures

Figures reproduced from arXiv: 2505.15493 by the authors.

**Figure 4.** Array output SINR versus number of snapshots with SNR [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗

**Figure 5.** Array output SINR versus ρ1/kS0kF at SNR = −10 dB with T = 100 snapshots. The parameter ρ1 defines the size of the uncertainty set for the INC matrix. The figure shows that the beamformer performance remains stable over a wide range of ρ1 values, indicating that precise tuning is not necessary for reliable operation. 10−3 10−2 10−1 100 101 102 γ1/ka0k −1.0 −0.8 −0.6 −0.4 −0.2 0.0 Output SINR [dB] Optimal SINR Propos… view at source ↗

Figures from the paper (1 more)

**Figure 6.** Figure 6: Array output SINR versus γ1/ka0k at SNR = −10 dB with T = 100 snapshots. The parameter γ1 controls the level of uncertainty in the steering vector model. The figure shows that the SINR is largely unaffected by changes in γ1, demonstrating the robustness of the proposed…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Wasserstein Distributionally Robust Adaptive Beamforming
eess.SP 2025-06 conditional novelty 4.0 of 10

Wasserstein distributionally robust beamforming with norm and quadratic costs reduces exactly to classical worst-case and ellipsoidal robust beamformers, making the framework a unifying perspective.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · cited by 1 Pith paper

[17]

Robust minimum variance beamform- ing under distributional uncertainty,

X. Zhang, Y . Li, N. Ge and J. Lu, “Robust minimum variance beamform- ing under distributional uncertainty,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Brisbane, Australia, Apr. 2015, pp. 2514–2518

work page 2015
[21]

A distributionally r obust minimum variance beamformer design,

B. Li, Y . Rong, J. Sun and K. L. Teo, “A distributionally r obust minimum variance beamformer design,” IEEE Signal Processing Letters , vol. 25, no. 1, pp. 105–109, Jan. 2018

work page 2018
[31]

Robust adaptive be amforming maximizing the worst-case SINR over distributional uncert ainty sets for random INC matrix and signal steering vector,

Y . Huang, W. Y ang and S. A. V orobyov, “Robust adaptive be amforming maximizing the worst-case SINR over distributional uncert ainty sets for random INC matrix and signal steering vector,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Singapore, Singapore, May 2022, pp. 4918–4922

work page 2022
[1]

Chapter 9 - Recent advance s in network beamforming,

S. ShahbazPanahi and Y . Jing, “Chapter 9 - Recent advance s in network beamforming,” in Academic Press Library in Signal Processing, V olume 7, R. Chellappa and S. Theodoridis, Eds., Academic Press, 2018 , pp. 403– 477

work page 2018
[2]

Beamforming: A versatil e approach to spatial ﬁltering,

B. D. V an V een and K. M. Buckley, “Beamforming: A versatil e approach to spatial ﬁltering,” in IEEE ASSP Magazine , vol. 5, no. 2, pp. 4–24, Apr. 1988

work page 1988
[3]

Chapter 12 - Adaptive and Robust Beamfor ming,

S. A. V orobyov, “Chapter 12 - Adaptive and Robust Beamfor ming,” in Academic Press Library in Signal Processing, V olume 3, A. M. Zoubir, M. Viberg, R. Chellappa, and S. Theodoridis, Eds., Elsevier, 2 014, pp. 503– 552

work page
[4]

Li and P

J. Li and P . Stoica, Robust Adaptive Beamforming , John Wiley & Sons, Hoboken, NJ, 2006

work page 2006
[5]

Convex optimization-based beamforming : from receive to transmit and network designs,

A. B. Gershman, N. D. Sidiropoulos, S. Shahbazpanahi, M. Bengtsson, and B. Ottersten, “Convex optimization-based beamforming : from receive to transmit and network designs,” IEEE Signal Processing Magazine , vol. 27, no. 3, pp. 62–75, May 2010

work page 2010

Show all 38 references

[6]

Principles of minimum variance robust a daptive beam- forming design,

S. A. V orobyov, “Principles of minimum variance robust a daptive beam- forming design,” Signal Processing , vol. 93, no. 12, pp. 3264–3277, Dec. 2013

2013
[7]

Signal waveform estimation in the presence of uncertainties about the steering vector,

O. Besson, A. A. Monakov, and C. Chalus, “Signal waveform estimation in the presence of uncertainties about the steering vector, ” IEEE Trans- actions on Signal Processing , vol. 52, no. 9, pp. 2432–2440, Sept. 2004

2004
[8]

Performance analysis of beamf ormers using generalized loading of the covariance matrix in the presenc e of random steering vector errors,

O. Besson and F. Vincent, “Performance analysis of beamf ormers using generalized loading of the covariance matrix in the presenc e of random steering vector errors,” IEEE Transactions on Signal Processing , vol. 53, no. 2, pp. 452–459, Feb. 2005

2005
[9]

On the relatio nship between robust minimum variance beamformers with probabil istic and worst-case distortionless response constraints,

S. A. V orobyov, H. Chen and A. B. Gershman, “On the relatio nship between robust minimum variance beamformers with probabil istic and worst-case distortionless response constraints,” IEEE Transactions on Signal Processing, vol. 56, no. 11, pp. 5719–5724, Nov. 2008

2008
[10]

Robus t adaptive beamforming using worst-case performance optimization: A solution to the signal mismatch problem,

S. A. V orobyov, A. B. Gershman, and Zhi-Quan Luo, “Robus t adaptive beamforming using worst-case performance optimization: A solution to the signal mismatch problem,” IEEE Transactions on Signal Processing , vol. 51, no. 2, pp. 313–324, Feb. 2003

2003
[11]

Robust beamforming via worst-case SINR maximization,

S. -J. Kim, A. Magnani, A. Mutapcic, S. P . Boyd and Z. -Q. L uo, “Robust beamforming via worst-case SINR maximization,” IEEE Transactions on Signal Processing, vol. 56, no. 4, pp. 1539–1547, Apr. 2008

2008
[12]

Robust ada ptive beamforming via worst-case SINR maximization with nonconv ex uncer- tainty sets,

Y . Huang, H. Fu, S. A. V orobyov and Z. -Q. Luo, “Robust ada ptive beamforming via worst-case SINR maximization with nonconv ex uncer- tainty sets,” IEEE Transactions on Signal Processing , vol. 71, pp. 218– 232, 2023

2023
[13]

Shapiro, D

A. Shapiro, D. Dentcheva, and A. Ruszczy´ nski, Lectures on Stochastic Programming: Modeling and Theory , MPS-SIAM series on optimization, Society for Industrial and Applied Mathematics, Philadelp hia, USA, 2nd Ed., 2014

2014
[14]

J. R. Birge and F. Louveaux, Introduction to Stochastic Programming , Springer, New Y ork, 2nd Ed., 2011

2011
[15]

Frameworks and results in di stributionally robust optimization,

H. Rahimian; S. Mehrotra, “Frameworks and results in di stributionally robust optimization,” Open Journal of Mathematical Optimization , vol. 3, no. 4, pp. 1–85, 2022

2022
[16]

Quantitative Risk Management Using Robust Optimization,

E. Delage, “Quantitative Risk Management Using Robust Optimization,” Lecture Notes, HEC Montr´ eal, Jan. 2021

2021
[18]

The S-procedure and duality relations in nonconvex problems of quadratic programming,

A. L. Fradkov and V . A. Y akubovich, “The S-procedure and duality relations in nonconvex problems of quadratic programming, ” V estnik Leningrad University, Leningrad, Russia, vol. 6, pp. 101–1 09, 1979. En- glish translation. Original Russian publication in V estni k Leningr...

1979
[19]

Distributionally rob ust chance- constrained minimum variance beamforming,

X. Zhang, Q. Feng, N. Ge and J. Lu, “Distributionally rob ust chance- constrained minimum variance beamforming,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Shanghai, China, Mar. 2016, pp. 2881–2885

2016
[20]

Distributionally ro bust joint chance constraints with second-order moment information,

S. Zymler, D. Kuhn, and B. Rustem, “Distributionally ro bust joint chance constraints with second-order moment information,” Mathematical Programming, Series A , vol. 137, pp. 167–198, 2013

2013
[22]

Robust adaptive beamforming based on inter- ference covariance matrix reconstruction and steering vec tor estimation,

Y . Gu and A. Leshem, “Robust adaptive beamforming based on inter- ference covariance matrix reconstruction and steering vec tor estimation,” IEEE Transactions on Signal Processing , vol. 60, no. 7, pp. 3881–3885, July 2012

2012
[23]

Robust adaptive beamformin g using a low-complexity shrinkage-based mismatch estimation algo rithm,

H. Ruan and R. C. de Lamare, “Robust adaptive beamformin g using a low-complexity shrinkage-based mismatch estimation algo rithm,” IEEE Signal Processing Letters , vol. 21, no. 1, pp. 60–64, Jan. 2014

2014
[24]

Robust adaptive beamformin g based on low-rank and cross-correlation techniques,

H. Ruan and R. C. de Lamare, “Robust adaptive beamformin g based on low-rank and cross-correlation techniques,” IEEE Transactions on Signal Processing, vol. 64, no. 15, pp. 3919–3932, Aug. 2016

2016
[25]

Robust adaptive beamforming based on power method process ing and spatial spectrum matching,

S. Mohammadzadeh, V . H. Nascimento, R. C. De Lamare and O . Kukrer, “Robust adaptive beamforming based on power method process ing and spatial spectrum matching,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Singapore, Singapore, May 2022, pp. ...

2022
[26]

Robust adaptive beam forming with a novel interference-plus-noise covariance matrix re construction method,

L. Huang, J. Zhang, X. Xu and Z. Y e, “Robust adaptive beam forming with a novel interference-plus-noise covariance matrix re construction method,” IEEE Transactions on Signal Processing , vol. 63, no. 7, pp. 1643–1650, April 2015

2015
[27]

Interferen ce- plus-noise covariance matrix reconstruction via spatial p ower spectrum sampling for robust adaptive beamforming,

Z. Zhang, W. Liu, W. Leng, A. Wang and H. Shi, “Interferen ce- plus-noise covariance matrix reconstruction via spatial p ower spectrum sampling for robust adaptive beamforming,” IEEE Signal Processing Letters, vol. 23, no. 1, pp. 121–125, Jan. 2016

2016
[28]

Robust adapt ive beamforming with precise main beam control,

B. Liao, C. Guo, L. Huang, Q. Li and H. C. So, “Robust adapt ive beamforming with precise main beam control,” IEEE Transactions on Aerospace and Electronic Systems, vol. 53, no. 1, pp. 345–356, Feb. 2017

2017
[29]

A bayesian approach to in- formed spatial ﬁltering with robustness against DOA estima tion errors,

S. Chakrabarty and E. A. P . Habets, “A bayesian approach to in- formed spatial ﬁltering with robustness against DOA estima tion errors,” IEEE/ACM Transactions on Audio, Speech, and Language Proce ssing, vol. 26, no. 1, pp. 145–160, Jan. 2018

2018
[30]

High-dimensional MVDR beamforming: Optimized solutions based on spiked random ma trix mod- els,

L. Y ang, M. R. McKay and R. Couillet, “High-dimensional MVDR beamforming: Optimized solutions based on spiked random ma trix mod- els,” IEEE Transactions on Signal Processing , vol. 66, no. 7, pp. 1933– 1947, April 2018

1933
[32]

H. H. Andersen, M. Hojbjerre, D. Sorensen, and P . S. Erik sen, Linear and Graphical Models for the Multivariate Complex Normal Di stribution, Lecture Notes in Statistics, Springer, New Y ork, NY , 1995

1995
[33]

Beck, First-Order Methods in Optimization , MPS-SIAM Series on Optimization, SIAM, Philadelphia, PA, 2017

A. Beck, First-Order Methods in Optimization , MPS-SIAM Series on Optimization, SIAM, Philadelphia, PA, 2017

2017
[34]

On duality theory of conic linear problems ,

A. Shapiro, “On duality theory of conic linear problems ,” in Semi- Inﬁnite Programming, M. ´A. Goberna and M. A. L ´ opez, Eds. Nonconvex Optimization and Its Applications, vol. 57, Boston, MA: Spr inger, 2001, pp. 135–165

2001
[35]

Nesterov and A

Y . Nesterov and A. Nemirovskii, Interior-Point Polynomial Algorithms in Convex Programming , SIAM, 1994

1994
[36]

Robust adaptive beamforming based on steering vector estimation w ith as little as possible prior information,

A. Khabbazibasmenj, S. A. V orobyov, and A. Hassanien, “ Robust adaptive beamforming based on steering vector estimation w ith as little as possible prior information,” IEEE Transactions on Signal Processing , vol. 60, no. 6, pp. 2974–2987, June 2012

2012
[37]

CVXPY: A Python-embedded model ing language for convex optimization,

S. Diamond and S. Boyd, “CVXPY: A Python-embedded model ing language for convex optimization,” J. Mach. Learn. Res. , vol. 17, no. 1, pp. 2909–2913, Jan. 2016

2016
[38]

CVX: Matlab software for discipli ned con- vex programming, version 2.0 beta,

M. Grant and S. Boyd, “CVX: Matlab software for discipli ned con- vex programming, version 2.0 beta,” Available: https://cvxr.com/cvx, Sep. 2013

2013

Pith tools

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

[1] [17]

Robust minimum variance beamform- ing under distributional uncertainty,

X. Zhang, Y . Li, N. Ge and J. Lu, “Robust minimum variance beamform- ing under distributional uncertainty,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Brisbane, Australia, Apr. 2015, pp. 2514–2518

work page 2015

[2] [21]

A distributionally r obust minimum variance beamformer design,

B. Li, Y . Rong, J. Sun and K. L. Teo, “A distributionally r obust minimum variance beamformer design,” IEEE Signal Processing Letters , vol. 25, no. 1, pp. 105–109, Jan. 2018

work page 2018

[3] [31]

Robust adaptive be amforming maximizing the worst-case SINR over distributional uncert ainty sets for random INC matrix and signal steering vector,

Y . Huang, W. Y ang and S. A. V orobyov, “Robust adaptive be amforming maximizing the worst-case SINR over distributional uncert ainty sets for random INC matrix and signal steering vector,” in Proc. IEEE Int. Conf. Acoustics, Speech, and Signal Processing (ICASSP) , Singapore, Singapore, May 2022, pp. 4918–4922

work page 2022

[4] [1]

Chapter 9 - Recent advance s in network beamforming,

S. ShahbazPanahi and Y . Jing, “Chapter 9 - Recent advance s in network beamforming,” in Academic Press Library in Signal Processing, V olume 7, R. Chellappa and S. Theodoridis, Eds., Academic Press, 2018 , pp. 403– 477

work page 2018

[5] [2]

Beamforming: A versatil e approach to spatial ﬁltering,

B. D. V an V een and K. M. Buckley, “Beamforming: A versatil e approach to spatial ﬁltering,” in IEEE ASSP Magazine , vol. 5, no. 2, pp. 4–24, Apr. 1988

work page 1988

[6] [3]

Chapter 12 - Adaptive and Robust Beamfor ming,

S. A. V orobyov, “Chapter 12 - Adaptive and Robust Beamfor ming,” in Academic Press Library in Signal Processing, V olume 3, A. M. Zoubir, M. Viberg, R. Chellappa, and S. Theodoridis, Eds., Elsevier, 2 014, pp. 503– 552

work page

[7] [4]

Li and P

J. Li and P . Stoica, Robust Adaptive Beamforming , John Wiley & Sons, Hoboken, NJ, 2006

work page 2006

[8] [5]

Convex optimization-based beamforming : from receive to transmit and network designs,

A. B. Gershman, N. D. Sidiropoulos, S. Shahbazpanahi, M. Bengtsson, and B. Ottersten, “Convex optimization-based beamforming : from receive to transmit and network designs,” IEEE Signal Processing Magazine , vol. 27, no. 3, pp. 62–75, May 2010

work page 2010