Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

The paper claims that increasing the proportion of near-field channels improves device activity detection in cell-free massive MIMO, and provides a distributed algorithm that converges to a stationary point of the maximum-likelihood problem

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

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

T0 review · deepseek-v4-flash

2026-08-04 16:12 UTC pith:QC7B33VE

load-bearing objection A practical distributed algorithm for hybrid near-far field activity detection, but the headline theoretical claim about near-field proportion is not actually proven; the convergence proof also has a suspect sign. the 4 major comments →

arxiv 2509.15162 v2 pith:QC7B33VE submitted 2025-09-18 eess.SP

A Unified Distributed Algorithm for Hybrid Near-Far Field Activity Detection in Cell-Free Massive MIMO

Jingreng Lei , Yang Li , Ziyue Wang , Qingfeng Lin , Ya-Feng Liu , Yik-Chung Wu This is my paper
classification eess.SP
keywords activity detectioncell-free massive MIMOhybrid near-far fieldcovariance-based detectiondistributed ADMMgrant-free random accessidentifiabilitycoordinate descent
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper tackles device activity detection in cell-free massive MIMO when some devices are in the near field of an access point and others in the far field. It builds a covariance-based maximum-likelihood formulation that captures the correlated near-field channels, and claims a theoretical result: the more near-field channels, the better the detection performance, because the covariance descriptors become more distinct. To make the method practical, it proposes a distributed algorithm in which access points solve local subproblems and exchange only detection results with the central processor, with a proof of convergence to a stationary point. If correct, the work would let operators exploit near-field propagation (for example, by enlarging Rayleigh distance through wavelength or array size) to improve massive machine-type access while keeping fronthaul load low.

Core claim

This paper establishes a covariance-based maximum-likelihood formulation for device activity detection in cell-free massive MIMO when users lie in both near-field and far-field regions of each AP, and proves that the columns of the covariance descriptor matrix Ψ_m become more orthogonal as the fraction of near-field channels grows (Proposition 1). Since identifiability of the activity vector is tied to the nullspace of these columns, the paper concludes that more near-field channels improve detection performance. It then proposes an ADMM-based distributed algorithm in which each AP solves a local coordinate-descent subproblem using a Sherman-Morrison-Woodbury update with Taylor expansion, an

What carries the argument

The central object is the covariance matrix C_m of the vectorized received signal at AP m, written as a sum over devices of Ξ_{m,n} ⊗ (s_n s_n^H) plus noise, where Ξ_{m,n} is the near-field channel covariance R_{m,n} or far-field g_{m,n} I_K. The argument turns on the columns ψ_{m,n} = vec(X_{m,n}X_{m,n}^H) of the matrix Ψ_m; Proposition 1 shows the cosine similarity between these columns is ordered by channel type because the matrix-correlation factor tr(Ξ_{m,n}Ξ_{m,n'}) / (||Ξ_{m,n}||_F ||Ξ_{m,n'}||_F) equals 1 for two far-field devices and is strictly smaller when either device is near-field. The distributed algorithm is carried by an ADMM consensus reformulation θ_m = a, with a local coo

Load-bearing premise

The paper's claim that more near-field channels improve detection rests on treating lower pairwise cosine similarity between covariance-descriptor columns as the mechanism for identifiability, but identifiability actually requires the global nullspace condition V ∩ D = ∅ (and also equality of the mean vectors), and the paper does not prove that the cosine-similarity ordering implies this global condition.

What would settle it

For a small system (e.g., M=2 APs, K=4 antennas, L=4 signatures, N=6 devices), compute the true identifiability condition by checking, for every pair of distinct activity vectors a and ã, whether the means and covariances of the received signal coincide. If a configuration with more near-field channels satisfies V ∩ D ≠ ∅ while a configuration with fewer near-field channels has V ∩ D = ∅, the proposed link between near-field proportion and identifiability fails.

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

If this is right

  • Near-field coverage becomes a resource: enlarging the Rayleigh distance by increasing carrier wavelength or array aperture should improve activity detection, offering a new design lever for grant-free massive access.
  • The fronthaul load of activity detection no longer depends on the number of antennas times signature length: only N-dimensional local activity vectors are exchanged, so large-array cell-free systems remain scalable.
  • The same algorithm, with parameters changed, covers single-cell and cell-free operation in pure near-field, pure far-field, or hybrid settings, unifying existing special cases under one implementation.
  • Under the stated conditions on the penalty parameter and the Lipschitz bounds, the distributed iterations are guaranteed to reach a stationary point of the consensus MLE problem, and simulations indicate 1-2 iterations suffice to match the centralized baseline.
  • Because only local detection results are sent to the CPU, the method pairs naturally with low-resolution fronthaul quantization; the paper reports near-ideal performance with 3-bit quantization.

Where Pith is reading between the lines

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

  • The identifiability analysis considers only covariance equality; a rigorous version must also enforce equality of the mean vectors, which depend on near-field LoS components. The true identifiability condition is therefore stricter than V ∩ D = ∅, and the stated performance gain from near-field channels may be either stronger or weaker than the cosine-similarity argument suggests.
  • The cosine-similarity ordering is a pairwise proxy; identifiability is a joint condition over all M APs. A direct test would compute the probability that V ∩ D = ∅ as a function of near-field fraction; the paper asserts but does not prove that the pairwise ordering tracks this probability.
  • The convergence theorem is nonconstructive: the Lipschitz constants and upper bounds are asserted to exist but not computed, so in practice ω and µ are tuned by trial. A future refinement could give explicit choices, turning the conditional guarantee into a prescription.
  • The communication analysis assumes the exchanged vectors have small dynamic range and are sparse; one could extend the approach to compress the exchanged local activity vectors further with learned or model-driven quantization, a step the paper leaves implicit.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper studies device activity detection in cell-free massive MIMO when channels are a hybrid of near-field and far-field propagation. It proposes a covariance-based maximum-likelihood formulation (10) that unifies single-cell/cell-free and near-field/far-field cases. Section III analyzes identifiability and claims, via Proposition 1 and the cosine similarity between columns of Ψ_m, that increasing the proportion of near-field channels improves detection performance. Section IV develops a distributed ADMM algorithm in which each AP solves a local detection subproblem by a coordinate-descent method with a Taylor-expanded surrogate objective; Algorithm 1 exchanges only local detection results with the CPU. Section V states convergence guarantees: Proposition 2 for the local CD method and Theorem 1 for the overall distributed algorithm. Simulations compare detection error, computation time, and fronthaul overhead against centralized and compressed-sensing baselines.

Significance. If the claims are correct, the paper would provide a useful unified formulation and a communication-efficient distributed algorithm for a practically motivated problem. The covariance-based likelihood with near-field LoS and correlated NLoS components is a reasonable extension of prior far-field covariance frameworks, and the idea of exchanging only local activity estimates rather than raw signals is attractive for cell-free systems. The proof of Proposition 1 is elementary and correct in its pairwise comparison. However, the paper's headline theoretical contribution, that more near-field channels enhance detection performance, currently rests on an identifiability heuristic rather than on the pairwise cosine bound, and the convergence theorems depend on unverified constants plus a sign error in the Hessian argument. These are load-bearing issues. The paper does provide extensive simulations, but no code or data; the empirical results are suggestive rather than conclusive.

major comments (4)
  1. [Section III, Eqs. (13)–(17), Proposition 1] The claimed relation between near-field proportion and detection performance is not established. Identifiability of a0 requires that no feasible ξ = ã − a0 gives both C_m(ã)=C_m(a0) and ¯y_m(ã)=¯y_m(a0) for every AP. Equations (13)–(15) enforce only covariance equality; the mean equality Σ_n ξ_n β_{m,n} b(r_{m,n}) ⊗ s_n = 0 for n∈U_m (and 0 for far-field) is absent from V in (15). Moreover, V∩D=∅ is sufficient, not necessary, for identifiability: a vector in V∩D may fail the mean equality. Proposition 1 bounds pairwise cosine similarities, but a global null vector of Ψ_m can exist with all pairwise cosines small when N exceeds the column dimension; pairwise coherence does not control sparse linear dependencies. Thus the abstract and Section III conclusion that increasing the proportion of near-field channels enhances detection performance follows from the 'more likely' heuristic after (1
  2. [Appendix C, proof of Theorem 1] The strong-convexity claim used for the decrease bound (64) has a sign error. The text states 'Since µ > 2 \tilde L_m, we have µI_N − ∇² f_m(θ_m) ⪰ \tilde L_m I_N, which guarantees the strong convexity of problem (21)'. Strong convexity of f_m(θ)+(µ/2)||θ−a||² is governed by µI_N + ∇² f_m(θ), not µI_N − ∇² f_m(θ). The printed inequality therefore does not imply the quadratic lower bound used in (64). The later assertion 'µI_N ⪰ \tilde L_m I_N ⪰ ∇² f_m' is also unproved. The proof of Theorem 1 should be repaired with correct Hessian bounds or replaced by a standard nonconvex ADMM argument.
  3. [Proposition 2 and Theorem 1] The convergence conditions are nonconstructive. Proposition 2 assumes a Lipschitz constant L_m for ∇U_m and upper bounds \barρ_2,\barρ_3,\barρ_4 for ρ_2,ρ_3,ρ_4, and Theorem 1 assumes a Lipschitz constant \tilde L_m for ∇f_m, but no expressions or procedures to compute these constants are provided. Consequently, the conditions ω ≥ L_m + ρ and µ > 2\tilde L_m cannot be verified or instantiated; Algorithm 2 only states that ω is 'properly selected'. The convergence guarantee is therefore conditional on unknown quantities and does not constitute the practical guarantee advertised in the abstract. Please provide explicit bounds or an implementable selection rule.
  4. [Algorithm 2 vs. Appendix B, Eqs. (58)–(61)] There is a mismatch between the algorithm and its convergence proof. Algorithm 2 updates coordinates according to a random permutation (line 4), i.e., each coordinate is updated exactly once per sweep. The proof of Proposition 2, however, treats the updated coordinate n_k as chosen i.i.d. uniformly at each step, as shown by Eq. (61): 'the expected value ... is (1/N)||v(θ^k_m)||²'. Random-permutation cyclic coordinate descent does not satisfy this per-iteration distribution. The proof should either be adapted to random-permutation sweeps or the algorithm should be changed to i.i.d. random coordinate selection.
minor comments (5)
  1. [Section II] The symbol D is used both for the array aperture in the Rayleigh distance 2D²/λ_c (Section II-A) and for the feasible deviation set D in Eq. (16). Please disambiguate these two uses.
  2. [Appendix A, Eq. (37)] The strict inequality ρ_NF-FF < 1 is justified by 'R_{m,n} is not a diagonal matrix'. This is not sufficient; the condition needed is that R_{m,n} is not proportional to the identity (equivalently, rank < K). Please clarify.
  3. [Section IV-D] The communication overhead count '2IMN real-valued numbers' appears to ignore quantization and compression. Also, the claim that few quantization bits suffice because entries lie in [0,1] applies to a and θ_m, but the dual variables λ_m in the exchanged vector µθ_m+λ_m may have larger dynamic range; please clarify.
  4. [Appendix B, Eq. (44)] The inequality (44) relies on d ∈ [−θ_{m,n}, 1−θ_{m,n}] ⊂ [−1,1] and on the chosen bound ρ; this should be stated explicitly. As written, the bound appears to hold only because d is restricted to this interval.
  5. [General] Typos and wording: 'dominanted' in Section IV-D; 'the the dynamic range' in Section IV-D; 'ininequality' in Appendix C; 'increasing λ_c can significantly expands' in Section VI-D. Figure 1 would also benefit from a precise caption explaining what is plotted.

Circularity Check

0 steps flagged

No significant circularity: the algorithm and Proposition 1 are derived from the model; self-citations are contextual, not load-bearing.

full rationale

The paper's central chain is a model-based derivation, not a fitted-input loop. The hybrid near/far-field covariance formulation (8)-(10) is constructed from the channel model, and the distributed algorithm is derived via an ADMM consensus reformulation; no parameter is fitted to data and then renamed as a prediction. Proposition 1 is proved directly from the definitions of Xi_m,n and the Cauchy-Schwarz inequality; it does not import the target conclusion. The step that deserves scrutiny is Section III's move from pairwise cosine similarity (17) to the identifiability condition V∩D=empty: the paper states, 'if the columns of Psi_m are closer to orthogonality, then the condition V∩D=empty is more likely to hold' and cites [33]. This is an informal heuristic: pairwise orthogonality is not equivalent to the global nullspace condition, and the mean vector in (8) is omitted from (13)-(15). That is a genuine correctness gap in the advertised monotonicity claim, but it is not circular--the paper does not use the desired conclusion as an input, nor does any equation in the proof reduce to itself. Similarly, Theorem 1 relies on nonconstructive Lipschitz constants and unspecified bounds for omega and mu; this is an incompleteness issue, not a self-referential reduction. Self-citations to the authors' earlier covariance-based activity-detection work ([17], [22], [30], [33]) supply the framework and the orthogonality heuristic, but Proposition 1 and the ADMM/CD algorithm are derived in this paper, and the simulation comparisons use external benchmarks. No load-bearing argument reduces to a self-citation or to a fitted value.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The paper introduces no new physical entities. It relies on standard near-field channel models and adds two algorithm hyperparameters (ω, µ) whose required values depend on unknown constants. The main unproven premise is the identifiability heuristic linking pairwise column orthogonality to detection performance.

free parameters (3)
  • ω (proximal weight in coordinate descent) = not specified
    Required in (29) to satisfy ω ≥ L_m + ρ with L_m and ρ unknown; chosen ad hoc to control convergence.
  • µ (ADMM penalty parameter) = not specified
    Required in (20) to satisfy µ > 2 L̃_m with L̃_m unknown; chosen by user.
  • γ (detection threshold) = simulation-dependent
    Threshold in â_n = I(a_n ≥ γ), set to balance PM and PF in simulations.
axioms (4)
  • domain assumption Near-field channel model in (2): h = β b(r) + Σ φ_l β̃_l b(r̃_l) with CN(βb, R) distribution.
    Taken from prior near-field communications literature [31], [32], not derived in this paper.
  • domain assumption Joint PDF across APs factorizes as product in (9).
    Standard cell-free assumption of statistically independent channels per AP.
  • ad hoc to paper Identifiability heuristic: closer-to-orthogonal columns of Ψ_m make V ∩ D = ∅ more likely.
    Asserted without proof in Section III and used to conclude improved detection performance.
  • ad hoc to paper Existence of Lipschitz constants L̃_m and upper bounds ρ̄_2, ρ̄_3, ρ̄_4.
    Needed for Propositions 2 and Theorem 1, but never quantified, making convergence conditions nonconstructive.

pith-pipeline@v1.3.0-alltime-deepseek · 20197 in / 14771 out tokens · 130528 ms · 2026-08-04T16:12:46.916211+00:00 · methodology

0 comments
read the original abstract

A great amount of endeavor has recently been devoted to activity detection for massive machine-type communications in cell-free multiple-input multiple-output (MIMO) systems. However, as the number of antennas at the access points (APs) increases, the Rayleigh distance that separates the near-field and far-field regions also expands, rendering the conventional assumption of far-field propagation alone impractical. To address this challenge, this paper establishes a covariance-based formulation that can effectively capture the statistical property of hybrid near-far field channels. Based on this formulation, we theoretically reveal that increasing the proportion of near-field channels enhances the detection performance. Furthermore, we propose a distributed algorithm, where each AP performs local activity detection and only exchanges the detection results to the central processing unit, thus significantly reducing the computational complexity and the communication overhead. Not only with convergence guarantee, the proposed algorithm is unified in the sense that it can handle single-cell or cell-free systems with either near-field or far-field devices as special cases. Simulation results validate the theoretical analyses and demonstrate the superior performance of the proposed approach compared with existing methods.

discussion (0)

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

Forward citations

Cited by 2 Pith papers

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

  1. Dualformer: Efficient Feature Extractor for Complex-valued Blind Communication Signal Analysis

    cs.LG 2026-06 unverdicted novelty 5.0

    Dualformer applies a parameter-sharing DualNN via Transformer patches to complex signals, claiming better results on AMR, SSR, and SSP tasks than baselines.

  2. Compositional Zero-Shot Recognition based on Tangent Space Disentanglement for Composite Modulation Signals

    eess.SP 2026-07 reject novelty 4.0

    TSDN recognizes unseen combinations of inner and outer modulations by log-transforming composite signals and learning layer-specific classifiers, but the core linearization is derived for positive real signals while t...

Reference graph

Works this paper leans on

37 extracted references · cited by 2 Pith papers

  1. [1]

    A survey of recent advances in optimization methods for wireless communications,

    Y .-F. Liu, T.-H. Chang, M. Hong, Z. Wu, A. M.-C. So, E. A. Jorswieck, and W. Yu, “A survey of recent advances in optimization methods for wireless communications,”IEEE J. Sel. Areas Commun., vol. 42, no. 11, pp. 2992–3031, Nov. 2024

    2024

  2. [2]

    Grant-free non-orthogonal multiple access for IoT: A survey,

    M. B. Shahab, R. Abbas, M. Shirvanimoghaddam, and S. J. Johnson, “Grant-free non-orthogonal multiple access for IoT: A survey,”IEEE Commun. Surveys Tuts., vol. 22, no. 3, pp. 1805–1838, May 2020

    2020

  3. [3]

    Heterogeneous transformer: A scale adaptable neural network architecture for device activity detection,

    Y . Li, Z. Chen, Y . Wang, C. Yang, B. Ai, and Y .-C. Wu, “Heterogeneous transformer: A scale adaptable neural network architecture for device activity detection,”IEEE Trans. Wireless Commun., vol. 22, no. 5, pp. 3432–3446, May 2023

    2023

  4. [4]

    Activity detection for massive connectivity under frequency offsets via first-order algorithms,

    Y . Li, M. Xia, and Y .-C. Wu, “Activity detection for massive connectivity under frequency offsets via first-order algorithms,”IEEE Trans. Wireless Commun., vol. 18, no. 3, pp. 1988–2002, Mar. 2019

    1988

  5. [5]

    Compressive-sensing-based grant-free massive access for 6G massive communication,

    Z. Gao, M. Ke, Y . Mei, L. Qiao, S. Chen, D. W. K. Ng, and H. V . Poor, “Compressive-sensing-based grant-free massive access for 6G massive communication,”IEEE IoT J., vol. 11, no. 5, pp. 7411–7435, Mar. 2024

    2024

  6. [6]

    Zhang, Q

    H. Zhang, Q. Lin, Y . Li, L. Cheng, and Y .-C. Wu, “Activity detection for massive connectivity in cell-free networks with unknown large-scale fading, channel statistics, noise variance, and activity probability: A bayesian approach,”IEEE Trans. Signal Process., vol. 72, pp. 942–957, Feb. 2024

    2024

  7. [7]

    Sparse activity detection for massive connectivity,

    Z. Chen, F. Sohrabi, and W. Yu, “Sparse activity detection for massive connectivity,”IEEE Trans. Signal Process., vol. 66, no. 7, pp. 1890– 1904, Apr. 2018

    1904

  8. [8]

    Compressive sensing- based adaptive active user detection and channel estimation: Massive access meets massive MIMO,

    M. Ke, Z. Gao, Y . Wu, X. Gao, and R. Schober, “Compressive sensing- based adaptive active user detection and channel estimation: Massive access meets massive MIMO,”IEEE Trans. Signal Process., vol. 68, pp. 764–779, Jan. 2020

    2020

  9. [9]

    Multi-cell sparse activity detection for massive random access: Massive MIMO versus cooperative MIMO,

    Z. Chen, F. Sohrabi, and W. Yu, “Multi-cell sparse activity detection for massive random access: Massive MIMO versus cooperative MIMO,” IEEE Trans. Wireless Commun., vol. 18, no. 8, pp. 4060–4074, Aug. 2019

    2019

  10. [10]

    Grant-free massive MTC-enabled massive MIMO: A compressive sensing approach,

    K. Senel and E. G. Larsson, “Grant-free massive MTC-enabled massive MIMO: A compressive sensing approach,”IEEE Trans. Commun., vol. 66, no. 12, pp. 6164–6175, Dec. 2018

    2018

  11. [11]

    Compressive sensing based joint activity and data detection for grant-free massive IoT access,

    Y . Mei, Z. Gao, Y . Wu, W. Chen, J. Zhang, D. W. K. Ng, and M. Di Renzo, “Compressive sensing based joint activity and data detection for grant-free massive IoT access,”IEEE Trans. Wireless Commun., vol. 21, no. 3, pp. 1851–1869, Mar. 2022

    2022

  12. [12]

    Joint activity detection and channel estimation in massive MIMO systems with angular domain enhancement,

    W. Chen, H. Xiao, L. Sun, and B. Ai, “Joint activity detection and channel estimation in massive MIMO systems with angular domain enhancement,”IEEE Trans. Wireless Commun., vol. 21, no. 5, pp. 2999– 3011, May 2022

    2022

  13. [13]

    Harnessing common sparsity for enhancing AMP-based activity detection and channel estimation,

    H. Zhang, Y . Li, Q. Lin, Y .-C. Wu, and H. Vincent Poor, “Harnessing common sparsity for enhancing AMP-based activity detection and channel estimation,”IEEE Trans. Signal Process., vol. 73, pp. 3220– 3236, Jul. 2025

    2025

  14. [14]

    Improved scaling law for activity detection in massive MIMO systems,

    S. Haghighatshoar, P. Jung, and G. Caire, “Improved scaling law for activity detection in massive MIMO systems,” inProc. IEEE Int. Symp. Inf. Theory (ISIT), 2018

    2018

  15. [15]

    Covariance based joint activity and data detection for massive random access with massive MIMO,

    Z. Chen, F. Sohrabi, Y .-F. Liu, and W. Yu, “Covariance based joint activity and data detection for massive random access with massive MIMO,” inProc. IEEE Int. Conf. Commun. (ICC), 2019

    2019

  16. [16]

    Sparsity constrained joint activity and data detection for massive access: A difference-of-norms penalty framework,

    Q. Lin, Y . Li, and Y .-C. Wu, “Sparsity constrained joint activity and data detection for massive access: A difference-of-norms penalty framework,” IEEE Trans. Wireless Commun., vol. 22, no. 3, pp. 1480–1494, Mar. 2023

    2023

  17. [17]

    Asynchronous activity detection for cell-free massive MIMO: From centralized to distributed algorithms,

    Y . Li, Q. Lin, Y .-F. Liu, B. Ai, and Y .-C. Wu, “Asynchronous activity detection for cell-free massive MIMO: From centralized to distributed algorithms,”IEEE Trans. Wireless Commun., vol. 22, no. 4, pp. 2477– 2492, Apr. 2023

    2023

  18. [18]

    Intelligent reflecting surface aided activity detection for massive access: Performance analysis and learning approach,

    Q. Lin, Y . Li, Y .-C. Wu, and R. Zhang, “Intelligent reflecting surface aided activity detection for massive access: Performance analysis and learning approach,”IEEE Trans. Wireless Commun., vol. 23, no. 11, pp. 16 935–16 949, Nov. 2024

    2024

  19. [19]

    Mixture of experts- augmented deep unfolding for activity detection in IRS-aided systems,

    Z. Ren, Q. Lin, J. Lei, Y . Li, and Y .-C. Wu, “Mixture of experts- augmented deep unfolding for activity detection in IRS-aided systems,” IEEE Wireless Communications Letters, vol. 14, no. 9, pp. 2912–2916, Jun. 2025

    2025

  20. [20]

    Phase transition analysis for covariance-based massive random access with massive MIMO,

    Z. Chen, F. Sohrabi, Y .-F. Liu, and W. Yu, “Phase transition analysis for covariance-based massive random access with massive MIMO,”IEEE Trans. Inf. Theory, vol. 68, no. 3, pp. 1696–1715, Mar. 2021

    2021

  21. [21]

    Coopera- tive activity detection: Sourced and unsourced massive random access paradigms,

    X. Shao, X. Chen, D. W. K. Ng, C. Zhong, and Z. Zhang, “Coopera- tive activity detection: Sourced and unsourced massive random access paradigms,”IEEE Trans. Signal Process., vol. 68, pp. 6578–6593, Nov. 2020

    2020

  22. [22]

    Clustering-based activity detection algorithms for grant-free random access in cell-free massive MIMO,

    U. K. Ganesan, E. Bj ¨ornson, and E. G. Larsson, “Clustering-based activity detection algorithms for grant-free random access in cell-free massive MIMO,”IEEE Trans. Commun., vol. 69, no. 11, pp. 7520– 7530, Nov. 2021

    2021

  23. [23]

    Communication- efficient activity detection for cell-free massive MIMO: An augmented model-driven end-to-end learning framework,

    Q. Lin, Y . Li, W.-B. Kou, T.-H. Chang, and Y .-C. Wu, “Communication- efficient activity detection for cell-free massive MIMO: An augmented model-driven end-to-end learning framework,”IEEE Trans. Wireless Commun., vol. 23, no. 10, pp. 12 888–12 903, Oct. 2024

    2024

  24. [24]

    A tutorial on near-field XL-MIMO communications toward 6G,

    H. Lu, Y . Zeng, C. You, Y . Han, J. Zhang, Z. Wang, Z. Dong, S. Jin, C.-X. Wang, T. Jiang, X. You, and R. Zhang, “A tutorial on near-field XL-MIMO communications toward 6G,”IEEE Commun. Surveys Tuts., vol. 26, no. 4, pp. 2213–2257, Apr. 2024

    2024

  25. [25]

    Near-field MIMO communications for 6G: Fundamentals, challenges, potentials, and future directions,

    M. Cui, Z. Wu, Y . Lu, X. Wei, and L. Dai, “Near-field MIMO communications for 6G: Fundamentals, challenges, potentials, and future directions,”IEEE Commun. Mag., vol. 61, no. 1, pp. 40–46, Jan. 2023

    2023

  26. [26]

    A tutorial on extremely large-scale MIMO for 6G: Fundamentals, signal processing, and applications,

    Z. Wang, J. Zhang, H. Du, D. Niyato, S. Cui, B. Ai, M. Debbah, K. B. Letaief, and H. V . Poor, “A tutorial on extremely large-scale MIMO for 6G: Fundamentals, signal processing, and applications,”IEEE Commun. Surveys Tuts., vol. 26, no. 3, pp. 1560–1605, Jan. 2024

    2024

  27. [27]

    Near-field communica- tions: What will be different?

    Y . Liu, J. Xu, Z. Wang, X. Mu, and L. Hanzo, “Near-field communica- tions: What will be different?”IEEE Wireless Commun., vol. 32, no. 2, pp. 262–270, Apr. 2025

    2025

  28. [28]

    Near-field communications with block-dominant compressed sensing: Fundamentals, approaches, and future directions,

    L. Lu, K. Ma, Y . Wang, and Z. Wang, “Near-field communications with block-dominant compressed sensing: Fundamentals, approaches, and future directions,”IEEE Commun. Mag., vol. 63, no. 6, pp. 138– 145, Jun. 2025

    2025

  29. [29]

    Non-bayesian ac- tivity detection, large-scale fading coefficient estimation, and unsourced random access with a massive MIMO receiver,

    A. Fengler, S. Haghighatshoar, P. Jung, and G. Caire, “Non-bayesian ac- tivity detection, large-scale fading coefficient estimation, and unsourced random access with a massive MIMO receiver,”IEEE Trans. Inf. Theory, vol. 67, no. 5, pp. 2925–2951, May 2021

    2021

  30. [30]

    Covariance-based activity detection in cooperative multi-cell massive MIMO: Scaling law and efficient algorithms,

    Z. Wang, Y .-F. Liu, Z. Wang, and W. Yu, “Covariance-based activity detection in cooperative multi-cell massive MIMO: Scaling law and efficient algorithms,”IEEE Trans. Inf. Theory, vol. 70, no. 12, pp. 8770– 8790, Dec. 2024

    2024

  31. [31]

    Beamfocusing optimization for near-field wideband multi-user communications,

    Z. Wang, X. Mu, and Y . Liu, “Beamfocusing optimization for near-field wideband multi-user communications,”IEEE Trans. Commun., vol. 73, no. 1, pp. 555–572, Jan. 2025

    2025

  32. [32]

    Near- field communications: A tutorial review,

    Y . Liu, Z. Wang, J. Xu, C. Ouyang, X. Mu, and R. Schober, “Near- field communications: A tutorial review,”IEEE Open J. Commun. Soc., vol. 4, pp. 1999–2049, Aug. 2023

    1999

  33. [33]

    Joint CFO and channel estimation for multiuser MIMO-OFDM systems with optimal training sequences,

    J. Chen, Y .-C. Wu, S. Ma, and T.-S. Ng, “Joint CFO and channel estimation for multiuser MIMO-OFDM systems with optimal training sequences,”IEEE Trans. Signal Process., vol. 56, no. 8, pp. 4008–4019, Aug. 2008

    2008

  34. [34]

    MLE-based device activity detection under Rician fading for massive grant-free access with perfect and imperfect synchronization,

    W. Liu, Y . Cui, F. Yang, L. Ding, and J. Sun, “MLE-based device activity detection under Rician fading for massive grant-free access with perfect and imperfect synchronization,”IEEE Trans. Wireless Commun., vol. 23, no. 8, pp. 8787–8804, Aug. 2024

    2024

  35. [35]

    User activity detection and channel estimation of spatially correlated channels via AMP in massive MTC,

    H. Djelouat, L. Marata, M. Leinonen, H. Alves, and M. Juntti, “User activity detection and channel estimation of spatially correlated channels via AMP in massive MTC,” inProc. 55th Asilomar Conf. Signals Syst. Comput., 2021

    2021

  36. [36]

    Orthogonal AMP for massive access in channels with spatial and temporal correlations,

    Y . Cheng, L. Liu, and L. Ping, “Orthogonal AMP for massive access in channels with spatial and temporal correlations,”IEEE J. Sel. Areas in Commun., vol. 39, no. 3, pp. 726–740, Mar. 2021

    2021

  37. [37]

    Nonlinear programming,

    D. P. Bertsekas, “Nonlinear programming,”J. Oper. Res. Soc., vol. 48, no. 3, pp. 334–334, 1997

    1997