Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

Pre-Equalization Aided Grant-Free Massive Access in Massive MIMO System

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

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

Pith's one-line read Pre-equalizing a single beacon antenna lets a massive MIMO base station detect user activity and data without any CSI, and the paper shows this beats pilot-based grant-free NOMA at equal access latency.

desk verdict A coherent mMIMO extension of beacon pre-equalization with a real comparison flaw: the 'same latency' plots charge the baselines only for pilots, not their full frames. read the letter →

arxiv 2502.06239 v2 pith:2DFFWYAU submitted 2025-02-10 eess.SP cs.ITmath.IT

Yueqing Wang , Yikun Mei , Zhen Gao , Ziwei Wan , Boyu Ning , De Mi , Sami Muhaidat This is my paper · ORCID
classification eess.SPcs.ITmath.IT
keywords grant-freemassiveaccessMIMOnon-orthogonalmultiplepre-equalizationapproximatemessagepassingactivitydetectionchannelestimationdata
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

Massive machine-type communication needs to identify which of many users are active and decode their data with minimal latency, but standard grant-free schemes first estimate a high-dimensional channel from pilot symbols, which costs time and spectrum. This paper argues that the pilot stage can be largely bypassed: the base station activates one beacon antenna, users estimate only that antenna's downlink channel and pre-equalize their uplink signals against it, and the receiver then treats the beacon antenna's received signal as a sparse linear model with known spreading codes and unknown data. From there, an iterative detector alternates data-aided channel estimation and fine data detection across all antennas, exploiting massive MIMO diversity. Simulation results show lower activity-detection error and bit error rate than state-of-the-art pilot-based mMIMO grant-free NOMA baselines at the same access latency, and also gains over the single-antenna beacon baseline. The reason a reader should care is that, if the scheme holds up, grant-free access can move a substantial part of the channel-estimation burden from the base station to the users and cut the latency floor for massive IoT.

What carries the argument

The machinery is the pre-equalized beacon antenna with a nulling threshold. Each user divides by its estimated beacon-antenna channel on subcarriers where the channel magnitude exceeds $h_0$ and sends zero where it is below, so the beacon antenna's channel is flattened and the received tensor there becomes $SX$ plus noise. This one identity converts a semi-blind problem--both channel and data unknown--into three standard linear models: coarse data detection via approximate message passing (AMP) with a spike-and-slab prior plus EM parameter learning and nearest-neighbor sparsity-pattern updates; data-aided channel estimation via GMMV-AMP in the virtual angular domain; and fine data detection via an LMMSE estimator using all antennas. The iteration between the second and third modules is what extracts the massive MIMO diversity gain.

What would settle it

Run the Section V simulation with a controlled mismatch between the channel the users estimate from the beacon and the channel they actually transmit over--for example, adding Gaussian estimation error of increasing variance or a fixed reciprocity calibration offset--and record ADEP and BER. If the coarse detection performance degrades sharply at small mismatches, the scheme's reliance on exact pre-equalization is the deciding factor; if it degrades gracefully, the method is robust.

Watch

Extended reading notes

Core claim

The paper's central claim is that pre-equalization, not pilot-based channel estimation, can carry the first step of grant-free massive access in massive MIMO. By having each user multiply its uplink signal by the reciprocal of its estimated channel to the beacon antenna, the beacon antenna's received signal over one frame reduces to $Y_{\eta,:,:}=SX+W_{\eta,:,:}$, where $S$ is the known spreading-code matrix and $X$ contains the unknown activity-and-data matrix. This makes joint activity and data detection a compressed-sensing problem solvable without any channel state information at the base station. The paper further claims that after this coarse detection, treating the estimated data as pilots lets the receiver estimate the equivalent channel of all antennas in the virtual angular domain and refine the data with LMMSE combining, and that iterating these last two steps harvests multi-antenna diversity. On its simulation settings, the scheme beats both the single-antenna beacon baseline and the pilot-based JADCE baselines under the same time-frequency overhead.

Load-bearing premise

The load-bearing premise is that each user's estimate of the beacon-antenna downlink channel is perfect and that downlink/uplink reciprocity holds exactly, so the pre-equalization truly flattens that channel; if estimation noise, calibration mismatch, or transmit-power limits break equation (5), the CSI-free coarse detection loses its foundation.

Editorial extensions

If this is right

  • The uplink pilot phase can be removed entirely from the random-access frame: the first activity-and-data decision is made from the beacon antenna alone, so access latency is set by the data frame rather than by preamble overhead.
  • Because the coarse detection step does not use the other antennas, activity detection performance is essentially the same as the single-antenna beacon scheme; the multi-antenna gain appears in channel estimation and fine data detection.
  • The data-aided channel estimation problem is sparse in the virtual angular domain, so standard compressed-sensing solvers apply, and the scheme is most attractive at very low frame lengths where pilot-based JADCE fails to converge.
  • When the frame length meets or exceeds the number of detected active users, the GMMV-AMP-based channel estimation becomes overdetermined and must be replaced by an LMMSE-style estimator, a regime the paper counts as covered.

Reading between the lines

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

  • The authors assume users know the beacon-antenna channel exactly; an obvious extension is to quantify ADEP and BER as functions of user-side estimation error or reciprocity calibration mismatch, which will likely show that the coarse detection step is the first place errors appear.
  • The fairness of the headline comparison rests on giving pilot-based baselines the same time-frequency resources for their pilots as the entire data frame of the proposed scheme; a different pilot-allocation policy for baselines could shift the crossover points.
  • The same beacon-antenna idea could be extended to several beacon antennas or to cell-free massive MIMO, trading a small pilot overhead for better immunity to deep fades on a single antenna.
  • Because the coarse detector uses only one antenna, activity detection is diversity-starved; a plausible modification is to add a second beacon antenna or to use the channels estimated by the data-aided CE module to re-check the activity set.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes a pre-equalization aided grant-free massive access scheme for massive MIMO systems. In the proposed scheme, the base station activates one beacon antenna to broadcast a beacon; each UE estimates the downlink channel from that antenna, pre-equalizes its uplink signal on the corresponding subcarriers, and transmits a spread-spectrum data frame without any uplink pilot. The BS then performs joint activity and data detection through three iterative modules: coarse data detection based on the beacon-antenna signal, data-aided channel estimation using the detected data as pilots, and fine data detection using all BS antennas. The authors claim via simulations that the scheme outperforms state-of-the-art massive MIMO grant-free NOMA schemes at the same access latency, and they provide simulation code for reproducibility.

Significance. If the claims were fully validated, the scheme would be a meaningful contribution to low-latency grant-free massive access: it removes the uplink pilot overhead, exploits mMIMO diversity in a semi-blind setting, and breaks the joint activity/data/channel estimation problem into three standard linear estimation sub-problems solvable by AMP-type algorithms. The paper also provides reproducible simulation code and compares against several baselines. However, the numerical support for the central "same access latency" claim is undermined by an inequitable comparison setup, and a few load-bearing modeling assumptions are either internally inconsistent or left untested. The algorithmic idea is still interesting and potentially correct, but the manuscript needs substantial revision before publication.

major comments (5)
  1. [Section V, Figs. 3–5] The claimed comparison at "the same access latency" is not supported as stated. The text explicitly says that for Baselines 1 and 2, which use a "pilot+data" frame, the time slot overhead T refers to the pilot signal only, whereas for the proposed scheme T is the total number of OFDM symbols in the entire uplink data frame. The baselines then require additional data symbols for the BER that is plotted, so the total latency of Baselines 1 and 2 exceeds T; conversely, the proposed scheme requires a beacon broadcast period before each frame (Section II) that is not charged to its latency. Thus the comparisons in Figs. 3 and 5 do not hold either the end-to-end latency or the total time-frequency resources fixed. The reported ADEP/BER gains could therefore be artifacts of unequal latency budgets rather than of the algorithmic design. Please redo the comparison under a common latency definition, for example by counting pilot-plus-data symbols for Baselines 1/2 and beacon-plus-data symbols for the proposed scheme, or by comparing against baselines at their total end-to-end frame length.
  2. [Section IV-A, Eq. (5)] Equation (5) assumes that the pre-equalized beacon-antenna channel is exactly H_η,:,: ◦ Θ = 1, so that Y_η,:,: = S X + W. However, the pre-equalization defined in Eq. (2) includes a nulling threshold h0: for subcarriers with |H_η,m,k| < h0, the pre-equalization factor θ_m,k is set to 0, and the product H_η,m,k θ_m,k is 0 rather than 1. Consequently, the received signal at the beacon antenna is not exactly S X + W on those subcarriers, and the coarse data detection model is invalid as stated. The paper needs to either incorporate the nulled subcarriers into the signal model (e.g., via a mask matrix) or justify that the fraction of nulled subcarriers is negligible under the simulation parameters, ideally by quantifying its effect on ADEP and BER.
  3. [Section III] The load-bearing assumption that UEs have perfect downlink channel estimates and perfect uplink/downlink reciprocity is stated only as "we suppose the CE at UEs' side is sufficiently accurate, i.e., Ĥ = H." This assumption is essential for Eq. (5) and for the entire coarse detection stage. The paper provides no sensitivity analysis with respect to channel estimation error, reciprocity calibration mismatch, or beacon SNR. Since user-side channel estimation is performed on a single beacon signal and the users are low-complexity IoT devices, this assumption is not automatically satisfied. Please add simulations that perturb the user-side channel estimate (e.g., with Gaussian estimation error or a bounded reciprocity mismatch) and show how ADEP/BER degrade, or discuss the beacon SNR and channel coherence conditions under which the assumption is reasonable.
  4. [Section IV-A, Eq. (7)] The AMP update for Z in Eq. (7) appears to contain an error. As written, it is Z_i_m,t = ∑_k s_m,k x̂_i_k,t − V_i_m,t / [σ^2 + V^{i-1}_m,t (Y_η,m,t − Z^{i-1}_m,t)]. In standard AMP, the residual term is V_i_m,t (Y_η,m,t − Z^{i-1}_m,t)/(σ^2 + V^{i-1}_m,t), i.e., the denominator is a scalar variance term and the residual is a separate factor. The current expression places the residual inside the denominator, leading to a complex, dimensionally inconsistent denominator. This error propagates to Algorithm 1 and the subsequent derivations in Eqs. (8)–(12). Please correct the formula or clarify the intended notation; as stated, the algorithm is not well-defined.
  5. [Section V, simulation parameters] The large-scale fading parameter g_k is used inconsistently. In Eq. (1) and Section II, g_k is a large-scale channel fading factor that multiplies the small-scale fading in the form √g_k H:,m,k, which implies g_k is a linear (power) gain. In Section V, the authors write "the large scale fading follows the Log-distance path loss model as g_k = 128.1 + 37.6 log10(d_k)", which is a formula for path loss in dB, not a linear gain factor. If interpreted literally, g_k is on the order of 100 or more, and √g_k is not a valid multiplicative channel amplitude; the power control factor p_k = 1/g_k would also be ill-defined. This needs to be corrected, likely by defining a dB path-loss PL_k and then setting g_k = 10^{-PL_k/10} (for power) or 10^{-PL_k/20} (for amplitude), with a matching power-control formula. As written, the simulation parameter setup is not reproducible.
minor comments (5)
  1. [General] The simulation results do not report error bars, confidence intervals, or the number of Monte Carlo realizations over which the ADEP/BER/NMSE curves are averaged. Since the channel and data realizations are random, please add this information to the figures or text.
  2. [Section V, Table I and text] There is a baseline-numbering inconsistency between the table and the text. In Table I, Baseline 3 is the pre-equalization aided JADD scheme with a single antenna and OAMP, and Baseline 4 is the SOMP variant. In the text, however, "Baseline 3" is first described as "Same as Baseline 1 except that JADCE is performed with SOMP", which appears to be a typo for Baseline 2. Please align the text with Table I.
  3. [Section III, power control] The definition of the average pre-equalization power p_e is not used consistently. The paper says E[|θ_m,k|^2] = p_e and sets the UE transmit power to ρ = 7 dBm, but it does not explain how p_e and the total transmit power constraint are related, nor whether the nulling operation changes the average power normalization. Please clarify the power model.
  4. [Section II, beacon overhead] The beacon broadcast period is mentioned but its duration, periodicity, and resource overhead are not quantified. This is relevant to the latency comparison in major comment 1, but also to the practical framing of the scheme.
  5. [Section V, Fig. 5a] The text notes that the proposed scheme's performance is not shown for T ≥ 50 because GMMV-AMP does not converge when T ≥ K̂_a, and that LMMSE could be used instead. Since T = 50 is within the plotted range in Fig. 5a (up to 90), please either include the LMMSE-based results for large T, or clearly indicate the range of T for which the proposed receiver is applicable.

Circularity Check

0 steps flagged · score 0.0 of 10

No material circularity: the proposed pre-equalization and iterative detection derivation is self-contained, and the equalized beacon-channel equation is a design identity rather than a prediction fitted to outputs.

full rationale

The paper's derivation chain is a constructive system-design argument: the UE pre-equalization factor is defined in (2) as the reciprocal of the estimated beacon-antenna channel, and equation (5) follows by substitution under the explicitly stated perfect-CE assumption; this is a design identity, not a circular prediction. The coarse DD, data-aided CE, and fine DD modules are standard linear/CS estimation stages whose inputs are received signals and spreading codes, not the claimed outputs. The data-aided CE does use detected data as pilots, but those detections come from the CSI-free coarse DD stage, and the final BER is evaluated against true transmitted symbols, so the loop is decision-directed estimation rather than a reduction of the conclusion to its premise. Prior work by overlapping authors ([4], [13]) supplies component algorithms and a baseline, but it is peer-reviewed, externally falsifiable, and not invoked as a uniqueness theorem or to forbid alternatives. The one serious concern in the paper is benchmarking, not circularity: Section V states that for Baselines 1 and 2 'the time slot overhead refers to the number of OFDM symbols within one frame T and the time slot overhead of the pilot signal, respectively,' and the beacon broadcast period preceding each uplink frame is not charged to the proposed scheme's latency. That unequal accounting threatens the 'same access latency' claim as a matter of experimental validity, but it does not make any derived quantity equivalent to its inputs by construction. Under the circularity standard of this review, no load-bearing step reduces to a self-citation or to a fitted parameter renamed as a prediction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 1 invented entities

The central claim rests on standard linear channel models and existing signal processing algorithms. The main free parameters are the nulling threshold, the data-estimated sparsity ratio and noise variance. The strongest assumptions are perfect user-side channel estimation and perfect reciprocity, which are stated in Section III but not tested in the simulations.

free parameters (3)
  • pre-equalization nulling threshold h0 = 0.2
    Equation (2) defines the threshold that decides whether pre-equalization is applied or the subcarrier is nulled. The paper sets h0 = 0.2 in the simulations without a sensitivity analysis, and the BER and NMSE results depend on this choice.
  • sparsity ratio gamma_k,t = estimated online
    In equations (12) and Algorithm 1, the sparsity ratio is learned via EM and then averaged over time slots, so it is effectively fitted to the received data rather than being a fixed known prior.
  • noise variance sigma^2 = estimated online
    The noise variance is learned by the EM algorithm in equation (12), making it a data-fitted parameter for the AMP-based coarse DD.
assumptions (4)
  • domain assumption Perfect downlink channel estimation at the UE and perfect uplink/downlink channel reciprocity
    Section III assumes the beacon antenna channel estimate equals the true channel, which makes pre-equalization exact. Reciprocity of the uplink and downlink channels is required for the pre-equalized uplink channel to be the inverse of the downlink channel.
  • domain assumption The equivalent CSI, meaning true channel times pre-equalization factors, is sparse in the virtual angular domain
    Section IV-B assumes the equivalent channel reserves the complete virtual angular domain sparsity of the accurate CSI, so that the GMMV-AMP algorithm can recover it from a small number of measurements.
  • domain assumption The data symbols of active users follow a known discrete constellation prior with a spike-and-slab sparsity prior
    Equation (6) assumes the prior distribution of transmitted symbols is a spike-and-slab distribution over the modulation constellation, which is accurate when users are active or zero otherwise.
  • domain assumption The one-ring channel model with a limited number of paths and limited angle spread
    Equation (1) models each UE channel as a sum of P paths with a ULA steering vector and delay, which is the standard physical channel model used to justify the angular-domain sparsity.
invented entities (1)
  • beacon antenna at the massive MIMO base station
    purpose: A single BS antenna broadcasts a beacon so that UEs can estimate its channel and pre-equalize their uplink signals; the pre-equalized beacon-antenna signal enables coarse joint activity and data detection without CSI.
    The beacon antenna is a protocol element, not a new physical entity. It requires no new hardware, but the claimed performance relies on its channel being perfectly known at the UE and on reciprocity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Pre-Equalization Aided Grant-Free Massive Access in Massive MIMO System." pith.science (2026). https://pith.science/paper/2DFFWYAU

@misc{pith2026250206239,
  author       = {Pith},
  title        = {Pith review of: Pre-Equalization Aided Grant-Free Massive Access in Massive MIMO System},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2DFFWYAU}},
  note         = {Machine review of arXiv:2502.06239}
}
read the original abstract

The spatial diversity and multiplexing advantages of massive multi-input-multi-output (mMIMO) can significantly improve the capacity of massive non-orthogonal multiple access (NOMA) in machine type communications. However, state-of-the-art grant-free massive NOMA schemes for mMIMO systems require accurate estimation of random access channels to perform activity detection and the following coherent data demodulation, which suffers from excessive pilot overhead and access latency. To address this, we propose a pre-equalization aided grant-free massive access scheme for mMIMO systems, where an iterative detection scheme is conceived. Specifically, the base station (BS) firstly activates one of its antennas (i.e., beacon antenna) to broadcast a beacon signal, which facilitates the user equipment (UEs) to perform downlink channel estimation and pre-equalize the uplink random access signal with respect to the channels associated with the beacon antenna. During the uplink transmission stage, the BS detects UEs' activity and data by using the proposed iterative detection algorithm, which consists of three modules: coarse data detection (DD), data-aided channel estimation (CE), and fine DD. In the proposed algorithm, the joint activity and DD is firstly performed based on the signals received by the beacon antenna. Subsequently, the DD is further refined by iteratively performing data-aided CE module and fine DD module using signals received by all BS antennas. Our simulation results demonstrate that the proposed scheme outperforms state-of-the-art mMIMO-based grant-free massive NOMA schemes with the same access latency. Simulation codes are provided to reproduce the results in this article: https://github.com/owenwang517/tvt-2025.

Figures

Figures reproduced from arXiv: 2502.06239 by the authors.

Figure 1
Figure 1. The proposed pre-equalization aided grant-free mas [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The BER and NMSE performance of different schemes [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Comparison of ADEP performance for different schemes versus time slot overhead T with M = 56, ρ = 7dBm. 10 15 20 25 30 35 40 Time slot overhead T 10-2 10-1 100 101 NMSE Baseline 1 Baseline 2 Proposed scheme [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: The BER performance of different schemes with [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

Discussion (0). Continue with ORCID 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

  1. TinyMA-IEI-PPO: Exploration Incentive-Driven Multi-Agent DRL with Self-Adaptive Pruning for Vehicular Embodied AI Agent Twins Migration

    cs.MA 2025-04 reject novelty 4.0 of 10

    A pruned multi-agent PPO algorithm with exploration incentives is designed to approximate the Stackelberg equilibrium of a vehicular AI twin migration game.

Reference graph

Works this paper leans on

19 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [13]

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

    Y . Mei et al. , “Compressive sensing-based joint activity and data detec - tion for grant-free massive IoT access,” IEEE Trans. Wireless Commun. , vol. 21, no. 3, pp. 1851-1869, March 2022

    work page 2022

  2. [1]

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

    Z. Gao et al. , “Compressive-sensing-based grant-free massive access f or 6G massive communication,” IEEE Internet Things J. , vol. 11, no. 5, pp. 7411-7435, March 2024

    work page 2024

  3. [2]

    Massive Access for 5G and Beyond,

    X. Chen et al., “Massive Access for 5G and Beyond,” IEEE J. Sel. Areas Commun., vol. 39, no. 3, pp. 615-637, March 2021

    work page 2021

  4. [3]

    Massive connectivity with massive MIMO—part I: device activity detection and channel estimation,

    L. Liu et al. , “Massive connectivity with massive MIMO—part I: device activity detection and channel estimation,” IEEE Trans. Signal Process. , vol. 66, no. 11, pp. 2933-2946, June 2018

    work page 2018

  5. [4]

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

    M. Ke et al. , “Compressive sensing-based adaptive active UE detection and channel estimation: Massive access meets massive MIMO, ” IEEE Trans. Signal Process. , vol. 68, pp. 764-779, Jan. 2020

    work page 2020

  6. [5]

    Massive access in cell-free massive MIMO-based internet of things: Cloud computing and edge computing paradigms,

    M. Ke et al. , “Massive access in cell-free massive MIMO-based internet of things: Cloud computing and edge computing paradigms,” IEEE J. Sel. Areas Commun. , vol. 39, no. 3, pp. 756-772, March 2021

    work page 2021

  7. [6]

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

    W. Chen et al., “Joint activity detection and channel estimation in massi ve MIMO systems with angular domain enhancement,” IEEE Trans. Signal Process., vol. 21, no. 5, pp. 2999-3011, May 2022

    work page 2022

  8. [7]

    Active terminal identification, channel estimation, and signal detection for grant-Free NOMA-OTFS in LEO satellite Internet-of- Things,

    X. Zhou et al. , “Active terminal identification, channel estimation, and signal detection for grant-Free NOMA-OTFS in LEO satellite Internet-of- Things,” IEEE Trans. Wireless Commun. , vol. 22, no. 4, pp. 2847-2866, April 2023

    work page 2023

Show all 19 references
  1. [8]

    Hybrid driven learning for joint activity detection and channel estimation in IRS-assisted massive connectivity,

    S. Zheng et al. , “Hybrid driven learning for joint activity detection and channel estimation in IRS-assisted massive connectivity, ” IEEE Trans. Signal Process. , early access, March 19, 2024

    2024

  2. [9]

    Random access with massive MIMO-OTFS in LEO satellite communications,

    B. Shen et al. , “Random access with massive MIMO-OTFS in LEO satellite communications,” IEEE J. Select. Areas Commun. , vol. 40, no. 10, pp. 2865–2881, 2022

    2022

  3. [10]

    Exploiting simultaneous low-rank and sparsity in delay- angular domain for millimeter-wave/terahertz wideband ma ssive access,

    X. Shao et al., “Exploiting simultaneous low-rank and sparsity in delay- angular domain for millimeter-wave/terahertz wideband ma ssive access,” IEEE Trans. Wireless Commun. , vol. 21, no. 4, pp. 2336–2351, 2022

    2022

  4. [11]

    OTFS signaling for uplink NOMA of heterogeneous mobility users,

    Y . Ge et al. , “OTFS signaling for uplink NOMA of heterogeneous mobility users,” IEEE Trans. Wireless Commun. , vol. 69, no. 5, pp. 3147–3161, 2021

    2021

  5. [12]

    Approximate m essage passing-based joint sser activity and data detection for NO MA,

    C. Wei, H. Liu, Z. Zhang, J. Dang and L. Wu, “Approximate m essage passing-based joint sser activity and data detection for NO MA,” in IEEE Communications Letters, vol. 21, no. 3, pp. 640-643, March 2 017

  6. [14]

    Spatially common sparsity based adaptive channel estimation and feedback for FDD massive MIMO,

    Z. Gao et al. , “Spatially common sparsity based adaptive channel estimation and feedback for FDD massive MIMO,” IEEE Trans. Signal Process., vol. 63, no. 23, pp. 6169-6183, Dec. 2015

    2015

  7. [15]

    Expectation-Maximization Gaussian-mixture approxi- mate message passing,

    J. P . Vila et al. , “Expectation-Maximization Gaussian-mixture approxi- mate message passing,” IEEE Trans. Signal Process. , vol. 61, no. 19, pp. 4658-4672, Oct 2013

    2013

  8. [16]

    Approximate message passing with nearest neighbor sparsity pattern learning,

    X. Meng et al. , “Approximate message passing with nearest neighbor sparsity pattern learning,” 2016, arXiv:1601.00543

    2016 arXiv

  9. [17]

    Message passing algorithms for compressed sens- ing: I. motivation and construction,

    D. L. Donoho et al. ,“Message passing algorithms for compressed sens- ing: I. motivation and construction,” in 2010 IEEE Informat ion Theory Workshop on Information Theory (ITW 2010, Cairo), 2010, pp. 1–5

    2010

  10. [18]

    On the noise robustness of simultaneous orthog- onal matching pursuit,

    J. -F. Determe et al. , “On the noise robustness of simultaneous orthog- onal matching pursuit,” IEEE Trans. Signal Process. , vol. 65, no. 4, pp. 864-875, 15 Feb.1, 2017

    2017

  11. [19]

    Quasi-synchronous random access for massive MIMO- based LEO satellite constellations,

    K. Ying et al. , “Quasi-synchronous random access for massive MIMO- based LEO satellite constellations,” IEEE J. Select. Areas Commun. , vol. 41, no. 6, pp. 1702-1722, June 2023

    2023

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.