pith. sign in

arxiv: 2605.27930 · v1 · pith:BRCM7KAUnew · submitted 2026-05-27 · 💻 cs.IT · eess.SP· math.IT

Optimization of CF-mMIMO Systems for the Coexistence between eMBB+ and mMTC+: From Analytical to GNN-Aided Designs

Sergi Liesegang , Lou Sala\"un , Chung Shue Chen , Stefano Buzzi This is my paper

Pith reviewed 2026-06-29 10:20 UTC · model grok-4.3

classification 💻 cs.IT eess.SPmath.IT
keywords cell-free massive MIMOeMBBmMTCpower controlgraph neural networkfinite blocklengthnon-orthogonal multiple accessenergy efficiency
0
0 comments X

The pith

Cell-free massive MIMO enables eMBB+ and mMTC+ coexistence via non-orthogonal access and GNN-approximated power control from statistical CSI.

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

The paper establishes that spreading low-rate mMTC+ transmissions across the shared time-frequency grid with eMBB+ users in a terminal-centric cell-free massive MIMO system supports efficient resource reuse. Closed-form achievable rate expressions are derived solely from statistical channel knowledge, incorporating finite-blocklength modeling for mMTC+ short packets under imperfect CSI. A power-control optimization problem is formulated to maximize the minimum energy efficiency of mMTC+ devices subject to eMBB+ quality-of-service constraints; this nonconvex problem is solved by sequential fractional programming and approximated in real time by a graph neural network with multi-head attention trained under an augmented Lagrangian loss. A sympathetic reader would care because the approach shows how heterogeneous services can share resources without orthogonal separation while meeting distinct rate and efficiency targets.

Core claim

By deriving closed-form rate expressions based solely on statistical channel knowledge for both services under imperfect CSI and finite blocklength for mMTC+, and solving the resulting power allocation problem with sequential fractional programming, the authors demonstrate that a graph neural network can approximate the optimal solution with near-optimal performance at significantly reduced computational complexity, enabling effective multiplexing of the two services.

What carries the argument

The graph neural network with multi-head attention that approximates the sequential fractional programming solution to the power-control optimization problem maximizing minimum mMTC+ energy efficiency subject to eMBB+ QoS constraints.

If this is right

  • The non-orthogonal spreading scheme enables simultaneous operation of eMBB+ and mMTC+ without dedicated orthogonal resources.
  • Closed-form rates based on statistical CSI alone suffice for optimization even when instantaneous CSI is imperfect.
  • The GNN with augmented Lagrangian training produces feasible solutions in real time at far lower complexity than sequential fractional programming.
  • Numerical evaluation confirms effective multiplexing while satisfying the heterogeneous service requirements.
  • Constraint satisfaction is preserved during GNN training through the augmented Lagrangian loss.

Where Pith is reading between the lines

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

  • The same statistical-CSI rate framework could be tested for robustness when user mobility increases channel variation rates.
  • The GNN approximation might allow power updates on a per-coherence-interval basis in large networks where model-based solvers become prohibitive.
  • Extending the non-orthogonal scheme to include downlink transmissions would require only re-deriving the rate expressions while reusing the same optimization and GNN structure.

Load-bearing premise

The closed-form rate expressions derived from statistical channel knowledge remain accurate enough to support the power-control optimization and GNN training under modeled imperfect CSI and finite-blocklength conditions.

What would settle it

A set of Monte Carlo simulations or over-the-air measurements in which the empirical achievable rates under the derived power allocations deviate by more than a few percent from the closed-form expressions, or in which the GNN solution gap to the model-based optimum exceeds the reported near-optimality margin.

Figures

Figures reproduced from arXiv: 2605.27930 by Chung Shue Chen, Lou Sala\"un, Sergi Liesegang, Stefano Buzzi.

Figure 1
Figure 1. Figure 1: Illustrative example of a terminal-centric CF-mMIMO setup, where subsets of Ms = 2 APs (out of M = 8), equipped with L = 3 antennas, serve Ku = 15 eMBB+ users and Kd = 10 mMTC+ devices simultaneously. II. SYSTEM MODEL In this work, we consider a CF-mMIMO deployment similar to the one presented in [1], where M APs, each equipped with L antennas, are connected to a central processing unit (CPU) through high-… view at source ↗
Figure 2
Figure 2. Figure 2: Illustrative example of the line-graph representation of a terminal [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Update of AP-user tensor h(t). A similar procedure holds for g(t). by applying ReLU activations followed by layer normalization to h˜(t), g˜(t), the concatenations of the intermediate tensors that account for the multi-headed attention mechanism, i.e., h˜ i(t) = X •∈Tu M C c=1 Φ eMBB+ •,c,t (i), g˜j (t) = X •∈Td M C c=1 Φ mMTC+ •,c,t (j), (38) where L is the concatenation operator and C ≥ 1 is the number o… view at source ↗
Figure 4
Figure 4. Figure 4: CDF of the EE of mMTC+ devices vs. spreading factor [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: CDF of the EE of mMTC+ devices vs. MA scheme (NOMA/OMA) [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 8
Figure 8. Figure 8: CDF of the rate of eMBB+ users vs. power control design. Dotted [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: CDF of the EE of mMTC+ devices vs. GNN training loss. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
read the original abstract

This paper investigates uplink multiple access for the coexistence of enhanced mobile broadband+ (eMBB+) and massive machine-type communications+ (mMTC+) in terminal-centric cell-free massive MIMO (CF-mMIMO) systems. We propose a non-orthogonal scheme in which low-rate mMTC+ transmissions are spread across the time-frequency grid shared with eMBB+ users, enabling efficient resource reuse. In the presence of imperfect channel state information, we derive closed-form expressions for the achievable rates of both services based solely on statistical channel knowledge. For mMTC+ devices, the analysis also incorporates finite blocklength (FBL) modeling to capture short-packet transmissions. To support heterogeneous service requirements, we formulate a power-control problem that maximizes the minimum energy efficiency of mMTC+ devices subject to quality-of-service constraints on eMBB+ users. The resulting nonconvex problem is solved via sequential fractional programming, accounting for both the Shannon and FBL regimes. To enable real-time operation, we further propose a graph neural network (GNN) with multi-head attention to approximate the model-based solution. Constraint satisfaction during training is enforced via an augmented Lagrangian loss. Numerical results demonstrate effective multiplexing of the two data services and show that the proposed GNN algorithm achieves near-optimal performance with a significantly lower computational complexity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper investigates uplink non-orthogonal multiple access for eMBB+ and mMTC+ coexistence in terminal-centric cell-free massive MIMO systems. It derives closed-form achievable rate expressions for both services from statistical channel knowledge only (Shannon for eMBB+, finite-blocklength for mMTC+), formulates a max-min energy-efficiency power-control problem subject to eMBB+ QoS constraints, solves it via sequential fractional programming, and proposes a multi-head-attention GNN to approximate the solution in real time using an augmented-Lagrangian loss for constraint satisfaction. Numerical results are claimed to show effective service multiplexing and near-optimal GNN performance at substantially lower complexity.

Significance. If the closed-form rate expressions remain accurate under the joint effects of imperfect CSI, pilot contamination, power disparity, and finite-blocklength coding, the work supplies an analytical foundation for heterogeneous-service power allocation in CF-mMIMO together with a practical low-complexity GNN surrogate. The use of purely statistical CSI and explicit FBL modeling, together with the constraint-aware GNN training, are concrete strengths that would support real-time deployment if the underlying approximations hold.

major comments (2)
  1. [Abstract and rate derivations] Abstract and rate-derivation sections: the central claim that the closed-form uplink rates (Shannon for eMBB+, FBL for mMTC+) derived solely from statistical CSI remain sufficiently tight to support both the sequential-fractional-programming power-control solution and the GNN training labels is load-bearing. Under the non-orthogonal coexistence model, the interference terms plus the finite-blocklength penalty are incorporated via bounding or expectation approximations; if these loosen materially when pilot contamination, power disparity, and short-packet coding are jointly present, the subsequent optimization and GNN surrogate become mis-calibrated. A concrete validation (e.g., comparison of the closed-form expressions against Monte-Carlo rates or exact FBL bounds for representative parameter sets) is required.
  2. [Power-control problem and GNN section] Power-control formulation and GNN training: the max-min EE objective and the augmented-Lagrangian loss both rely on the same closed-form rate expressions. Any looseness identified in the rate approximations directly propagates into the claimed near-optimality of the GNN; therefore the numerical results demonstrating "near-optimal performance" cannot be accepted without the tightness verification above.
minor comments (2)
  1. [Notation and system model] Notation for the two services (eMBB+ vs. mMTC+) and the distinction between statistical CSI and instantaneous CSI should be introduced once and used consistently throughout.
  2. [Numerical results] Figure captions and axis labels should explicitly state whether the plotted rates are the closed-form expressions or Monte-Carlo realizations, and whether the GNN curves include the augmented-Lagrangian penalty term.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments highlighting the importance of validating our closed-form rate expressions. We address each major comment below and will revise the manuscript to incorporate the requested comparisons.

read point-by-point responses

  1. Referee: [Abstract and rate derivations] Abstract and rate-derivation sections: the central claim that the closed-form uplink rates (Shannon for eMBB+, FBL for mMTC+) derived solely from statistical CSI remain sufficiently tight to support both the sequential-fractional-programming power-control solution and the GNN training labels is load-bearing. Under the non-orthogonal coexistence model, the interference terms plus the finite-blocklength penalty are incorporated via bounding or expectation approximations; if these loosen materially when pilot contamination, power disparity, and short-packet coding are jointly present, the subsequent optimization and GNN surrogate become mis-calibrated. A concrete validation (e.g., comparison of the closed-form expressions against Monte-Carlo rates or exact FBL bounds for representative parameter sets) is required.

    Authors: We agree that a direct comparison of the closed-form expressions to Monte-Carlo simulations is necessary to confirm tightness under the combined effects of imperfect CSI, pilot contamination, power disparity, and finite-blocklength coding. In the revised manuscript, we will add explicit numerical validation (new figures or tables) for representative parameter sets, including varying levels of pilot contamination and power disparity. This will substantiate the approximations used in the rate derivations. revision: yes

  2. Referee: [Power-control problem and GNN section] Power-control formulation and GNN training: the max-min EE objective and the augmented-Lagrangian loss both rely on the same closed-form rate expressions. Any looseness identified in the rate approximations directly propagates into the claimed near-optimality of the GNN; therefore the numerical results demonstrating "near-optimal performance" cannot be accepted without the tightness verification above.

    Authors: We acknowledge the dependency: the GNN near-optimality claims rely on the accuracy of the same rate expressions. Following the addition of the rate validation requested in the first comment, we will update the numerical results section to demonstrate GNN performance using the verified expressions, ensuring the near-optimality claims are properly supported. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation chain is self-contained

full rationale

The paper derives closed-form uplink rate expressions from statistical CSI (including FBL for mMTC+), formulates a max-min EE power-control problem using those expressions, solves it via sequential fractional programming, and trains a GNN surrogate with augmented Lagrangian loss to approximate the model-based solution. No quoted step reduces a claimed prediction or result to a fitted parameter, self-citation, or input by construction; the rate expressions are analytically derived rather than fitted, and the GNN approximates an independently solved optimization rather than recycling its own outputs as labels. This is the standard non-circular workflow for model-based optimization followed by learned approximation.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Abstract-only review; no explicit free parameters, invented entities, or detailed axioms are visible beyond standard domain modeling assumptions.

axioms (2)
  • domain assumption Imperfect CSI is modeled via statistical channel knowledge only
    Basis for all closed-form rate derivations stated in abstract
  • domain assumption Finite blocklength regime applies to mMTC+ short-packet transmissions
    Incorporated into mMTC+ rate expressions

pith-pipeline@v0.9.1-grok · 5794 in / 1389 out tokens · 31908 ms · 2026-06-29T10:20:40.707245+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

45 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    Coexistence of eMBB+ and mMTC+ in uplink cell-free massive MIMO networks,

    S. Liesegang and S. Buzzi, “Coexistence of eMBB+ and mMTC+ in uplink cell-free massive MIMO networks,” in2025 IEEE Wireless Communications and Networking Conference (WCNC), 2025, pp. 1–6. [2]6G: The next horizon, Huawei, White Paper, Jan. 2022. [Online]. Available: https://www.huawei.com/en/huaweitech/future-technologies/ 6g-white-paper [3]Ericsson mobili...

    2025

  2. [2]

    6G internet of things: A comprehensive survey,

    D. C. Nguyenet al., “6G internet of things: A comprehensive survey,” IEEE Internet Things J., vol. 9, no. 1, pp. 359–383, 2022

    2022

  3. [3]

    Massive MIMO with spatially correlated Rician fading channels,

    ¨O. ¨Ozdoganet al., “Massive MIMO with spatially correlated Rician fading channels,”IEEE Trans. Commun., vol. 67, no. 5, pp. 3234–3250, 2019

    2019

  4. [4]

    Demir, E

    ¨O. Demir, E. Bj ¨ornson, and L. Sanguinetti,Foundations of User-Centric Cell-Free Massive MIMO. Now Publishers, 2021

    2021

  5. [5]

    Why user-centric cell-free distributed MIMO systems will be the disruptive 6G technology,

    S. Buzziet al., “Why user-centric cell-free distributed MIMO systems will be the disruptive 6G technology,”IEEE Commun. Mag., pp. 1–7, early access, 2026

    2026

  6. [6]

    Next-generation multiple access: From basic principles to modern architectures,

    E. A. Jorswieck, “Next-generation multiple access: From basic principles to modern architectures,”Proc. IEEE, vol. 112, no. 9, pp. 1149–1178, 2024

    2024

  7. [7]

    Next-generation multiple access with cell-free massive MIMO,

    M. Mohammadiet al., “Next-generation multiple access with cell-free massive MIMO,”Proc. IEEE, vol. 112, no. 9, pp. 1372–1420, 2024

    2024

  8. [8]

    M. K. Simon, J. K. Omura, R. A. Scholtz, and B. K. Levitt,Spread Spectrum Communications Handbook. McGraw-Hill, 1994, vol. 2

    1994

  9. [9]

    Energy and rate allocation for massive multiple access with interference cancelation,

    F. Molinaet al., “Energy and rate allocation for massive multiple access with interference cancelation,”IEEE Access, vol. 10, pp. 74 825–74 840, 2022

    2022

  10. [10]

    A survey of energy-efficient techniques for 5G networks and challenges ahead,

    S. Buzziet al., “A survey of energy-efficient techniques for 5G networks and challenges ahead,”IEEE J. Sel. Areas Commun., vol. 34, no. 4, pp. 697–709, 2016

    2016

  11. [11]

    Channel coding rate in the finite blocklength regime,

    Y . Polyanskiyet al., “Channel coding rate in the finite blocklength regime,”IEEE Trans. Inf. Theory, vol. 56, no. 5, pp. 2307–2359, 2010

    2010

  12. [12]

    A globally optimal energy-efficient power control framework and its efficient implementation in wireless interference networks,

    B. Matthiesenet al., “A globally optimal energy-efficient power control framework and its efficient implementation in wireless interference networks,”IEEE Trans. Signal Process., vol. 68, pp. 3887–3902, 2020

    2020

  13. [13]

    Toward a 6G AI-native air interface,

    J. Hoydiset al., “Toward a 6G AI-native air interface,”IEEE Commun. Mag., vol. 59, no. 5, pp. 76–81, 2021

    2021

  14. [14]

    Learning power allocation for multi-cell-multi- user systems with heterogeneous graph neural networks,

    J. Guo and C. Yang, “Learning power allocation for multi-cell-multi- user systems with heterogeneous graph neural networks,”IEEE Trans. Wireless Commun., vol. 21, no. 2, pp. 884–897, 2022

    2022

  15. [15]

    Graph neural networks for wireless communications: From theory to practice,

    Y . Shenet al., “Graph neural networks for wireless communications: From theory to practice,”IEEE Trans. Wireless Commun., vol. 22, no. 5, pp. 3554–3569, 2023

    2023

  16. [16]

    Graph neural network aided power control in partially connected cell-free massive MIMO,

    S. Mishraet al., “Graph neural network aided power control in partially connected cell-free massive MIMO,”IEEE Trans. Wireless Commun., vol. 23, no. 9, pp. 12 412–12 423, 2024

    2024

  17. [17]

    Learning optimal resource allocations in wireless systems,

    M. Eisenet al., “Learning optimal resource allocations in wireless systems,”IEEE Trans. Signal Process., vol. 67, no. 10, pp. 2775–2790, 2019

    2019

  18. [18]

    Distributed optimization and statistical learning via the alternating direction method of multipliers,

    S. Boydet al., “Distributed optimization and statistical learning via the alternating direction method of multipliers,”Foundations and Trends® in Machine learning, vol. 3, no. 1, pp. 1–122, 2011

    2011

  19. [19]

    5G wireless network slicing for eMBB, URLLC, and mMTC: A communication-theoretic view,

    P. Popovskiet al., “5G wireless network slicing for eMBB, URLLC, and mMTC: A communication-theoretic view,”IEEE Access, vol. 6, pp. 55 765–55 779, 2018

    2018

  20. [20]

    Network slicing for eMBB, URLLC, and mMTC: An uplink rate-splitting multiple access approach,

    Y . Liuet al., “Network slicing for eMBB, URLLC, and mMTC: An uplink rate-splitting multiple access approach,”IEEE Trans. Wireless Commun., vol. 23, no. 3, pp. 2140–2152, 2024

    2024

  21. [21]

    Coexistence mechanism between eMBB and uRLLC in 5G wireless networks,

    A. K. Bairagiet al., “Coexistence mechanism between eMBB and uRLLC in 5G wireless networks,”IEEE Trans. Commun., vol. 69, no. 3, pp. 1736–1749, 2021

    2021

  22. [22]

    On the coexistence of eMBB and URLLC in multi-cell massive MIMO,

    G. Interdonatoet al., “On the coexistence of eMBB and URLLC in multi-cell massive MIMO,”IEEE O. J. Commun. Soc., vol. 4, pp. 1040– 1059, 2023

    2023

  23. [23]

    Scalable eMBB/URLLC-enabled cell- free massive-MIMO with hardware impairments,

    G. Femenias and F. Riera-Palou, “Scalable eMBB/URLLC-enabled cell- free massive-MIMO with hardware impairments,”IEEE Trans. Veh. Technol., pp. 1–16, early access, 2026

    2026

  24. [24]

    Two applications of deep learning in the physical layer of communication systems [lecture notes],

    E. Bjornson and P. Giselsson, “Two applications of deep learning in the physical layer of communication systems [lecture notes],”IEEE Signal Process. Mag., vol. 37, no. 5, pp. 134–140, 2020

    2020

  25. [25]

    Learning resilient radio resource management policies with graph neural networks,

    N. NaderiAlizadehet al., “Learning resilient radio resource management policies with graph neural networks,”IEEE Trans. Signal Process., vol. 71, pp. 995–1009, 2023

    2023

  26. [26]

    Power control in cell-free massive MIMO networks for UA Vs URLLC under the finite blocklength regime,

    M. Elwekeilet al., “Power control in cell-free massive MIMO networks for UA Vs URLLC under the finite blocklength regime,”IEEE Trans. Commun., vol. 71, no. 2, pp. 1126–1140, 2023. [29]Further advancements for E-UTRA physical layer aspects, 3GPP, Tech- nical Report 36.814 v9.2.0 2017

    2023

  27. [27]

    Scalable cell-free massive MIMO systems,

    E. Bj ¨ornson and L. Sanguinetti, “Scalable cell-free massive MIMO systems,”IEEE Trans. Commun., vol. 68, no. 7, pp. 4247–4261, 2020

    2020

  28. [28]

    Distribution of inner product of complex Gaussian random vectors and its applications,

    R. K. Mallik and N. C. Sagias, “Distribution of inner product of complex Gaussian random vectors and its applications,”IEEE Trans. Commun., vol. 59, no. 12, pp. 3353–3362, 2011

    2011

  29. [29]

    The dispersion of nearest-neighbor decoding for additive non-gaussian channels,

    J. Scarlettet al., “The dispersion of nearest-neighbor decoding for additive non-gaussian channels,”IEEE Trans. Inf. Theory, vol. 63, no. 1, pp. 81–92, 2017

    2017

  30. [30]

    Bandwidth partitioning in decentralized wireless networks,

    N. Jindalet al., “Bandwidth partitioning in decentralized wireless networks,”IEEE Trans. Wireless Commun., vol. 7, no. 12, pp. 5408– 5419, 2008

    2008

  31. [31]

    Boyd and L

    S. Boyd and L. Vandenberghe,Convex Optimization. Cambridge Univ. Press, 2004

    2004

  32. [32]

    Algorithms for generalized fractional programming,

    J.-P. Crouzeix and J. A. Ferland, “Algorithms for generalized fractional programming,”Mathematical Programming, vol. 52, no. 1, pp. 191–207, 1991

    1991

  33. [33]

    On nonlinear fractional programming,

    W. Dinkelbach, “On nonlinear fractional programming,”Management science, vol. 13, no. 7, pp. 492–498, 1967

    1967

  34. [34]

    Globally optimal energy-efficient power control and receiver design in wireless networks,

    A. Zapponeet al., “Globally optimal energy-efficient power control and receiver design in wireless networks,”IEEE Trans. Signal Process., vol. 65, no. 11, pp. 2844–2859, 2017

    2017

  35. [35]

    Design of RIS-aided mMTC+ networks for rate maximization under the finite blocklength regime with imperfect channel knowledge,

    S. Lieseganget al., “Design of RIS-aided mMTC+ networks for rate maximization under the finite blocklength regime with imperfect channel knowledge,”IEEE Commun. Lett., vol. 29, no. 11, pp. 2511–2515, 2025

    2025

  36. [36]

    Avriel,Nonlinear programming: Analysis and methods

    M. Avriel,Nonlinear programming: Analysis and methods. Courier Corporation, 2003

    2003

  37. [37]

    CVX: MATLAB software for disciplined convex programming, version 2.2.2,

    M. Grant and S. Boyd, “CVX: MATLAB software for disciplined convex programming, version 2.2.2,” http://cvxr.com/cvx, 2024

    2024

  38. [38]

    Majorization-minimization algorithms in signal pro- cessing, communications, and machine learning,

    Y . Sunet al., “Majorization-minimization algorithms in signal pro- cessing, communications, and machine learning,”IEEE Trans. Signal Process., vol. 65, no. 3, pp. 794–816, 2017

    2017

  39. [39]

    Attention is all you need,

    A. Vaswaniet al., “Attention is all you need,”Advances in Neural Information Processing Systems, vol. 30, 2017

    2017

  40. [40]

    When machine learning meets wireless cellular networks: Deployment, challenges, and applications,

    U. Challitaet al., “When machine learning meets wireless cellular networks: Deployment, challenges, and applications,”IEEE Commun. Mag., vol. 58, no. 6, pp. 12–18, 2020

    2020

  41. [41]

    Decoupled Weight Decay Regularization

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” arXiv:1711.05101, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 2019

  42. [42]

    Y . C. Eldar, A. Goldsmith, D. G¨und¨uz, and H. V . Poor,Machine learning and wireless communications. Cambridge Univ. Press, 2022

    2022

  43. [43]

    A general framework for scalable UE-AP association in user-centric cell-free massive MIMO based on recurrent neural networks,

    G. Di Gennaroet al., “A general framework for scalable UE-AP association in user-centric cell-free massive MIMO based on recurrent neural networks,”IEEE Trans. Commun., vol. 74, pp. 3103–3119, 2026

    2026

  44. [44]

    Scalable integrated sensing and communications for multi-target detection and tracking in cell-free massive MIMO:A unified framework,

    S. Lieseganget al., “Scalable integrated sensing and communications for multi-target detection and tracking in cell-free massive MIMO:A unified framework,”IEEE Trans. Commun., vol. 74, pp. 2777–2793, 2026

    2026

  45. [45]

    Ghosh and R

    A. Ghosh and R. Ratasuk,Essentials of LTE and LTE-A. Cambridge Univ. Press, 2011

    2011