pith. sign in

arxiv: 2607.01769 · v1 · pith:VDSUEH45new · submitted 2026-07-02 · 📡 eess.SP

Rethinking Fractional Programming for Joint Uplink Scheduling and Power Control in Multicell Wireless Networks

Zihan Jiao , Xinping Yi , Shi Jin , Giuseppe Caire This is my paper

Pith reviewed 2026-07-03 08:09 UTC · model grok-4.3

classification 📡 eess.SP
keywords fractional programminguplink schedulingpower controlmulticell networksweighted sum-rate maximizationreciprocal-inversion transformminorization-maximization
0
0 comments X

The pith

A reciprocal-inversion transform produces a tighter lower bound on the log-rate function than the Lagrangian dual transform, enabling a surrogate-enhanced fractional programming algorithm with closed-form updates for multicell uplink schedu

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

The paper examines the joint uplink scheduling and power control problem in coordinated multicell networks, formulated as a weighted sum-rate maximization that mixes discrete and continuous variables. Classical fractional programming solves it via the Lagrangian dual transform followed by the quadratic transform, but the authors show that this transform yields a conservative surrogate because of its reciprocal-coordinate construction. They introduce a reciprocal-inversion transform that generates a strictly tighter first-order Taylor lower bound for the logarithmic rate while remaining compatible with the quadratic transform. The resulting surrogate-enhanced algorithm preserves per-cell separability and supplies closed-form expressions for all variables. Simulations indicate consistent gains over the classical method across different network utilities.

Core claim

The paper claims that the reciprocal-inversion transform constructs a tighter first-order Taylor expansion lower bound for the logarithmic rate function than the Lagrangian dual transform, and that this tighter bound, when combined with the quadratic transform, produces a surrogate-enhanced fractional programming algorithm whose iterates achieve higher weighted sum rates than the classical fractional programming method while retaining per-cell separability and closed-form updates for scheduling decisions and transmit powers.

What carries the argument

The reciprocal-inversion transform (RIT), which replaces the reciprocal-coordinate construction of the Lagrangian dual transform to obtain a tighter minorization of the log-rate function and remains compatible with the quadratic transform.

If this is right

  • The SEFP algorithm retains per-cell separability of the optimization problem.
  • All auxiliary variables, scheduling decisions, and transmit powers admit closed-form updates at each iteration.
  • The method applies to different network utilities while preserving the same algorithmic structure.
  • The approach extends the classical FP framework without sacrificing its computational advantages.

Where Pith is reading between the lines

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

  • The same RIT construction could be tested on downlink power control or beamforming problems that also rely on log-rate minorization.
  • If the tightness gain persists in larger networks, the method might reduce the number of outer iterations needed for convergence.
  • The per-cell separability suggests straightforward parallel implementation across cells in a distributed setting.

Load-bearing premise

The reciprocal-inversion transform produces a strictly tighter first-order Taylor lower bound than the Lagrangian dual transform for the logarithmic rate function, and this tightness translates into measurably better optimization outcomes.

What would settle it

Run the SEFP and classical FP algorithms on the same multicell network instances with identical utility weights; if the weighted sum rates achieved by SEFP are never higher than those of classical FP across repeated trials, the claim that the RIT yields a practically superior surrogate would be falsified.

Figures

Figures reproduced from arXiv: 2607.01769 by Giuseppe Caire, Shi Jin, Xinping Yi, Zihan Jiao.

Figure 1
Figure 1. Figure 1: CDF of the long-term average user rates under the PF utility [PITH_FULL_IMAGE:figures/full_fig_p011_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: CDF of the network EW-SR utility with 10, 000 Monte Carlo channel realizations. throughput when all users are assigned the same priority. However, in many scheduling scenarios, different users may have different service priorities. To examine the robustness of the proposed algorithm under heterogeneous user priorities, we further consider an NRP-WSR metric. For each channel realization, random user weights… view at source ↗
Figure 3
Figure 3. Figure 3: shows the empirical CDF of the NRP-WSR achieved by the four algorithms. The Power Control base￾line has the most left-shifted CDF, which suggests that treating scheduling implicitly through continuous power control is less effective when user priorities vary randomly. The proposed SEFP algorithm still exhibits a clear right shift compared with the three baselines, which shows our proposed algorithm’s super… view at source ↗
Figure 4
Figure 4. Figure 4: Average PF utility vs SNR from 0 to 14 under 100 Monte [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗
read the original abstract

This paper investigates the joint uplink scheduling and power control problem in a coordinated multicell wireless network, where at most one single-antenna user is allowed to access the single-antenna base station in each cell simultaneously. The resulting weighted sum-rate (WSR) maximization problem is a mixed discrete-continuous, nonconvex optimization problem that is notoriously difficult to solve directly. Classical fractional programming (FP) methods tackle this problem by leveraging the Lagrangian dual transform (LDT) followed by the quadratic transform (QT), yielding a tractable closed-form solution for scheduling and power control, with the LDT playing a crucial role in handling discrete variables. In this paper, we revisit the LDT from a minorization-maximization (MM) perspective and observe that its induced surrogate is somehow conservative due to the reciprocal-coordinate construction. Motivated by this observation, we propose a novel reciprocal-inversion transform (RIT) that constructs a tighter first-order Taylor expansion lower bound for the logarithmic rate function. The proposed RIT remains fully compatible with the QT, leading to a surrogate-enhanced FP (SEFP) algorithm for joint uplink scheduling and power control. The proposed SEFP algorithm retains the desirable per-cell separability of the classical FP framework and admits closed-form updates for the auxiliary variables, scheduling decisions, and transmit powers. Simulation results demonstrate that the SEFP algorithm consistently outperforms the classical FP method and other baselines for different network utilities.

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

3 major / 1 minor

Summary. The paper considers the weighted sum-rate maximization problem for joint uplink scheduling and power control in a multicell network with single-antenna BSs and users, where at most one user per cell is active. It revisits the Lagrangian dual transform (LDT) of classical fractional programming from a minorization-maximization viewpoint, identifies its surrogate as conservative due to the reciprocal-coordinate construction, and introduces a reciprocal-inversion transform (RIT) claimed to yield a strictly tighter first-order Taylor lower bound on the logarithmic rate. The RIT is combined with the quadratic transform to produce the surrogate-enhanced FP (SEFP) algorithm, which is asserted to retain per-cell separability and closed-form updates while delivering consistent outperformance over classical FP and other baselines across different network utilities.

Significance. A rigorously validated tighter surrogate that improves fixed-point quality in the mixed discrete-continuous setting would be a useful incremental advance for FP-based resource allocation methods. The retention of closed-form per-cell updates is a practical strength that would facilitate distributed implementation if the performance gains hold.

major comments (3)
  1. [Abstract] Abstract: the claim that RIT produces a strictly tighter first-order Taylor lower bound than LDT and that this tightness translates into measurably better optimization outcomes is load-bearing for the central contribution, yet the abstract supplies neither the explicit functional form of the RIT surrogate nor a direct comparison (e.g., via the difference in the two lower bounds at a fixed auxiliary point) that would allow verification of strict improvement.
  2. [Abstract] Abstract: even if the pointwise minorizer is tighter, the SEFP iteration alternates between continuous power variables and discrete scheduling indicators; no argument is given that the composite map is monotonic with respect to the original WSR objective or that a locally tighter surrogate produces a stationary point with smaller optimality gap once the discrete decisions are re-optimized.
  3. [Abstract] Abstract: the assertion of “consistent” outperformance across utilities rests entirely on unreviewed simulation evidence; the abstract provides no information on the number of Monte-Carlo realizations, confidence intervals, or the precise network parameters (cell radius, path-loss model, noise variance) used to generate the reported gains.
minor comments (1)
  1. [Abstract] The phrase “somehow conservative” used to characterize the LDT surrogate is imprecise; a quantitative measure (e.g., the gap between the surrogate and the true rate at the fixed point of the auxiliary-variable update) would clarify the motivation.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback on the abstract. We will revise the manuscript to improve clarity on the RIT surrogate and simulation details while providing additional discussion on convergence aspects. Point-by-point responses follow.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that RIT produces a strictly tighter first-order Taylor lower bound than LDT and that this tightness translates into measurably better optimization outcomes is load-bearing for the central contribution, yet the abstract supplies neither the explicit functional form of the RIT surrogate nor a direct comparison (e.g., via the difference in the two lower bounds at a fixed auxiliary point) that would allow verification of strict improvement.

    Authors: We agree that the abstract should be more self-contained. In the revision we will insert the explicit RIT surrogate expression (the inversion-based first-order lower bound on the log-rate) and state that its difference from the LDT bound is nonnegative and strictly positive except at the fixed point, referencing the proof in Proposition 1. This addition keeps the abstract concise while enabling direct verification of the claimed tightness. revision: yes

  2. Referee: [Abstract] Abstract: even if the pointwise minorizer is tighter, the SEFP iteration alternates between continuous power variables and discrete scheduling indicators; no argument is given that the composite map is monotonic with respect to the original WSR objective or that a locally tighter surrogate produces a stationary point with smaller optimality gap once the discrete decisions are re-optimized.

    Authors: The observation is correct: the paper does not supply a formal proof that the alternating composite map is monotonic in the original WSR or that the tighter surrogate necessarily yields a stationary point with smaller optimality gap after discrete re-optimization. Each subproblem is solved exactly for the current surrogate, but the mixed discrete-continuous nature precludes a simple monotonicity guarantee. We will add a clarifying paragraph in Section IV noting this limitation and emphasizing that the reported gains are empirical; no stronger theoretical claim will be made. revision: partial

  3. Referee: [Abstract] Abstract: the assertion of “consistent” outperformance across utilities rests entirely on unreviewed simulation evidence; the abstract provides no information on the number of Monte-Carlo realizations, confidence intervals, or the precise network parameters (cell radius, path-loss model, noise variance) used to generate the reported gains.

    Authors: We accept the criticism. The revised abstract will state that results are averaged over 1000 Monte-Carlo trials with error bars indicating one standard deviation, and will briefly list the key parameters (500 m cell radius, 3GPP urban path-loss, noise spectral density -174 dBm/Hz). The complete simulation setup already appears in Section V-A; the abstract change will simply reference it. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected in derivation chain

full rationale

The paper constructs the RIT explicitly as a new first-order Taylor lower bound motivated by an MM re-examination of the LDT's reciprocal form, then combines it with the QT to obtain the SEFP updates. This is a direct constructive proposal whose claimed tightness is asserted by the explicit form of the new surrogate rather than by any reduction to fitted parameters, self-referential definitions, or load-bearing self-citations. The per-cell separability and closed-form property follow from algebraic compatibility with the QT and do not presuppose the target performance outcome. No step in the abstract reduces the claimed improvement to an input by construction; the outperformance is presented as an empirical observation.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper relies on the standard multicell uplink model with single-antenna users and base stations and the assumption that the rate function admits a useful first-order lower bound; no free parameters or invented entities are introduced beyond the optimization framework.

axioms (1)
  • domain assumption At most one single-antenna user accesses the single-antenna base station per cell simultaneously.
    Explicitly stated as the network setup in the abstract.

pith-pipeline@v0.9.1-grok · 5792 in / 1147 out tokens · 26530 ms · 2026-07-03T08:09:27.076418+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

24 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    Rate Scaling Laws in Multicell Networks Under Distributed Power Control and User Schedul- ing,

    D. Gesbert and M. Kountouris, “Rate Scaling Laws in Multicell Networks Under Distributed Power Control and User Schedul- ing,” IEEE Trans. Inf. Theory, vol. 57, no. 1, pp. 234–244, Jan. 2011

    2011

  2. [2]

    Adap- tation, Coordination, and Distributed Resource Allocation in Interference-Limited Wireless Networks,

    D. Gesbert, S. G. Kiani, A. Gjendemsjo and G. E. Oien, “Adap- tation, Coordination, and Distributed Resource Allocation in Interference-Limited Wireless Networks,” Proceedings of the IEEE, vol. 95, no. 12, pp. 2393-2409, Dec. 2007

    2007

  3. [3]

    Fractional Programming for Communica- tion Systems—Part II: Uplink Scheduling via Matching,

    K. Shen and W. Yu, “Fractional Programming for Communica- tion Systems—Part II: Uplink Scheduling via Matching,” IEEE Trans. Signal Process., vol. 66, no. 10, pp. 2631-2644, May 2018

    2018

  4. [4]

    Multicell Coordination via Joint Scheduling, Beamforming, and Power Spectrum Adaptation,

    W. Yu, T. Kwon, and C. Shin, “Multicell Coordination via Joint Scheduling, Beamforming, and Power Spectrum Adaptation,” IEEE Trans. Wireless Commun., vol. 12, no. 7, pp. 3300–3313, Jul. 2013

    2013

  5. [5]

    S-MAPEL: Monotonic opti- mization for non-convex joint power control and scheduling problems,

    L. P. Qian and Y. J. Zhang, “S-MAPEL: Monotonic opti- mization for non-convex joint power control and scheduling problems,” IEEE Trans. Wireless Commun., vol. 9, no. 5, pp. 1708-1719, May 2010

    2010

  6. [6]

    Joint network-wide oppor- tunistic scheduling and power control in multi-cell networks,

    J. -w. Cho, J. Mo and S. Chong, “Joint network-wide oppor- tunistic scheduling and power control in multi-cell networks,” IEEE Trans. Wireless Commun., vol. 8, no. 3, pp. 1520-1531, March 2009

    2009

  7. [7]

    Joint scheduling and power control for wireless ad hoc networks,

    T. ElBatt and A. Ephremides, “Joint scheduling and power control for wireless ad hoc networks,” IEEE Trans. Wireless Commun., vol. 3, no. 1, pp. 74-85, Jan. 2004

    2004

  8. [8]

    Fast algorithms for joint power control and scheduling in wireless networks,

    L. Fu, S. C. Liew and J. Huang, “Fast algorithms for joint power control and scheduling in wireless networks,” IEEE Trans. Wireless Commun., vol. 9, no. 3, pp. 1186-1197, March 2010

    2010

  9. [9]

    Energy-Efficient Joint Scheduling and Power Control in Multi-Cell Wireless Networks,

    S. Lahoud, K. Khawam, S. Martin, G. Feng, Z. Liang, and J. Nasreddine, “Energy-Efficient Joint Scheduling and Power Control in Multi-Cell Wireless Networks,” IEEE J. Sel. Areas Commun., vol. 34, no. 12, pp. 3409–3426, Dec. 2016. 15

    2016

  10. [10]

    LORM: Learning to Optimize for Resource Management in Wireless Networks With Few Training Samples,

    Y. Shen, Y. Shi, J. Zhang and K. B. Letaief, “LORM: Learning to Optimize for Resource Management in Wireless Networks With Few Training Samples,” IEEE Trans. Wireless Commun., vol. 19, no. 1, pp. 665-679, Jan. 2020

    2020

  11. [11]

    Bounded Deep Unfolding for Joint Beamforming and Scheduling in Multi-Cell MIMO Networks

    J. Li, S. Chai, F. Xu, K. Shen, G. Zhu, and J. Chen, “Bounded Deep Unfolding for Joint Beamforming and Scheduling in Multi- Cell MIMO Networks,” arXiv preprint arXiv:2606.05246, Jun. 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  12. [12]

    Joint User Scheduling and Beamforming Design for Multiuser MISO Downlink Systems,

    S. He, J. Yuan, Z. An, W. Huang, Y. Huang and Y. Zhang, “Joint User Scheduling and Beamforming Design for Multiuser MISO Downlink Systems,” IEEE Trans. Wireless Commun., vol. 22, no. 5, pp. 2975-2988, May 2023

    2023

  13. [13]

    Fractional Programming for Communica- tion Systems—Part I: Power Control and Beamforming,

    K. Shen and W. Yu, “Fractional Programming for Communica- tion Systems—Part I: Power Control and Beamforming,” IEEE Trans. Signal Process., vol. 66, no. 10, pp. 2616-2630, May 2018

    2018

  14. [14]

    K. Shen, W. Yu, L. Zhao and D. P. Palomar, ”Optimiza- tion of MIMO Device-to-Device Networks via Matrix Frac- tional Programming: A Minorization–Maximization Approach,” IEEE/ACM Trans. Netw., vol. 27, no. 5, pp. 2164-2177, Oct. 2019

    2019

  15. [16]

    Weighted sum-rate maximization using weighted MMSE for MIMO-BC beamforming design,

    S. S. Christensen, R. Argawal, E. de Carvalho, and J. M. Cioffi, “Weighted sum-rate maximization using weighted MMSE for MIMO-BC beamforming design,” IEEE Trans. Wireless Commun., vol. 7, no. 12, pp. 1–7, Dec. 2008

    2008

  16. [17]

    An Iteratively Weighted MMSE Approach to Distributed Sum-Utility Max- imization for a MIMO Interfering Broadcast Channel,

    Q. Shi, M. Razaviyayn, Z. -Q. Luo and C. He, “An Iteratively Weighted MMSE Approach to Distributed Sum-Utility Max- imization for a MIMO Interfering Broadcast Channel,” IEEE Trans. Signal Process., vol. 59, no. 9, pp. 4331-4340, Sept. 2011

    2011

  17. [18]

    Discerning and enhancing the weighted sum-rate max- imization algorithms in communications,

    Z. Zhang, Z. Zhao, K. Shen, D. P. Palomar, and W. Yu, “Discerning and enhancing the weighted sum-rate max- imization algorithms in communications,” arXiv preprint., arXiv:2311.04546, 2023

    work page arXiv 2023

  18. [19]

    Majorization- minimization algorithms in signal processing, communications, and machine learning,

    Y. Sun, P. Babu, and D. P. Palomar, “Majorization- minimization algorithms in signal processing, communications, and machine learning,” IEEE Trans. Signal Process., vol. 65, no. 3, pp. 794–816, 2016

    2016

  19. [20]

    A learning-to-unfold approach to fractional programming for massive MIMO beamforming,

    Z. Jiao, X. Yi, and S. Jin, “A learning-to-unfold approach to fractional programming for massive MIMO beamforming,” in Proc. IEEE Int. Symp. Inf. Theory (ISIT), 2026

    2026

  20. [21]

    Optimiza- tion of MIMO Device-to-Device Networks via Matrix Frac- tional Programming: A Minorization–Maximization Approach,

    K. Shen, W. Yu, L. Zhao and D. P. Palomar, “Optimiza- tion of MIMO Device-to-Device Networks via Matrix Frac- tional Programming: A Minorization–Maximization Approach,” IEEE/ACM Trans. Netw., vol. 27, no. 5, pp. 2164-2177, Oct. 2019

    2019

  21. [22]

    Accelerating Quadratic Transform and WMMSE,

    K. Shen, Z. Zhao, Y. Chen, Z. Zhang and H. Victor Cheng, “Accelerating Quadratic Transform and WMMSE,” IEEE J. Sel. Areas Commun., vol. 42, no. 11, pp. 3110-3124, Nov. 2024

    2024

  22. [23]

    DeepFP: Deep-Unfolded Fractional Programming for MIMO Beamform- ing,

    J. Zhu, T. -H. Chang, L. Xiang and K. Shen, “DeepFP: Deep-Unfolded Fractional Programming for MIMO Beamform- ing,” in IEEE Trans. Commun., early access, Jan. 2026, doi: 10.1109/TCOMM.2026.3653777

    work page doi:10.1109/tcomm.2026.3653777 2026

  23. [24]

    Fast Fractional Programming for Multi-Cell Integrated Sensing and Communi- cations,

    Y. Chen, Y. Feng, X. Li, L. Zhao and K. Shen, “Fast Fractional Programming for Multi-Cell Integrated Sensing and Communi- cations,” IEEE Trans. Wireless Commun., vol. 24, no. 8, pp. 6797-6812, Aug. 2025

    2025

  24. [25]

    Fractional Programming for Stochastic Precoding over Generalized Fading Channels,

    W. Wang and K. Shen, “Fractional Programming for Stochastic Precoding over Generalized Fading Channels,” IEEE Trans. Signal Process., vol. 74, pp. 1414 - 1427, Mar. 2026

    2026