Pith. sign in

REVIEW 3 major objections 5 minor 4 cited by

This first survey and tutorial argues LLMs, not heuristics or DRL, are the right engine for 6G's combinatorial wireless optimization problems, organized as a formulate-solve-verify pipeline.

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 ↗

A survey and tutorial that organizes LLM-enabled wireless network optimization into formulation, solution, and verification stages, with case studies drawn from the authors' own prior papers.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Useful survey with a sensible pipeline taxonomy; the quantitative case studies are self-cited and poorly controlled, so the 'transformative' framing overstates the evidence. the 3 major comments →

arxiv 2509.05946 v1 pith:XVC6BSNP submitted 2025-09-07 cs.NI

Large Language Models for Next-Generation Wireless Network Management: A Survey and Tutorial

Bisheng Wei , Ruihong Jiang , Ruichen Zhang , Yinqiu Liu , Dusit Niyato , Yaohua Sun , Yang Lu , Yonghui Li
show 4 more authors
Shiwen Mao Chau Yuen Marco Di Renzo Mugen Peng
This is my paper
classification cs.NI
keywords large language modelswireless network optimization6G networksresource allocationintent networkingretrieval-augmented generationchain-of-thought promptingUAV trajectory optimization
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.

The reading

Wireless network optimization is outgrowing its traditional tools — exhaustive methods explode combinatorially, heuristics stall in dynamic environments, and deep reinforcement learning needs retraining as conditions drift. This paper is the first survey and tutorial devoted specifically to LLM-enabled optimization in wireless networks, and its central claim is that LLMs can carry all three stages: translate natural-language intents into formal optimization models, guide solution search in collaboration with classical solvers, and verify that solutions respect physical constraints. The paper reports concrete gains — up to 16.6–24.7% reduction in average age-of-information for UAV routing over greedy baselines and 11.45% higher data rate than PPO on an LEO satellite problem — and uses them to argue that LLM-enabled optimization is a coherent approach worth standardizing. Why it matters: if the claim holds, network management shifts from expert-crafted models to prompt-driven, human-aligned, self-verifying pipelines ready for 6G's dynamic conditions.

Core claim

The paper claims LLM-enabled optimization is a distinct, viable paradigm for wireless networks, distinguished from ILP, heuristics, and DRL by interpreting natural-language intents, decomposing non-convex combinatorial problems, and refining solutions without retraining. Its organizational claim is a three-stage pipeline — formulation, solution, verification — proposed as the field's roadmap. The evidence backbone is three case studies: a satellite RSMA formulation system completing symbolic modeling within six interaction rounds; LLM-guided UAV routing (LAURA) cutting average maximum AoI by 16.6% and 24.7% versus Greedy for 30 and 40 sensors; and an LEO RSMA intent-networking loop convergin

What carries the argument

The central organizing object is the formulation–solution–verification pipeline for LLM-enabled wireless optimization. The load-bearing mechanisms inside it: natural-language-to-mathematical translation, where LLMs convert user intents into solver-compatible MILP/constraint-programming models; chain-of-thought prompting and retrieval-augmented generation, which decompose problems and pull domain knowledge from curated corpora; solver feedback loops and LLM-guided heuristics (population initialization, heuristic generation, local-search repair, automated parameter tuning); and verification layers — constraint/syntax checking, prompt-based self-verification, ensemble voting, and simulation-gui

Load-bearing premise

The survey's positive conclusions depend on a handful of experimental results, most of them the authors' own prior work, reported without error bars or statistical tests; if those results do not reproduce, the case that LLM-enabled optimization beats traditional methods loses its quantitative support.

What would settle it

Rerun the LAURA UAV routing pipeline on the same sensor configurations (N=30 and N=40) across many random seeds and compare the distribution of average maximum AoI against Greedy; rerun the LEO RSMA intent-networking loop against PPO the same way. If the reported 16.6–24.7% and 11.45% advantages disappear once seed-to-seed variance is counted, or if LLM-generated formulations on a standard benchmark (e.g., the ACCORD-90K or CO-Bench instances) yield worse optimality gaps than expert formulations, the survey's central quantitative narrative loses its support.

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

If this is right

  • Optimization formulation stops being an expert-only activity: operators could state goals in natural language and receive solver-ready models reconfigurable by prompt edits.
  • Hybrid LLM-plus-solver architectures become the default design pattern for wireless optimization, with LLMs owning semantics and solvers owning numeric rigor.
  • Verification becomes a standard third stage, with simulation-in-the-loop repair safety-checking LLM solutions against physical-layer constraints before deployment.
  • Multi-agent systems with token-efficient communication make large-scale, decentralized scheduling and slicing feasible in latency-sensitive environments.
  • The open-source frameworks and benchmarks the paper catalogs give the community ready-made infrastructure to test and compare LLM optimization methods on wireless problems.

Where Pith is reading between the lines

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

  • If the formulation-stage results hold across domains, the bottleneck shifts from modeling expertise to knowledge retrieval and prompt design; a testable prediction, left implicit in the paper, is that richer RAG corpora (standards-style documents) should reduce verification-stage repairs.
  • The headline numerical claims should be the first target for replication: LAURA's AoI advantage and the RSMA loop's PPO gain are reported without variance or significance tests, so a seeded rerun with distributional reporting would settle the strength of the survey's evidence — a concern the paper itself registers under trustworthy verification.
  • The pipeline structure implies a natural scaling path: as small language models and edge deployment mature, the formulate-solve-verify loop could move on-device, turning the case-study prototypes into closed-loop network controllers — listed among future directions but not developed.
  • The same pipeline may transfer to adjacent discrete-decision domains where humans express intent (edge offloading, slicing, vehicular scheduling), though the survey's evidence is limited to the three presented scenarios.
Share X Bluesky LinkedIn Reddit HN

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

3 major / 5 minor

Summary. The paper is a survey and tutorial on using large language models (LLMs) for optimization problems in wireless networks. It proposes a formulation–solution–verification pipeline taxonomy, contrasts LLM-driven optimization with classical heuristics and DRL, reviews techniques such as prompting, RAG, solver collaboration, verification, and multi-agent frameworks, and tabulates open-source frameworks and datasets. Three case studies are developed in depth: LLM-assisted optimization formulation for satellite communications, LLM-guided UAV routing for AoI minimization (LAURA), and LLM-enabled RSMA sum-rate maximization in LEO satellite networks. The paper closes with challenges and future research directions.

Significance. If its central claims hold, the paper fills a genuine gap: it is one of the first surveys devoted specifically to LLM-enabled optimization in wireless networks, and its pipeline taxonomy is a useful organizing device for a fast-growing literature. The paper is strong on breadth: it covers a wide range of recent work, includes clear figures and comparative tables, and provides pointers to open-source tools and benchmarks. The main weakness is that the survey's positive quantitative narrative relies on a small number of self-cited, unreplicated case studies that are presented without statistical or computational-budget controls. The survey is therefore valuable as a structured entry point to the field, but its strongest empirical claims are not yet supported.

major comments (3)
  1. [§V.B.4 and §V.C.4, Figs. 11 and 13] The central quantitative evidence for LLM superiority is not reproducible or statistically grounded. The AoI reductions (16.6–24.7% vs Greedy) and the 11.45% rate improvement vs PPO are presented without error bars, number of seeds, hyperparameters, or compute budgets. Both LLM pipelines are evolutionary/prompt-refinement loops (Steps 1–3 in V.B.3; Steps 1–4 in V.C.3), so the reported gains may reflect additional search effort rather than LLM semantic reasoning. The baselines are not matched in number of fitness evaluations, generations, or wall-clock time, and no ablation isolates the LLM contribution (e.g., a genetic algorithm with the same initialization or a random mutator with the same evaluation budget). These results underpin the 'super improvement' claim in §I.A and the comparative entries in Table II, so the overstatements should be tempered and the experimental details/ablation
  2. [§V.A–V.C, Refs. [122], [156], [195], [210]] The three featured case studies are drawn almost entirely from the authors' own prior work: LAURA (B. Wei et al., [122]), satellite modeling (R. Zhang et al., [156], [195]), and RSMA intent networking (R. Zhang et al., [210]). No independent replication or third-party comparison is reported. For a survey whose contribution is to synthesize evidence for LLM-enabled wireless optimization, this self-cited evidence base is a circular support for the central narrative. The manuscript should either incorporate independent results from other groups, or clearly and repeatedly state that these demonstrations are author-provided proofs of concept rather than an empirical survey of the literature. The current wording, e.g., 'markedly outperforming traditional manual modeling techniques' and 'fundamentally transforms', oversells the strength of the evidence.
  3. [§II.C, Table II] Table II and the surrounding discussion overstate LLM capabilities. The entry '✓ Real-time learning capabilities; adapts to dynamic network states [85]' is not accurate: LLMs do not update their parameters in real time; they perform inference with prompts, retrieval, and iterative refinement. The cited reference [85] is a survey of digital twins and does not establish real-time learning for LLMs. Similarly, 'Designed for complexity and scale' is a design aspiration rather than a demonstrated property. These claims are load-bearing because they define the central distinction between LLM-driven and traditional optimization paradigms. They should be reworded to describe what is currently evidenced: flexible inference-time adaptation, prompt-based reconfiguration, and retrieval-augmented context handling.
minor comments (5)
  1. [§V.A.1, §V.B.1, §V.C.1] Cross-references to open-source toolkits say 'detailed in Section V', but the open-source frameworks are presented in Section IV. For example, V.B.1 refers to 'ReEvo [123] and HeuriGym [184]' as being in Section V. These should be corrected to Section IV.
  2. [Throughout (title, abstract, Section I)] The phrase 'optimization problem' is repeatedly used where 'optimization problems' or 'optimization' is intended (e.g., the abstract: 'dedicated explicitly to LLM-enabled optimization problem in wireless networks'). This is a recurring grammar/language issue that should be cleaned up.
  3. [Fig. 4] The figure label 'Refinement over RHLF' appears to be a typo for 'RLHF'. Please correct.
  4. [§V.A.4, Fig. 9] The 'interaction rounds' metric is not formally defined, and the claim that chunk size 2000/3000 'achieve great interaction quality' is based on a single illustrative experiment. Consider defining the metric and clearly labeling this as a qualitative/illustrative result.
  5. [§V.B.4] The text says the LLM framework 'consistently achieves superior performance' but Fig. 11 shows only three values of N and no statistical comparison across multiple runs. Please add error bars or weaken the wording.

Circularity Check

0 steps flagged

No significant circularity: the paper is a survey/tutorial, and its self-cited case studies are reported experimental evidence rather than predictions derived from the survey's framework.

full rationale

This manuscript is a survey and tutorial, not a derivation. Its central contribution is a taxonomy of LLM-enabled optimization pipelines (formulation, solution, verification) and a synthesis of existing results. The case studies in Section V (LAURA for UAV routing, RSMA for LEO satellite networks, and LLM-based optimization formulation) are explicitly drawn from prior cited papers, several of which are by the same authors. Citing one's own prior work in a survey is normal; it becomes circular only if the survey's central claim is justified exclusively by those citations or if the cited results are secretly assumed by the framework. Here the survey's quantitative statements (e.g., 16.6–24.7% AoI reductions in Fig. 11 and 11.45% rate gain in Fig. 13) are empirical findings reported in the cited papers, not consequences of the survey's own equations. The paper does not fit any parameter and then rename it a prediction, and it invokes no uniqueness theorem or ansatz from self-citation. The broader narrative is also supported by independent external work, including OptiMUS [47], LLM-OptiRA [48], ReEvo [123], and Habib et al. [86] (17.25% throughput improvement over DRL). Furthermore, the paper itself flags the central limitations in Section VI—LLMs' lack of causal grounding in wireless physics, feasibility risks, and the need for verification—so it does not present LLM-enabled optimization as a forced or definitionally guaranteed result. Concerns about unmatched compute budgets or missing error bars in the self-cited case studies are legitimate reproducibility/correctness concerns, but they are not circularity: no specific reduction of a prediction to its input is exhibited. Under the required standard, no circular step can be identified.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The survey's central claim rests on the validity of the cited literature and the completeness of its taxonomy, both of which are assumed rather than demonstrated. It introduces no new entities, parameters, or formal results.

axioms (3)
  • domain assumption LLMs possess the capabilities the survey attributes to them in wireless optimization (e.g., translating natural language intents into correct mathematical models).
    This is the premise of the entire survey, invoked in Section II.B and III.A, and supported only by cited examples, many self-authored.
  • domain assumption The cited experimental results, especially those from the authors' own prior papers, are accurate and representative.
    The survey treats the LAURA, satellite modeling, and RSMA results (Section V) as evidence for its claims, but these are single-run comparisons without independent replication.
  • domain assumption The taxonomy of formulation, solution, and verification is a complete and useful decomposition of LLM-enabled wireless optimization.
    This organizing structure is proposed in Section III and used throughout, but it is not derived from first principles or validated against the literature.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Large Language Models for Next-Generation Wireless Network Management: A Survey and Tutorial." pith.science (2026). https://pith.science/paper/XVC6BSNP

@misc{pith2026250905946,
  author       = {Pith},
  title        = {Pith review of: Large Language Models for Next-Generation Wireless Network Management: A Survey and Tutorial},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XVC6BSNP}},
  note         = {Machine review of arXiv:2509.05946}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The rapid advancement toward sixth-generation (6G) wireless networks has significantly intensified the complexity and scale of optimization problems, including resource allocation and trajectory design, often formulated as combinatorial problems in large discrete decision spaces. However, traditional optimization methods, such as heuristics and deep reinforcement learning (DRL), struggle to meet the demanding requirements of real-time adaptability, scalability, and dynamic handling of user intents in increasingly heterogeneous and resource-constrained network environments. Large language models (LLMs) present a transformative paradigm by enabling natural language-driven problem formulation, context-aware reasoning, and adaptive solution refinement through advanced semantic understanding and structured reasoning capabilities. This paper provides a systematic and comprehensive survey of LLM-enabled optimization frameworks tailored for wireless networks. We first introduce foundational design concepts and distinguish LLM-enabled methods from conventional optimization paradigms. Subsequently, we critically analyze key enabling methodologies, including natural language modeling, solver collaboration, and solution verification processes. Moreover, we explore representative case studies to demonstrate LLMs' transformative potential in practical scenarios such as optimization formulation, low-altitude economy networking, and intent networking. Finally, we discuss current research challenges, examine prominent open-source frameworks and datasets, and identify promising future directions to facilitate robust, scalable, and trustworthy LLM-enabled optimization solutions for next-generation wireless networks.

Figures

Figures reproduced from arXiv: 2509.05946 by Bisheng Wei, Chau Yuen, Dusit Niyato, Marco Di Renzo, Mugen Peng, Ruichen Zhang, Ruihong Jiang, Shiwen Mao, Yang Lu, Yaohua Sun, Yinqiu Liu, Yonghui Li.

Figure 1
Figure 1. Figure 1: Overall organization of this survey on LLM-enabled optimization problem in wireless networks. We first present the evolution and core foundations [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An overview of LLM-enabled optimization problem. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A comparison of features for typical optimization problem in wireless [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: An illustration of how LLMs are integrated into the process of solving optimization. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Illustration of optimization formulation using the LLM-enabled framework in Open Ran systems. Part A shows the user’s natural language input for [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of a multi-agent framework applied to an optimization pipeline, where a complex task is partitioned into subtasks handled by domain [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Architecture of an LLM-enabled optimization formulation system for satellite communication networks, integrating semantic parsing, RAG, and [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Illustration of satellite communication modeling using the LLMs. Part A depicts the user’s natural language description and the generative AI agent’s [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Relationship between chunk size and interaction number for LLM [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: System model with one UAV, one data center and [PITH_FULL_IMAGE:figures/full_fig_p021_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of the average AoI between the LLM-enabled framework [PITH_FULL_IMAGE:figures/full_fig_p021_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: The system model and the workflow of the LLM-enabled optimizing transmission strategies. The population is updated iteratively through a [PITH_FULL_IMAGE:figures/full_fig_p023_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Comparison of convergence speed between the LLM-enabled [PITH_FULL_IMAGE:figures/full_fig_p023_13.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 4 Pith papers

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

  1. 6G Needs Agents: Toward Agentic AI-Native Networks for Autonomous Intelligence

    cs.NI 2026-05 unverdicted novelty 6.0

    6G networks need LLM-based agents in a layered semantic control plane to achieve autonomous intelligence, with empirical results showing that heterogeneous deployment across device-edge-core is required due to inheren...

  2. Smart-TCP: An Agentic AI-based Autonomous and Adaptive TCP Protocol

    cs.NI 2025-11 conditional novelty 6.0

    An LLM with a deterministic arithmetic tool can drive TCP state/flag decisions and complete simulated sessions, where pure LLMs cannot.

  3. When Large Language Models Meet UAV Projects: An Empirical Study from Developers' Perspective

    cs.SE 2025-09 conditional novelty 6.0

    The first empirical taxonomy of LLM tasks in UAVs, with an academia-industry comparison and survey, shows LLMs are used mainly for planning and interaction, not direct control.

  4. Agents Should Replace Narrow Predictive AI as the Orchestrator in 6G AI-RAN

    cs.NI 2026-05 unverdicted novelty 4.0

    Position paper proposes replacing fragmented narrow AI models with LLMs as the cognitive orchestrator in the RAN Intelligent Controller for Level 5 autonomous 6G networks.

Reference graph

Works this paper leans on

273 extracted references · 21 canonical work pages · cited by 4 Pith papers · 20 internal anchors

  1. [1]

    5g-advanced toward 6g: Past, present, and future,

    W. Chen, X. Lin, J. Lee,et al., “5g-advanced toward 6g: Past, present, and future,”IEEE Journal on Selected Areas in Communications, vol. 41, no. 6, pp. 1592–1619, 2023

    2023

  2. [2]

    6g wireless systems: Vision, requirements, challenges, insights, and opportunities,

    H. Tataria, M. Shafi, A. F. Molisch,et al., “6g wireless systems: Vision, requirements, challenges, insights, and opportunities,”Proceedings of the IEEE, vol. 109, no. 7, pp. 1166–1199, 2021

    2021

  3. [3]

    6g internet of things: A comprehensive survey,

    D. C. Nguyen, M. Ding, P. N. Pathirana,et al., “6g internet of things: A comprehensive survey,”IEEE Internet of Things Journal, vol. 9, no. 1, pp. 359–383, 2022

    2022

  4. [4]

    A survey on integrated sensing, communication, and computation,

    D. Wen, Y . Zhou, X. Li,et al., “A survey on integrated sensing, communication, and computation,”IEEE Communications Surveys & Tutorials, pp. 1–1, 2024

    2024

  5. [5]

    Urllc in beyond 5g and 6g networks: An interference management perspective,

    M. U. A. Siddiqui, H. Abumarshoud, L. Bariah,et al., “Urllc in beyond 5g and 6g networks: An interference management perspective,”IEEE Access, vol. 11, pp. 54 639–54 663, 2023

    2023

  6. [6]

    Decision-theoretic distributed channel selection for opportunistic spectrum access: Strategies, chal- lenges and solutions,

    Y . Xu, A. Anpalagan, Q. Wu,et al., “Decision-theoretic distributed channel selection for opportunistic spectrum access: Strategies, chal- lenges and solutions,”IEEE Communications Surveys & Tutorials, vol. 15, no. 4, pp. 1689–1713, 2013

    2013

  7. [7]

    Resource allocation for ultra- dense networks: A survey, some research issues and challenges,

    Y . Teng, M. Liu, F. R. Yu,et al., “Resource allocation for ultra- dense networks: A survey, some research issues and challenges,”IEEE Communications Surveys & Tutorials, vol. 21, no. 3, pp. 2134–2168, 2019

    2019

  8. [8]

    Joint task offloading and resource allocation for multi-server mobile-edge computing networks,

    T. X. Tran and D. Pompili, “Joint task offloading and resource allocation for multi-server mobile-edge computing networks,”IEEE Transactions on Vehicular Technology, vol. 68, no. 1, pp. 856–868, 2018

    2018

  9. [9]

    Ultra-high rate-reliability fairness in grant-free massive urllc noma system: Joint power and channel allocation using meta-heuristic search,

    M. Katwe, K. Singh, C.-P. Li,et al., “Ultra-high rate-reliability fairness in grant-free massive urllc noma system: Joint power and channel allocation using meta-heuristic search,”IEEE Transactions on Vehicular Technology, vol. 72, no. 12, pp. 15 899–15 915, 2023

    2023

  10. [10]

    A survey on mobile anchor node as- sisted localization in wireless sensor networks,

    G. Han, J. Jiang, C. Zhang,et al., “A survey on mobile anchor node as- sisted localization in wireless sensor networks,”IEEE Communications Surveys & Tutorials, vol. 18, no. 3, pp. 2220–2243, 2016

    2016

  11. [11]

    On the complexity of joint subcarrier and power allocation for multi-user ofdma systems,

    Y .-F. Liu and Y .-H. Dai, “On the complexity of joint subcarrier and power allocation for multi-user ofdma systems,”IEEE Transactions on Signal Processing, vol. 62, no. 3, pp. 583–596, 2014

    2014

  12. [12]

    Optimal resource allocation for multicarrier noma in short packet communications,

    J. Chen, L. Zhang, Y .-C. Liang,et al., “Optimal resource allocation for multicarrier noma in short packet communications,”IEEE Transactions on Vehicular Technology, vol. 69, no. 2, pp. 2141–2156, 2020

    2020

  13. [13]

    A survey on modeling and optimizing multi-objective systems,

    J.-H. Cho, Y . Wang, R. Chen,et al., “A survey on modeling and optimizing multi-objective systems,”IEEE Communications Surveys & Tutorials, vol. 19, no. 3, pp. 1867–1901, 2017

    1901

  14. [14]

    Embracing complexity: Agent-based modeling for hetnets design and optimization via concur- rent reinforcement learning algorithms,

    M. Ibrahim, U. S. Hashmi, M. Nabeel,et al., “Embracing complexity: Agent-based modeling for hetnets design and optimization via concur- rent reinforcement learning algorithms,”IEEE Transactions on Network and Service Management, vol. 18, no. 4, pp. 4042–4062, 2021

    2021

  15. [15]

    Adaptive antenna array assisted dynamic channel allocation techniques,

    J. S. Blogh, P. J. Cherriman, and L. Hanzo, “Adaptive antenna array assisted dynamic channel allocation techniques,”IEEE Journal on Selected Areas in Communications, vol. 19, no. 2, pp. 305–311, 2002

    2002

  16. [16]

    Energy-efficient deep reinforcement learning assisted resource allocation for 5g-ran slicing,

    Y . Azimi, S. Yousefi, H. Kalbkhani,et al., “Energy-efficient deep reinforcement learning assisted resource allocation for 5g-ran slicing,” IEEE Transactions on Vehicular Technology, vol. 71, no. 1, pp. 856– 871, 2021

    2021

  17. [17]

    Network slicing for embb, urllc, and mmtc: An uplink rate-splitting multiple access approach,

    Y . Liu, B. Clerckx, and P. Popovski, “Network slicing for embb, urllc, and mmtc: An uplink rate-splitting multiple access approach,”IEEE Transactions on Wireless Communications, vol. 23, no. 3, pp. 2140– 2152, 2023

    2023

  18. [18]

    Joint optimization strategy of computa- tion offloading and resource allocation in multi-access edge computing environment,

    H. Li, H. Xu, C. Zhou,et al., “Joint optimization strategy of computa- tion offloading and resource allocation in multi-access edge computing environment,”IEEE Transactions on Vehicular Technology, vol. 69, no. 9, pp. 10 214–10 226, 2020

    2020

  19. [19]

    Intelligent reflecting surface enhanced wireless network via joint active and passive beamforming,

    Q. Wu and R. Zhang, “Intelligent reflecting surface enhanced wireless network via joint active and passive beamforming,”IEEE transactions on wireless communications, vol. 18, no. 11, pp. 5394–5409, 2019

    2019

  20. [20]

    Federated edge learning for 6g: Foundations, methodologies, and applications,

    M. Tao, Y . Zhou, Y . Shi,et al., “Federated edge learning for 6g: Foundations, methodologies, and applications,”Proceedings of the IEEE, 2024

    2024

  21. [21]

    Space-air-ground integrated wireless networks for 6g: Basics, key technologies and future trends,

    Y . Xiao, Z. Ye, M. Wu,et al., “Space-air-ground integrated wireless networks for 6g: Basics, key technologies and future trends,”IEEE Journal on Selected Areas in Communications, 2024

    2024

  22. [22]

    Toward 6g optical fronthaul: A survey on enabling technologies and research perspectives,

    A. Fayad, T. Cinkler, and J. Rak, “Toward 6g optical fronthaul: A survey on enabling technologies and research perspectives,”IEEE Communications Surveys & Tutorials, vol. 27, no. 1, pp. 629–666, 2024

    2024

  23. [23]

    A comprehensive survey of large ai models for future communications: Foundations, applications and challenges,

    F. Jiang, C. Pan, L. Dong,et al., “A comprehensive survey of large ai models for future communications: Foundations, applications and challenges,”arXiv preprint arXiv:2505.03556, 2025

    Pith/arXiv arXiv 2025

  24. [24]

    Large language model (llm) for telecommunications: A comprehensive survey on principles, key tech- niques, and opportunities,

    H. Zhou, C. Hu, Y . Yuan,et al., “Large language model (llm) for telecommunications: A comprehensive survey on principles, key tech- niques, and opportunities,”IEEE Communications Surveys & Tutorials, 2024

    2024

  25. [25]

    Deepseek-inspired exploration of rl-based llms and synergy with wireless networks: A survey,

    Y . Qiao, P.-N. Tran, J. S. Yoon,et al., “Deepseek-inspired exploration of rl-based llms and synergy with wireless networks: A survey,”arXiv preprint arXiv:2503.09956, 2025

    arXiv 2025

  26. [26]

    Large language models- empowered wireless networks: Fundamentals, architecture, and chal- lenges,

    L. U. Khan, M. Guizani, S. Muhaidat,et al., “Large language models- empowered wireless networks: Fundamentals, architecture, and chal- lenges,”arXiv preprint arXiv:2506.10651, 2025

    Pith/arXiv arXiv 2025

  27. [27]

    Wirelessllm: Empowering large language models towards wireless intelligence,

    J. Shao, J. Tong, Q. Wu,et al., “Wirelessllm: Empowering large language models towards wireless intelligence,”arXiv preprint arXiv:2405.17053, 2024

    Pith/arXiv arXiv 2024

  28. [28]

    Nextg-gpt: Leveraging genai for advancing wireless networks and communication research,

    A. M. Nazar, M. Y . Selim, D. Qiao,et al., “Nextg-gpt: Leveraging genai for advancing wireless networks and communication research,” arXiv preprint arXiv:2505.19322, 2025

    Pith/arXiv arXiv 2025

  29. [29]

    Large language models for wireless networks: an overview from the prompt engineering perspective,

    H. Zhou, C. Hu, D. Yuan,et al., “Large language models for wireless networks: an overview from the prompt engineering perspective,”IEEE Wireless Communications, 2025

    2025

  30. [30]

    A survey on large language models for communication, network, and service manage- ment: Application insights, challenges, and future directions,

    G. O. Boateng, H. Sami, A. Alagha,et al., “A survey on large language models for communication, network, and service manage- ment: Application insights, challenges, and future directions,”IEEE Communications Surveys & Tutorials, 2025

    2025

  31. [31]

    A survey on intelligent network operations and performance optimization based on large language models,

    S. Long, J. Tan, B. Mao,et al., “A survey on intelligent network operations and performance optimization based on large language models,”IEEE Communications Surveys & Tutorials, pp. 1–1, 2025

    2025

  32. [32]

    Mobile edge intelligence for large language models: A contemporary survey,

    G. Qu, Q. Chen, W. Wei,et al., “Mobile edge intelligence for large language models: A contemporary survey,”IEEE Communications Surveys & Tutorials, 2025

    2025

  33. [33]

    Generative ai for space-air-ground integrated networks,

    R. Zhang, H. Du, D. Niyato,et al., “Generative ai for space-air-ground integrated networks,”IEEE Wireless Communications, vol. 31, no. 6, pp. 10–20, 2024

    2024

  34. [34]

    Empowering near-field communications in low-altitude economy with llm: Fundamentals, potentials, solutions, and future directions,

    Z. Xu, T. Zheng, and L. Dai, “Empowering near-field communications in low-altitude economy with llm: Fundamentals, potentials, solutions, and future directions,”arXiv preprint arXiv:2506.17067, 2025

    Pith/arXiv arXiv 2025

  35. [35]

    Machine learning for large-scale optimization in 6g wireless networks,

    Y . Shi, L. Lian, Y . Shi,et al., “Machine learning for large-scale optimization in 6g wireless networks,”IEEE Communications Surveys & Tutorials, vol. 25, no. 4, pp. 2088–2132, 2023

    2088

  36. [36]

    Large language models for combinatorial optimization: A systematic review,

    F. D. Ros, M. Soprano, L. D. Gaspero,et al., “Large language models for combinatorial optimization: A systematic review,” 2025. [Online]. Available: https://arxiv.org/abs/2507.03637 26

    arXiv 2025

  37. [37]

    An overview of machine learning-based techniques for solving optimization problems in communications and signal processing,

    H. Dahrouj, R. Alghamdi, H. Alwazani,et al., “An overview of machine learning-based techniques for solving optimization problems in communications and signal processing,”IEEE Access, vol. 9, pp. 74 908–74 938, 2021

    2021

  38. [38]

    A comprehensive survey of linear, integer, and mixed-integer programming approaches for optimizing resource allocation in 5g and beyond networks,

    N. Ejaz and S. Choudhury, “A comprehensive survey of linear, integer, and mixed-integer programming approaches for optimizing resource allocation in 5g and beyond networks,” 2025. [Online]. Available: https://arxiv.org/abs/2502.15585

    Pith/arXiv arXiv 2025

  39. [39]

    A lagrangian-relaxation-based approach for service function chain dynamic orchestration for the internet of things,

    Y . Liu, Y . Lu, X. Li,et al., “A lagrangian-relaxation-based approach for service function chain dynamic orchestration for the internet of things,”IEEE Internet of Things Journal, vol. 8, no. 23, pp. 17 071– 17 089, 2021

    2021

  40. [40]

    Time-varying convex optimization: Time-structured algorithms and applications,

    A. Simonetto, E. Dall’Anese, S. Paternain,et al., “Time-varying convex optimization: Time-structured algorithms and applications,”Proceed- ings of the IEEE, vol. 108, no. 11, pp. 2032–2048, 2020

    2032

  41. [41]

    Spatio-temporal mixing for compu- tational offloading in satellite edge networks with channel uncertainty,

    S. Zhang, H. Yang, F. Wang,et al., “Spatio-temporal mixing for compu- tational offloading in satellite edge networks with channel uncertainty,” IEEE Transactions on Wireless Communications, vol. 24, no. 7, pp. 6151–6165, 2025

    2025

  42. [42]

    Deep reinforcement learning for internet of things: A comprehensive survey,

    W. Chen, X. Qiu, T. Cai,et al., “Deep reinforcement learning for internet of things: A comprehensive survey,”IEEE Communications Surveys & Tutorials, vol. 23, no. 3, pp. 1659–1692, 2021

    2021

  43. [43]

    Reinforcement learning for intelligent healthcare systems: A review of challenges, applications, and open research issues,

    A. A. Abdellatif, N. Mhaisen, A. Mohamed,et al., “Reinforcement learning for intelligent healthcare systems: A review of challenges, applications, and open research issues,”IEEE Internet of Things Journal, vol. 10, no. 24, pp. 21 982–22 007, 2023

    2023

  44. [44]

    Foundational challenges in assuring alignment and safety of large language models,

    U. Anwar, A. Saparov, J. Rando,et al., “Foundational challenges in assuring alignment and safety of large language models,” 2024. [Online]. Available: https://arxiv.org/abs/2404.09932

    Pith/arXiv arXiv 2024

  45. [45]

    A survey of integrating generative artificial intelligence and 6g mobile services: Architectures, solutions, technologies and outlooks,

    Y .-J. Liu, H. Du, X. Xu,et al., “A survey of integrating generative artificial intelligence and 6g mobile services: Architectures, solutions, technologies and outlooks,”IEEE Transactions on Cognitive Commu- nications and Networking, 2025

    2025

  46. [46]

    Ai-generated network design: A diffusion model-based learning approach,

    Y . Huang, M. Xu, X. Zhang,et al., “Ai-generated network design: A diffusion model-based learning approach,”IEEE Network, vol. 38, no. 3, pp. 202–209, 2024

    2024

  47. [47]

    Optimus: Scalable op- timization modeling with (mi)lp solvers and large language models,

    A. AhmadiTeshnizi, W. Gao, and M. Udell, “Optimus: Scalable op- timization modeling with (mi)lp solvers and large language models,” arXiv preprint arXiv:2402.10172, 2024

    Pith/arXiv arXiv 2024

  48. [48]

    Llm-optira: Llm-driven optimization of resource allocation for non-convex problems in wireless communi- cations,

    X. Peng, Y . Liu, Y . Cang,et al., “Llm-optira: Llm-driven optimization of resource allocation for non-convex problems in wireless communi- cations,”arXiv preprint arXiv:2505.02091, 2025

    arXiv 2025

  49. [49]

    Towards better chain-of-thought prompting strategies: A survey,

    Z. Yu, L. He, Z. Wu,et al., “Towards better chain-of-thought prompting strategies: A survey,”arXiv preprint arXiv:2310.04959, 2023

    Pith/arXiv arXiv 2023

  50. [50]

    Automatic chain of thought prompting in large language models,

    Z. Zhang, A. Zhang, M. Li,et al., “Automatic chain of thought prompting in large language models,”arXiv preprint arXiv:2210.03493, 2022

    Pith/arXiv arXiv 2022

  51. [51]

    Chain-of-thought for large language model-empowered wireless communications,

    X. Wang, J. Zhu, R. Zhang,et al., “Chain-of-thought for large language model-empowered wireless communications,” 2025. [Online]. Available: https://arxiv.org/abs/2505.22320

    Pith/arXiv arXiv 2025

  52. [52]

    World models for cognitive agents: Transforming edge intelligence in future networks,

    C. Zhao, R. Zhang, J. Wang,et al., “World models for cognitive agents: Transforming edge intelligence in future networks,”arXiv preprint arXiv:2506.00417, 2025

    Pith/arXiv arXiv 2025

  53. [53]

    A survey on large language models with some insights on their capabilities and limitations,

    A. Matarazzo and R. Torlone, “A survey on large language models with some insights on their capabilities and limitations,”arXiv preprint arXiv:2501.04040, 2025

    Pith/arXiv arXiv 2025

  54. [54]

    Artificial intelligence for safety-critical systems in industrial and transportation domains: A survey,

    J. Perez-Cerrolaza, J. Abella, M. Borg,et al., “Artificial intelligence for safety-critical systems in industrial and transportation domains: A survey,”ACM Computing Surveys, vol. 56, no. 7, pp. 1–40, 2024

    2024

  55. [55]

    Real-time prediction of non-stationary wireless channels,

    M. A. A. Careem and A. Dutta, “Real-time prediction of non-stationary wireless channels,”IEEE Transactions on Wireless Communications, vol. 19, no. 12, pp. 7836–7850, 2020

    2020

  56. [56]

    An overview for designing 6g networks: Technologies, spectrum management, enhanced air interface and ai/ml optimization,

    D. Sharma, V . Tilwari, and S. Pack, “An overview for designing 6g networks: Technologies, spectrum management, enhanced air interface and ai/ml optimization,”IEEE Internet of Things Journal, 2024

    2024

  57. [57]

    Graph reinforcement learning for combinatorial optimization: A survey and unifying per- spective,

    V .-A. Darvariu, S. Hailes, and M. Musolesi, “Graph reinforcement learning for combinatorial optimization: A survey and unifying per- spective,”arXiv preprint arXiv:2404.06492, 2024

    Pith/arXiv arXiv 2024

  58. [58]

    Deep reinforcement learning-based intelligent reflecting surface for secure wireless communications,

    H. Yang, Z. Xiong, J. Zhao,et al., “Deep reinforcement learning-based intelligent reflecting surface for secure wireless communications,” IEEE Transactions on Wireless Communications, vol. 20, no. 1, pp. 375–388, 2021

    2021

  59. [59]

    Deep reinforcement learning for user association and resource allocation in heterogeneous cellular networks,

    N. Zhao, Y .-C. Liang, D. Niyato,et al., “Deep reinforcement learning for user association and resource allocation in heterogeneous cellular networks,”IEEE Transactions on Wireless Communications, vol. 18, no. 11, pp. 5141–5152, 2019

    2019

  60. [60]

    Balancing qos and security in the edge: Existing practices, challenges, and 6g opportunities with machine learning,

    Z. M. Fadlullah, B. Mao, and N. Kato, “Balancing qos and security in the edge: Existing practices, challenges, and 6g opportunities with machine learning,”IEEE Communications Surveys & Tutorials, vol. 24, no. 4, pp. 2419–2448, 2022

    2022

  61. [61]

    Joint resource allocation and trajectory optimization with qos in uav-based noma wireless networks,

    Y . Li, H. Zhang, K. Long,et al., “Joint resource allocation and trajectory optimization with qos in uav-based noma wireless networks,” IEEE Transactions on Wireless Communications, vol. 20, no. 10, pp. 6343–6355, 2021

    2021

  62. [62]

    Toward safe and accelerated deep reinforcement learning for next-generation wireless networks,

    A. M. Nagib, H. Abou-zeid, and H. S. Hassanein, “Toward safe and accelerated deep reinforcement learning for next-generation wireless networks,”IEEE Network, vol. 37, no. 2, pp. 182–189, 2022

    2022

  63. [63]

    Joint resource alloca- tion and transceiver design for sum-rate maximization under latency constraints in multicell mu-mimo systems,

    I. M. Braga, R. P. Antonioli, G. Fodor,et al., “Joint resource alloca- tion and transceiver design for sum-rate maximization under latency constraints in multicell mu-mimo systems,”IEEE Transactions on Communications, vol. 69, no. 7, pp. 4569–4584, 2021

    2021

  64. [64]

    Quantum optimiza- tion: Potential, challenges, and the path forward,

    A. Abbas, A. Ambainis, B. Augustino,et al., “Quantum optimiza- tion: Potential, challenges, and the path forward,”arXiv preprint arXiv:2312.02279, 2023

    Pith/arXiv arXiv 2023

  65. [65]

    Joint optimization of resource allocation and trajectory control for mobile group users in fixed- wing uav-enabled wireless network,

    X. Yan, X. Fang, C. Deng,et al., “Joint optimization of resource allocation and trajectory control for mobile group users in fixed- wing uav-enabled wireless network,”IEEE Transactions on Wireless Communications, vol. 23, no. 2, pp. 1608–1621, 2023

    2023

  66. [66]

    Resource allocation and task scheduling in fog computing and internet of everything environments: A taxonomy, review, and future directions,

    B. Jamil, H. Ijaz, M. Shojafar,et al., “Resource allocation and task scheduling in fog computing and internet of everything environments: A taxonomy, review, and future directions,”ACM Computing Surveys (CSUR), vol. 54, no. 11s, pp. 1–38, 2022

    2022

  67. [67]

    Semantic-aware vision-assisted inte- grated sensing and communication: Architecture and resource alloca- tion,

    Y . Lu, W. Mao, H. Du,et al., “Semantic-aware vision-assisted inte- grated sensing and communication: Architecture and resource alloca- tion,”IEEE Wireless Communications, vol. 31, no. 3, pp. 302–308, 2024

    2024

  68. [68]

    Energy-quality-time optimized task mapping on dvfs-enabled multicores,

    L. Mo, A. Kritikakou, and O. Sentieys, “Energy-quality-time optimized task mapping on dvfs-enabled multicores,”IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, vol. 37, no. 11, pp. 2428–2439, 2018

    2018

  69. [69]

    Joint optimization of trajectory and user association via reinforcement learning for uav-aided data collection in wireless networks,

    G. Chen, X. B. Zhai, and C. Li, “Joint optimization of trajectory and user association via reinforcement learning for uav-aided data collection in wireless networks,”IEEE Transactions on Wireless Com- munications, vol. 22, no. 5, pp. 3128–3143, 2022

    2022

  70. [70]

    Large vision model-enhanced digital twin with deep reinforcement learning for user association and load balancing in dynamic wireless networks,

    Z. Tao, W. Xu, and X. You, “Large vision model-enhanced digital twin with deep reinforcement learning for user association and load balancing in dynamic wireless networks,” 2025. [Online]. Available: https://arxiv.org/abs/2410.07611

    Pith/arXiv arXiv 2025

  71. [71]

    Wireless large ai model: Shaping the ai-native future of 6g and beyond,

    F. Zhu, X. Wang, X. Li,et al., “Wireless large ai model: Shaping the ai-native future of 6g and beyond,”arXiv preprint arXiv:2504.14653, 2025

    Pith/arXiv arXiv 2025

  72. [72]

    Next-generation wi-fi net- works with generative ai: Design and insights,

    J. Wang, X. Fang, D. Niyato,et al., “Next-generation wi-fi net- works with generative ai: Design and insights,”arXiv preprint arXiv:2408.04835, 2024

    Pith/arXiv arXiv 2024

  73. [73]

    Advancing transformer architecture in long-context large language models: A comprehensive survey,

    Y . Huang, J. Xu, J. Lai,et al., “Advancing transformer architecture in long-context large language models: A comprehensive survey,”arXiv preprint arXiv:2311.12351, 2023

    Pith/arXiv arXiv 2023

  74. [74]

    A practical survey on zero-shot prompt design for in-context learning,

    Y . Li, “A practical survey on zero-shot prompt design for in-context learning,”arXiv preprint arXiv:2309.13205, 2023

    Pith/arXiv arXiv 2023

  75. [75]

    Language models are few-shot learners,

    B. Mann, N. Ryder, M. Subbiah,et al., “Language models are few-shot learners,”arXiv preprint arXiv:2005.14165, vol. 1, p. 3, 2020

    Pith/arXiv arXiv 2005

  76. [76]

    Lora: Low-rank adaptation of large language models

    E. J. Hu, Y . Shen, P. Wallis,et al., “Lora: Low-rank adaptation of large language models.”ICLR, vol. 1, no. 2, p. 3, 2022

    2022

  77. [77]

    Empowering large language models in wireless communication: A novel dataset and fine-tuning framework,

    Y . Lin, R. Zhang, W. Huang,et al., “Empowering large language models in wireless communication: A novel dataset and fine-tuning framework,”IEEE Transactions on Communications, pp. 1–1, 2025

    2025

  78. [78]

    Large language models (llms) assisted wireless network deployment in urban settings,

    N. Sevim, M. Ibrahim, and S. Ekin, “Large language models (llms) assisted wireless network deployment in urban settings,” 2024. [Online]. Available: https://arxiv.org/abs/2405.13356

    Pith/arXiv arXiv 2024

  79. [79]

    Chain-of-thought prompt- ing elicits reasoning in large language models,

    J. Wei, X. Wang, D. Schuurmans,et al., “Chain-of-thought prompt- ing elicits reasoning in large language models,”Advances in neural information processing systems, vol. 35, pp. 24 824–24 837, 2022

    2022

  80. [80]

    Retrieval-augmented generation for large language models: A survey,

    Y . Gao, Y . Xiong, X. Gao,et al., “Retrieval-augmented generation for large language models: A survey,”arXiv preprint arXiv:2312.10997, vol. 2, no. 1, 2023

    Pith/arXiv arXiv 2023

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.