Pith. sign in

REVIEW 4 major objections 5 minor 3 cited by

Retrieval-Augmented Generation for Mobile Edge Computing via Large Language Model

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A retrieval-augmented LLM can cut mobile edge offloading latency by up to 86 percent, beating deep RL baselines.

desk verdict The RAG-for-MEC hook is timely, but the reported latency gains look like artifacts of unconstrained LLM outputs, and the 'retrieval' is just a per-user parameter lookup. read the letter →

arxiv 2412.20820 v1 pith:7SQ7SASU submitted 2024-12-30 eess.SP cs.ET

Runtao Ren , Yinyu Wu , Xuhui Zhang , Jinke Ren , Yanyan Shen , Shuqiang Wang , Kim-Fung Tsang This is my paper · ORCID
classification eess.SPcs.ET
keywords retrieval-augmentedgenerationmobileedgecomputingtaskoffloadingresourceallocationlargelanguagemodelslatencyminimizationdeepreinforcementlearningBi-encoderretrieval
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to show that retrieval-augmented generation (RAG) can replace deep reinforcement learning as the decision engine for computation offloading in mobile edge computing. It formulates a latency-minimization problem over the offloading ratio, transmit power, and server resource allocation, then proposes a three-module pipeline: a vector database stores each user's computing capability, a retrieval module matches the current query to the most relevant stored configurations, and an LLM generates the offloading decisions from a prompt that combines retrieved context with current system state. The reported experiments claim consistent latency reductions over DQN, DDPG, and PPO baselines, with improvements of 57%, 86%, 30%, and 42% under varying user computing capability, server capability, transmit power, and data volume. If true, the value is practical: an LLM-driven controller could adapt to new conditions by retrieval and re-prompting instead of retraining.

What carries the argument

The central object is the RAG pipeline of three modules. The storage module encodes each user's computing cycles per second $f_k$ into a vector $v_k = \text{Encode}(f_k)$; the retrieval module computes cosine similarity $\text{Sim}(q,v_k) = q \cdot v_k / (\|q\|\|v_k\|)$ between a query built from user identities and data volumes and the stored vectors; the generation module concatenates the retrieved computing capability $C$, the edge server's capability $F$, the data volumes $D$, bandwidth $B$, and noise power $\sigma^2$ into a prompt $P = (K\|D\|C\|F\|B\|\sigma^2)$ from which the LLM outputs decisions $L_k = \text{LLM}(P)$. This pipeline is what lets the system convert a stored-knowledge retrieval problem into a text-generation problem, and it carries the paper's entire argument.

What would settle it

Run the proposed RAG pipeline on a small instance (e.g., two users, one server, one time slot), verify whether the returned $\alpha_k$, $p_k$, and $\beta_k$ satisfy constraint (12d) and the energy constraint (12e), and compare the achieved average latency to an exhaustive-search solution of P1. If feasibility fails or the gap to the optimum exceeds the reported margins, the central claim collapses.

Watch

Extended reading notes

Core claim

The paper claims that the latency-minimization problem P1 for a multi-user MEC system can be solved by a retrieval-augmented generation pipeline: store each user's local computing capability as a vector, retrieve the best-matching configuration for the current query, concatenate it with the current data volumes, server capability, bandwidth and noise into a prompt, and let an LLM output the offloading ratio, transmit power, and server allocation ratio. In experiments, this pipeline reports 57%, 86%, 30%, and 42% lower average latency than DQN, DDPG, and PPO baselines under varying user computing capability, server capability, transmit power, and data volume, respectively.

Load-bearing premise

The load-bearing premise is that the LLM's raw output already satisfies the constraints of problem P1 (the sum of server shares below 1 and each user's energy budget, among others) and is close to optimal, even though the pipeline performs no feasibility check, projection, or optimality verification.

Editorial extensions

If this is right

  • A retrieval-augmented LLM can produce offloading decisions without training, beating deep RL baselines in average latency across varying user computing capability, server capability, transmit power, and data volume.
  • The method's reported latency improvements are 57%, 86%, 30%, and 42% on the four test datasets, suggesting the benefit holds across different sources of system heterogeneity.
  • Retrieval hit rates between 0.855 and 0.948 and MRR values between 0.846 and 0.948 indicate that stored computing-capability configurations are matched to users with high accuracy, supporting the claim that retrieved context is relevant.
  • Because adaptation happens through retrieval and prompting rather than gradient updates, the approach can respond to changing system parameters without retraining.
  • Since RAG provides a traceable path from retrieved information to the generated decision, the offloading choices are more interpretable than those of deep RL policies.

Reading between the lines

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

  • The paper's approach effectively replaces constrained optimization with a single shot of LLM prompting; adding a constraint-projection or feasibility-check layer after generation would likely improve reliability when the method is scaled to larger systems.
  • The retrieval module only indexes user computing capability, so the reported gains may be dominated by the retrieval of that one parameter; retrieving other time-varying context such as channel gains, server load history, or energy budgets could matter more in truly dynamic scenarios.
  • Because the LLM is used to output continuous-valued decisions, the method depends on the model's numerical compliance; a cheap testable improvement would be to reject and re-prompt when outputs violate constraints.
  • The comparison is only against deep RL baselines; comparing against a classical optimization heuristic or a convex relaxation would isolate whether the gains come from the RAG/LLM approach or from the problem structure itself.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a retrieval-augmented generation (RAG) framework for mobile edge computing (MEC), where a large language model (LLM) generates task offloading decisions—offloading ratio, transmit power, and server resource allocation—to minimize average latency. A vector knowledge base stores each user's computing capability, a Bi-encoder retrieves the user's own configuration, and GPT-4o/Qwen2.5-turbo generate decisions from the retrieved context. The authors evaluate the method on four data sets and report latency improvements of 57%, 86%, 30%, and 42% over DQN, DDPG, and PPO baselines. The central claim is that retrieval-augmented LLM decisions outperform deep reinforcement learning in average latency.

Significance. If established, the result would be a notable demonstration of LLM-based resource allocation in MEC. The paper has strengths: it does not fit free parameters to the test set, it evaluates several LLM and embedding configurations, and it formulates a clear constrained optimization problem with explicit constraints. These strengths, however, are outweighed by the absence of any feasibility guarantee or verification for the LLM outputs, a retrieval mechanism that only returns the user's own stored parameter, and a lack of statistical or optimality baselines. The reported gains are therefore not supported by the presented evidence. The paper is best viewed as a preliminary position that could become a serious contribution only after substantial reworking of both the method and the evaluation.

major comments (4)
  1. [Section IV.B, Eq. (15); Section III.D, Eqs. (12a)–(12e)] The generation module directly outputs decisions, but the paper does not describe any projection onto the feasible set of problem P1. In particular, constraint (12d), Σ_k β_k(t) ≤ 1, and the energy budget (12e) are not enforced at inference time. If the LLM returns β_k(t)=1 for all k, Eq. (10) gives each user the full server capacity F, artificially reducing edge latency in Figs. 6–9. The authors must either verify that the actual stored LLM outputs satisfy (12a)–(12e) and report the verification, or add a projection/penalty step and recompute all latency comparisons.
  2. [Section IV.B, Eqs. (13)–(14); Table II] The retrieval step is self-referential: it encodes each user's own f_k into the vector database and then retrieves that same configuration using the user's identity K. Consequently, Table II's hit rate and MRR measure the trivial task of retrieving exactly the item just stored, not the utility of retrieved context for decision quality. No experiment compares the full RAG pipeline against the same LLM without retrieval or with retrieval of a different user's configuration, so the paper does not establish that retrieval contributes to the reported latency gains. This is a central claim of the title and abstract, and the current experimental design cannot support it.
  3. [Section V.D, Figs. 6–9] No variance, confidence intervals, or multiple-seed results are reported, and no classical optimization baseline (e.g., exhaustive search on a coarse grid, convex relaxation, or greedy allocation) is included. Because LLM generation is stochastic, the reported 57%, 86%, 30%, and 42% improvements cannot be distinguished from sampling noise. Without an optimality reference, the claim that the method 'optimizes' latency is unquantified and unverifiable.
  4. [Section IV.B, Eq. (15)] The notation L_k = LLM(P) is ambiguous: the text says the module 'output optimization decisions,' but the equation writes L_k, which was already defined in Eq. (11) as the latency objective. The manuscript must specify exactly which quantities the LLM returns (α_k, β_k, p_k) and how they are parsed from the generated text before the numerical evaluation can be reproduced or trusted.
minor comments (5)
  1. [Section II.D heading] The heading 'Genrative Learning for Computation Offloading' contains a typo and should read 'Generative Learning for Computation Offloading.'
  2. [Section III.A and Eq. (1)] The text says 'Let f_m denote the computing cycles per second of the kth MU,' but Eq. (1) uses f_k. The notation should be made consistent throughout the paper.
  3. [Section IV.B, Eq. (13); Table I] Eq. (13) stores v_k = Encode(f_k), while Table I expresses user computing capability as ϕ/f_k in seconds per bit. The relationship between f_k and ϕ/f_k should be clarified, and the units in Table I should be stated explicitly.
  4. [Section V.C] The DDPG action space is described as 'constrained to [0, 1],' but no mechanism is described that enforces the coupled constraint Σ_k β_k(t) ≤ 1. To ensure a fair comparison, the baseline implementations should be described with the same feasibility treatment as the proposed method.
  5. [Section V.D, Figs. 6–9] The figure captions do not state the units of the y-axis (latency) or the averaging procedure used to produce each curve, which reduces reproducibility.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor circular framing: the RAG retrieval step returns the user's own stored f_k by identity, so the 'retrieval augmentation' is a self-lookup; the latency comparisons themselves are empirical and not fitted.

  1. self definitional [Section IV.B, Storage and Retrieval Modules, Eqs. (13)-(14)]
    "The retrieval module retrieves relevant computing capability information based on the identity information of the MUs. Initially, the system generates a query q based on the identities K = (k1, k2, ..., kn) and data volume D = (D1, D2, ..., Dn) of all MUs. Then the Bi-encoder model is utilized to retrieve the information of local computational capability C related to the sending MUs: C = Sim(q, vk) = q · vk / (||q||||vk||)."

    The stored vectors are defined as v_k = Encode(f_k), and the query q is built from the identities K of the same users (plus their data volumes D). Hence the retrieved C is, by construction, the local computing capability of the querying user already indexed by K; it is a self-lookup rather than an externally retrieved piece of knowledge. The prompt P = (K||D||C||F||B||σ2) then feeds the LLM the same f_k implicit in K, so the 'retrieval augmentation' that motivates the method reduces to identity-based retrieval of the input itself. The HR/MRR results in Table II consequently measure how well the embedding recovers the stored configuration that was put into the database from the same users, not the value of an external knowledge source.

full rationale

The central latency claims are not statistically forced: no parameter is fitted to the test data, and the reported 57%, 86%, 30%, and 42% improvements are computed from the physical latency expression in Eq. (11) after the LLM produces decisions. No load-bearing self-citation chain or imported uniqueness theorem appears in the derivation. The only concrete circularity is in the retrieval module: Eq. (13) stores Encode(f_k) for each user, Eq. (14) forms the query from those users' identities K (and D), and the 'retrieved' C is therefore the user's own f_k by design. The paper then presents this self-lookup as context-aware retrieval and evaluates it with MRR/HR, which is a self-definitional framing rather than a genuine augmentation from an independent source. The separate concern that Eq. (15)'s raw LLM outputs may violate constraints (12d) and (12e) with no projection or feasibility check is a validity and fairness issue about the experiments, not a circularity of the derivation, so it does not raise the score beyond 2. Overall, the paper's main empirical comparison is self-contained against external DRL baselines; only the RAG retrieval contribution is circular in its construction.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The optimization model is standard and uses standard physical/dataset parameters. The main unspoken premise is that an LLM can produce feasible, near-optimal decisions for a non-convex optimization problem from a text prompt, which is an ad hoc assumption.

assumptions (3)
  • ad hoc to paper LLM-generated alpha, beta, p lie in the feasible set of P1 and are near-optimal.
    Section IV.B Generation Module: the LLM outputs decisions directly with no constraint projection or verification, yet Eq. (11) and the latency results treat them as valid solutions.
  • domain assumption Rician fading channel model with kappa=50 and free-space path loss at g0=1e-5 governs the simulation.
    Section V.A Experimental Setup; the channel model affects all latency computations.
  • domain assumption Tasks are divisible and local/edge execution latencies combine as Eq. (11).
    Section III uses the max of local and offloaded+edge latency; this is a modeling choice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Retrieval-Augmented Generation for Mobile Edge Computing via Large Language Model." pith.science (2026). https://pith.science/paper/7SQ7SASU

@misc{pith2026241220820,
  author       = {Pith},
  title        = {Pith review of: Retrieval-Augmented Generation for Mobile Edge Computing via Large Language Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7SQ7SASU}},
  note         = {Machine review of arXiv:2412.20820}
}
read the original abstract

The rapid evolution of mobile edge computing (MEC) has introduced significant challenges in optimizing resource allocation in highly dynamic wireless communication systems, in which task offloading decisions should be made in real-time. However, existing resource allocation strategies cannot well adapt to the dynamic and heterogeneous characteristics of MEC systems, since they are short of scalability, context-awareness, and interpretability. To address these issues, this paper proposes a novel retrieval-augmented generation (RAG) method to improve the performance of MEC systems. Specifically, a latency minimization problem is first proposed to jointly optimize the data offloading ratio, transmit power allocation, and computing resource allocation. Then, an LLM-enabled information-retrieval mechanism is proposed to solve the problem efficiently. Extensive experiments across multi-user, multi-task, and highly dynamic offloading scenarios show that the proposed method consistently reduces latency compared to several DL-based approaches, achieving 57% improvement under varying user computing ability, 86% with different servers, 30% under distinct transmit powers, and 42% for varying data volumes. These results show the effectiveness of LLM-driven solutions to solve the resource allocation problems in MEC systems.

Figures

Figures reproduced from arXiv: 2412.20820 by the authors.

Figure 1
Figure 1. Multiuser MEC system model. A. Local Computing Model Let αk(t) ∈ [0, 1] denote the task offloading ratio, where the part of αk(t)Dk(t) is offloaded to the edge server for remote processing, and the other part (1 − αk(t))Dk(t) is processed locally. Let fm denote the computing cycles per second of the kth MU, ϕ denote the computing cycles required to process [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. The framework of Bi-encoder. The second stage of RAG involves leveraging the generative capabilities of LLMs. In this architecture, LLMs act as the reasoning and decision-making engine, synthesizing retrieved information with the task-specific query to produce optimized solutions. LLMs in RAG are not confined to predefined rules or static algorithms. Instead, they dynamically adapt to the input context by integratin… view at source ↗
Figure 2
Figure 2. The framework of RAG. The retrieval component of RAG relies on dense retrieval techniques, which use vector representations to capture the semantic relationship between user queries and pre-stored con￾figurations in a knowledge base. Unlike traditional keyword￾based methods, dense retrieval encodes both queries and doc￾uments into high-dimensional vectors, enabling more accurate and context-aware matching. The Bi-en… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The framework of RAG for the MEC system. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The prompt of LLM for decision task offloading strategies through a large amount of training time and data. The state space includes user task data volume, channel gain, user computing capability, and server comput￾ing capability. The action space comprises user offloa…
Figure 7
Figure 7. Figure 7: Performance comparison of different algorithms under [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 6
Figure 6. Figure 6: Performance comparison of different algorithms under [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 9
Figure 9. Figure 9: Performance comparison of different algorithms in [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

  1. ORAN-GUIDE: RAG-Driven Prompt Learning for LLM-Augmented Reinforcement Learning in O-RAN Network Slicing

    cs.LG 2025-05 reject novelty 4.0 of 10

    ORAN-GUIDE couples a domain-specific LLM prompt generator with a frozen GPT-2 encoder and learnable prompt tokens to improve multi-agent SAC sample efficiency in O-RAN slicing.

  2. HybridRAG-based LLM Agents for Low-Carbon Optimization in Low-Altitude Economy Networks

    cs.NI 2025-06 reject novelty 3.0 of 10

    HybridRAG merges keyword, vector, and graph retrieval to let an LLM formulate carbon-emission optimization problems for multi-UAV MEC networks, and R2DSAC solves them with a diffusion-regularized SAC plus neuron pruni...

  3. From Large AI Models to Agentic AI: A Tutorial on Future Intelligent Communications

    cs.AI 2025-05 conditional novelty 2.0 of 10

    This paper is a broad tutorial on applying LAMs and agentic AI to 6G, largely restating existing research rather than introducing new results.

Reference graph

Works this paper leans on

46 extracted references · 44 canonical work pages · cited by 3 Pith papers

  1. [1]

    Collaborative cloud and edge computing for latency minimization,

    J. Ren, G. Yu, Y . He, and G. Y . Li, “Collaborative cloud and edge computing for latency minimization,” IEEE Trans. Veh. Technol., vol. 68, no. 5, pp. 5031–5044, May 2019

    work page 2019

  2. [2]

    Mobile edge computing: A survey on architec- ture and computation offloading,

    P. Mach and Z. Becvar, “Mobile edge computing: A survey on architec- ture and computation offloading,” IEEE Commun. Surv. Tuts. , vol. 19, no. 3, pp. 1628–1656, 3rd Quart. 2017

    work page 2017

  3. [3]

    An edge- computing based architecture for mobile augmented reality,

    J. Ren, Y . He, G. Huang, G. Yu, Y . Cai, and Z. Zhang, “An edge- computing based architecture for mobile augmented reality,” IEEE Network, vol. 33, no. 4, pp. 162–169, Jul./Aug. 2019

    work page 2019

  4. [4]

    Computation offloading and resource allocation in wireless cellular networks with mobile edge computing,

    C. Wang, C. Liang, F. R. Yu, Q. Chen, and L. Tang, “Computation offloading and resource allocation in wireless cellular networks with mobile edge computing,” IEEE Trans. Wireless Commun., vol. 16, no. 8, pp. 4924–4938, Aug. 2017

    work page 2017

  5. [5]

    Dynamic computation offloading for mobile-edge computing with energy harvesting devices,

    Y . Mao, J. Zhang, and K. B. Letaief, “Dynamic computation offloading for mobile-edge computing with energy harvesting devices,” IEEE J. Sel. Areas Commun. , vol. 34, no. 12, pp. 3590–3605, Dec. 2016

    work page 2016

  6. [6]

    Energy-efficient resource allocation for mobile-edge computation offloading,

    C. You, K. Huang, H. Chae, and B.-H. Kim, “Energy-efficient resource allocation for mobile-edge computation offloading,” IEEE Trans. Wire- less Commun., vol. 16, no. 3, pp. 1397–1411, Mar. 2017

    2017

  7. [7]

    Resource allocation in cognitive radio-enabled UA V communication,

    S. K. Nobar, M. H. Ahmed, Y . Morgan, and S. A. Mahmoud, “Resource allocation in cognitive radio-enabled UA V communication,”IEEE Trans. on Cogn. Commun. Netw. , vol. 8, no. 1, pp. 296–310, Mar. 2022

    work page 2022

  8. [8]

    Learning to hybrid offload in space-air-ground integrated mobile edge computing for IoT networks,

    X. Zhang, W. Liu, H. Xing, Z. Jin, W. Zang, S. Wang, Y . Shen, and L. Xue, “Learning to hybrid offload in space-air-ground integrated mobile edge computing for IoT networks,” in Proc. IEEE Int. Conf. CYBER Technol. Autom., Control, Intell. Syst. (CYBER) , Qinhuangdao, China, Jul. 2023, pp. 836–841

    work page 2023

Show all 46 references
  1. [9]

    Computation offloading for mo- bile edge computing: A deep learning approach,

    S. Yu, X. Wang, and R. Langar, “Computation offloading for mo- bile edge computing: A deep learning approach,” in Proc. IEEE Int. Symp. Person. Indoor Mobile Radio Commun. (PIMRC) , Montreal, QC, Canada, Oct. 2017, pp. 1–6

    2017

  2. [10]

    Applications of deep reinforcement learning in communications and networking: A survey,

    N. C. Luong, D. T. Hoang, S. Gong, D. Niyato, P. Wang, Y .-C. Liang, and D. I. Kim, “Applications of deep reinforcement learning in communications and networking: A survey,” IEEE Commun. Surv. Tuts., vol. 21, no. 4, pp. 3133–3174, 4th Quart. 2019

    2019

  3. [11]

    DRL based data of- floading for intelligent reflecting surface aided mobile edge computing,

    X. Zhang, Y . Shen, B. Yang, W. Zang, and S. Wang, “DRL based data of- floading for intelligent reflecting surface aided mobile edge computing,” in Proc. IEEE Wireless Commun. Networking Conf. (WCNC) , Nanjing China, Mar. 2021, pp. 1–7

    2021

  4. [12]

    Deep reinforcement learning-based dynamic resource management for mobile edge computing in industrial internet of things,

    Y . Chen, Z. Liu, Y . Zhang, Y . Wu, X. Chen, and L. Zhao, “Deep reinforcement learning-based dynamic resource management for mobile edge computing in industrial internet of things,” IEEE Trans. Ind. Informat., vol. 17, no. 7, pp. 4925–4934, Jul. 2021

    2021

  5. [13]

    Deep reinforcement learning for multi-hop offloading in UA V-assisted edge computing,

    N. T. Hoa, D. Van Dai, L. H. Lan, N. C. Luong, D. Van Le, and D. Niyato, “Deep reinforcement learning for multi-hop offloading in UA V-assisted edge computing,” IEEE Trans. Veh. Technol. , vol. 72, no. 12, pp. 16 917–16 922, Dec. 2023

    2023

  6. [14]

    MADDPG-based joint service placement and task offloading in MEC empowered air–ground integrated networks,

    J. Du, Z. Kong, A. Sun, J. Kang, D. Niyato, X. Chu, and F. R. Yu, “MADDPG-based joint service placement and task offloading in MEC empowered air–ground integrated networks,” IEEE Internet Things J. , vol. 11, no. 6, pp. 10 600–10 615, Mar. 2024

    2024

  7. [15]

    Deep learning empowered task offloading for mobile edge computing in urban informatics,

    K. Zhang, Y . Zhu, S. Leng, Y . He, S. Maharjan, and Y . Zhang, “Deep learning empowered task offloading for mobile edge computing in urban informatics,” IEEE Internet Things J., vol. 6, no. 5, pp. 7635–7647, Oct. 2019

    2019

  8. [16]

    A survey on computation offloading in edge systems: From the perspective of deep reinforcement learning approaches,

    P. Peng, W. Lin, W. Wu, H. Zhang, S. Peng, Q. Wu, and K. Li, “A survey on computation offloading in edge systems: From the perspective of deep reinforcement learning approaches,” Comput. Sci. Rev., vol. 53, p. 100656, Aug. 2024

    2024

  9. [17]

    Large language model enhanced multi-agent systems for 6G communications,

    F. Jiang, Y . Peng, L. Dong, K. Wang, K. Yang, C. Pan, D. Niyato, and O. A. Dobre, “Large language model enhanced multi-agent systems for 6G communications,” IEEE Wireless Commun., Dec. 2024

    2024

  10. [18]

    Large language models (LLMs) inference offloading and resource allocation in cloud-edge computing: An active inference approach,

    Y . He, J. Fang, F. R. Yu, and V . C. Leung, “Large language models (LLMs) inference offloading and resource allocation in cloud-edge computing: An active inference approach,”IEEE Trans. Mobile Comput., vol. 23, no. 12, pp. 11 253–11 264, Dec. 2024

    2024

  11. [19]

    Large language model-based wireless network design,

    K. Qiu, S. Bakirtzis, I. Wassell, H. Song, J. Zhang, and K. Wang, “Large language model-based wireless network design,” IEEE Wireless Commun. Lett., vol. 13, no. 12, pp. 3340–3344, Dec. 2024

    2024

  12. [20]

    Retrieval-augmented generation for knowledge-intensive NLP tasks,

    P. Lewis, E. Perez, A. Piktus, F. Petroni, V . Karpukhin, N. Goyal, H. Kuttler, M. Lewis, W.-T. Yih, T. Rocktaschel, S. Riedel, and D. Kiela, “Retrieval-augmented generation for knowledge-intensive NLP tasks,” in Proc. Adv. Neural Inf. Proces. Syst. (NIPS), vol. 33, Virtual, O...

    2020

  13. [21]

    Game theory for compu- tation offloading and resource allocation in edge computing: A survey,

    M. Zamzam, T. El-Shabrawy, and M. Ashour, “Game theory for compu- tation offloading and resource allocation in edge computing: A survey,” in Proc. Novel Intell. Lead. Emerg. Sci. Conf. (NILES) , Giza, Egypt, Oct. 2020, pp. 47–53

    2020

  14. [22]

    Joint optimization of task offloading and resource allocation in mobile edge computing system,

    J. Huang, Y . Du, Y . Zheng, and X. Zhang, “Joint optimization of task offloading and resource allocation in mobile edge computing system,” in Proc. Int. Conf. Consum. Electron. Comput. Eng. (ICCECE), Guangzhou, China, Jan. 2023, pp. 229–232

    2023

  15. [23]

    Adaptive computation offloading and resource allocation strategy in a mobile edge computing environment,

    T. Zhao, D. Xiaomei, Y . Feng, B. Sunitha, X. Xueli, and P. Yi, “Adaptive computation offloading and resource allocation strategy in a mobile edge computing environment,” Inf. Sci., vol. 537, pp. 116–131, Oct. 2020

    2020

  16. [24]

    Learning for com- putation offloading in mobile edge computing,

    T. Q. Dinh, Q. D. La, T. Q. S. Quek, and H. Shin, “Learning for com- putation offloading in mobile edge computing,” IEEE Trans. Commun., vol. 66, no. 12, pp. 6353–6367, Dec. 2018

    2018

  17. [25]

    Resource trading in blockchain-based industrial internet of things,

    H. Yao, T. Mai, J. Wang, Z. Ji, C. Jiang, and Y . Qian, “Resource trading in blockchain-based industrial internet of things,” IEEE Trans. Ind. Informat., vol. 15, no. 6, pp. 3602–3609, Jun. 2019

    2019

  18. [26]

    Task offloading and resource allocation for mobile edge computing by deep reinforcement learning based on SARSA,

    T. Alfakih, M. M. Hassan, A. Gumaei, C. Savaglio, and G. Fortino, “Task offloading and resource allocation for mobile edge computing by deep reinforcement learning based on SARSA,” IEEE Access , vol. 8, pp. 54 074–54 084, 2020

    2020

  19. [27]

    Collaborative learning of communication routes in edge-enabled multi-access vehicular environment,

    C. Wu, Z. Liu, F. Liu, T. Yoshinaga, Y . Ji, and J. Li, “Collaborative learning of communication routes in edge-enabled multi-access vehicular environment,” IEEE Trans. on Cogn. Commun. Netw. , vol. 6, no. 4, pp. 1155–1165, Dec. 2020

    2020

  20. [28]

    Deep reinforcement learning for online computation offloading in wireless powered mobile-edge computing networks,

    L. Huang, S. Bi, and Y .-J. A. Zhang, “Deep reinforcement learning for online computation offloading in wireless powered mobile-edge computing networks,” IEEE Trans. Mobile Comput. , vol. 19, no. 11, pp. 2581–2593, Nov. 2020

    2020

  21. [29]

    Deep reinforcement learning for the computation offloading in MIMO-based edge computing,

    A. Sadiki, J. Bentahar, R. Dssouli, A. En-Nouaary, and H. Otrok, “Deep reinforcement learning for the computation offloading in MIMO-based edge computing,” Ad Hoc Netw. , vol. 141, Mar. 2023

    2023

  22. [30]

    Cybertwin-driven multi-intelligent reflecting surfaces aided vehicular edge computing leveraged by deep reinforcement learning,

    X. Zhang, H. Xing, W. Zang, Z. Jin, and Y . Shen, “Cybertwin-driven multi-intelligent reflecting surfaces aided vehicular edge computing leveraged by deep reinforcement learning,” in Proc. IEEE Veh. Technol. Conf. (VTC), London, The United Kingdom, Sep. 2022, pp. 1–7

    2022

  23. [31]

    Deep reinforcement learning for energy-efficient computation offloading in mobile-edge computing,

    H. Zhou, K. Jiang, X. Liu, X. Li, and V . C. M. Leung, “Deep reinforcement learning for energy-efficient computation offloading in mobile-edge computing,” IEEE Internet Things J. , vol. 9, no. 2, pp. 1517–1530, Jan. 2022

    2022

  24. [32]

    Latency-aware resource allocation for mobile edge generation and computing via deep reinforcement learning,

    Y . Wu, X. Zhang, J. Ren, H. Xing, Y . Shen, and S. Cui, “Latency-aware resource allocation for mobile edge generation and computing via deep reinforcement learning,” IEEE Networking Lett. , to appear, 2024

    2024

  25. [33]

    Generative AI and ChatGPT: Applications, challenges, and AI-human collaboration,

    F. Fui-Hoon Nah, R. Zheng, J. Cai, K. Siau, and L. Chen, “Generative AI and ChatGPT: Applications, challenges, and AI-human collaboration,” J. Inf. Technol. Case Appl. Research , vol. 25, no. 3, pp. 277–304, Jul. 2023

    2023

  26. [34]

    Energy-efficient dynamic offloading and resource scheduling in mobile cloud computing,

    S. Guo, B. Xiao, Y . Yang, and Y . Yang, “Energy-efficient dynamic offloading and resource scheduling in mobile cloud computing,” in Proc. IEEE Int. Conf. Computer Commun. (INFOCOM) , San Francisco, CA, United states, Apr. 2016, pp. 1–9

    2016

  27. [35]

    Backscatter- assisted computation offloading for energy harvesting IoT devices via policy-based deep reinforcement learning,

    Y . Xie, Z. Xu, Y . Zhong, J. Xu, S. Gong, and Y . Wang, “Backscatter- assisted computation offloading for energy harvesting IoT devices via policy-based deep reinforcement learning,” in Proc. IEEE/CIC Int. Conf. Commun. Workshops China (ICCC Workshops) , Changchun, China, Aug...

    2019

  28. [36]

    Joint optimization of trajectory and communication resource allocation for unmanned surface vehicle enabled maritime wireless networks,

    C. Zeng, J.-B. Wang, C. Ding, H. Zhang, M. Lin, and J. Cheng, “Joint optimization of trajectory and communication resource allocation for unmanned surface vehicle enabled maritime wireless networks,” IEEE Trans. Commun., vol. 69, no. 12, pp. 8100–8115, Dec. 2021

    2021

  29. [37]

    Intelli- gent reflecting surface aided mobile edge computing with rate-splitting multiple access,

    Y . Wu, X. Zhang, H. Xing, W. Zang, S. Wang, and Y . Shen, “Intelli- gent reflecting surface aided mobile edge computing with rate-splitting multiple access,” in Proc. IEEE Veh. Technol. Conf. (VTC) , Singapore, Jun. 2024, pp. 1–6

    2024

  30. [38]

    Robust 3D-trajectory and time switching optimization for dual- UA V-enabled secure communications,

    W. Wang, X. Li, R. Wang, K. Cumanan, W. Feng, Z. Ding, and O. A. Dobre, “Robust 3D-trajectory and time switching optimization for dual- UA V-enabled secure communications,” IEEE J. Sel. Areas Commun. , vol. 39, no. 11, pp. 3334–3347, Nov. 2021

    2021

  31. [39]

    Joint trajectory design and resource allocation in UA V-enabled heterogeneous MEC systems,

    W. Liu, H. Wang, X. Zhang, H. Xing, J. Ren, Y . Shen, and S. Cui, “Joint trajectory design and resource allocation in UA V-enabled heterogeneous MEC systems,” IEEE Internet Things J. , vol. 11, no. 19, pp. 30 817– 30 832, Oct. 2024

    2024

  32. [40]

    Latency minimization for intelligent reflecting surface aided mobile edge computing,

    T. Bai, C. Pan, Y . Deng, M. Elkashlan, A. Nallanathan, and L. Hanzo, “Latency minimization for intelligent reflecting surface aided mobile edge computing,” IEEE J. Sel. Areas Commun. , vol. 38, no. 11, pp. 2666–2682, Nov. 2020. 11

    2020

  33. [41]

    wChain: A fast fault-tolerant blockchain protocol for multihop wireless networks,

    M. Xu, C. Liu, Y . Zou, F. Zhao, J. Yu, and X. Cheng, “wChain: A fast fault-tolerant blockchain protocol for multihop wireless networks,” IEEE Trans. Wireless Commun. , vol. 20, no. 10, pp. 6915–6926, Oct. 2021

    2021

  34. [42]

    Interference aware cooperative routing for edge computing-enabled 5G networks,

    A. Waqas, H. Mahmood, and N. Saeed, “Interference aware cooperative routing for edge computing-enabled 5G networks,” IEEE Sensors J. , vol. 22, no. 4, pp. 3777–3784, Feb. 2022

    2022

  35. [43]

    Retrieval augmented generation using engi- neering design knowledge,

    L. Siddharth and J. Luo, “Retrieval augmented generation using engi- neering design knowledge,” Knowl. Based Syst. , vol. 303, p. 112410, Nov. 2024

    2024

  36. [44]

    C- pack: Packed resources for general Chinese embeddings,

    S. Xiao, Z. Liu, P. Zhang, N. Muennighoff, D. Lian, and J.-Y . Nie, “C- pack: Packed resources for general Chinese embeddings,” in Proc. Int. ACM SIGIR Conf. Res. Dev. Inf. Retr. (SIGIR), Washington, DC, United states, Jul. 2024, pp. 641–649

    2024

  37. [45]

    Distilling knowledge for fast retrieval-based chat-bots,

    A. Vakili Tahami, K. Ghajar, and A. Shakery, “Distilling knowledge for fast retrieval-based chat-bots,” in Proc. Int. ACM SIGIR Conf. Res. Dev. Inf. Retr. (SIGIR), China, Jul. 2020, pp. 2081–2084

    2020

  38. [46]

    Evaluating retrieval quality in retrieval- augmented generation,

    A. Salemi and H. Zamani, “Evaluating retrieval quality in retrieval- augmented generation,” in Proc. Int. ACM SIGIR Conf. Res. Dev. Inf. Retr. (SIGIR), Washington, DC, United states, Jul. 2024, pp. 2395–2400

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.