Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

UAV-assisted Joint Mobile Edge Computing and Data Collection via Matching-enabled Deep Reinforcement Learning

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

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

Pith's one-line read A joint UAV-assisted edge-computing and data-collection system can be optimized in real time by removing user association from the reinforcement-learning action space and resolving it with a two-phase matching game, yielding lower latency…

desk verdict Useful combination of SAC and swap matching for a genuinely new joint MEC-DC scenario, but the empirical claim needs better reproducibility. read the letter →

arxiv 2502.07388 v1 pith:4BJTPIPD submitted 2025-02-11 cs.NE

Boxiong Wang , Hui Kang , Jiahui Li , Geng Sun , Zemin Sun , Jiacheng Wang , Dusit Niyato This is my paper · ORCID
classification cs.NE
keywords UAV-assistedmobileedgecomputingUAVdatacollectionjointMEC-DCoptimizationdeepreinforcementlearningsoftactor-criticmatchinggameuserassociationco-channelinterference
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

The paper tries to show that a UAV network can serve two conflicting kinds of ground users at once—those needing low-latency edge computing and those needing bulk data pickup—without treating the two as separate problems. It formulates a joint optimization that minimizes total MEC latency and maximizes collected data volume under co-channel interference, then rewrites the problem as a Markov decision process whose user-association decision is removed from the reinforcement-learning action space. The association is instead computed by a matching game with externalities, while a soft actor-critic agent controls UAV movement and user transmit powers. Simulations indicate this hybrid approach beats several benchmark algorithms on latency, collected data volume, task completion rate, and reward stability.

What carries the argument

The central object is a two-phase matching-based association (TMA) strategy embedded in a soft actor-critic loop. User association is modeled as a one-to-many matching game with externalities, meaning each ground user's achievable rate depends on the other users' associations through co-channel interference. Phase I builds a preliminary assignment using distance-based then rate-based evaluation; Phase II performs swap matching, accepting a swap only when it strictly improves the system sum rate, and is claimed to converge to a stable matching. That stable matching replaces the discrete association action in the MDP, so the actor network outputs only continuous actions: UAV movement distance and direction, and ground-user transmit powers.

What would settle it

Construct a small instance with, say, three ground users and two UAVs under path-loss interference, enumerate all swap-blocking pairs, and check whether the while-loop can revisit a previously seen matching or terminate at an assignment that a different swap order would improve. A simpler numerical check is to run TMA from several random initial matchings on the same field and test whether the final sum rate varies across seeds; if it does, the convergence claim does not imply a unique or globally optimal association.

Watch

Extended reading notes

Core claim

The central claim is that the joint MEC-DC uplink problem, despite being mixed-integer, non-convex, and dynamically coupled through interference, can be decomposed into a continuous-control part and a combinatorial association part without losing coordination. The continuous part is learned by a soft actor-critic algorithm; the association is settled by a two-phase swap-matching procedure that first seeds a Gale-Shapley-style assignment and then accepts only utility-improving swaps. The MEC and DC subsystems are separated onto different UAVs, and the single DC UAV is treated as a special one-sided matching case. Simulation with three MEC UAVs, one DC UAV, 35 ground users, and time-varying task arrivals shows the proposed algorithm reducing MEC latency and increasing collected data volume relative to the benchmark algorithms considered.

Load-bearing premise

The whole scheme leans on the assumption that the repeated swap matching always converges to a stable assignment because there are finitely many users and drones, even though every swap changes interference for everyone else; if the swap loop can cycle or settle in a poor local optimum, the reported advantage disappears.

Editorial extensions

If this is right

  • The MEC and DC subsystems can be coordinated by a single controller rather than optimized separately, so operators do not need to solve a bi-level problem.
  • Because association is computed by a bounded swap routine instead of sampled by the reinforcement-learning policy, the discrete action-space dimension is removed, which the paper identifies as the main source of training difficulty.
  • The stability of the matching routine provides a deterministic association update at every step, so the learning agent only needs to handle continuous trajectory and power decisions.
  • The simulation results claim lower MEC latency, higher collected data volume, and better task completion and DC rates than all compared benchmarks under varying numbers of MEC users and varying per-UAV service capacity.
  • The proposed TMA strategy is reported to be practical in runtime, completing association computation in about $2\times10^{-2}$ seconds per slot, which is far shorter than the slot duration.

Reading between the lines

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

  • The authors do not explore this implication, but the action-reduction trick is portable: any mixed-integer MEC or DC problem where association couples with interference could replace a discrete reinforcement-learning head with a swap-matching routine, at the cost of re-establishing stability for that problem.
  • Because the DC UAV is treated as a one-sided matching with an empty slot, the same TMA machinery could extend to multiple DC UAVs by running swap matching over the DC side, a case the paper does not simulate.
  • A testable consequence the authors leave implicit is that the gap between the proposed algorithm and the baselines should widen as co-channel interference grows, since the matching routine's rate thresholds filter out harmful associations; varying the interference level directly would confirm this.
  • The paper does not analyze what happens when a swap changes the rates of users not involved in the swap; a natural extension is to update utilities inside each swap evaluation and check whether the monotone-improvement property still holds.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper considers a multi-UAV-assisted joint mobile edge computing (MEC) and data collection (DC) system in which a set of MEC-UAVs serves delay-sensitive tasks and a single DC-UAV collects data from freshness-insensitive users, with co-channel interference among UAVs. The authors formulate a bi-objective mixed-integer non-convex optimization problem, reformulate it as a Markov decision process with a reduced action space in which user association is handled by a two-phase matching-based association (TMA) strategy, and propose a soft actor-critic (SAC) algorithm augmented with TMA (SAC-TMA) to jointly optimize UAV movement, user transmit power, and user association. Simulation results over several scenarios are reported to show that SAC-TMA outperforms greedy and DRL-based benchmarks in terms of MEC latency reward, collected data volume, task completion rate, and DC rate.

Significance. If the empirical claims are reproducible and the convergence of the matching step is properly established, the paper provides a useful integration of matching-based combinatorial user association with continuous-action deep reinforcement learning for a scenario (joint MEC and DC with mutual interference) that, according to the authors' literature survey, has not been jointly studied. The system and channel models are standard, the action-space reduction idea is sensible, and the complexity analysis and running-time evaluation of the TMA strategy are welcome additions. The paper's central claim, however, is empirical, and the evidence as presented is not yet sufficient to support it: the reward coefficients are unspecified, the comparison plots lack error bars or significance tests, and the matching convergence argument is asserted rather than proved. These issues are fixable but currently prevent verification of the headline performance gains.

major comments (4)
  1. [§V-A, Eqs. (18)–(22), and §VI] The penalty coefficients ρ, δ_p, and ϱ in Eqs. (19)–(21) are never assigned numerical values, and Table III and Table IV do not list them. Since the total reward r(t) = r_l(t) + r_d(t) + r_p(t) is a weighted sum of a latency term, a DC-volume term, and penalties, the relative weighting of the two objectives is unspecified. An independent implementation cannot reproduce the learned policy or the reported trade-off between MEC latency and collected data. The authors should report the exact values used for all reward coefficients and, ideally, a sensitivity study over these coefficients.
  2. [§VI-B, Figs. 3, 5–7] The comparison plots show single point curves with no error bars, confidence intervals, or significance tests; only Fig. 8 displays spread over three seeds. The headline statements that SAC-TMA 'outperforms' the baselines are therefore not supported as statistical claims, particularly in cases where the margins appear small (e.g., the latency reward at larger MEC-user counts in Fig. 3). The authors should report mean ± standard deviation over multiple random seeds and network topologies, and perform a basic significance test or state the number of independent runs used to generate each point.
  3. [§V-A, Definition 3 and Algorithm 3] The convergence of the swap-matching process to a stable matching is asserted on the basis of finiteness and citations [46], [51], but the matching here has externalities: after a swap, the communication rates of all users change, so the utility U(X) may change for pairs not involved in the swap. The paper does not prove that the sequence of swap operations is monotone in a way that rules out cycles, nor does it prove convergence to a local optimum under externalities. Algorithm 3 also does not explicitly state that U(X_m^{m'}) is computed under the swapped association before deciding whether to accept the swap (Step 10 says only 'Calculate Eq. (16)'). This is load-bearing because TMA feeds a deterministic association into the SAC loop; if the swap process can cycle, the effective environment transition is not well-defined. The authors should either provide a convergence proof for the one-to-many swap matching with externalities, or weaken the stability claim to a finite-improvement termination statement with explicit utility recomputation.
  4. [§V-A, Eqs. (7), (9), and (17)] The latency reward r_l(t) in Eq. (17) uses the term T_i^f(t), which is not defined: Eq. (7) defines T_{i,m}^f(t) for transmission latency, and Eq. (9) defines T_i(t) as the sum of transmission and computation latencies. The note in Eq. (17) says T_i^f(t) is 'the latency of MEC task offloading and executing at this time step,' but the notation is inconsistent with Eqs. (7)–(9). Moreover, the optimized objective f_1 in Eq. (13) sums T_i(t), while the reward r_l uses a different quantity. This inconsistency obscures what the agent actually optimizes and makes the latency figures in Section VI hard to interpret. The authors should define T_i^f(t) precisely and align the reward with the declared objective f_1.
minor comments (5)
  1. [§VI-A] The text says 'The number of UAVs is M = 4,' but M was previously used for the number of MEC users; the total number of UAVs should use a distinct symbol (e.g., N_U + 1) to avoid confusion with the number of MEC users.
  2. [§V-A, Eq. (20)] In the sentence defining B_x^u(t) and B_y^u(t), the text writes 'Bx_u(t) and Bx_u(t)' where the second should be B_y^u(t).
  3. [§VI-A] The baseline list says the proposed algorithm is compared with four benchmarks, but then lists six methods (Distance-Greedy, SAC-TMA-greedy, TD3-TMA, PPO-TMA, MSE-PPO, Dual-timescales). The count should be corrected.
  4. [§III-D, Eq. (15b)] Constraint (15b) uses T_max_{m,f} while the task model in Section III-D defines the maximum tolerance time as t_max_{m,f}; the notation should be unified.
  5. [§III-E, Eq. (12)] The summation limits in Eq. (12) are written as 'P T_{t=0}' and 'P T_{t=0}'; the index should start at t = 1 to be consistent with the time-step definition in Section III-A.

Circularity Check

2 steps flagged · score 3.0 of 10

Partial circularity in evaluation: latency and sum-reward figures report the SAC training reward itself, and TMA's swap-based superiority over its own initializers is entailed by its utility-improving acceptance rule; independent metrics (task completion rate, DC rate, energy, runtime) keep the central claim partially grounded.

  1. self definitional [Section VI-B-1 (Comparison Results), Fig. 3 caption; Eqs. (17) and (22)]
    "The performance of latency is measured by the latency reward defined in Eq. (17). (Fig. 3 caption); Eq. (22): r(t)=rl(t)+rd(t)+rp(t)"

    The abstract claims SAC-TMA 'can effectively reduce the system latency,' and Fig. 3's latency panel measures performance by the 'total latency reward' — precisely rl(t) from Eq. (17), the latency component of the training reward r(t)=rl(t)+rd(t)+rp(t) that SAC maximizes (Eq. 22). High rl for the trained agent is therefore, by construction, a consequence of maximizing the training objective (and Figs. 4, 6, 8 report r(t) itself as 'sum reward'), so the latency and cumulative-reward evidence partly re-states the optimized quantity rather than an independent latency measure such as f1 in Eq. (13). The reduction is partial: task completion rate (Eq. 10), DC rate (Eq. 12), and average energy are independent metrics, and baselines share the same reward.

  2. self definitional [Section VI-B-4 (Effectiveness Analysis); Eq. (16); Section I contribution list]
    "the swap operation is only performed when the performance of the solution will improve. Therefore, the solution obtained by the swap algorithm will not be inferior to any of the previous solutions."

    The contribution list claims the proposed TMA strategy 'outperforms traditional matching-based algorithms and random strategies,' and the supporting Fig. 10 compares strategies using the utility U(X) of Eq. (16). But U is exactly the objective the swap rule greedily maximizes: every accepted swap strictly increases U and the matching set is finite, so the swap-based final matching has U at least as large as its initialization (random or GS-based) by construction. Hence the superiority of swap variants over non-swap strategies on U is entailed by the algorithm's own acceptance rule, not an empirical finding. The non-circular parts of the claim — feasibility in running time and the rate-swap vs. distance-swap contrast — are separately reported in Fig. 10.

full rationale

Two partial self-definitional reductions are present, but neither fully determines the central claim. First, the headline latency result is demonstrated with the latency reward of Eq. (17), which is a component of the training reward r(t) in Eq. (22) that SAC is trained to maximize; high rl for the trained agent therefore partly re-reports the optimized objective, and the 'sum reward' curves in Figs. 4, 6, and 8 are the training reward itself. The counterweight is that the paper also evaluates genuinely independent metrics — MEC task completion rate (Eq. 10), DC rate (Eq. 12), and average UAV energy — and all learning baselines are trained against the same reward, so the head-to-head comparison retains meaning. Second, the TMA effectiveness analysis measures strategies by the utility U(X) of Eq. (16), the very quantity the swap rule monotonically improves; the paper itself states that the swap solution 'will not be inferior to any of the previous solutions,' so the claimed superiority over random/GS strategies on U is a built-in property of the acceptance rule rather than an empirical discovery. The non-circular elements of the TMA claim — per-slot running time and the rate-swap vs. distance-swap contrast — are reported in Fig. 10. No load-bearing self-citation chain exists: the convergence citations [46] and [51] are external to the author list, and convergence of Algorithm 3 is in fact self-contained under Definition 3 via strict U-improvement over a finite matching set, so the externality concern raised by the reviewer is a correctness nuance, not circularity. Reproducibility gaps (reward coefficients sigma, rho, delta_p, and vartheta are never assigned numerical values; only Fig. 8 shows seed spread) are evidence-quality concerns and are excluded from the circularity score per the review rules. The central derivations (MDP reformulation, SAC losses, complexity analysis) are otherwise self-contained, so the paper is not systematically circular; the score of 3 reflects the two partial by-construction reductions in the evaluation evidence.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard UAV channel and energy models plus a set of hand-chosen reward-shaping constants; the matching convergence is asserted rather than proven.

free parameters (4)
  • Out-of-bounds penalty factor rho = not specified
    Defined in Eq. (20) as a penalty factor when UAVs cross the boundary; no numeric value is given in Table III, so the reward shaping is under-specified and hand-chosen.
  • Energy penalty coefficient delta_p = not specified
    Defined in Eq. (21); no value is provided, and it directly scales the energy violation penalty in the reward.
  • Collision penalty varrho = not specified
    Defined in Eq. (19) as a positive constant; no value is given, so its contribution to the reward cannot be reproduced.
  • DC reward decay coefficient sigma = 0.5 if storage limit reached, else 1
    Hand-chosen discount in Eq. (18) to discourage data loss at the storage limit; value is stated but its choice is not justified.
assumptions (5)
  • domain assumption Probabilistic LoS channel model with environmental constants lambda1, lambda2 and excessive losses eta_LoS, eta_NLoS taken from prior literature.
    Eq. (1)-(2) assume this channel model; the simulation parameters are inherited from [25], [36], [53].
  • domain assumption Each GU can be served by at most one UAV, and UAVs use OFDMA so intra-UAV interference is absent while inter-UAV co-channel interference is present.
    This defines the association constraints (15e)-(15g) and the interference term in Eq. (3).
  • domain assumption Task and data generation follow an intermittent model with probability P_g,f = delta_g * (t - eta_g) in Eq. (6).
    The stochastic dynamics of the MDP are built on this model; the concrete distribution is a design choice.
  • ad hoc to paper The swap matching process converges to a stable matching in the one-to-many matching with externalities because the number of players is finite.
    Stated after Algorithm 3 in Section V-A without a formal proof; externalities can cause cycles, so convergence is not guaranteed by finiteness alone.
  • ad hoc to paper The scalar reward r(t)=rl+rd+rp with hand-chosen penalty coefficients faithfully represents the two objectives in problem P.
    Eq. (22) defines the training objective; the weights (rho, delta_p, varrho, sigma) are not derived from the problem and are not fully specified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of UAV-assisted Joint Mobile Edge Computing and Data Collection via Matching-enabled Deep Reinforcement Learning." pith.science (2026). https://pith.science/paper/4BJTPIPD

@misc{pith2026250207388,
  author       = {Pith},
  title        = {Pith review of: UAV-assisted Joint Mobile Edge Computing and Data Collection via Matching-enabled Deep Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4BJTPIPD}},
  note         = {Machine review of arXiv:2502.07388}
}
read the original abstract

Unmanned aerial vehicle (UAV)-assisted mobile edge computing (MEC) and data collection (DC) have been popular research issues. Different from existing works that consider MEC and DC scenarios separately, this paper investigates a multi-UAV-assisted joint MEC-DC system. Specifically, we formulate a joint optimization problem to minimize the MEC latency and maximize the collected data volume. This problem can be classified as a non-convex mixed integer programming problem that exhibits long-term optimization and dynamics. Thus, we propose a deep reinforcement learning-based approach that jointly optimizes the UAV movement, user transmit power, and user association in real time to solve the problem efficiently. Specifically, we reformulate the optimization problem into an action space-reduced Markov decision process (MDP) and optimize the user association by using a two-phase matching-based association (TMA) strategy. Subsequently, we propose a soft actor-critic (SAC)-based approach that integrates the proposed TMA strategy (SAC-TMA) to solve the formulated joint optimization problem collaboratively. Simulation results demonstrate that the proposed SAC-TMA is able to coordinate the two subsystems and can effectively reduce the system latency and improve the data collection volume compared with other benchmark algorithms.

Figures

Figures reproduced from arXiv: 2502.07388 by the authors.

Figure 1
Figure 1. UAV-assisted joint MEC-DC system. environment. To investigate this effect, Seid et al. [40] consid￾ered the inter-cell interference among UAVs and proposed an approach based on multi-agent deep reinforcement learning (MADRL) to ensure the QoS requirements of IoT devices or users while reducing the total computing cost of their considered network. However, this work assumes that UAV clusters provide MEC services to u… view at source ↗
Figure 2
Figure 2. Schematic of the proposed SAC-TMA algorithm. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Effect of MEC users number on latency and DC performance (The performance of latency is measured by the latency reward defined in Eq. (17)). [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Sum reward under different numbers of MEC users with [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Effect of Nmax u on latency reward, DC reward, task complete rate, and DC rate. 1 2 3 4 5 6 7 8 9 10 200 300 400 500 600 700 Average sum reward Number of N max u SAC-TMA SAC-TMA-greedy TD3-TMA [54] PPO-TMA [55] MSE-PPO [56] Dual-timescales [57] Greedy [PITH_FULL_IMAGE…
Figure 6
Figure 6. Figure 6: Sum reward under different numbers of maximum service capacity [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Average energy consumption of one UAV during a step. [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Training reward using random seeds 0, 1, and 2 (The curves represent [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 10
Figure 10. Figure 10: Comparison of system sum rate and total association time for 300 [PITH_FULL_IMAGE:figures/full_fig_p017_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Toward Realization of Low-Altitude Economy Networks: Core Architecture, Integrated Technologies, and Future Directions

    cs.NI 2025-04 conditional novelty 4.0 of 10

    A survey that organizes low-altitude economy networks into a layered architecture and argues that multi-technology integration is the key enabler.

Reference graph

Works this paper leans on

57 extracted references · 54 canonical work pages · cited by 1 Pith paper

  1. [46]

    Resource management and reflection optimization for intelligent reflecting surface assisted multi-access edge computing using deep reinforcement learning,

    Z. Wang, Y . Wei, Z. Feng, F. R. Yu, and Z. Han, “Resource management and reflection optimization for intelligent reflecting surface assisted multi-access edge computing using deep reinforcement learning,” IEEE Trans. Wirel. Commun., vol. 22, no. 2, pp. 1175–1186, 2023

    work page 2023

  2. [51]

    An efficient matching game approach to association formation in uav-enabled hierarchical distributed learning,

    X. Huo, H. Zhang, Z. Wang, H. Yan, and C. Liu, “An efficient matching game approach to association formation in uav-enabled hierarchical distributed learning,” IEEE Trans. Cybern. , 2024

    work page 2024

  3. [1]

    Fundamental design aspects of UA V-enabled MEC systems: A review on models, challenges, and future opportunities,

    M. H. Adnan, Z. A. Zukarnain, and O. A. Amodu, “Fundamental design aspects of UA V-enabled MEC systems: A review on models, challenges, and future opportunities,” Comput. Sci. Rev., vol. 51, p. 100615, 2024

    work page 2024

  4. [2]

    A tutorial on extremely large-scale MIMO for 6G: Fundamentals, signal processing, and applications,

    Z. Wang, J. Zhang, H. Du, D. Niyato, S. Cui, B. Ai, M. Debbah, K. B. Letaief, and H. V . Poor, “A tutorial on extremely large-scale MIMO for 6G: Fundamentals, signal processing, and applications,” IEEE Commun. Surveys Tuts., vol. 26, no. 3, pp. 1560–1605, 3rd quarter, 2024

    work page 2024

  5. [3]

    UA V trajectory optimization for large-scale and low-power data collection: An attention-reinforced learning scheme,

    Y . Zhu, B. Yang, M. Liu, and Z. Li, “UA V trajectory optimization for large-scale and low-power data collection: An attention-reinforced learning scheme,” IEEE Trans. Wirel. Commun., vol. 23, no. 4, pp. 3009– 3024, 2024

    work page 2024

  6. [4]

    3- d trajectory optimization and communication resources allocation in UA V-assisted IoT networks for sustainable industry 5.0,

    P. Du, Y . Shi, H. Cao, S. Garg, G. Kaddoum, and M. Alrashoud, “3- d trajectory optimization and communication resources allocation in UA V-assisted IoT networks for sustainable industry 5.0,” IEEE Trans. Consumer Electron., vol. 70, no. 1, pp. 1423–1433, 2024

    work page 2024

  7. [5]

    Dual UA V cluster-assisted maritime physical layer secure communications via collaborative beamforming,

    J. Huang, A. Wang, G. Sun, J. Li, J. Wang, H. Du, and D. Niyato, “Dual UA V cluster-assisted maritime physical layer secure communications via collaborative beamforming,” IEEE Internet Things J. , pp. 1–1, 2024

    work page 2024

  8. [6]

    UA V swarm-enabled collaborative secure relay communications with time-domain colluding eavesdropper,

    C. Zhang, G. Sun, Q. Wu, J. Li, S. Liang, D. Niyato, and V . C. M. Leung, “UA V swarm-enabled collaborative secure relay communications with time-domain colluding eavesdropper,” IEEE Trans. Mob. Comput., vol. 23, no. 9, pp. 8601–8619, 2024

    work page 2024

Show all 57 references
  1. [7]

    Multi- objective aerial collaborative secure communication optimization via generative diffusion model-enabled deep reinforcement learning,

    C. Zhang, G. Sun, J. Li, Q. Wu, J. Wang, D. Niyato, and Y . Liu, “Multi- objective aerial collaborative secure communication optimization via generative diffusion model-enabled deep reinforcement learning,” IEEE Trans. Mob. Comput., pp. 1–18, 2024

    2024

  2. [8]

    A comprehensive overview on 5G-and-beyond networks with UA Vs: From communications to sensing and intelligence,

    Q. Wu, J. Xu, Y . Zeng, D. W. K. Ng, N. Al-Dhahir, R. Schober, and A. L. Swindlehurst, “A comprehensive overview on 5G-and-beyond networks with UA Vs: From communications to sensing and intelligence,” IEEE J. Sel. Areas Commun. , vol. 39, no. 10, pp. 2912–2945, 2021

    2021

  3. [9]

    Outage analysis of UA V-aided networks with underlaid ambient backscatter communications,

    X. Jiang, M. Sheng, N. Zhao, J. Liu, D. Niyato, and F. R. Yu, “Outage analysis of UA V-aided networks with underlaid ambient backscatter communications,” IEEE Trans. Wirel. Commun. , vol. 22, no. 11, pp. 7492–7505, 2023

    2023

  4. [10]

    UA V-assisted connectivity enhancement algorithms for multiple isolated sensor networks in agricultural Internet of things,

    J. Pei, H. Chen, and L. Shu, “UA V-assisted connectivity enhancement algorithms for multiple isolated sensor networks in agricultural Internet of things,” Comput. Networks, vol. 207, p. 108854, 2022

    2022

  5. [11]

    UA V-assisted sleep scheduling algorithm for energy-efficient data collection in agricultural Internet of things,

    M. Zhou, H. Chen, L. Shu, and Y . Liu, “UA V-assisted sleep scheduling algorithm for energy-efficient data collection in agricultural Internet of things,” IEEE Internet Things J., vol. 9, no. 13, pp. 11 043–11 056, 2022

    2022

  6. [12]

    UA V-based MEC-assisted automated traffic management scheme using blockchain,

    M. Masuduzzaman, A. Islam, K. Sadia, and S. Y . Shin, “UA V-based MEC-assisted automated traffic management scheme using blockchain,” Future Gener. Comput. Syst. , vol. 134, pp. 256–270, 2022. 18

    2022

  7. [13]

    Monitoring road traffic with a UA V-based system,

    M. Elloumi, R. Dhaou, B. Escrig, H. Idoudi, and L. A. Sa ¨ıdane, “Monitoring road traffic with a UA V-based system,” in Proc. IEEE WCNC, 2018, pp. 1–6

    2018

  8. [14]

    Joint task offloading and resource allocation in aerial-terrestrial UA V networks with edge and fog computing for post-disaster rescue,

    G. Sun, L. He, Z. Sun, Q. Wu, S. Liang, J. Li, D. Niyato, and V . C. M. Leung, “Joint task offloading and resource allocation in aerial-terrestrial UA V networks with edge and fog computing for post-disaster rescue,” IEEE Trans. Mob. Comput. , vol. 23, no. 9, pp. 8582–8600, 2024

    2024

  9. [15]

    Uav-based real-time survivor detection system in post-disaster search and rescue operations,

    J. Dong, K. Ota, and M. Dong, “Uav-based real-time survivor detection system in post-disaster search and rescue operations,” IEEE Journal on Miniaturization for Air and Space Systems , vol. 2, no. 4, pp. 209–219, 2021

    2021

  10. [16]

    Reliable and energy-efficient communications via collaborative beamforming for UA V networks,

    X. Zheng, G. Sun, J. Li, S. Liang, Q. Wu, M. Yin, D. Niyato, and V . C. M. Leung, “Reliable and energy-efficient communications via collaborative beamforming for UA V networks,” IEEE Trans. Wirel. Commun., vol. 23, no. 10, pp. 13 235–13 251, 2024

    2024

  11. [17]

    Multi-objective optimization for multi-uav-assisted mobile edge computing,

    G. Sun, Y . Wang, Z. Sun, Q. Wu, J. Kang, D. Niyato, and V . C. M. Leung, “Multi-objective optimization for multi-uav-assisted mobile edge computing,” IEEE Trans. Mob. Comput. , vol. 23, no. 12, pp. 14 803– 14 820, 2024

    2024

  12. [18]

    TJCCT: A two-timescale approach for UA V-assisted mobile edge computing,

    Z. Sun, G. Sun, Q. Wu, L. He, S. Liang, H. Pan, D. Niyato, C. Yuen, and V . C. M. Leung, “TJCCT: A two-timescale approach for UA V-assisted mobile edge computing,” IEEE Trans. Mob. Comput. , pp. 1–18, 2024

    2024

  13. [19]

    Robust com- putation offloading and trajectory optimization for multi-UA V-assisted MEC: A multiagent DRL approach,

    B. Li, R. Yang, L. Liu, J. Wang, N. Zhang, and M. Dong, “Robust com- putation offloading and trajectory optimization for multi-UA V-assisted MEC: A multiagent DRL approach,” IEEE Internet Things J. , vol. 11, no. 3, pp. 4775–4786, 2024

    2024

  14. [20]

    Multi- objective optimization for data collection in UA V-assisted agricultural IoT,

    L. Liu, A. Wang, G. Sun, J. Li, H. Pan, and T. Q. S. Quek, “Multi- objective optimization for data collection in UA V-assisted agricultural IoT,” IEEE Trans. Veh. Technol., pp. 1–17, 2024

    2024

  15. [21]

    Max-min fair 3D trajectory design and transmission scheduling for solar-powered fixed-wing UA V-assisted data collection,

    C. Sun, X. Xiong, Z. Zhai, W. Ni, T. Ohtsuki, and X. Wang, “Max-min fair 3D trajectory design and transmission scheduling for solar-powered fixed-wing UA V-assisted data collection,”IEEE Trans. Wirel. Commun., vol. 22, no. 12, pp. 8650–8665, 2023

    2023

  16. [22]

    Privacy- aware and security-enhanced efficient matchmaking encryption,

    J. Sun, G. Xu, T. Zhang, X. Yang, M. Alazab, and R. H. Deng, “Privacy- aware and security-enhanced efficient matchmaking encryption,” IEEE Trans. Inf. Forensics Secur., vol. 18, pp. 4345–4360, 2023

    2023

  17. [23]

    Privacy-preserving fine-grained data sharing with dynamic service for the cloud-edge IoT,

    J. Sun, Y . Bao, W. Qiu, R. Lu, S. Zhang, Y . Guan, and X. Cheng, “Privacy-preserving fine-grained data sharing with dynamic service for the cloud-edge IoT,” IEEE Trans. Dependable Secure Comput. , 2024

    2024

  18. [24]

    Joint optimization on trajectory, computation and communication resources in information freshness sensitive MEC system,

    H. Li, J. Zhang, H. Zhao, Y . Ni, J. Xiong, and J. Wei, “Joint optimization on trajectory, computation and communication resources in information freshness sensitive MEC system,” IEEE Trans. Veh. Technol. , vol. 73, no. 3, pp. 4162–4177, 2024

    2024

  19. [25]

    Deep reinforcement learning based resource allocation in multi-UA V- aided MEC networks,

    J. Chen, X. Cao, P. Yang, M. Xiao, S. Ren, Z. Zhao, and D. O. Wu, “Deep reinforcement learning based resource allocation in multi-UA V- aided MEC networks,” IEEE Trans. Commun., vol. 71, no. 1, pp. 296– 309, 2023

    2023

  20. [26]

    Service time maximization for data collection in multi-UA V-aided networks,

    J. Dandapat, N. Gupta, S. Agarwal, and B. Kumbhani, “Service time maximization for data collection in multi-UA V-aided networks,” IEEE Trans. Intell. Veh., vol. 9, no. 1, pp. 328–337, 2024

    2024

  21. [27]

    Secure video offloading in multi-uav-enabled MEC networks: A deep reinforcement learning approach,

    T. Zhao, F. Li, and L. He, “Secure video offloading in multi-uav-enabled MEC networks: A deep reinforcement learning approach,” IEEE Internet Things J., vol. 11, no. 2, pp. 2950–2963, 2024

    2024

  22. [28]

    Multi- UA V-enabled load-balance mobile-edge computing for IoT networks,

    L. Yang, H. Yao, J. Wang, C. Jiang, A. Benslimane, and Y . Liu, “Multi- UA V-enabled load-balance mobile-edge computing for IoT networks,” IEEE Internet Things J. , vol. 7, no. 8, pp. 6898–6908, 2020

    2020

  23. [29]

    Joint task offloading and resource allocation in uav-enabled mobile edge computing,

    Z. Yu, Y . Gong, S. Gong, and Y . Guo, “Joint task offloading and resource allocation in uav-enabled mobile edge computing,” IEEE Internet Things J., vol. 7, no. 4, pp. 3147–3159, 2020

    2020

  24. [30]

    Multi-uav-enabled mobile-edge computing for time-constrained IoT applications,

    C. Zhan, H. Hu, Z. Liu, Z. Wang, and S. Mao, “Multi-uav-enabled mobile-edge computing for time-constrained IoT applications,” IEEE Internet Things J. , vol. 8, no. 20, pp. 15 553–15 567, 2021

    2021

  25. [31]

    Computation bits maximization in UA V-assisted MEC networks with fairness constraint,

    X. Zhou, L. Huang, T. Ye, and W. Sun, “Computation bits maximization in UA V-assisted MEC networks with fairness constraint,”IEEE Internet Things J., vol. 9, no. 21, pp. 20 997–21 009, 2022

    2022

  26. [32]

    Multiagent reinforcement learning in controlling offloading ratio and trajectory for multi-uav mobile-edge computing,

    W. Lee and T. Kim, “Multiagent reinforcement learning in controlling offloading ratio and trajectory for multi-uav mobile-edge computing,” IEEE Internet Things J. , vol. 11, no. 2, pp. 3417–3429, 2024

    2024

  27. [33]

    Energy and latency efficient joint communication and computation optimization in a multi- UA V-assisted MEC network,

    F. Pervez, A. Sultana, C. Yang, and L. Zhao, “Energy and latency efficient joint communication and computation optimization in a multi- UA V-assisted MEC network,” IEEE Trans. Wirel. Commun. , vol. 23, no. 3, pp. 1728–1741, 2024

    2024

  28. [34]

    Joint UA V placement optimization, resource allocation, and computation offloading for thz band: A DRL approach,

    H. Wang, H. Zhang, X. Liu, K. Long, and A. Nallanathan, “Joint UA V placement optimization, resource allocation, and computation offloading for thz band: A DRL approach,” IEEE Trans. Wirel. Commun., vol. 22, no. 7, pp. 4890–4900, 2023

    2023

  29. [35]

    Deep reinforcement learning based latency minimization for mobile edge computing with virtualization in maritime UA V communication network,

    Y . Liu, J. Yan, and X. Zhao, “Deep reinforcement learning based latency minimization for mobile edge computing with virtualization in maritime UA V communication network,”IEEE Trans. Veh. Technol., vol. 71, no. 4, pp. 4225–4236, 2022

    2022

  30. [36]

    Multi- objective optimization for UA V-assisted wireless powered IoT networks based on extended DDPG algorithm,

    Y . Yu, J. Tang, J. Huang, X. Zhang, D. K. C. So, and K. Wong, “Multi- objective optimization for UA V-assisted wireless powered IoT networks based on extended DDPG algorithm,” IEEE Trans. Commun. , vol. 69, no. 9, pp. 6361–6374, 2021

    2021

  31. [37]

    UA V trajectory planning with interference awareness in UA V-enabled time-constrained data collection systems,

    K. Liu and J. Zheng, “UA V trajectory planning with interference awareness in UA V-enabled time-constrained data collection systems,” IEEE Trans. Veh. Technol., vol. 73, no. 2, pp. 2799–2815, 2024

    2024

  32. [38]

    Mec-assisted real-time data acquisition and processing for UA V with general missions,

    Y . Zeng and J. Tang, “Mec-assisted real-time data acquisition and processing for UA V with general missions,” IEEE Trans. Veh. Technol., vol. 72, no. 1, pp. 1058–1072, 2023

    2023

  33. [39]

    Learning-based multi-UA V assisted data acquisition and computation for information freshness in WPT enabled space-air-ground PIoT,

    J. Liu, X. Zhao, P. Qin, S. Geng, Z. Chen, and H. Zhou, “Learning-based multi-UA V assisted data acquisition and computation for information freshness in WPT enabled space-air-ground PIoT,” IEEE Trans. Netw. Sci. Eng., vol. 11, no. 1, pp. 48–63, 2024

    2024

  34. [40]

    Multi- agent DRL for task offloading and resource allocation in multi-UA V enabled IoT edge network,

    A. M. Seid, G. O. Boateng, B. Mareri, G. Sun, and W. Jiang, “Multi- agent DRL for task offloading and resource allocation in multi-UA V enabled IoT edge network,” IEEE Trans. Netw. Serv. Manage. , vol. 18, no. 4, pp. 4531–4547, 2021

    2021

  35. [41]

    Service caching based aerial cooperative computing and resource allocation in multi-uav enabled MEC systems,

    G. Zheng, C. Xu, M. Wen, and X. Zhao, “Service caching based aerial cooperative computing and resource allocation in multi-uav enabled MEC systems,” IEEE Trans. Veh. Technol., vol. 71, no. 10, pp. 10 934– 10 947, 2022

    2022

  36. [42]

    Caching in the sky: Proactive deployment of cache-enabled unmanned aerial vehicles for optimized quality-of-experience,

    M. Chen, M. Mozaffari, W. Saad, C. Yin, M. Debbah, and C. S. Hong, “Caching in the sky: Proactive deployment of cache-enabled unmanned aerial vehicles for optimized quality-of-experience,” IEEE J. Sel. Areas Commun., vol. 35, no. 5, pp. 1046–1061, 2017

    2017

  37. [43]

    Mobile-edge com- puting: Partial computation offloading using dynamic voltage scaling,

    Y . Wang, M. Sheng, X. Wang, L. Wang, and J. Li, “Mobile-edge com- puting: Partial computation offloading using dynamic voltage scaling,” IEEE Trans. Commun. , vol. 64, no. 10, pp. 4268–4282, 2016

    2016

  38. [44]

    Modeling and analysis of stochastic mobile-edge computing wireless networks,

    Y . Gu, Y . Yao, C. Li, B. Xia, D. Xu, and C. Zhang, “Modeling and analysis of stochastic mobile-edge computing wireless networks,” IEEE Internet Things J. , vol. 8, no. 18, pp. 14 051–14 065, 2021

    2021

  39. [45]

    Deep reinforcement learning-based contract incentive mechanism for joint sensing and computation in mobile crowdsourcing networks,

    N. Zhao, Y . Pei, Y .-C. Liang, and D. Niyato, “Deep reinforcement learning-based contract incentive mechanism for joint sensing and computation in mobile crowdsourcing networks,” IEEE Internet Things J., 2023

    2023

  40. [47]

    Throughput maximization for multiedge multiuser edge computing systems,

    Y . Deng, Z. Chen, X. Chen, and Y . Fang, “Throughput maximization for multiedge multiuser edge computing systems,” IEEE Internet Things J., vol. 9, no. 1, pp. 68–79, 2022

    2022

  41. [48]

    Many-to-one matching markets with externalities among firms,

    K. Bando, “Many-to-one matching markets with externalities among firms,” J. Math. Econ. , vol. 48, no. 1, pp. 14–20, 2012

    2012

  42. [49]

    Many-to-many matching with externalities for device-to-device communications,

    J. Zhao, Y . Liu, K. K. Chai, Y . Chen, and M. Elkashlan, “Many-to-many matching with externalities for device-to-device communications,” IEEE Wirel. Commun. Lett., vol. 6, no. 1, pp. 138–141, 2017

    2017

  43. [50]

    Peer effects and stability in matching markets,

    E. Bodine-Baron, C. Lee, A. Chong, B. Hassibi, and A. Wierman, “Peer effects and stability in matching markets,” in Proc. SAGT, vol. 6982, 2011, pp. 117–129

    2011

  44. [52]

    Soft actor- critic algorithms and applications,

    T. Haarnoja, A. Zhou, K. Hartikainen, G. Tucker, S. Ha, J. Tan, V . Kumar, H. Zhu, A. Gupta, P. Abbeel, and S. Levine, “Soft actor- critic algorithms and applications,” CoRR, vol. abs/1812.05905, 2018

    2018 arXiv

  45. [53]

    Energy-constrained UA V data collection systems: NOMA and OMA,

    X. Mu, Y . Liu, L. Guo, J. Lin, and Z. Ding, “Energy-constrained UA V data collection systems: NOMA and OMA,” IEEE Trans. Veh. Technol., vol. 70, no. 7, pp. 6898–6912, 2021

    2021

  46. [54]

    Addressing function approxi- mation error in actor-critic methods,

    S. Fujimoto, H. van Hoof, and D. Meger, “Addressing function approxi- mation error in actor-critic methods,” CoRR, vol. abs/1802.09477, 2018

    2018 arXiv

  47. [55]

    Proximal policy optimization algorithms,

    J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,” CoRR, vol. abs/1707.06347, 2017

    2017 arXiv

  48. [56]

    UA V-enabled fair offloading for MEC networks: A DRL approach based on actor-critic parallel architecture,

    W. Li, S. Li, H. Shi, W. Yan, and Y . Zhou, “UA V-enabled fair offloading for MEC networks: A DRL approach based on actor-critic parallel architecture,” Appl. Intell., vol. 54, no. 4, pp. 3529–3546, 2024

    2024

  49. [57]

    Dual- timescales optimization of task scheduling and resource slicing in satellite-terrestrial edge computing networks,

    T. Huang, Z. Fang, Q. Tang, R. Xie, T. Chen, and F. R. Yu, “Dual- timescales optimization of task scheduling and resource slicing in satellite-terrestrial edge computing networks,” IEEE Trans. Mob. Com- put., vol. 23, no. 12, pp. 14 111–14 126, 2024

    2024

Pith tools

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