Pith. sign in

REVIEW 4 major objections 6 minor 45 references

Letting a UAV-mounted edge server reposition its antennas within a small region cuts total system energy by about a third compared with fixed antennas.

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 ↗

Jointly tuning antenna positions, transmit powers, and CPU allocation on a movable-antenna drone edge server cuts simulated system energy by up to ~34% versus fixed-antenna baselines.

T0 review reviewed 2026-08-01 challenge →

load-bearing objection Novel combination of movable antennas on a UAV edge server with movement energy/latency in an energy-minimization problem; the qualitative claim is plausible, but the closed-form derivations are shaky and the 33.8% headline number likely oversells the benefit because the FPA baseline is under-specified. the 4 major comments →

arxiv 2607.23214 v1 pith:LPMLQMOY submitted 2026-07-25 cs.IT eess.SPmath.IT

Movable-Antenna Assisted Energy Minimization in UAV-Enabled Mobile Edge Computing Systems

Jiang Chen , Chunjie Wang , Xuhui Zhang , Yanyan Shen , Kejiang Ye , Chengzhong Xu This is my paper
classification cs.IT eess.SPmath.IT
keywords movable antennaUAVmobile edge computingenergy minimizationalternating optimizationbeamformingresource allocationlow-altitude network
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

This paper tries to establish that allowing the antennas on a UAV-mounted edge server to shift positions within a small region reduces total system energy (CE transmission, UAV computation, and antenna movement) compared to fixed or random antenna arrays. It claims that jointly optimizing CPU allocation, transmit power, receive beamforming, and antenna positions yields roughly 33.8% energy savings over a fixed-position array and 15.7% over a random-position array under a 1 MHz bandwidth with 10-million-bit tasks. The paper also claims that a four-element movable array matches or beats a six-element fixed array, suggesting hardware efficiency. This matters for battery-constrained consumer devices and UAVs in low-altitude networks, where energy is the key bottleneck.

Core claim

In a UAV-enabled mobile edge computing system where a hovering UAV serves ground consumer devices, the paper demonstrates through simulation that optimizing antenna positions in addition to classic resource allocation substantially lowers total energy consumption. The mechanism is that better array gain improves the uplink rate, which shortens the offload transmission time, thereby letting the UAV run its CPU at a lower frequency while still meeting the latency deadline; this cuts both communication and computation energy. Under a 1 MHz bandwidth and 10-million-bit load per device, the proposed scheme reduces total energy by about 33.8% versus a fixed-position antenna array and 15.7% versus

What carries the argument

The central object is the alternating optimization (AO) algorithm based on block coordinate descent, which decomposes the non-convex problem into three subproblems: convex computation-resource allocation, transmit power and receive beamforming reformulated with the quadratic transform and WMMSE, and antenna positions optimized by particle swarm optimization with penalty constraints. The position-dependent array gain is carried by the steering vector a_m(x, θ_m) = [e^{j2π x_1 cosθ_m/λ}, ..., e^{j2π x_K cosθ_m/λ}]^T, which plugs into the LoS channel model and determines the achievable rate r_m, the offload time T_m^tran, and ultimately the UAV CPU frequency f_m^uav.

Load-bearing premise

The entire energy benefit is computed assuming pure free-space line-of-sight channels with a single angle of arrival and perfect position-dependent channel knowledge at the optimizer; if real low-altitude links include multipath, scattering, or channel estimation error, the movable-antenna advantage shrinks and the 4-beats-6 inversion may not survive.

What would settle it

Run the same alternating-optimization scheme on a simulated Rician fading channel (varying K-factor from, say, 10 dB down to 0 dB) or with measured low-altitude UAV channel data, and compare the optimized total energy against the fixed-antenna scheme; if the energy gap drops below a few percent or reverses, the LoS-only channel model is the reason.

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

If this is right

  • If the claimed savings hold, operators can deploy smaller movable-antenna arrays instead of larger fixed arrays, reducing hardware cost, weight, and energy footprint on UAV platforms.
  • The link between antenna positioning and computation energy (better rate → shorter offload time → lower CPU frequency) suggests that MA tuning can be a new lever in MEC energy minimization beyond conventional power and resource allocation.
  • The proposed AO algorithm is guaranteed to converge because the objective is non-increasing and the feasible set is bounded, making the approach implementable in practice.
  • The advantage of movable antennas grows with heavier task loads and more devices, where interference and energy pressure are highest, indicating the technology is most valuable in dense low-altitude consumer scenarios.
  • Energy savings are achieved without violating latency constraints, as the simulations show monotonic improvement in total energy across a range of task loads, computation intensities, and bandwidths.

Where Pith is reading between the lines

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

  • The LoS-only channel model (free-space path loss with a single angle of arrival) is the load-bearing physical assumption; if real links have substantial Rician/multipath components or channel estimation error, the movable-antenna gain should shrink — a testable extension is to rerun the same optimization under a Rician fading model and observe where the energy gap versus fixed antennas falls below
  • The paper counts antenna movement time and energy in the latency and objective, but it does not account for the energy and time cost of learning or tracking the channel state needed to reposition antennas; folding that overhead in would likely reduce the net benefit, especially in fast-changing channels.
  • The '4 movable beats 6 fixed' result suggests a design frontier: given a target energy budget, one could optimize the number of antennas, the movable span L, and the UAV trajectory jointly, rather than treating antenna count as fixed.
  • Since the UAV hovers at a fixed position in the model, a natural multi-slot extension would couple MA positioning with UAV trajectory planning, where macro-scale movement and micro-scale antenna tuning could yield additional savings in dynamic user distributions.
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

4 major / 6 minor

Summary. The paper considers a UAV-enabled mobile edge computing system in which the UAV carries K movable antennas and serves M single-antenna CE devices that offload identical-size tasks under a full-offloading policy. The authors formulate a total-energy minimization problem (P1) over computation resource allocation F, CE transmit powers P, receive beamforming vectors W, and MA positions X, subject to power, latency, CPU, and antenna-spacing constraints. The proposed solution is an alternating-optimization (BCD) algorithm: (P2) for F is solved as a convex problem with CVX; (P3) for P and W is handled by the quadratic transform and WMMSE, with closed-form updates stated in Theorems 1 and 2; and (P5) for X is solved by PSO with penalty terms. Simulations benchmark the proposed scheme against fixed-position antenna (FPA) and random-position antenna (RPA) baselines, reporting substantial energy savings (e.g., 33.8% vs FPA and 15.7% vs RPA at 1 MHz in Fig. 9) and showing that a K=4 MA array can outperform a K=6 FPA array.

Significance. If the results are correct, the paper would provide a useful demonstration that movable-antenna position optimization can reduce energy consumption in UAV-aided MEC, potentially enabling smaller antenna arrays to match or beat larger fixed arrays. The system model is clearly laid out, and the inclusion of antenna-movement energy and latency in the formulation is a strength. The qualitative direction of the result is plausible: under LoS-dominated channels, moving antennas to improve array gain reduces transmission time, which propagates into lower CPU-frequency energy. However, several load-bearing technical points need attention before the quantitative claims can be accepted: the closed-form power update in Theorem 1 omits cross-user interference terms, the closed-form beamformer in Theorem 2 uses an incorrect interference covariance, the convergence proof assumes a monotonicity property that PSO does not guarantee, and the FPA/RPA baselines are too vaguely specified to support the headline savings figures.

major comments (4)
  1. [§IV-B, Theorem 1, Eqs. (30)–(33)] The derivative in (32) is not the full derivative of the Lagrangian (30) with respect to p^new_m. Each rate surrogate \tilde r_n (n≠m) contains the interference term |I_n|^2 (p^new_m)^2 |w_n^H h_m|^2, so ∂(Σ_n α_n² \tilde r_n)/∂p^new_m includes cross-terms of the form −2B G_n |I_n|^2 p^new_m |w_n^H h_m|^2 (and analogous ϖ_n terms). Equation (32) contains only the n=m terms. Consequently, setting (32) to zero and solving yields (33)/(19), which is not a KKT point of (P4.2) for M>1. The power update is coupled across users and does not admit the simple per-user closed form claimed. The authors should either derive the correct coupled update, use a numerical solver for (P4.2) and state that the theorem is only a single-user special case, or remove the closed-form claim.
  2. [§IV-B, Theorem 2, Eq. (21)] The denominator of the closed-form beamformer contains the matrix Σ_{j=1}^M p_j h_m^* h_m^T + σ² I, which depends only on the desired user's channel vector h_m. The correct interference-plus-noise covariance is Σ_j p_j h_j h_j^H (with the desired-user term included or excluded depending on convention). As written, the expression ignores all other users' channel directions and is not the optimum of (P4.3). This invalidates Theorem 2 as a general MMSE-type beamformer. The proof and equation should be corrected, or the theorem should be stated for a single-user scenario.
  3. [§IV-D, Proposition 1 and Appendix C, Eq. (39)] The convergence proof relies on the inequality Φ(F^{l+1},P^{l+1},W^{l+1},X^{l}) ≥ Φ(F^{l+1},P^{l+1},W^{l+1},X^{l+1}), which assumes the PSO update is guaranteed to produce a non-increasing objective. The PSO velocity update (26)–(28) is stochastic and does not guarantee monotone improvement at every iteration. Moreover, the fitness (29) uses additive penalties for constraints (9d) and (9f) rather than exact projection, so the accepted particle may be infeasible. The chain (37)–(40) therefore does not establish convergence of Algorithm 2. The proof can be repaired by retaining the global-best feasible particle and accepting an update only when it improves the true objective; otherwise the convergence claim should be presented as empirical.
  4. [§V, Figs. 4–9 (especially Fig. 9)] The FPA and RPA baselines are not sufficiently specified. FPA 'fixes the array location' but the paper does not report what geometry is used (e.g., uniform λ/2 spacing centered in [0,L], or the initial X of Algorithm 2). RPA draws random positions with no stated seed or averaging. Since the proposed scheme's only additional degree of freedom over FPA is the position vector X, comparing against an arbitrary fixed X can inflate the reported savings: the headline 33.8% improvement in Fig. 9 may measure the weakness of the baseline rather than the value of antenna movability. The authors should (i) state the FPA geometry explicitly, (ii) add a fixed-position optimized baseline obtained by running the same AO with X held at the best of a small grid/search over translations, and (iii) report means and standard errors over multiple PSO/RPA random seeds. Without this, the quantitative central cl
minor comments (6)
  1. [§IV-B, Eq. (22)] The first displayed equality for \tilde r_m is dimensionally inconsistent: it writes (√p_m w_m^H h_m) divided by a real denominator plus 1, while the final expression is the standard WMMSE rate surrogate. The intermediate line should use |√p_m w_m^H h_m|² in the numerator (or a similar squared magnitude).
  2. [§IV-B, Algorithm 1 and Eqs. (13), (32)] The variable is called α in (13) and in the Lagrangian (30), but the derivative in (32) uses y_m. Please unify notation. Also, after differentiation in (32), the term Σ_j p^new_j |w_m^H h_j|² should be p^new_m |w_m^H h_m|² only, since the derivative of (p^new_j)² selects j=m.
  3. [§IV-B, Problem (P4.3)] The objective of (P4.3) is written as a single-user expression 2α_m√(l p_m) − α_m² \tilde r_m, but it should be summed over m. As printed, the problem is ambiguous about whether each w_m is optimized independently or jointly.
  4. [§IV-D, Appendix B] In the proof of Theorem 2, the sentence 'which yields the optimal transmit power as' should read 'optimal receive beamforming as'.
  5. [§V, all figures] No error bars or random-seed information is reported. Since PSO and RPA are stochastic, the claimed performance gaps (e.g., K=4 MA vs K=6 FPA in Figs. 4–6) should be accompanied by confidence intervals or multiple-run averages.
  6. [§III-A, Eqs. (1)–(2)] The channel model assumes pure free-space LoS with a rank-one steering vector and perfect position-dependent CSI. The quantitative gains are computed entirely under this model. A brief discussion (or a robustness simulation) of Rician/multipath channels and channel-estimation error would clarify the domain of validity of the claimed energy savings.

Circularity Check

0 steps flagged

No circular derivation; central result is an optimization-simulation comparison with no fitted-input/prediction reduction.

full rationale

The derivation chain is self-contained. Problem P1 is built from explicit channel, communication, computing, and movement-energy models (Eqs. (1)-(8)); the AO algorithm solves P2 (convex CPU allocation), P3 via quadratic transform/WMMSE with KKT closed forms (Theorems 1-2, Appendices A-B), and P5 via PSO. Convergence is proven by monotone non-increasing objective over a bounded feasible set (Appendix C). No empirical parameters are fitted and no quantity that appears as an input is later relabeled as a prediction; the reported 33.8%/15.7% savings are simulation outputs under stated parameters. The LoS channel model (Eq. (2)) is an explicit assumption, not a consequence of the results. Several references are to the authors' prior work, but none is load-bearing in the derivation: [45] is invoked only for PSO pseudo-code details ('the detailed procedure of the PSO-based algorithm is omitted here. Please refer to the Algorithm 3 in [45]' — Section IV-C), while the paper's Eqs. (24)-(28) already specify the update rules; Lemma 1 rests on the external fractional-programming result [43]; the other self-citations are contextual. The FPA/RPA benchmark construction may understate the fixed-antenna baseline (an optimized fixed geometry, or adopting the MA-converged positions at zero movement energy, is not simulated), but that is a benchmark-fairness/correctness concern rather than a circular reduction; it does not make any derived equation equal to its input by construction.

Axiom & Free-Parameter Ledger

4 free parameters · 7 axioms · 0 invented entities

The central claim rests on standard optimization machinery (quadratic transform, WMMSE, BCD) applied to a stylized physical model. No new entities are invented. The hand-chosen PSO/penalty/threshold knobs, the LoS perfect-CSI channel model, and the fixed-hover/full-offload/identical-task assumptions govern the magnitude of the reported savings, so these are the items a reader should audit before accepting the headline energy numbers.

free parameters (4)
  • PSO hyperparameters (population, iterations, inertia, learning factors) = U=200, i_max=200, ς_max=0.9, ς_min=0.4, ϱ=μ=1.5
    Hand-chosen in Sec. V; they control the quality of the antenna-position subproblem solution and therefore the reported energy numbers. No sensitivity analysis is provided.
  • PSO penalty coefficients = ψ1=ψ2=1000
    Chosen in Sec. V for the adaptive penalty in Eq. (29); the trade-off between objective value and constraint feasibility in the position search depends on them.
  • Convergence thresholds and iteration caps = ε, j_max, l_max (unspecified)
    Algorithms 1-2 terminate on ε, j_max, l_max (Sec. IV), but the values are never given; the reported converged energy depends on this unreported stopping rule.
  • Simulation operating point = H=20 m, τ=50 s, l=10^7 bits, B=1.45 MHz, ρ=-30 dB, σ²=-80 dBm, Ē=0.175 J/m, v=0.2 m/s, L=4λ, c=1000 cycles/bit
    The headline 33.8%/15.7% savings (Fig. 9) are computed at this operating point, which has large latency slack (τ=50 s vs ~8 s compute time) and strong LoS geometry — favorable conditions for MA gains to convert into CPU-frequency reductions.
axioms (7)
  • standard math Quadratic transform equivalence for sum-of-ratios minimization (Lemma 1)
    Invoked to reformulate (P3) into (P4); the proof is delegated to [43] and not re-derived.
  • standard math WMMSE rate transformation: r_m equals expression (22) when G_m, I_m follow (15)-(16)
    Basis for P4.1; standard from [44], but the paper's chain (22)-(23) is dense and the change of variables p^new_m=√p_m is asserted without guarding its domain.
  • domain assumption LoS-only free-space channel with path-loss exponent 2 and perfect CSI (Eqs. (1)-(2))
    The entire simulated MA benefit is produced under this model; no multipath, Rician fading, or channel-estimation error is considered anywhere in the paper.
  • domain assumption Linear movement model: E^mov_k=Ē|x_k−x0_k|, T^mov_k=|x_k−x0_k|/v, with move-then-transmit-then-compute sequential slots (Eqs. (8), (9d))
    The single-slot decomposition and max_k T^mov_k latency term assume antenna movement is strictly sequential to transmission and computation and costs linearly in distance.
  • domain assumption Full offload, identical task volume l per device, CPU energy ξc(f)^2 l (Sec. III-C)
    No local computing, homogeneous tasks, and a frequency-squared compute model; these shape the objective (P1) and the transmission-vs-computation trade-off the results rest on.
  • domain assumption UAV hovers at fixed horizontal position q; trajectory not part of the optimization (Sec. III-A)
    The UAV coordinate q is an input; macro-scale mobility, a major lever in the cited UAV-MEC literature, is excluded from the problem.
  • ad hoc to paper PSO rules (26)-(28) with penalty (29) produce a usable solution to (P5)
    The PSO procedure is explicitly 'omitted here' and deferred to Algorithm 3 in [45] (self-citation); no per-iteration decrease or feasibility guarantee, yet Proposition 1 relies on the X-update decreasing Φ.

reviewed 2026-08-01 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Movable-Antenna Assisted Energy Minimization in UAV-Enabled Mobile Edge Computing Systems." pith.science (2026). https://pith.science/paper/LPMLQMOY

@misc{pith2026260723214,
  author       = {Pith},
  title        = {Pith review of: Movable-Antenna Assisted Energy Minimization in UAV-Enabled Mobile Edge Computing Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LPMLQMOY}},
  note         = {Machine review of arXiv:2607.23214}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Driven by the exponential growth of latency-sensitive applications, mobile edge computing (MEC) has emerged as a pivotal paradigm, yet mitigating its substantial energy consumption remains critical. This paper explores a movable-antenna (MA) assisted energy minimization scheme in an uncrewed aerial vehicle (UAV)-enabled MEC system, where a UAV equipped with an MA array serves as an edge server to process tasks offloaded from terrestrial consumer electronics (CE) devices. To minimize the total system energy consumption, we jointly optimize computation resource allocation, CE transmit power, receive beamforming, and MA positions. To tackle the resulting non-convex problem with coupled variables, a robust alternating optimization algorithm based on the block coordinate descent method is developed. The problem is iteratively decomposed into three subproblems. In particular, the subproblem of transmit power and receive beamforming is reformulated and optimized using the quadratic transform technique, while the MA array positions are optimized via the particle swarm optimization algorithm. Numerical simulations verify that the proposed scheme achieves substantial energy savings over conventional benchmarks.

Figures

Figures reproduced from arXiv: 2607.23214 by Chengzhong Xu, Chunjie Wang, Jiang Chen, Kejiang Ye, Xuhui Zhang, Yanyan Shen.

Figure 1
Figure 1. Figure 1: The system model of the UAV-enabled MEC system. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The sequential steps of Algorithm 2. term into the fitness evaluation, which steers the optimization toward valid regions, i.e., δ(P u (i) ) = X M m=1 (E tran m (P u (i) ) + E com m (P u (i) ) +X K k=1 E mov k (P u (i) ) + ψ1|Ψ1(P u (i) )| + ψ2|Ψ2(P u (i) )|, (29) the fitness value of a particle, represented by δ(P u (i) ), incor￾porates penalty coefficients ψ1 and ψ2 to address constraint breaches. Specif… view at source ↗
Figure 3
Figure 3. Figure 3: The convergence performance. Algorithm 2 AO-based Algorithm for Solving Problem (P1). Input: An initial feasible solution {F(0) , P(0) ,W(0) , X(0) , }, 1: Initialize: iteration index l = 1, maximum iteration number lmax, accuracy threshold ε > 0, iteration index of the PSO i = 1, and maximum iteration number of the PSO imax; Compute the initial objective function value of problem P1, i.e.Φ(0)(F(0) , P(0) … view at source ↗
Figure 5
Figure 5. Figure 5: Minimum total energy consumption versus the computation intensity [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Minimum total energy consumption versus the maximum tolerable [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 9
Figure 9. Figure 9: Minimum total energy consumption versus the bandwidth for task [PITH_FULL_IMAGE:figures/full_fig_p009_9.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

45 extracted references · 2 linked inside Pith

  1. [1]

    Movable-antenna array empowered ISAC systems for low-altitude economy,

    Z. Kuang, W. Liu, C. Wang, Z. Jin, J. Ren, X. Zhang, and Y . Shen, “Movable-antenna array empowered ISAC systems for low-altitude economy,” inProc. IEEE/CIC Int. Conf. Commun. China Workshops (ICCC Workshops), Hangzhou, China, Aug. 2024, pp. 776–781

    2024

  2. [2]

    Low-altitude wireless networks: A survey,

    J. Wu, Y . Yang, W. Yuan, W. Liu, J. Wang, T. Mao, L. Zhou, Y . Cui, F. Liu, G. Sunet al., “Low-altitude wireless networks: A survey,”arXiv preprint arXiv:2509.11607, 2025

    Pith/arXiv arXiv 2025

  3. [3]

    Joint beamforming and resource coordination for RIS-aided ISAC over secure LAWNs,

    C. Wang, X. Zhang, W. Liu, J. Ren, H. Xing, S. Wang, Y . Shen, and K. Ye, “Joint beamforming and resource coordination for RIS-aided ISAC over secure LAWNs,”IEEE Trans. Consum. Electron., vol. 72, no. 2, pp. 3365–3377, May 2026. 11

    2026

  4. [4]

    Wireless human activity recognition: A survey of recent advances and a novel zero-shot LLM-based inference framework,

    Z. Xing, X. Zhang, W. Liu, H. Xing, W. Zhao, G. Zhang, S. Wang, Y . Shen, and K.-F. Tsang, “Wireless human activity recognition: A survey of recent advances and a novel zero-shot LLM-based inference framework,”IEEE Trans. Consum. Electron., to appear, 2026

    2026

  5. [5]

    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,” inProc. IEEE Int. Conf. CYBER Technol. Autom., Control, Intell. Syst. (CYBER), Qinhuangdao, China, Jul. 2023, pp. 836–841

    2023

  6. [6]

    From ground to sky: Architectures, applications, and challenges shaping low-altitude wireless networks,

    W. Yuan, Y . Cui, J. Wang, F. Liu, G. Sun, T. Xiang, J. Xu, S. Jin, D. Niy- ato, S. Coleriet al., “From ground to sky: Architectures, applications, and challenges shaping low-altitude wireless networks,”arXiv preprint arXiv:2506.12308, 2025

    Pith/arXiv arXiv 2025

  7. [7]

    Age of information minimization in UA V-enabled IoT networks via federated reinforcement learning,

    X. Zhang, H. Xing, Y . Shen, J. Xu, and S. Cui, “Age of information minimization in UA V-enabled IoT networks via federated reinforcement learning,”IEEE Trans. Wireless Commun., vol. 24, no. 9, pp. 7923–7939, Sep. 2025

    2025

  8. [8]

    Latency optimization for resource allocation in mobile-edge computation offloading,

    J. Ren, G. Yu, Y . Cai, and Y . He, “Latency optimization for resource allocation in mobile-edge computation offloading,”IEEE Trans. Wireless Commun., vol. 17, no. 8, pp. 5506–5519, Aug. 2018

    2018

  9. [9]

    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,” inProc. IEEE V eh. Technol. Conf., London, UK, Sep. 2022, pp. 1–7

    2022

  10. [10]

    Incentive-aware blockchain-assisted in- telligent edge caching and computation offloading for IoT,

    Q. Wang, S. Chen, and M. Wu, “Incentive-aware blockchain-assisted in- telligent edge caching and computation offloading for IoT,”Engineering, vol. 31, pp. 127–138, 2023

    2023

  11. [11]

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

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

    2020

  12. [12]

    AoI-aware UA V-enabled marine MEC networks with integrated sensing, computa- tion, and communication,

    W. Liu, Z. Jin, X. Zhang, W. Zang, S. Wang, and Y . Shen, “AoI-aware UA V-enabled marine MEC networks with integrated sensing, computa- tion, and communication,” inProc. IEEE/CIC Int. Conf. Commun. China Workshops (ICCC Workshops), Dalian, China, Aug. 2023, pp. 1–6

    2023

  13. [13]

    Joint 3-D trajectory and resource optimization in multi-UA V-enabled IoT networks with wireless power transfer,

    W. Luo, Y . Shen, B. Yang, S. Wang, and X. Guan, “Joint 3-D trajectory and resource optimization in multi-UA V-enabled IoT networks with wireless power transfer,”IEEE Internet Things J., vol. 8, no. 10, pp. 7833–7848, May 2021

    2021

  14. [14]

    Unmanned-aerial-vehicle- assisted wireless networks: Advancements, challenges, and solutions,

    M. Dai, N. Huang, Y . Wu, J. Gao, and Z. Su, “Unmanned-aerial-vehicle- assisted wireless networks: Advancements, challenges, and solutions,” IEEE Internet Things J., vol. 10, no. 5, pp. 4117–4147, Mar. 2023

    2023

  15. [15]

    Multiple UA Vs as relays: Multi-hop single link versus multiple dual-hop links,

    Y . Chen, N. Zhao, Z. Ding, and M.-S. Alouini, “Multiple UA Vs as relays: Multi-hop single link versus multiple dual-hop links,”IEEE Trans. Wireless Commun., vol. 17, no. 9, pp. 6348–6359, Sep. 2018

    2018

  16. [16]

    Resource and trajectory optimization for secure communications in dual unmanned aerial vehicle mobile edge computing systems,

    W. Lu, Y . Ding, Y . Gao, S. Hu, Y . Wu, N. Zhao, and Y . Gong, “Resource and trajectory optimization for secure communications in dual unmanned aerial vehicle mobile edge computing systems,”IEEE Trans. Ind. Informat., vol. 18, no. 4, pp. 2704–2713, Apr. 2022

    2022

  17. [17]

    Utility-efficient edge-cloud collaborative space-air-ground integrated internet of things,

    T. Zhang, Y . Wu, W. Liu, C. Wang, Z. Jin, X. Zhang, and Y . Shen, “Utility-efficient edge-cloud collaborative space-air-ground integrated internet of things,” inProc. IEEE Int. Conf. Commun. Technol. (ICCT), Chengdu, China, Oct. 2024, pp. 389–394

    2024

  18. [18]

    Computing over space: Status, challenges, and opportunities,

    Y . Liu, Y . Han, H. Li, S. Gu, J. Qiu, and T. Li, “Computing over space: Status, challenges, and opportunities,”Engineering, vol. 54, pp. 20–25, 2025

    2025

  19. [19]

    UA V-enabled fluid antenna systems for multi-target wireless sensing over LAWCNs,

    X. Zhang, W. Liu, C. Wang, J. Ren, H. Xing, S. Wang, and Y . Shen, “UA V-enabled fluid antenna systems for multi-target wireless sensing over LAWCNs,”IEEE Trans. Consum. Electron., vol. 72, no. 1, pp. 1146–1155, Feb. 2026

    2026

  20. [20]

    Fluid antenna system—part I: Preliminaries,

    K.-K. Wong, W. K. New, X. Hao, K.-F. Tong, and C.-B. Chae, “Fluid antenna system—part I: Preliminaries,”IEEE Commun. Lett., vol. 27, no. 8, pp. 1919–1923, Aug. 2023

    1919

  21. [21]

    Movable antennas for wireless commu- nication: Opportunities and challenges,

    L. Zhu, W. Ma, and R. Zhang, “Movable antennas for wireless commu- nication: Opportunities and challenges,”IEEE Commun. Mag., vol. 62, no. 6, pp. 114–120, Jun. 2024

    2024

  22. [22]

    Fluid antenna systems meet low-altitude wireless networks: Fundamentals, opportunities, and future directions,

    W. Liu, X. Zhang, C. Wang, J. Ren, W. Yuan, and C. You, “Fluid antenna systems meet low-altitude wireless networks: Fundamentals, opportunities, and future directions,”IEEE Internet Things Mag., vol. 9, no. 3, pp. 55–62, May 2026

    2026

  23. [23]

    Performance analysis and optimization for movable antenna aided wideband communications,

    L. Zhu, W. Ma, Z. Xiao, and R. Zhang, “Performance analysis and optimization for movable antenna aided wideband communications,” IEEE Trans. Wireless Commun., vol. 23, no. 12, pp. 18 653–18 668, Dec. 2024

    2024

  24. [24]

    QoS-aware integrated sensing, communication, and control with movable antenna,

    Y . Wang, Z. Wu, J. Chen, C. Wang, X. Zhang, and Y . Shen, “QoS-aware integrated sensing, communication, and control with movable antenna,” inProc. IEEE Int. Conf. Commun. Technol. (ICCT), Shenyang, China, Oct. 2025, pp. 565–570

    2025

  25. [25]

    UA V-enabled ISAC with fluid antennas for low-altitude wireless networks,

    W. Liu, X. Zhang, J. Ren, W. Yuan, C. You, and S. Li, “UA V-enabled ISAC with fluid antennas for low-altitude wireless networks,”IEEE Trans. Commun., vol. 74, pp. 9529–9546, 2026

    2026

  26. [26]

    Secure transmission for multi-UA V-assisted mobile edge computing based on reinforcement learning,

    W. Lu, Y . Mo, Y . Feng, Y . Gao, N. Zhao, Y . Wu, and A. Nallanathan, “Secure transmission for multi-UA V-assisted mobile edge computing based on reinforcement learning,”IEEE Trans. Netw. Sci. Eng., vol. 10, no. 3, pp. 1270–1282, May-Jun. 2023

    2023

  27. [27]

    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

  28. [28]

    Joint multi-domain resource allocation and trajectory optimization in UA V-assisted maritime IoT networks,

    L. P. Qian, H. Zhang, Q. Wang, Y . Wu, and B. Lin, “Joint multi-domain resource allocation and trajectory optimization in UA V-assisted maritime IoT networks,”IEEE Internet Things J., vol. 10, no. 1, pp. 539–552, Jan. 2023

    2023

  29. [29]

    Energy-efficient multi-UA V-enabled MEC systems over space- air-ground integrated networks,

    W. Liu, X. Zhang, C. Wang, J. Ren, Z. Xing, B. Yang, S. Wang, and Y . Shen, “Energy-efficient multi-UA V-enabled MEC systems over space- air-ground integrated networks,”IEEE Trans. V eh. Technol., to appear, 2026

    2026

  30. [30]

    Joint resource allocation and trajectory optimization for multi-UA V-assisted multi-access mobile edge computing,

    X. Qin, Z. Song, Y . Hao, and X. Sun, “Joint resource allocation and trajectory optimization for multi-UA V-assisted multi-access mobile edge computing,”IEEE Wireless Commun. Lett., vol. 10, no. 7, pp. 1400– 1404, Jul. 2021

    2021

  31. [31]

    Energy minimization for UA V-enabled covert offloading MEC systems,

    Y . Zhou, Z. Ma, P. Fan, A. Al-nahari, P. L. Yeoh, B. Vucetic, and Y . Li, “Energy minimization for UA V-enabled covert offloading MEC systems,”IEEE Trans. V eh. Technol., vol. 75, no. 2, pp. 3306–3311, Feb. 2026

    2026

  32. [32]

    Full-duplex integrated sensing, communication, and computation over low-altitude wireless networks,

    Y . Chen, W. Liu, X. Zhang, J. Ren, H. Xing, S. Wang, Y . Shen, and K.-F. Tsang, “Full-duplex integrated sensing, communication, and computation over low-altitude wireless networks,”IEEE Trans. Consum. Electron., vol. 72, no. 1, pp. 925–937, Feb. 2026

    2026

  33. [33]

    Movable antenna-enhanced wireless communications: General architec- tures and implementation methods,

    B. Ning, S. Yang, Y . Wu, P. Wang, W. Mei, C. Yuen, and E. Bj ¨ornson, “Movable antenna-enhanced wireless communications: General architec- tures and implementation methods,”IEEE Wireless Commun., vol. 32, no. 5, pp. 108–116, May 2025

    2025

  34. [34]

    Latency minimization for multi-AA V-enabled ISCC systems with movable an- tenna,

    Y . Chen, W. Liu, C. Wang, Y . Wu, X. Zhang, and Y . Shen, “Latency minimization for multi-AA V-enabled ISCC systems with movable an- tenna,” inProc. IEEE Int. Conf. Commun. Technol. (ICCT), Shenyang, China, Oct. 2025, pp. 1232–1238

    2025

  35. [35]

    Secure wireless communication via movable-antenna array,

    G. Hu, Q. Wu, K. Xu, J. Si, and N. Al-Dhahir, “Secure wireless communication via movable-antenna array,”IEEE Signal Process. Lett., vol. 31, pp. 516–520, 2024

    2024

  36. [36]

    Movable antenna enabled integrated sensing and communication,

    W. Lyu, S. Yang, Y . Xiu, Z. Zhang, C. Assi, and C. Yuen, “Movable antenna enabled integrated sensing and communication,”IEEE Trans. Wireless Commun., vol. 24, no. 4, pp. 2862–2875, Apr. 2025

    2025

  37. [37]

    UA V-enabled wireless networks with movable-antenna array: Flexible beamforming and trajectory design,

    W. Liu, X. Zhang, H. Xing, J. Ren, Y . Shen, and S. Cui, “UA V-enabled wireless networks with movable-antenna array: Flexible beamforming and trajectory design,”IEEE Wireless Commun. Lett., vol. 14, no. 3, pp. 566–570, Mar. 2025

    2025

  38. [38]

    Movable- antenna-enhanced wireless-powered mobile-edge computing systems,

    P. Chen, Y . Yang, B. Lyu, Z. Yang, and A. Jamalipour, “Movable- antenna-enhanced wireless-powered mobile-edge computing systems,” IEEE Internet Things J., vol. 11, no. 21, pp. 35 505–35 518, Nov. 2024

    2024

  39. [39]

    Fluid antenna for mobile edge computing,

    Y . Zuo, J. Guo, B. Sheng, C. Dai, F. Xiao, and S. Jin, “Fluid antenna for mobile edge computing,”IEEE Commun. Lett., vol. 28, no. 7, pp. 1728–1732, Jul. 2024

    2024

  40. [40]

    Movable-antenna array enhanced beam- forming: Achieving full array gain with null steering,

    L. Zhu, W. Ma, and R. Zhang, “Movable-antenna array enhanced beam- forming: Achieving full array gain with null steering,”IEEE Commun. Lett., vol. 27, no. 12, pp. 3340–3344, Dec. 2023

    2023

  41. [41]

    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,” inProc. IEEE/CIC Int. Conf. Commun. China Workshops (ICCC Workshops), Changchun, China, Aug. 2019, pp. 65–70

    2019

  42. [42]

    ETC: Enhanc- ing transportation computation with IRS-enabled wireless powered MEC systems,

    L. Xue, C. Wang, Y . Shen, S. Li, S. Wang, and X. Guan, “ETC: Enhanc- ing transportation computation with IRS-enabled wireless powered MEC systems,”IEEE Internet Things J., vol. 12, no. 19, pp. 40 837–40 850, Oct. 2025

    2025

  43. [43]

    Fractional programming for communication systems—part I: Power control and beamforming,

    K. Shen and W. Yu, “Fractional programming for communication systems—part I: Power control and beamforming,”IEEE Trans. Signal Process., vol. 66, no. 10, pp. 2616–2630, May 2018

    2018

  44. [44]

    Generalized MMSE beamform- ing for downlink MIMO systems,

    H. Lee, I. Sohn, D. Kim, and K. B. Lee, “Generalized MMSE beamform- ing for downlink MIMO systems,” inProc. IEEE Int. Conf. Commun. (ICC), Kyoto, Japan, Jul. 2011, pp. 1–6

    2011

  45. [45]

    Movable-antenna empowered AA V-enabled data collection over low- altitude wireless networks,

    X. Zhang, W. Liu, J. Ren, C. Wang, H. Xing, Y . Shen, and S. Cui, “Movable-antenna empowered AA V-enabled data collection over low- altitude wireless networks,”IEEE Trans. Netw. Sci. Eng., vol. 13, pp. 4506–4523, 2026

    2026

This paper was first reviewed by deepseek-v4-flash on August 1, 2026.