Pith. sign in

REVIEW 4 major objections 6 minor 3 cited by

LLM-guided DRL for Multi-tier LEO Satellite Networks with Hybrid FSO/RF Links

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a truncated-quantile-critic DRL algorithm, augmented with dynamic action masking and LLM-driven hyperparameter tuning, jointly optimizes downlink rate and handover frequency in a LEO-HAP-ground network, cutting…

desk verdict A genuinely new combination of TQC, action masking, and LLM-driven hyperparameter tuning for satellite handover, but the reported gains rest on a dimensionally broken channel model and no statistics. read the letter →

arxiv 2505.11978 v1 pith:4TUMM6EG submitted 2025-05-17 cs.NI eess.SP

Jiahui Li , Geng Sun , Zemin Sun , Jiacheng Wang , Yinqiu Liu , Ruichen Zhang , Dusit Niyato , Shiwen Mao This is my paper · ORCID
classification cs.NIeess.SP
keywords LEOsatellitenetworkshybridFSO/RFlinkshandoverdeepreinforcementlearningtruncatedquantilecriticsdynamicactionmaskingLLM-guidedhyperparametertuninghigh-altitudeplatforms
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's aim is to make LEO-satellite internet work where terrestrial infrastructure is weak, by placing high-altitude platforms (HAPs) as relays and using free-space optical (FSO) links from satellites to the HAP and radio frequency (RF) links from the HAP to ground clusters. It casts the joint choice of satellite handover, subcarrier allocation, and user selection as a time-varying, non-convex optimization problem, then solves it online with a deep reinforcement learning agent. The agent is a truncated quantile critic (TQC) algorithm modified in two ways: it masks out LEO satellites that are not currently visible, and it lets a large language model (LLM) adjust training hyperparameters from recent reward history. In simulation, this LTQC-DAM agent converges faster than baselines and reduces handover frequency by 17.69% relative to standard TQC, with a 0.44% improvement in average downlink rate. A sympathetic reader would take the paper as evidence that meta-level LLM control of DRL hyperparameters can stabilize learning in a volatile networking environment, not as a claim about real-world deployment.

What carries the argument

The load-bearing object is TQC (truncated quantile critics), an off-policy distributional reinforcement learning method that represents the return distribution with quantile critics and discards the highest quantiles during backup to reduce overestimation bias. Around it the paper adds two mechanisms: a binary visibility mask $M_t$ that restricts the policy's output distribution to satellites visible at time $t$, converting a large invalid action space into a small valid one; and a meta-optimization function $F_{\mathrm{LLM}}$ that maps a window of recent episode rewards and normalized training progress to bounded updates of hyperparameters. The mask is what makes exploration efficient, and the LLM controller is what adapts training dynamics to reward volatility. The paper argues that these two additions are what let the agent learn from a reward function $r[t]=\eta R_{\mathrm{total}}(t)-\zeta\mathds{1}(s_t\neq s_{t+1})$ that mixes a noisy rate term with a handover penalty.

What would settle it

Correct Eq. (9) to use a proper RF free-space path-loss term and rerun the 1000-episode training; if the 17.69% handover advantage or the 0.44% rate advantage shrinks or reverses, the reported gains are artifacts of the unit error rather than properties of LTQC-DAM.

Watch

Extended reading notes

Core claim

The central claim is that combining dynamic action masking with LLM-guided hyperparameter tuning turns TQC into an algorithm that outperforms both value-based and actor-critic baselines in a simulated three-tier satellite downlink. The paper states that LTQC-DAM achieves a 0.44% improvement in the average downlink transmission rate $f_1$ and a 17.69% improvement in the handover-frequency objective $f_2$ compared with standard TQC, and that it converges faster than SAC, TD3, PPO, DQN, and TQC in the same environment. The mechanism is that visibility masks eliminate exploration of impossible satellite choices, while the LLM meta-controller adapts parameters such as learning rate, entropy coefficient, truncation parameter, and exploration decay in bounded steps based on a window of episode rewards. The paper also reports that the five LLMs it compares are not interchangeable: the best one works because it makes gradual, context-aware adjustments rather than large oscillatory ones.

Load-bearing premise

Every performance claim rests on the assumption that the simulated channel environment, configured with parameters from [15], is accurate enough that rankings inside the simulator transfer to real LEO-HAP-ground links; the paper's own Eq. (9) puts that assumption under strain by writing the RF path loss with the FSO wavelength and mixing a decibel loss with a linear fading variable.

Editorial extensions

If this is right

  • A single DRL policy can make satellite handover, subcarrier allocation, and user selection decisions online, without solving the non-convex joint problem from scratch at each time slot.
  • Restricting actions to visible satellites cuts wasted exploration, so the same masking logic should transfer to any LEO constellation where the visible-satellite set changes quickly.
  • If the simulation rankings carry over to deployment, operators can reduce handover-related overhead by roughly 18% while keeping downlink throughput essentially unchanged (0.44% gain).
  • Because the FSO feed rate bounds the total RF rate, handover decisions and subcarrier allocation must be optimized jointly; optimizing either tier alone would miss the bottleneck cascade the paper models.

Reading between the lines

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

  • Beyond the paper, ablating the two mechanisms separately would determine whether the 17.69% handover gain comes from masking, from LLM tuning, or from their interaction; the paper reports only the combined algorithm.
  • The paper's own numbers imply the LLM controller's main practical value is training stability and handover reduction, not raw throughput, since the rate gain is only 0.44%.
  • The LLM-guiding scheme is formulated generically over reward histories, so the same controller could tune SAC, TD3, or PPO; the paper demonstrates it only on TQC.
  • A cheap falsification of the LLM's causal role would be to compare LLM-guided tuning against a hand-designed annealing schedule with the same bounds; the paper does not include that baseline.
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 / 6 minor

Summary. The paper studies a three-tier LEO-HAP-ground downlink system in which LEO satellites connect to an HAP via FSO links and the HAP serves ground user clusters via OFDM-based RF links. The authors formulate a joint optimization problem over satellite selection, subcarrier allocation, and user selection that maximizes cumulative downlink rate and minimizes the number of satellite handovers. They propose LTQC-DAM, a truncated quantile critics (TQC) algorithm enhanced with dynamic action masking and LLM-driven adaptive hyperparameter tuning, and evaluate it in a custom simulator against SAC, TD3, PPO, DQN, and standard TQC. The paper reports faster convergence, a 0.44% improvement in the rate objective f1 and a 17.69% improvement in the handover objective f2 relative to standard TQC, and it identifies DeepSeek as the best-performing LLM controller among five tested models.

Significance. The problem is well motivated and the multi-tier architecture with hybrid FSO/RF links is relevant to current non-terrestrial network research. The MDP formulation and the dynamic action masking idea are clear and reasonable, and the complexity analysis is a useful addition. If the reported results were reproducible and statistically robust, the paper would provide a credible demonstration of LLM-guided hyperparameter adaptation in DRL for satellite networking. However, the physical layer model contains a dimensional inconsistency in the HAP-to-ground RF channel, and the evaluation lacks seeds, error bars, and ablations, so the central performance claims are not currently supported.

major comments (4)
  1. [Section III-D2, Eq. (9)] The HAP-to-ground RF path-loss coefficient C_HC,i(t) is written as G_HC + R_i + 1/2(20 lg lambda_F - 10 eta lg d_HC,i(t) - 20 lg 4 pi), which uses the FSO wavelength lambda_F in an RF link and adds dB-scale gain terms to a coefficient that Eq. (8) multiplies directly with a Nakagami fading variable and that Eq. (10) squares to compute rate. This is dimensionally inconsistent: dB quantities cannot be added to a linear coefficient and then multiplied, and using 1550 nm instead of an RF wavelength of roughly 0.1 m shifts the distance-dependent loss by about 50 dB. Because this coefficient enters R_RFi(t) and the reward r[t] in Eq. (20), the reported f1 and f2 margins may be artifacts of the model. The same issue appears in Eq. (5), where hl is a sum of dB gains/losses that is then multiplied by the linear fading term ha(t) in Eq. (4). Please correct both channel expressions (e.g., convert dB to linear units and use the RF wavelength in Eq. (9)) and rerun all simulations.
  2. [Section V-B, Figs. 3-4] The evaluation reports no random seeds, error bars, or statistical tests. In particular, the 0.44% improvement in f1 over standard TQC reported in Section V-B2 is very likely within run-to-run noise for a stochastic DRL algorithm trained in a simulator with random HAP mobility, channel fading, and user placement. Please report mean and standard deviation over at least 5-10 independent seeds and perform significance tests; without this, the central claim that LTQC-DAM outperforms all baselines is not established.
  3. [Section IV-C2 and Section V-C] The LLM-guided hyperparameter tuning is not ablated against a random scheduler or a simple non-LLM heuristic. All baselines use fixed hyperparameters, so the observed gains could come from any adaptive adjustment rather than from LLM reasoning. Additionally, the selection of DeepSeek as 'best' is made post hoc from the same convergence curves in Fig. 3(c) without hold-out evaluation. Please add ablations that isolate each contribution: (i) LTQC with random hyperparameter schedules sampled from the same bounds, (ii) LTQC with a hand-coded rule-based scheduler, and (iii) LTQC-DAM without action masking.
  4. [Section IV-A3, Eqs. (19)-(20)] The action a[t] specifies st+1, n(t+1), and u(t+1), but the reward r[t] is defined in terms of R_total(t) and I(st != st+1). Unless R_total(t) is understood as the rate achieved after applying a[t], the agent is rewarded for the previous time slot's rate together with the current handover decision, creating an off-by-one coupling between actions and rewards. Please clarify the intended causal ordering or redefine the reward using R_total(t+1).
minor comments (6)
  1. [Section V-B2] The phrase '17.69% improvement in f2' should state the direction explicitly; since f2 is a handover count to be minimized, the figure is presumably a reduction in handover frequency, not an increase.
  2. [Section III-D2, Eq. (9)] Use lambda_RF rather than lambda_F in the RF path-loss expression and write 20 log10(...) instead of 20 lg(...) to make the expression unambiguous.
  3. [Section V-A1] Please provide the complete LLM prompt template, API settings (e.g., temperature, max tokens), and the number of LLM calls per training run; without these details the LLM-tuning component is not reproducible.
  4. [Algorithm 1, line 22] The reward history window H_e = {r[e-k], ..., r[e]} is undefined for e < k; specify how the window is initialized in the early episodes.
  5. [Section V-C, Fig. 5] The explanation that DeepSeek is best because of 'no auxiliary loss load balancing' and 'multi-head latent attention' is not supported by any quantitative analysis in the paper; either remove this speculation or substantiate it with ablations.
  6. [References, [45]] Reference [45] is not the original TD3 paper; cite Fujimoto et al., 'Addressing Function Approximation Error in Actor-Critic Methods,' ICML 2018, or justify the alternative citation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: LTQC-DAM is evaluated on the same simulator used for training, but no predicted quantity is equal to an input by construction.

full rationale

The paper's derivation chain is self-contained in the relevant sense. The MDP formulation (Eqs. (18)-(20)) defines state, action, and reward directly from problem (P1); the proposed algorithm combines standard TQC [37] with a dynamic visibility mask (Eqs. (24)-(27)) and an LLM meta-controller (Eqs. (28)-(30)). None of these steps defines an output in terms of the quantity it is later claimed to predict: hyperparameters are tuned online from reward history and then the converged cumulative reward is reported, which is standard DRL evaluation rather than a fitted-parameter-renamed-prediction. The channel models are imported from external references [15] and [33], and the only overlapping-author citations ([32], [34], [35]) support mobility-model background, DRL background, and complexity accounting; none is load-bearing as a uniqueness theorem or as an ansatz that forces the central result. The DeepSeek comparison is a post-hoc observation from the same simulation runs, not a derived prediction, so it carries at most hindsight bias, not circularity. The dimensional inconsistency in Eq. (9), where the RF loss coefficient uses the FSO wavelength and mixes dB with linear fading, is a channel-model validity risk that would affect all compared algorithms; it is a correctness concern, not a circularity in the derivation chain. Thus no circular step is identifiable from the paper's own equations and citations.

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

The central claims rest on standard stochastic channel models and an unvalidated simulator; the main custom choices are the reward weights and hyperparameter schedules, which are tuned by hand or by the LLM. No physical entities are introduced, so the invented-entity burden is zero. The chief unstated cost is that all numerical results depend on a simulator whose fidelity is not demonstrated.

free parameters (4)
  • Reward weights eta and zeta = not reported
    The reward in Eq. (20) is a weighted sum of rate and handover penalty; without eta and zeta, the reported handover reduction is not reproducible, and the values affect the f1/f2 trade-off.
  • Initial TQC hyperparameters (learning rate, discount, truncation k, soft update, decay schedule) = 1e-4, 0.999, 2, 0.005, 0.3
    Hand-set in Section V-A and then modified by the LLM; performance depends on these starting values.
  • Exploration noise parameters epsilon0 and e_decay = not fully reported; decay 0.3 mentioned
    The epsilon schedule in Eq. (26) governs exploration; epsilon0 is never given.
  • LLM hyperparameter bounds [theta_min, theta_max] = not reported
    The LLM can only adjust parameters within bounds (Eq. (30)); these bounds are essential to reproduce the tuning behavior.
assumptions (5)
  • domain assumption Gamma-Gamma fading model is valid for LEO-to-HAP FSO links
    Adopted from [15] in Eq. (4); no validation for the 500-1000 km link geometry in this paper.
  • domain assumption Nakagami-m fading is valid for all HAP-to-ground cluster links
    Used in Eq. (8) for every cluster; no dependence on terrain or weather is modeled.
  • domain assumption Gauss-Markov mobility model captures real HAP drift
    Section III-C uses it with parameters alpha, mu_v, sigma that are not fit to HAP wind data.
  • ad hoc to paper The training simulator is a faithful proxy for the real system
    All experimental claims in Section V come from this simulator; there is no field data or hardware testbed.
  • standard math Two-body circular-orbit Keplerian dynamics are sufficient
    Section III-B assumes circular orbits with epsilon=0; standard idealization for LEO constellation studies.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLM-guided DRL for Multi-tier LEO Satellite Networks with Hybrid FSO/RF Links." pith.science (2026). https://pith.science/paper/4TUMM6EG

@misc{pith2026250511978,
  author       = {Pith},
  title        = {Pith review of: LLM-guided DRL for Multi-tier LEO Satellite Networks with Hybrid FSO/RF Links},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4TUMM6EG}},
  note         = {Machine review of arXiv:2505.11978}
}
read the original abstract

Despite significant advancements in terrestrial networks, inherent limitations persist in providing reliable coverage to remote areas and maintaining resilience during natural disasters. Multi-tier networks with low Earth orbit (LEO) satellites and high-altitude platforms (HAPs) offer promising solutions, but face challenges from high mobility and dynamic channel conditions that cause unstable connections and frequent handovers. In this paper, we design a three-tier network architecture that integrates LEO satellites, HAPs, and ground terminals with hybrid free-space optical (FSO) and radio frequency (RF) links to maximize coverage while maintaining connectivity reliability. This hybrid approach leverages the high bandwidth of FSO for satellite-to-HAP links and the weather resilience of RF for HAP-to-ground links. We formulate a joint optimization problem to simultaneously balance downlink transmission rate and handover frequency by optimizing network configuration and satellite handover decisions. The problem is highly dynamic and non-convex with time-coupled constraints. To address these challenges, we propose a novel large language model (LLM)-guided truncated quantile critics algorithm with dynamic action masking (LTQC-DAM) that utilizes dynamic action masking to eliminate unnecessary exploration and employs LLMs to adaptively tune hyperparameters. Simulation results demonstrate that the proposed LTQC-DAM algorithm outperforms baseline algorithms in terms of convergence, downlink transmission rate, and handover frequency. We also reveal that compared to other state-of-the-art LLMs, DeepSeek delivers the best performance through gradual, contextually-aware parameter adjustments.

Figures

Figures reproduced from arXiv: 2505.11978 by the authors.

Figure 1
Figure 1. The considered multi-tier hybrid satellite downlink communication [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Framework of the LTQC-DAM algorithm for the considered multi-tier hybrid satellite downlink communication system. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Convergence comparison of different algorithms. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Average optimization objective performance comparison of different [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: Visualization of hyperparameter adaptation patterns across different [PITH_FULL_IMAGE:figures/full_fig_p012_5.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. LLM-Aided Joint Secrecy Precoding and Trajectory for RSMA-Based Heterogeneous UAV Networks

    cs.NI 2025-07 conditional novelty 5.0 of 10

    A hierarchical framework combining semidefinite relaxation for secure precoding with LLM-guided multi-agent reinforcement learning for drone trajectories is claimed to outperform existing baselines in secrecy rate and...

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

    cs.NI 2025-09 conditional novelty 4.0 of 10

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

  3. Toward Edge General Intelligence with Agentic AI and Agentification: Concepts, Technologies, and Future Directions

    cs.NI 2025-08 conditional novelty 4.0 of 10

    A survey that organizes agentic AI for 6G edge networks into four pillars, compactness, efficiency, knowledge and reasoning, and migration, and illustrates them with prior case studies.

Reference graph

Works this paper leans on

47 extracted references · 38 canonical work pages · cited by 3 Pith papers

  1. [1]

    Sustainable UA V mobility support in integrated terrestrial and non-terrestrial networks,

    F. Wang, S. Zhang, J. Shi, Z. Li, and T. Q. S. Quek, “Sustainable UA V mobility support in integrated terrestrial and non-terrestrial networks,” IEEE Trans. Wirel. Commun. , vol. 23, no. 11, pp. 17 115–17 128, 2024

    work page 2024

  2. [2]

    Modeling and anal- ysis of non-terrestrial networks by spherical stochastic geometry: A survey,

    R. Wang, M. A. Kishk, and M.-S. Alouini, “Modeling and anal- ysis of non-terrestrial networks by spherical stochastic geometry: A survey,” IEEE Commun. Surv. Tutor . , 2025, early access, doi: 10.1109/COMST.2025.3562489

    arXiv 2025

  3. [3]

    Modeling and analysis of downlink communications in a heterogeneous LEO satellite network,

    C. Choi, “Modeling and analysis of downlink communications in a heterogeneous LEO satellite network,” IEEE Trans. Wirel. Commun. , vol. 23, no. 8, pp. 8588–8602, 2024

    work page 2024

  4. [4]

    RIS-empowered LEO satellite networks for 6G: Promising usage scenarios and future directions,

    M. Toka, B. Lee, J. Seong, A. Kaushik, J. Lee, J. Lee, N. Lee, W. Shin, and H. V . Poor, “RIS-empowered LEO satellite networks for 6G: Promising usage scenarios and future directions,” IEEE Commun. Mag., vol. 62, no. 11, pp. 128–135, 2024

    work page 2024

  5. [5]

    Communication-efficient federated learning for LEO constellations integrated with HAPs using hybrid NOMA-OFDM,

    M. Elmahallawy, T. Luo, and K. Ramadan, “Communication-efficient federated learning for LEO constellations integrated with HAPs using hybrid NOMA-OFDM,” IEEE J. Sel. Areas Commun. , vol. 42, no. 5, pp. 1097–1114, 2024

    work page 2024

  6. [6]

    Performance analysis of mixed FSO/RF system for satellite-terrestrial relay network,

    Q. Sun, Q. Hu, Y . Wu, X. Chen, J. Zhang, and M. L ´opez-Ben´ıtez, “Performance analysis of mixed FSO/RF system for satellite-terrestrial relay network,” IEEE Trans. V eh. Technol., vol. 73, no. 8, pp. 11 378– 11 393, 2024

    work page 2024

  7. [7]

    Space-air-ground FSO networks for high-throughput satellite communications,

    R. Samy, H. Yang, T. Rakia, and M. Alouini, “Space-air-ground FSO networks for high-throughput satellite communications,” IEEE Commun. Mag., vol. 61, no. 3, pp. 82–87, 2023

    work page 2023

  8. [8]

    Accelerating handover in mobile satellite network,

    J. Wu, S. Su, X. Wang, J. Zhang, and Y . Gao, “Accelerating handover in mobile satellite network,” in Proc. IEEE INFOCOM, 2024, pp. 531–540

    work page 2024

Show all 47 references
  1. [9]

    Enhancing robust- ness in hybrid FSO/RF systems: A feedback-free approach leveraging deep learning for real-time power allocation,

    H. Zeng, H. Wang, K. Wang, X. Yu, and Z. Zhang, “Enhancing robust- ness in hybrid FSO/RF systems: A feedback-free approach leveraging deep learning for real-time power allocation,” IEEE Internet Things J. , 2025, early access, doi: 10.1109/jiot.2025.3556401

    2025

  2. [10]

    Free space optical semantic communication for satellite remote sens- ing image transmission,

    W. Chen, C. Ju, T. Yuan, Y . Zhan, M. Zhang, and D. Wang, “Free space optical semantic communication for satellite remote sens- ing image transmission,” IEEE Trans. Commun. , 2025, early access, doi:10.1109/TCOMM.2025.3562356

    2025

  3. [11]

    Multi-layer airborne FSO systems: Per- formance analysis and optimization,

    M. Elamassie and M. Uysal, “Multi-layer airborne FSO systems: Per- formance analysis and optimization,” IEEE Trans. Commun. , vol. 73, no. 4, pp. 2522–2537, 2025

    2025

  4. [12]

    Blind reconciliation with protograph LDPC code extension for FSO- based satellite QKD systems,

    C. T. Nguyen, H. D. Le, V . V . Mai, P. V . Trinh, and A. T. Pham, “Blind reconciliation with protograph LDPC code extension for FSO- based satellite QKD systems,” in Proc. IEEE CSNDSP, 2024, pp. 17–22

    2024

  5. [13]

    FSO-based space-air-ground integrated vehicular networks: Cooperative HARQ with rate adaptation,

    H. D. Le, H. D. Nguyen, C. T. Nguyen, and A. T. Pham, “FSO-based space-air-ground integrated vehicular networks: Cooperative HARQ with rate adaptation,” IEEE Trans. Aerosp. Electron. Syst., vol. 59, no. 4, pp. 4076–4091, 2023

    2023

  6. [14]

    On the design of FSO-based satellite systems using incremental redundancy hybrid ARQ protocols with rate adaptation,

    H. D. Le and A. T. Pham, “On the design of FSO-based satellite systems using incremental redundancy hybrid ARQ protocols with rate adaptation,” IEEE Trans. V eh. Technol. , vol. 71, no. 1, pp. 463–477, 2022

    2022

  7. [15]

    Deep reinforcement learning-based energy efficiency optimization for ris- aided integrated satellite-aerial-terrestrial relay networks,

    M. Wu, K. Guo, X. Li, Z. Lin, Y . Wu, T. A. Tsiftsis, and H. Song, “Deep reinforcement learning-based energy efficiency optimization for ris- aided integrated satellite-aerial-terrestrial relay networks,” IEEE Trans. Commun., vol. 72, no. 7, pp. 4163–4178, 2024

    2024

  8. [16]

    Hybrid SAG-FSO/SH- FSO/RF transmission for next-generation satellite communication sys- tems,

    R. Samy, H. Yang, T. Rakia, and M. Alouini, “Hybrid SAG-FSO/SH- FSO/RF transmission for next-generation satellite communication sys- tems,” IEEE Trans. V eh. Technol. , vol. 72, no. 11, pp. 14 255–14 267, 2023

    2023

  9. [17]

    Adaptive-combining-based hybrid FSO/RF satellite com- munication with and without HAPS,

    S. Shah, M. Siddharth, N. Vishwakarma, R. Swaminathan, and A. S. Madhukumar, “Adaptive-combining-based hybrid FSO/RF satellite com- munication with and without HAPS,” IEEE Access , vol. 9, pp. 81 492– 81 511, 2021

    2021

  10. [18]

    Uplink massive access in mixed RF/FSO satellite-aerial-terrestrial networks,

    Q. Huang, M. Lin, W.-P. Zhu, J. Cheng, and M.-S. Alouini, “Uplink massive access in mixed RF/FSO satellite-aerial-terrestrial networks,” IEEE Trans. Commun. , vol. 69, no. 4, pp. 2413–2426, 2021

    2021

  11. [19]

    On secure uplink transmission in hybrid RF-FSO cooperative satellite-aerial-terrestrial networks,

    Y . Ma, T. Lv, G. Pan, Y . Chen, and M. Alouini, “On secure uplink transmission in hybrid RF-FSO cooperative satellite-aerial-terrestrial networks,” IEEE Trans. Commun., vol. 70, no. 12, pp. 8244–8257, 2022

    2022

  12. [20]

    RIS assisted UA V for weather-dependent satellite terrestrial integrated network with hybrid FSO/RF systems,

    X. Li, Y . Li, X. Song, L. Shao, and H. Li, “RIS assisted UA V for weather-dependent satellite terrestrial integrated network with hybrid FSO/RF systems,” IEEE Photonics J. , vol. 15, no. 5, pp. 1–17, 2023

    2023

  13. [21]

    Physical layer se- curity of HAPS-based space–air–ground-integrated network with hybrid FSO/RF communication,

    V . Bankey, S. Sharma, S. R, and A. S. Madhukumar, “Physical layer se- curity of HAPS-based space–air–ground-integrated network with hybrid FSO/RF communication,” IEEE Trans. Aerosp. Electron. Syst. , vol. 59, no. 4, pp. 4680–4688, 2023

    2023

  14. [22]

    On the design of RIS–UA V relay-assisted hybrid FSO/RF satellite–aerial–ground integrated net- work,

    T. V . Nguyen, H. D. Le, and A. T. Pham, “On the design of RIS–UA V relay-assisted hybrid FSO/RF satellite–aerial–ground integrated net- work,” IEEE Trans. Aerosp. Electron. Syst. , vol. 59, no. 2, pp. 757–771, 2023

    2023

  15. [23]

    Reliable terabits feeder link for very high-throughput satellite systems with SAG-FSO transmission,

    R. Samy, H.-C. Yang, T. Rakia, and M.-S. Alouini, “Reliable terabits feeder link for very high-throughput satellite systems with SAG-FSO transmission,” IEEE Wireless Commun. , vol. 31, no. 2, pp. 112–116, 2024

    2024

  16. [24]

    Integrating LEO satellites and multi-UA V reinforcement learning for hybrid FSO/RF non-terrestrial networks,

    J.-H. Lee, J. Park, M. Bennis, and Y .-C. Ko, “Integrating LEO satellites and multi-UA V reinforcement learning for hybrid FSO/RF non-terrestrial networks,” IEEE Trans. V eh. Technol. , vol. 72, no. 3, pp. 3647–3662, 2023

    2023

  17. [25]

    Parallel FSO-RF transmissions for high-throughput remote access with satellite commu- nications,

    R. Samy, H.-C. Yang, T. Rakia, and M.-S. Alouini, “Parallel FSO-RF transmissions for high-throughput remote access with satellite commu- nications,” IEEE Trans. Aerosp. Electron. Syst. , vol. 59, no. 6, pp. 9417– 9426, 2023

    2023

  18. [26]

    Effect of correlated turbulence on integrated SAG-FSO/SH-FSO/RF transmission for satellite communications,

    R. Samy, H. Ahmed, H.-C. Yang, and M.-S. Alouini, “Effect of correlated turbulence on integrated SAG-FSO/SH-FSO/RF transmission for satellite communications,” in Proc. IEEE PIMRC , 2024, pp. 1–6

    2024

  19. [27]

    Anticipating optical availability in hybrid RF/FSO links using RF beacons and deep learning,

    M. Ibrahim, A. Ahmad, S. Ekin, P. LoPresti, S. Altunc, O. Kegege, and J. F. O’Hara, “Anticipating optical availability in hybrid RF/FSO links using RF beacons and deep learning,” IEEE Trans. Mach. Learn. Commun. Netw., vol. 2, pp. 1369–1388, 2024

    2024

  20. [28]

    Latency versus reliability in LEO mega-constellations: Terrestrial, aerial, or space relay?

    G. Pan, J. Ye, J. An, and M. Alouini, “Latency versus reliability in LEO mega-constellations: Terrestrial, aerial, or space relay?” IEEE Trans. Mob. Comput., vol. 22, no. 9, pp. 5330–5345, 2023

    2023

  21. [29]

    Satellite orbits: Models, methods, and applications,

    O. Montenbruck, E. Gill, and F. Lutze, “Satellite orbits: Models, methods, and applications,” Appl. Mech. Rev , vol. 55, no. 2, pp. B27– B28, 2002

    2002

  22. [30]

    Ultra-dense LEO satellite constellations: How many LEO satellites do we need?

    R. Deng, B. Di, H. Zhang, L. Kuang, and L. Song, “Ultra-dense LEO satellite constellations: How many LEO satellites do we need?” IEEE Trans. Wirel. Commun. , vol. 20, no. 8, pp. 4843–4857, 2021

    2021

  23. [31]

    Combined control of coverage area and HAPS deployment in hy- brid FSO/RF SAGIN,

    K. Mashiko, Y . Kawamoto, N. Kato, K. Yoshida, and M. Ariyoshi, “Combined control of coverage area and HAPS deployment in hy- brid FSO/RF SAGIN,” IEEE Trans. V eh. Technol., 2025, early access, doi:10.1109/TVT.2025.3546335

    2025

  24. [32]

    Aerial reliable collaborative communications for terrestrial mobile users via evolutionary multi-objective deep reinforcement learning,

    G. Sun, J. Xiao, J. Li, J. Wang, J. Kang, D. Niyato, and S. Mao, “Aerial reliable collaborative communications for terrestrial mobile users via evolutionary multi-objective deep reinforcement learning,” IEEE Trans. Mob. Comput., 2025, early access, doi:10.1109/TMC.2025.3536093

    2025

  25. [33]

    Covert communi- cations in STAR-RIS assisted NOMA IoT networks over nakagami-m fading channels,

    Q. Li, D. Xu, K. Zhang, K. Navaie, and Z. Ding, “Covert communi- cations in STAR-RIS assisted NOMA IoT networks over nakagami-m fading channels,” IEEE Internet Things J. , vol. 11, no. 12, pp. 22 456– 22 470, 2024

    2024

  26. [34]

    Collaborative ground-space communications via evolutionary multi-objective deep reinforcement learning,

    J. Li, G. Sun, Q. Wu, D. Niyato, J. Kang, A. Jamalipour, and V . C. M. Leung, “Collaborative ground-space communications via evolutionary multi-objective deep reinforcement learning,” IEEE J. Sel. Areas Com- mun., vol. 42, no. 12, pp. 3395–3411, 2024

    2024

  27. [35]

    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. , vol. 24, no. 4, pp. 3041–3058, 2025

    2025

  28. [36]

    Quantum-enhanced DRL optimization for DoA estimation and task offloading in ISAC systems,

    A. Paul, K. Singh, A. Kaushik, C. Li, O. A. Dobre, M. D. Renzo, and T. Q. Duong, “Quantum-enhanced DRL optimization for DoA estimation and task offloading in ISAC systems,” IEEE J. Sel. Areas Commun. , vol. 43, no. 1, pp. 364–381, 2025

    2025

  29. [37]

    Controlling overestimation bias with truncated mixture of continuous distributional quantile critics,

    A. Kuznetsov, P. Shvechikov, A. Grishin, and D. P. Vetrov, “Controlling overestimation bias with truncated mixture of continuous distributional quantile critics,” in Proc. ICML, vol. 119, 2020, pp. 5556–5566

    2020

  30. [38]

    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

  31. [39]

    Truncated quantile critics algorithm for cryptocurrency portfolio optimization,

    L. Xiao, X. Wei, Y . Xu, X. Xu, K. Gong, H. Li, and F. Zhang, “Truncated quantile critics algorithm for cryptocurrency portfolio optimization,” in Proc. IEEE SMC , 2023, pp. 2208–2213

    2023

  32. [40]

    A closer look at invalid action masking in policy gradient algorithms,

    S. Huang and S. Onta ˜n´on, “A closer look at invalid action masking in policy gradient algorithms,” in Proc. FLAIRS, 2022

    2022

  33. [41]

    Meta-reinforcement learning in non-stationary and dynamic environments,

    Z. Bing, D. Lerch, K. Huang, and A. C. Knoll, “Meta-reinforcement learning in non-stationary and dynamic environments,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 45, no. 3, pp. 3476–3491, 2023

    2023

  34. [42]

    Large language models empower multimodal integrated sensing and communication,

    L. Cheng, H. Zhang, B. Di, D. Niyato, and L. Song, “Large language models empower multimodal integrated sensing and communication,” IEEE Commun. Mag. , vol. 63, no. 5, pp. 190–197, 2025

    2025

  35. [43]

    Large language models for networking: Applications, enabling techniques, and challenges,

    Y . Huang, H. Du, X. Zhang, D. Niyato, J. Kang, Z. Xiong, S. Wang, and T. Huang, “Large language models for networking: Applications, enabling techniques, and challenges,” IEEE Netw. , vol. 39, no. 1, pp. 235–242, 2025

    2025

  36. [44]

    Downlink coverage and rate analysis of low earth orbit satellite con- stellations using stochastic geometry,

    N. Okati, T. Riihonen, D. Korpi, I. Angervuori, and R. Wichman, “Downlink coverage and rate analysis of low earth orbit satellite con- stellations using stochastic geometry,” IEEE Trans. Commun. , vol. 68, no. 8, pp. 5120–5134, 2020

    2020

  37. [45]

    Twin-delayed DDPG: A deep reinforcement learning technique to model a continuous movement of an intelligent robot agent,

    S. Dankwa and W. Zheng, “Twin-delayed DDPG: A deep reinforcement learning technique to model a continuous movement of an intelligent robot agent,” in Proc. ICVISP, 2019, pp. 1–5

    2019

  38. [46]

    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

  39. [47]

    A theoretical analysis of deep Q-Learning,

    J. Fan, Z. Wang, Y . Xie, and Z. Yang, “A theoretical analysis of deep Q-Learning,” in Proc. L4DC, vol. 120, 2020, pp. 486–489

    2020

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.