Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript introduces a novel hybrid prediction-axiom dual-driven port selection algorithm tailored for Fluid Antenna Systems (FAS) in 6G high-mobility scenarios. This paper is interesting and this reviewer has the following comments:

1. In this paper, the port selection is considered for the channel estimation. Alternatively, the port selection can also be used for additional information transmission (e.g., index modulation [1], [2]), which is able to further reduce the outage probability (refer to the FAS with index modulation). To improve the novelty of the proposed method, it is better to introduce the index modulation related FAS systems ([1], [2], etc.) in Introduction, which is quite suitable for the proposed method.

[1] Index Modulation Multiple Access for 6G Communications: Principles, Applications, and Challenges, IEEE Network, 2023.

[2] A Survey on Spatial Modulation in Emerging Wireless Systems: Research Progresses and Applications, IEEE JSAC, 2019.

2. The description of the UAMA predictor and the information-theoretic scoring mechanism could include more detailed mathematical expressions and explanations to facilitate understanding.

3. It can be seen from Table 3 that the complexity of the proposed scheme is higher than other conventional schemes. The authors should elaborate how to reduce the computational complexity when N is a very large number.

4. Provide more in-depth discussion on why the proposed algorithm outperforms baseline methods (e.g., Figs. 4 and 5), particularly focusing on the trade-offs between outage probability, computational overhead, and robustness.

5. Please improve the resolutions of all figures.

Author Response

Comments 1: In this paper, the port selection is considered for the channel estimation. Alternatively, the port selection can also be used for additional information transmission (e.g., index modulation [1], [2]), which is able to further reduce the outage probability (refer to the FAS with index modulation). To improve the novelty of the proposed method, it is better to introduce the index modulation related FAS systems ([1], [2], etc.) in Introduction, which is quite suitable for the proposed method. 

[1] Index Modulation Multiple Access for 6G Communications: Principles, Applications, and Challenges, IEEE Network, 2023.

[2] A Survey on Spatial Modulation in Emerging Wireless Systems: Research Progresses and Applications, IEEE JSAC, 2019.

Response 1: Thank you for pointing this out. We concur with this observation. Consequently, we have cited these two articles, designated as [14] and [15] respectively. The amendments are located on page 2, lines 82–92 of the revised draft.

Comments 2: The description of the UAMA predictor and the information-theoretic scoring mechanism could include more detailed mathematical expressions and explanations to facilitate understanding.

Response 2: Thank you for pointing this out. We concur with this observation. Consequently, we have first revised the description of the UAMA, notably incorporating essential formulae to elucidate the methodology. These amendments are highlighted in red on pages 4, lines 175–242 of the revised manuscript. Secondly, to enhance comprehension of the information score mechanism, we have supplemented the analysis of parameter settings in Section 3. These modifications appear on pages 2, lines 269–280 and 307–320 of the revised manuscript. We have elaborated in detail on the conditions required for achieving optimal port selection (located on lines 321-343 of page 8 in the revised manuscript) and the operational process of the mechanism (located on lines 347-371 of page 8 in the revised manuscript).

Comments 3: It can be seen from Table 3 that the complexity of the proposed scheme is higher than other conventional schemes. The authors should elaborate how to reduce the computational complexity when N is a very large number.

Response 3: Thank you for pointing this out. We concur with this observation. Considering that this section of the discussion serves to introduce considerations regarding the future development of the algorithm, the amendments have been incorporated into the concluding section, located on page 18, lines 613–620 of the revised draft.

Comments 4: Provide more in-depth discussion on why the proposed algorithm outperforms baseline methods (e.g., Figs. 4 and 5), particularly focusing on the trade-offs between outage probability, computational overhead, and robustness.

Response 4: We are most grateful for your observation. We concur with this comment. The discussion pertaining to Figure 4 is located on lines 416–433 of page 10 in the revised manuscript, whilst the discussion concerning Figure 5 is situated on lines 438–455 of page 11 in the revised manuscript.

Comments 5: Please improve the resolutions of all figures.

Response 5: We unanimously concur with this suggestion and have consequently reprocessed all 13 images in the document to high-resolution format. You may view the revised images at any point within the draft.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Overall, the paper presents an interesting hybrid framework that combines channel prediction and information-theoretic scoring for port selection in fluid antenna systems, but several methodological and experimental issues limit the clarity, rigor, and practical impact of the work.

1. The paper lacks a rigorous theoretical analysis of the proposed information score, and it is unclear under what conditions maximizing this score approximates optimal SNR-based port selection.
2. The UAMA predictor is insufficiently described from a training perspective, with missing details on dataset generation, training procedure, convergence behavior, and generalization to unseen channel statistics.
3. The manuscript does not present three-dimensional radiation patterns of the antenna, which makes it difficult to fully understand and evaluate the radiation characteristics and spatial performance of the proposed design.
4. The discrepancies between simulated and measured results are not systematically analyzed, and the lack of quantitative error discussion weakens the credibility of the experimental validation.
5. More closely related defected-structure-based antenna designs should be discussed and compared, including those reported in doi:10.1109/OJAP.2023.3298773 and doi:10.1017/S1759078720001099.
6. The novelty and unique technical contributions of the work are not clearly and explicitly articulated, making it hard to distinguish the proposed approach from existing methods and prior art.

Author Response

Comments 1: The paper lacks a rigorous theoretical analysis of the proposed information score, and it is unclear under what conditions maximizing this score approximates optimal SNR-based port selection.

Response 1: Thank you for pointing this out. We concur with this observation and have incorporated a rigorous discussion of the theoretical analysis of the information score and the conditions for approximate optimality on pages 7, lines 321–343 of the revised manuscript. This clarifies that under the conditions of Gaussian prediction error distribution, wide-band channel stability, and ergodicity, maximising the information score approximates optimal SNR port selection.

Comments 2: The UAMA predictor is insufficiently described from a training perspective, with missing details on dataset generation, training procedure, convergence behavior, and generalization to unseen channel statistics.

Response 2: Thank you for pointing this out. We concur with this observation and have supplemented the revised manuscript on page 5, lines 207–242, with a detailed description of the UAMA predictor training perspective. This includes the dataset generation, training workflow, convergence behaviour, and generalisation capability to unseen channel statistics.

Comments 3: The manuscript does not present three-dimensional radiation patterns of the antenna, which makes it difficult to fully understand and evaluate the radiation characteristics and spatial performance of the proposed design.

Response 3: We are most grateful for your observations regarding the shortcomings in our MIMO simulation antenna design approach. We concur with this observation. Given that this paper primarily validates algorithmic performance rather than the antenna design itself, no additional three-dimensional diagrams have been included in this revision. However, to effectively characterise spatial characteristics, we have incorporated a two-dimensional radiation pattern diagram (Figure 9) covering four frequency points within the relevant band. Analysis of the radiation pattern has been enhanced on pages 14, lines 511–528 to elucidate spatial performance.

Comments 4: The discrepancies between simulated and measured results are not systematically analyzed, and the lack of quantitative error discussion weakens the credibility of the experimental validation.

Response 4: Thank you for pointing this out. We concur with this observation and acknowledge that the MIMO antenna design scheme (Figure 8) is currently constrained to HFSS simulation results without experimental validation. Therefore, to ensure the rigour and comprehensiveness of the validation methodology, Table 5 and a rigorous quantitative error analysis have been added to lines 550-576 on page 16 of the revised manuscript. This systematically compares the algorithm's performance across three antenna types, introducing the mean interruption probability, mean accuracy rate, and standard deviation of interruption probability as quantitative error metrics. Their calculation methods are detailed in the first paragraph beneath the table.

Comments 5: More closely related defected-structure-based antenna designs should be discussed and compared, including those reported in doi:10.1109/OJAP.2023.3298773 and doi:10.1017/S1759078720001099.

Response 5: Thank you for highlighting this point. We concur with this observation and agree that it is essential to cite and analyse these two papers to enrich our work. A discussion and comparison of the relevant defect structure antenna designs [19,20] has been incorporated on pages 17, lines 577–594 of the revised manuscript. This includes a brief analysis of their design advantages and indicates that future algorithms may be tested on similar systems.

Comments 6: The novelty and unique technical contributions of the work are not clearly and explicitly articulated, making it hard to distinguish the proposed approach from existing methods and prior art.

Response 6: Thank you for highlighting this point. We concur with this observation. Supplementary Table 4 provides a direct quantitative comparison with existing methods, located on page 12, lines 460–482 of the revised manuscript. This quantitative comparison more clearly demonstrates the innovation and unique technical contributions of the work. Finally, this comparison is also referenced in the conclusion section, on page 17, lines 600–605.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

1. It is unclear why the authors specifically consider measurements from 30% of the ports. Further clarification is required on how this measurement ratio is determined and whether it is fixed or adaptive depending on system conditions.

2. It would be beneficial to clarify the target application scenarios of the manuscript. Although the authors emphasize high-mobility environments, the simulations only consider a speed of 30 km/h, which may not fully represent highly dynamic scenarios.

3. Although the hybrid algorithm is addressed at the end of Section 3, a more explicit explanation of the overall procedure for solving the formulated problem would improve clarity. In particular, it appears that the solution is obtained using a greedy strategy; however, the explicit step-by-step procedure and its rationale should be more clearly described.

4. Related to the above comments, the adjustable parameters, such as the forgetting factor beta in (12) and the weighting parameter lambda in (16), are not sufficiently explained. The physical meaning, rationale, and criteria for selecting these parameters should be clarified.

5. As understood by the reviewer, the symbol lambda used in Section 3 denotes a different quantity from the lambda presented in Table 2 of Section 4.

6. Although several simulation results are presented in Section 4, no direct performance comparisons with existing or related methods are provided, making it difficult to assess the actual performance improvement of the proposed approach. Beyond the Big-O complexity analysis in Table 3, quantitative performance comparisons would strengthen the manuscript.

7. There are several minor typographical issues in the manuscript. For example, in the expression “fisher information matrix Fmeasures,” a space is missing after the symbol F.

Author Response

Comments 1: It is unclear why the authors specifically consider measurements from 30% of the ports. Further clarification is required on how this measurement ratio is determined and whether it is fixed or adaptive depending on system conditions.

Response 1: Thank you very much for pointing this out. We concur with this observation. The measurement ratio of 30% was selected to ensure a fair comparison with the baseline method [5]. This ratio is fixed; specifically, for a system with N=20 ports, the measurement port indices are predefined as {0, 3, 7, 11, 15, 19}, corresponding to uniform sampling at the 0%, 20%, 40%, 60%, 80%, and 100% positions on the antenna aperture. This fixed proportion ensures consistency with the benchmark settings in Reference [5] (see Table 3), thereby enabling a fair assessment of the proposed algorithm's effectiveness under sparse observation conditions. This has been explicitly clarified on lines 376–383 of page 9 in the revised manuscript.

Comments 2: It would be beneficial to clarify the target application scenarios of the manuscript. Although the authors emphasize high-mobility environments, the simulations only consider a speed of 30 km/h, which may not fully represent highly dynamic scenarios.

Response 2: Thank you for highlighting this point. We concur with this observation. 30 km/h represents a practical benchmark for typical urban mobility. We have clarified this choice and its limitations on lines 34–42 of page 1 and lines 400–412 of page 10 in the revised draft, noting that algorithmic design—such as Doppler-aware phase modelling—may support future extension to higher-speed scenarios.

Comments 3: Although the hybrid algorithm is addressed at the end of Section 3, a more explicit explanation of the overall procedure for solving the formulated problem would improve clarity. In particular, it appears that the solution is obtained using a greedy strategy; however, the explicit step-by-step procedure and its rationale should be more clearly described.

Response 3: We concur with this observation. The overall greedy strategy of the algorithm and its step-by-step procedure—encompassing sparse measurement, UAMA prediction, information score computation, and port selection—have been described with greater clarity on lines 347–371 of page 8 in the revised manuscript.

Comments 4: Related to the above comments, the adjustable parameters, such as the forgetting factor beta in (12) and the weighting parameter Y in (16), are not sufficiently explained. The physical meaning, rationale, and criteria for selecting these parameters should be clarified.

Response 4: Thank you for pointing this out. We concur with this observation. The physical significance, rationale for selection, and empirical basis for the Forgetting Factor β (Formula 12) and weighting parameter Y (formerly denoted as λ, now amended) have been supplemented with explanatory notes on pages 6 (lines 269–280) and 7 (lines 303–324) of the revised draft respectively.

Comments 5: As understood by the reviewer, the symbol lambda used in Section 3 denotes a different quantity from the lambda presented in Table 2 of Section.

Response 5: Thank you very much for pointing this out. To avoid confusion, the revised version now uniformly uses Y to denote the weighting parameter in the information score (page 7, line 304), while λ specifically refers to wavelength. This amendment has been applied throughout the entire text.

Comments 6: Although several simulation results are presented in Section 4, no direct performance comparisons with existing or related methods are provided, making it difficult to assess the actual performance improvement of the proposed approach. Beyond the Big-O complexity analysis in Table 3, quantitative performance comparisons would strengthen the manuscript.

Response 6: Thank you for pointing this out. We concur with this observation. Table 4 on page 12, lines 460–482 of the revised manuscript provides a quantitative performance comparison with the existing method [5], demonstrating specific improvements such as a two-order-of-magnitude reduction in the interruption probability gap. The addition of Table 5 on page 16, lines 550–594 further quantifies the algorithm's test results, thereby enhancing the credibility of the performance evaluation.

Comments 7: There are several minor typographical issues in the manuscript. For example, in the expression “fisher information matrix Fmeasures,” a space is missing after the symbol F.

Response 7: Thank you for pointing this out. We concur with this observation. We have carefully proofread the entire text and corrected the typesetting error in "fisher information matrix F-measures". This amendment is located on page 6, line 258 of the revised manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

This paper is well revised. No further comments.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been substantially revised and is now suitable for publication.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have carefully addressed the reviewer’s comments.
The reviewer has no further comments or questions.