Uncertainty-Aware Federated Reinforcement Learning for Optimizing Accuracy and Energy in Heterogeneous Industrial IoT
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper introduces an Uncertainty-Aware Federated Reinforcement Learning (UA-FedRL) method, but it would benefit from a clearer explanation of its novelty compared to existing techniques. Explicitly detailing the contribution and positioning the work within the broader context of FL and IoT research would strengthen the paper's impact.
The discussion on handling non-IID datasets could be more detailed. For instance, providing a mathematical formulation or more in-depth analysis of how UA-FedRL addresses data heterogeneity compared to other methods would make this section more robust.
While the paper mentions superior performance compared to other approaches, it would be beneficial to include a more comprehensive comparison with state-of-the-art techniques. Including a table that compares the performance of UA-FedRL with these methods across multiple metrics (e.g., accuracy, communication overhead, and computation time) would add depth.
The Predictive Weighted Average Aggregation (PWA) method is briefly mentioned. A more thorough explanation, including the mathematical model, rationale behind the weight adjustment, and the impact on overall model performance, would enhance the reader's understanding of its significance.
The paper relies on simulations for evaluation. It would be beneficial to provide more details about the simulation environment, including network settings, IoT device heterogeneity, and the distribution of stragglers. Additionally, discussing the potential challenges of implementing UA-FedRL in real-world IoT environments could make the research more applicable and relevant.
The paper mentions the UA-FedRL method's ability to handle heterogeneous devices but does not discuss its scalability in large-scale IoT networks. An analysis of how the method scales with an increasing number of devices or network complexity would provide a more comprehensive evaluation.
The paper reports the accuracy of the UA-FedRL on MNIST and CIFAR-10 datasets. Including a discussion on the statistical significance of these results, such as confidence intervals or p-values, would provide stronger evidence for the method's efficacy and reliability.
While the results are promising, it is important to acknowledge the potential limitations of the UA-FedRL method, such as its dependency on specific network conditions or types of data. Additionally, suggesting directions for future research, such as optimizing the PWA method or exploring other aggregation techniques, would show a forward-looking perspective and encourage further investigation.
Please avoid citing sources that were published before to 2019. Cite current research that are really pertinent to your topic. The study also lacks sufficient citations. Another critical step is to compare the topic of the article to other relevant recent publications or works in order to widen the research's repercussions beyond the issue. Authors can use and depend on these essential works while addressing the topic of their paper and current issues.
Zanbouri, Kouros, et al. "A GSO‐based multi‐objective technique for performance optimization of blockchain‐based industrial Internet of things." International Journal of Communication Systems: e5886.
Heidari, Arash, et al. "A reliable method for data aggregation on the industrial internet of things using a hybrid optimization algorithm and density correlation degree." Cluster Computing (2024): 1-19.
Comments on the Quality of English LanguageMinor editing of English language required.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper proposes a UA-FedRL method for IoT scenarios. Also, it introduces a PWA method for a proper weigth aggregation in IoT scenarios. The methodology is well described, as well as the experiments and their results.
Please improve figure 3. It is not posible to read the labels inside the figure. Also, figure 3 contains SDN technologies. What is its relation in your proposed method?
Check equation 21, is it correct?
Fix acronyms. For example DRL was defined in line 161, and its first appearance was in line 150.
In methodology section. Why the authors did not consider latency in communication problem formulation?
In Experiments section. What simulator was employed to emulate Raspberry Pi hardware characteristics?
What network parameters were configured in the mininet simulator? Network configuration is close related with SDN in figure 3?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsWell revised.
