Sustainable Optimizing Performance and Energy Efficiency in Proof of Work Blockchain: A Multilinear Regression Approach

1st comment of the reviewer 1

There are some issues in the abstract i.e.,

i. The initial sentences of the abstract should give background of the study for example,

Provide more specifics on models and techniques,

ii. The abstract is too long in my opinion. Summarize the key contributions in two to three

sentences,

iii. Include quantitative performance metrics,

iv. Streamlining the writing to communicate the main ideas as concisely as possible.

As recommended, the abstract has been revised and edited.

2nd comment of the reviewer 1

Both primary objectives and main contributions in the introduction section are listed with

numbers, I will suggest to represent points with bullet point for clear readability.

As recommended, the introduction has been edited.

3rd comment of the reviewer 1

Please discuss potential challenges and limitations of implementing such a system, such as

issues with data accuracy and accessibility, and potential resistance to the use of Machine

Learning approaches in practical.

As recommended, the implication of the conclusion  has been revised and edited.

4th comment of the reviewer 1

Please improve the quality of figure 3 and figure 5.

As recommended, all figures were ensure quality of 600 dpi.

5th comment of the reviewer 1

The paper used “Multilinear Regression Analysis” and Blockchain framework. There are state-ofthe-art (SOTA) models to analyze datasets for example Multi-layer Perceptron Network (https://doi.org/10.3390/diagnostics12102539), and utilize Blockchain frameworks (10.1109/ACCESS.2020.2975233), for such kind of data, which has a simple structure and

demonstrates superior performance. Authors are requested to study such SOTA models and discuss the integration of AI and Blockchain for smart sustainable infrastructures.

Multilinear regression analysis is utilized in this study due to its notable capacity to elucidate the connections between variables, with a specific focus on GPU optimization in blockchain systems. This study's model focuses not on AI-factor analysis or prediction. Although state-of-the-art (SOTA) models, such as multi-layer perceptron networks, are robust in prediction tasks, they may not be as practical or indispensable for the study's research objectives as the explanatory model. However, additional descriptions were revised in the section 5.3.

6th comment of the reviewer 1

No complexity analysis is done to assess the computational overhead of the proposed approach.

Section 5.2 of the discussion was revised and edited with complexity analysis, as suggested.

7th comment of the reviewer 1

Multilinear regression basics are explained, but more justification for its suitability over other

methods could help.

A revised section 6.3 Future Work, a paragraph concerning alternative methods was appended.

8th comment of the reviewer 1

No evaluation on imbalanced or proprietary large-scale real-world datasets to truly test

capabilities.

As described in sections 3.2 and 3.3, this study employs and assesses real-world data relating to ETASH algorithms that mine a block of ETC coins. As a result, section 3.2 was revised to be more explicit and understandable. Consequently, the dataset was obtained from a proprietary real-world dataset of significant magnitude.

9th comment of the reviewer 1

No parameter tuning is reported to potentially improve the model's efficiency and effectiveness

further.

As described in sections 3.2 and 3.3, 3.4, 3.5, and 3.6, especially in 4.3 were demonstrated how features were optimized (tuned) with various settings. Then, multicollinearity analysis used to tune by reducing correlation reduction to select only the significant features regarding to energy and performance efficiency. Each test was tuned for the best VIF’s value of 1.000 for the best model.

10th comment of the reviewer 1

Please discuss in the methodology section that approach for aspect category label assignment is

manual or automatic?

As a result of the absence of prediction and AI-factor analysis, this study did not employ any category labeling techniques at all.

11th comment of the reviewer 1

Readability could be improved with more consistent verb tenses, fewer abbreviations, and use

of active voice. Usage of commas could be improved in a few long and complex sentences to

enhance readability.

In addition to a double-check, as the reviewer suggested, the manuscript was revised to incorporate two additional reviewer comments as well. MDPI Editing will also provide a response regarding the final publication.

12th comment of the reviewer 1

The conclusion section is too wide, please make “Limitations and Future work” as a separate

section before conclusion section. Conclusion shall be short, and precise including few sentences

from, “Theoretical Contributions”, “practical implications”, and “Limitations and Future work”.

Regarding the sustainability journal, the conclusion was composed in adherence to the style guide for writing. Furthermore, this will provide additional concluding arguments that are both comprehensive and distinct. Two additional evaluators proposed that the remains be maintained.

Memorandum of preliminary remarks

Comments #5: The paper used “Multilinear Regression Analysis” and Blockchain framework. There are state-ofthe-art (SOTA) models to analyze datasets for example Multi-layer Perceptron Network (https://doi.org/10.3390/diagnostics12102539), and utilize Blockchain frameworks (10.1109/ACCESS.2020.2975233), for such kind of data, which has a simple structure and demonstrates superior performance. Authors are requested to study such SOTA models and discuss the integration of AI and Blockchain for smart sustainable infrastructures.

Comments #7: Multilinear regression basics are explained, but more justification for its suitability over other methods could help.

Comment #8, #9, #10

SOTA's recommendation was in no way directly applicable to this investigation. Nonetheless, the suggestion could be amended and expanded for AI-feature configuration, and the reviewer would be well-advised to include it in Section 5.3. Deep neural networks (DNN), recognition, and recommendation systems have all implemented the SOTA.

We revised section 6.3 Future Work, a paragraph concerning alternative methods was appended.

The reviewer was required to have comprehended the work's methodology and necessity. This research analyzes and applies empirical data about ETASH algorithms on different GPU architectures with various configuration features, which are utilized to mine a block of ETC coins. This research illustrated how features were tuned (optimized) using a variety of parameters. Then, multicollinearity analysis is employed to select only the energy-efficient and performance-enhancing features that are significant through correlation reduction. Every test was optimized to achieve the ideal model's optimal VIF value of 1.000. This study was not the subject of the manuscript to make predictions. Due to the non-utilization of prediction and AI-factor analysis, no category labeling techniques were applied in this study. 

1st comment of the reviewer 1

There’s a huge gap in the claims in the abstract and rest of the paper. Please revise the abstract

to summarize the motivation (1-3 sentences), background (1-3 sentences), contribution/novelty

(1-3 sentences), quantitative/qualitative results (1-3 sentences) and concluding remarks (1-3

sentences).

The abstract has been revised and edited per the recommendation. However, the abstract was presented considerably more effectively than the reviewer recommended. Sentences one and two served as inspiration. The third and fourth sentences addressed novelty and objectives, respectively. The fifth and sixth sentences addressed the methodology and framework, respectively. Seventh through tenth sentences contained the results and findings. The final clause comprised the concluding remarks.

2nd comment of the reviewer 1

In the last review report, I suggested to discuss potential challenges and limitations of

implementing such a system, such as issues with data accuracy and accessibility, and potential

resistance to the use of Machine Learning approaches in practical. Subsection 6.3 “Limitation and

Future Work” is given in the conclusion section. Conclusion section looks like a chapter of a

student’s thesis. In principle, conclusion section must not be so long. I would once again request

authors to study suggested papers for example, (https://doi.org/10.3390/diagnostics12102539),

(10.1109/ACCESS.2020.2975233) to structure the paper carefully. Please avoid extra rigorous text

for better readability.

In the context of the 7th comment from the initial review, we have made revisions to section 6.3 Future Work by including an additional paragraph that addresses alternative methods. In the reviewer's opinion, the revised conclusion was compared to a "student's thesis," which also recommended the most applicable and comparable article. Nevertheless, the thesis style entails an error-free conclusion, and numerous articles have employed this style. Inaccurate or misleading academic information, methodology, analysis, data, outcome, or conclusion were absent. Diverse manifestations of writing styles comprise it. Furthermore, two reviewers expressed their concurrence with the amended particulars. Consequently, we adhered to most of the reviewing comments as much as possible if it made the manuscript accurate and appropriate.

3rd comment of the reviewer 1

I requested to add complexity analysis to assess the computational overhead of the proposed

approach. The authors have added text only describing the computational complexity analysis in

subsection 5.2, secondly the text is ambiguous e.g.,

First paragraph mentioned that this study conducted computational complexity analysis of the

ETASH mining software emphasizing on three elements, i.e., GPU Memory, energy power

consumption measured in mVDC, and overall electricity consumption measured in watts per hour.

Second paragraph explaining the same assessment parameters.

Third paragraph explaining the same assessment parameters with paraphrased text. Similarly,

fourth and fifth paragraphs explaining the computational complexity of ETASH mining software.

The basic question was, ETASH has its own computational complexity then how the authors can

claim novelty in this study? Please avoid extra text and link complexity analysis to the current

study.

A unique central idea was introduced in each paragraph. The introductory paragraph composed the initial portion. In the second paragraph, the computational space complexity was described. The third paragraph elaborated on the complexities of case limitations and capabilities. Energy and computation were illustrated in the fourth paragraph via various configuration complexities. The final paragraph elaborated on PoW's ecological friendliness. Thus, it is a distinct and precise detail.

The reviewer must fully understand the study's concept, primary areas of focus, the crucial PoW infrastructure, and GPU-relevant features for sustainable blockchain technology before conducting the initial evaluation (excluding prediction and recognition). Turnitin software for the manuscript furnished proof of 9% similarity with references. Billion upon billions of articles were dissimilar to this one. The remaining two reviewers expressed gratitude and comprehension, concurring with the contributions of this research to the field.

4th comment of the reviewer 1

This comment is not addressed: “In my opinion, the paper contains excessive extraneous material.

Please stick to the innovative points as much as possible.”

We have responded to 12 of the 13 comments that were initially reviewed, omitting the thirteenth comment due to its subjective nature (reviewer opinion). Since the new methodology, framework, and findings have been presented, it is imperative that readers comprehend and emulate the approach for alternative approaches, thus unnecessary material is required. Moreover, two reviewers concurred with the amended particulars. Consequently, we adhered to the majority of the reviewing comments.