Hybrid Nearest-Neighbor Ant Colony Optimization Algorithm for Enhancing Load Balancing Task Management
Round 1
Reviewer 1 Report
This work deals with load balancing in distributed computing systems to enhance load scheduling mechanism based on Ant Colony Optimization. The proposed ACO-based approach rejects dissatisfied requests before scheduling, reducing the solution dimensions of the nearest neighbour and hence saving computing time. The claim is then sufficiently addressed by the results provided. However, there are some serious aspects that need to be improved as follows:
Please try to categorize your ref, rather than talking about each ref one by one.
A short intro on the nature-based optimizations and mentioning some of them, like particle swarm optimizatiuon is necessary. The great contribution of this type of optimization to solving real-world engineering problems, as explained in Design of a dielectric phase-correcting structure for an EBG resonator antenna using particle swarm optimization, and elaborated further in Multiobjective particle swarm optimization to design a time-delay equalizer metasurface for an electromagnetic band-gap resonator antenna.
Apart from optimizing solutions, it should be mentioned that nature-based optimization is also capable to design novel engineering components; for example, phase shiters can be designed based on nature-based approaches as explained in Simulation-driven particle swarm optimization of spatial phase shifters. The artificial magnetic conductor is another component that can be designed by a nature based optimization as discussed in Design of an artificial magnetic conductor surface using an evolutionary algorithm
Some effective methods in improving ant colony performance have been missing in the literature. There are some novel pheromone update strategy to improve the functionality of ant colony optimization algorithms as explained in An improved model of ant colony optimization using a novel pheromone update strategy. Another method is based on modification in the learning phase of AntNet, as explained in AntNet with reward-penalty reinforcement learning. These effective methods need to be covered.
Please explain Table 5 in more detail.
The quality of most pictures is sub-standard. Please replace pictures with higher quality pictures.
Author Response
Thank you so much for your valuable comments. Please find our answers as anattachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
Paper Summary: This paper proposed an Hybrid Nearest-Neighbour Ant Colony Optimization (ACO-NN) algorithm for enhancing load balancing task management in distributed systems. Experiments have been carried out to demonstrate the effectiveness of the proposed algorithm.
Strengths of the paper: The motivation behind the work seems good and relatively appropriate. The structure of the paper is good, and title is appropriate. The authors evaluated the proposed algorithm on various datasets. The references provided are applicable.
Weaknesses of the paper: Although the subject of the study seems relatively appropriate, the entire manuscript is difficult to follow and not technically sound. The poor writing, numerous grammatical and typographical errors, Inappropriate or no use of punctuation marks, incomplete sentences that pervade the entire manuscript (from the Introduction to Conclusion section) have hindered smooth comprehension of the details presented therein.
For example,
Line 54 – 78 should be revised for grammatical errors and academic writing.
On Line 128, T1’s demanding memory should be 1 GB.
Equation 7 was never referenced in-text.
Conclusions and Recommendations: In conclusion, numerous grammatical errors and non-adherence to academic writing style have impeded full comprehension of the paper's technical content as well as the flow and readability of the manuscript. Furthermore, a thorough revision of all sections is needed to enhance the flow, readability, and grammar in the manuscript. In this regard, it would help if the authors sought assistance from a native English speaker or a professional editing service. The present form of the paper is not suitable for publication due to its present weaknesses.
Author Response
Thank you so much for your valuable comments. Please find our answers as an attachment
Author Response File: Author Response.docx
Reviewer 3 Report
This paper proposed a hybrid ACO-NN scheme which based on the meta-heuristic ACO and an approximate NN approach to establish a dynamic load balancing algorithm for distributed systems.
Problems and concerns:
- Several state-of-the-art schemes should be investigated rather than the traditional schemes, and the advantage of the proposed scheme against the existing schemes should be clearly explained in the introduction part.
- For section 2.1, I have several concerns. Firstly, the authors mentioned that “the scheduler can be thought of like an ant”, in traditional ACO, the ant is moving, I cannot understand how/why the scheduler needs to move. Secondly, please elaborate the process of pheromone update since it is very important for the whole algorithm. Thirdly, please elaborate how to estimate the time of executing all tasks.
- For the Pre-approve procedure, the authors deny the packet which needs large memory in exchange for reducing the computing time the ACO scheduling. However, I did not see how to process the denied packets in this paper, dropped?
- The authors should introduce more parameters setting before running experiments.
- For all results, the authors should explain and analyze the reasons in detail rather than only illustrating the experimental results. The performance of designed scheme does not convince the reader without explicit results analysis.
Author Response
Thank you so much for your valuable comments. Please find our answers as an attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The paper has been improved; however, there are still some aspects that need further improvements.
The quality of text in the flowchart in Fig. 1 is really low, please make a new flowchart
Lines 83 to 86: please include this reference please include different staregies for ANT colony enhancement and include the following refs: An improved model of ant colony optimization using a novel pheromone update strategy. Please include AntNet with reward-penalty reinforcement learning as well.
Data on both vertical and horizontal axes in Fig. 13 should be re-written due to the low quality
Fig. 14 has low resolution
Lines 34 -78: when you are discussing swarm intelligence, please add some of the latest applications of swarm optimization in different engineering disciplines. For example; Multi-objective particle swarm optimization for the realization of a low profile bandpass frequency selective surface. Another application is simulation-driven particle swarm optimization of spatial phase shifters.
Author Response
Our thank valuable remarks. Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
The paper seems improved compared to the previous draft.
However, there still exists areas that needs improvement:
The Introduction section (i.e., section 2) still needs revision to enhance flow and smooth comprehension.
The first paragraph of section 2 (Methods) needs revision as the first two statements can be revised as one.
Grammatical errors and incomplete texts still exist in some parts of the paper. On line 194, “…pheromone according the load…” should be “…pheromone according to the load…”. Also, grammatical error or incomplete text exist on lines 204, 226,231, 256, etc.
In addition, the text quality of axis labels in the plots should be improved.
In conclusion, numerous grammatical errors and non-adherence to academic writing style have impeded full comprehension of the paper's technical content as well as the flow and readability of the manuscript. Furthermore, a thorough revision of all sections is still needed to enhance the flow, readability, and grammar in the manuscript. In this regard, it would help if the authors sought assistance from a native English speaker or a professional editing service. The present form of the paper is not suitable for publication due to its present weaknesses.
Author Response
Our thank you for your valuable remarks. Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
In this revision, I can see that the authors did make some changes to address my previous comments, but some of their responses are not satisfactory. My detailed comments are as follows:
- Regarding my previous Comment #1, several state-of-the-art researches in this domain should be investigated and compared with the proposed scheme to verify the effectiveness. But, I cannot see the comparisons in this paper, especially in the performance evaluation section.
- Regarding my previous Comment #2 (point 2), the authors provided the operation of global pheromone update. However, as far as I know that Eq. (9) has some problem from common sense. Here, (1-ρg) × τt means the remaining pheromone at time t due to the pheromone decay, you also mentioned this in this section. But, ρg × τt is self-contradictory and cannot be understood. In my understanding, here has three types for updating the pheromone (i.e., ant-density, ant-quantity and ant-cycle). Clarifying this is very important, because the whole proposal in this paper will be impractical, otherwise.
- Regarding my previous Comment #3 (point 3), according to the authors’ supplement, they did not estimate the time of executing all tasks. They actually record the time, this should be clarify clearly.
As the paper still have fundamental issues with its motivation and practicalness, I cannot suggest to consider it in Applied Sciences.
Author Response
Our thank you for your valuable remarks. Please see the attachment.
Author Response File: Author Response.docx
Round 3
Reviewer 1 Report
All concerns have been addressed. The paper can be accepted now
Reviewer 2 Report
The paper has been improved significantly and my comments were addressed.
Reviewer 3 Report
The authors well answered all my concerns.
