Machine Learning Prediction Approach to Enhance Congestion Control in 5G IoT Environment

Round  1

Reviewer 1 Report

A new machine learning model is proposed in this paper. The model is based on predicting the optimal node in 5G IoT networks. The claim is that the model can help improving congestion control and enhance the overall

network performance. The approach is well defined and the paper is well written. The Related Works section is very comprehensive. However, it will be great to elaborate on why DT concept of machine learning capabilities appears to be an improved approach. 

p.p1 {margin: 0.0px 0.0px 0.0px 0.0px; font: 10.0px Helvetica}

 In the Performance Evaluation section, the simulation setup is clearly described. However, it will be very helpful if the authors consider comparing their DT approach with other approaches, or at least elaborate on DT's advantages over other approaches. 

Finally, it will be helpful to at least briefly elaborate on how DT can possibly be implemented in hardware.  

I also suggest making the section headers larger for greater visibility. 

Author Response

Thank you for your thoughtful and thorough review of our manuscript.

Major comments:

1)   It will be great to elaborate on why DT concept of machine learning capabilities appears to be an improved approach? 

Response: We appreciate your comments. As you commented, To elaborate why DT concept of machine learning capabilities appears, the justification have been mentioned into the related work section and Effect of DT prediction section of our revised manuscript (page 5, line 221-226, page 15, line 508-524, and revised Table 3).

2)   It will be very helpful if the authors consider comparing their DT approach with other approaches, or at least elaborate on DT's advantages over other approaches?

Response: Thank you very much for your comment. As you recommended. We compared between various DTs approaches based on many performance criteria to find the optimal DT to predict the optimal alternative. These results have been incorporated into and the Effect of DT prediction section of our revised manuscript (page 16, line 533-536, and revised Table 3)

3)    It will be helpful to at least briefly elaborate on how DT can possibly be implemented in hardware? 

Response: We appreciate your comments. As you recommended. we have considered the following considerations the number of UE WSN 20 to 30, the experiment simulation times were 50 and 180 seconds respectively. The experiment based on real concepts such as 5G Environment, SCTP protocol and a real problem statement relates to Congestion Control mechanisms. The combination of improvement Congestion Control in 5G IoT Environment based on decision tree machine learning is considered the first attempt throughout this paper, we have mentioned in the future work we will implement the optimal prediction techniques on hardware and test in real testbed environments. The simulation setup have been incorporated into the Performance evaluation section and simulation setup section of our revised manuscript (page 13, line 422-429, line 431-44, figure 4, page, figure 9, page 17 , figure 10, and page 18).

4)    I also suggest making the section headers larger for greatervisibility?

Response: Thank you very much for your comment. As you commented. We have been made the changes as long as the journal-style formats are permitted.

Author Response File: Author Response.docx

Reviewer 2 Report

In this paper, the authors proposed a new machine learning model based on a decision tree algorithm to predict optimal enhancement congestion control in 5G IoT networks. The model is implemented on a training dataset to determine the optimal parameter settings in 5G environment and evaluated in simulation experiments. Generally speaking, the topic is interesting and the paper is technical solid. Also, the algorithm shows good performance in simulation. However, it would be better if the authors can provide evaluation on real test bed.

Author Response

Major comments:

1)    It would be better if the authors can provide evaluation on real test bed?

Response: Thank you very much for your comment. As you recommended. We have considered the following considerations the number of UE WSN 20 to 30, the experiment simulation times were 50 and 180 seconds respectively. The experiment based on real concepts such as 5G Environment, SCTP protocol and a real problem statement relates to Congestion Control mechanisms. The combination of improvement Congestion Control in 5G IoT Environment based on decision tree machine learning is considered the first attempt throughout this paper, we have mentioned in the future work part we will implement the optimal prediction techniques on hardware and test in real testbed environments. The simulation setup have been incorporated into the Performance evaluation section and simulation setup section of our revised manuscript (page 13, line 422-429, line 431-44, figure 4, and page 13).

Author Response File: Author Response.docx

Reviewer 3 Report

This submission presents a study for the application of machine learning within the context of congestion control in 5G-based IoT. At the first look, such a networking-related work does not meet the scope of the journal as it is not related to the science of electronics and its applications. Maybe a hardware implementation of the solution would make it more pertinent to the journal, but only a simulation is presented. In addition, the assessment is not fully convincing as the proposed approach should be compared with the main related works within the current literature. Why using the decision tree (DT) algorithm? A justification of this use is needed. The use of this approach is not extremely novel as many previous works with a similar approach are available in the current literature.

Author Response

Thank you for your thoughtful and thorough review of our manuscript.

Major comments:

1)    Maybe a hardware implementation of the solution would make it more pertinent to the journal, but only a simulation is presented?

Response: We appreciate your comments. As you commented. We have considered the following considerations the number of UE WSN 20 to 30, the experiment simulation times were 50 and 180 seconds respectively. The experiment based on real concepts such as 5G Environment, SCTP protocol and a real problem statement relates to Congestion Control mechanisms. The combination of improvement Congestion Control in 5G IoT Environment based on decision tree machine learning is considered the first attempt throughout this paper, we have mentioned in the future work part we will implement the optimal prediction techniques on hardware and test in real testbed environments. The simulation setup have been incorporated into the Performance evaluation section and simulation setup section of our revised manuscript (page 13, line 422-429, line 431-44, figure 4, and page 13)

2)    The assessment is not fully convincing as the proposed approach should be compared with the main related works within the current literature?

Response: Thank you very much for your comment. As you recommended. We compared between various DTs approaches based on many performance criteria to find the optimal DT to predict the optimal alternative. These results have been incorporated into and the Effect of DT prediction section of our revised manuscript (page 16, line 533-536, and revised Table 3)

3)    Why using the decision tree (DT) algorithm? A justification of this use is needed?

Response: We appreciate your comments. As you commented. To justify why decision trees are particularly helpful, the justification have been mentioned into the related work section and Effect of DT prediction section of our revised manuscript (page 5 line 221-226, page 15, line 508-524, and revised Table 3).

4)    The use of this approach is not extremely novel as many previous works with a similar approach are available in the current literature?

Response: Thank you very much for your comment. As you recommended, we have mentioned that most of the previous studies frequently used the decision tree machine learning to improve the network status without considering the congestion control mechanism over the recent environments specifically 5G IoT environments. The studies in our literature only implemented machine learning capabilities in classifiers to improve network status without considering the current mechanisms. In this particular case, the authors in [38] “A decision-tree-based on-line flow table compressing method in software-defined networks” presented a proposed a solution for the flow table congestion problem, thus, it’s not a solution for the congestion control mechanism problem over the recent environments specifically 5G IoT environments. The justification has been mentioned in the related work section and Effect of DT prediction section of our revised manuscript (page 5 line 221-226, page 15, line 508-524, and revised Table 3).

Author Response File: Author Response.docx