Resource Allocation for Network Slicing in RAN Using Case-Based Reasoning

Round 1

Reviewer 1 Report

After reviewing your work, I think you should considerably improve some parts of your paper. Specifically, you should improve the motivation of your work and the organization of the introduction. You should also introduce some key concepts instead of using them directly in the paper. The reader could feel lost in some parts of the text. More details are provided in the specific comments:

(Abstract) The statement “In this research, case-based reasoning  … the best bandwidth ratio between slices” is too direct. At this point, I do not know what “query case” means. The same question with “within the library”.

(Introduction): Bad organization. You need to use different paragraph to express different ideas.

(Introduction): You need to improve the motivation of your work. When you mention “whereas, there are some challenges to allocate resources”, you need to indicate and describe the specific challenges.

(Introduction): You need first to focus on a specific scenario for the resource allocation, and then describe the state-of-the-art works and justify why they do not solve the considered problem.

(Introduction): You need to explain the concept of case library (line 38 in page 2).

(Introduction): You need to provide the meaning of CBR. Same question for KNN.

(System model): In line 61 of page 2, you need to define the variable a.

(System model): You need to explain why do you split your cell in different sectors and the meaning of segment.

(System model): Could you directly generalize your model to N slices (instead of two)? In the first paragraph you consider two slices, but then in the page three you consider a set S of slices.

(System model): Eq. 1 is not well-expressed. The interference from neighbor cells should not be included as noise power. You need to explicitly define such interference in this equation.

(System model): In Eq. 2, you should use other letter for variable a. It is different in comparison with the variable a explained in the second paragraph (line 67 in page 2).

(System model): From line 77 to 87 in page 3 should be described in depth. You directly mention the concepts “query case” and “CBR library case” without any explanation. The reader could feel lost at this point. You should present these concepts at the beginning of this paragraph.

(Numerical Results): Define RMSE.

(Numerical Results): Use grids for figures 4 and 5.

(Numerical Results): Instead of using figure 6 and 7 to show how the prediction matches with the test samples, you should the distribution of the difference between the predicted and the real values. To that end, I suggest you to use a whisker and box plot to represent the difference between these values.

(Numerical Results): Define QUR.

(Conclusions): You need to add a first paragraph to summarize the context of your work: network slicing, 5G, radio resource allocation.

Author Response

Dear reviewer:

         thanks for your review, the replay please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper is well written, the topic is presented clearly, and simulation results support the proposed solution.

Three minor comments:

1)   In the formula (1) the bit rate is set equal to the Shannon limit, which is an ideal upper bound of the data rate it can be achieved. I suggest replacing the sign “=” with “<”.

2) The size n of the case library is a critical parameter for the time execution of your algorithm, you should comment on how large n should be as a function of the number of slices, users and cells.

3) I suggest adding the number of cells and the modulation format to the parameters of table 3.

Two typos spotted:

1) Misplaced “.” when citing Guijarro [3]

2) The acronym DRL is defined twice. The first time erroneously for deterministic policy gradient method and the second time correctly for deep reinforcement learning

 

 

 

 

Author Response

Dear reviewer:

        I have revised the paper according to your suggestion, please see the attachment.

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript addresses a machine learning-based approach for detecting the best bandwidth ratio in a case-based reasoning network slicing. The work is promising, however, I have the following concerns that need to be addressed:

1- Define symbols before being used as in page 2, line 45 for the "CBR" and other similar cases.

2- Page 5, line 139, the authors said "the larger the value of ae1", what is this symbol?

3-add a reference to CV-KNN and O-KNN and briefly explain the differences with respect to conventional KNN

4-page 11, line 184: change  OKNN to O-KNN

5- what is the reason behind using the term TQUR for exhaustive search, PQUR for O-KNN, and HQUR for CV-KNN
6-The work lacks a comparison with state of the art in the results section

Author Response

Dear reviewer:

        I have revised the paper, please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have properly addressed all the reviewers' comments. 

Reviewer 3 Report

The authors have addressed my comments. i have not other concerns