A Dynamic Service Reconfiguration Method for Satellite–Terrestrial Integrated Networks

Round 1

Reviewer 1 Report

This paper studies the service migration and reconfiguration scheme in STINs by considering the external and internal environment changing factors. The authors propose a Fuzzy logic and Quantum Genetic Algorithm (FQGA) to select the migration location and path of services. The paper is well-written and contributes to the field of dynamic resource management in STINs. However, the paper can be improved by considering the following revisions/improvements:

1. Sections 2.1 and 2.2. In the related work section, the authors present the existing research on intelligent service migration. It is unclear how the proposed algorithm solves (or goes beyond) the limitations of the existing works.
2. Section 2.1. Deep Q-learning techniques usually outperform the iterative-based algorithms because they are model-free and contain minimal assumptions of the problem formulation. It is worth including 2-3 references studying the Deep reinforcement learning in radio resource management problems and highlight the superiority of Machine learning based optimization in several management tasks against stochastic and Genetic Algorithms. For example, it is worth including the following works comparing DQL with iterative optimization algorithms: a) Ahmed, K. I., Tabassum, H., & Hossain, E. (2019). Deep learning for radio resource allocation in multi-cell networks. IEEE Network, 33(6), 188-195. b) Giannopoulos, A., Spantideas, S., Tsinos, C., & Trakadas, P. (2021, June). Power Control in 5G Heterogeneous Cells Considering User Demands Using Deep Reinforcement Learning. In IFIP International Conference on Artificial Intelligence Applications and Innovations (pp. 95-105). Springer, Cham.
3. Which is the delay of the Migration and Reconfiguration Execution stage (of Fig. 2)? Does this period produce delays to the remote users?
4. The authors could discuss the trajectory prediction schemes in order to avoid possible delays. For instance, if the users’ trajectory can be predicted, their scheme could benefit from knowing the target destination to avoid time-consuming ‘handovers’.
5. The authors have to include a limitation about not using Reinforcement learning schemes in their selected schemes (future work?).
6. Is there any specific reason for using the parameters of Table 3? Please include related criteria for clarification or references that use same or similar simulation parameter.
7. How the weights (w1 and w2) of the objective function were set? What’s the effect of changing between different weights/coefficients? Please clarify.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

In this article, the authors propose a dynamic service reconfiguration method for satellite-terrestrial Integrated networks. For the proof of concept, they used a mathematical model, algorithms, and simulations. They did elaborate on an excellent literature survey, compared their method to the most eminent ones, and proved their proposed method has better performance. Their paper is well structured and follows all the required steps. I would suggest to the authors to add a small paragraph describing their plans for future work.

The authors should check their paper for similarities as I have run through "SafeAssign," and it revealed a rate of 4%.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

In the abstract, please formally state what is the purpose of the article, the research method, the methodology, the main findings, and the main implications of the study;

Please avoid lumped, concentrated references [2,3]. It is better for the reader to individuate what article has contributed with what content.

The introduction is too long and lacks relevant parts. The mentions of related works must be concentrated in the second section. The introduction must only contextualize (what you have done) and justify (why). You may also derive a research gap, but further positioning of your article in front of others must be concentrated in section 2.

Please formally state what is the purpose of the article, the research method, the methodology, the main findings, and the main implications of the study, as I have asked you to do in the abstract.

Figure 1 must be positioned immediately after the first mention. All the text explaining the figure must follow, not anticipate the figure.

The same is true for figure 2 and the rest of the exhibits.

Equations must be mentioned in the text and evocated by texts such as …. As presented in Equation (1) ….

Please consider unifying sections 2, 3, and sections 4.1 and 4.2. I wonder if the problem formulation together with current section 5 does not deserve a unique section. There lays the most relevant content of your study.

Simulation and final remarks are fine. Please just add two or three paragraphs on the implications of your study and your findings. Who is the agent more favored or who can take more profit or advantage with your study?

Please hire a native or professional proofreading, as there are part of tedious or difficult reading for the international reader

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors satisfactorily addressed most issues.