Visualisation of Control Software for Cyber-Physical Systems
Round 1
Reviewer 1 Report
This paper proposes a human-readable graphical representation of a controller for a software-based control system that is useful for enabling a human designer to have a qualitative measure of the controller's behavior. The idea is interesting but, in my opinion, a major effort needs to be made to explain Figures 3-8. What does the shape and each color (green, purple, red, ...) of each figure mean. Why do the figures have that shape and that color?
Minor comments:
- Please use "." or "s.t." in the quantifier scope. For example, \exists x . P instead of \exists x P.
- line 173-174: I don't understand the meaning of the phrase "and with \exists x_i ..."
- line 202: I don't understand what F is. It appears in this line and it is neither defined nor used to explain semantics.
Author Response
This paper proposes a human-readable graphical representation of a controller for a software-based control system that is useful for enabling a human designer to have a qualitative measure of the controller's behavior. The idea is interesting but, in my opinion, a major effort needs to be made to explain Figures 3-8.
What does the shape and each color (green, purple, red, ...) of each figure mean. Why do the figures have that shape and that color?
***ANS***: Thanks for your remark. We added a new sentence explaining our usage of colors (lines 411--418). As for the shape, it directly comes from the input controller being fed to Algorithm 1 (and by the fact that the empty action set is always white).
Minor comments:
- Please use "." or "s.t." in the quantifier scope. For example, \exists x . P instead of \exists x P.
***ANS***: Thanks for your suggestion, we implemented it throughout our paper.
- line 173-174: I don't understand the meaning of the phrase "and with \exists x_i ..."
***ANS***: Thanks for your remark. We rephrased the sentence, we hope it is more clear now (it is now on lines 172-173).
- line 202: I don't understand what F is. It appears in this line and it is neither defined nor used to explain semantics.
***ANS***: Thanks for your remark, we rephrased the sentence, we hope it is more clear now (it is now on lines 202-203).
Author Response File: 
Author Response.pdf
Reviewer 2 Report
I've one only comment that should be delivered to the authors.
They write that the number of bits encoding states and actions is usually between 15 and 25 and between 1 and 4, respectively. I suggest to support this sentence with some more details or by quoting some paper in the literature that argument that his is really the case.
Author Response
I've one only comment that should be delivered to the authors.
They write that the number of bits encoding states and actions is usually between 15 and 25 and between 1 and 4, respectively. I suggest to support this sentence with some more details or by quoting some paper in the literature that argument that his is really the case.
***ANS***: Thanks for your remark, we rephrased the sentence, we hope it is more clear now (lines 63-76).
Author Response File: Author Response.pdf
Reviewer 3 Report
In this manuscript authors presented a graphical human-readable representation of a controller K for a software-based control system.
The scientific contribution of this work is very reduced. The proposed approach is interesting when I have a set of parameters that can be turned, and I can see the influence of them in the graphical tool.
The graphical representation allows us to choose the best configuration for an input model.
However, algorithms 1 and 2 are not described in detail to understand how the system can be used for any input model and parameters. Additionally, the experimental evaluation that authors shown is not enough to conclude that it can be used for any input model and how can we change input parameters and see the output influence in the graphical output.
other models/use cases must be evaluated to demonstrate the capabilities of the proposed tool.
In terms of paper organization, the abstract must be rewritten to be clear what is the scope of this work and what are the main results. The initial part of the current version must be reduced...
In terms of related work section, I suggest moving it to section 3 before describing the proposed tool. This section must describe the state-of-the-art in terms of visualization tools and highlight the main contribution of the current proposal.
Author Response
In this manuscript authors presented a graphical human-readable representation of a controller K for a software-based control system.
The scientific contribution of this work is very reduced. The proposed approach is interesting when I have a set of parameters that can be turned, and I can see the influence of them in the graphical tool.
The graphical representation allows us to choose the best configuration for an input model.
***ANS***: Thanks for your remark. This is out of the scope of the current paper, as we already have a controller in input and we do not focus on changing parameters in order to obtain a new one. However, it is an interesting area of further research, so we added it to the future work session (lines 558-560).
However, algorithms 1 and 2 are not described in detail to understand how the system can be used for any input model and parameters. Additionally, the experimental evaluation that authors shown is not enough to conclude that it can be used for any input model and how can we change input parameters and see the output influence in the graphical output.
other models/use cases must be evaluated to demonstrate the capabilities of the proposed tool.
In terms of paper organization, the abstract must be rewritten to be clear what is the scope of this work and what are the main results. The initial part of the current version must be reduced...
***ANS***: Thanks for your remark. As discussed above, the suggested discussion is out of the scope of the current paper, where we deal with a fixed controller and we want to give qualitative insights about it. We have rewritten the abstract as required, so as to make our focus more explicit.
In terms of related work section, I suggest moving it to section 3 before describing the proposed tool. This section must describe the state-of-the-art in terms of visualization tools and highlight the main contribution of the current proposal.
***ANS***: Thanks for your suggestion, we moved the section as requested.
Author Response File: Author Response.pdf
Reviewer 4 Report
Starting with the title, in my opinion it does not reflect the content of the paper. Why refer to cyber-physical systems when the methodology can be applied to any control plant. The second point that must be clarified is what are the real contributions of the visualization component in the are of control systems. For example, focusing on Figure 3, what are the messages transmitted by this Figure? First, the axes need to be well identified, there are no units identified, second, what is the relevant meaning of colors? Left side, significant green area, what does that mean? a line at iL = 0, violet, what is it, and the variation of the gradient on the right, what does it mean? Questions like this needs to be clarified. Another question is, how do I act on the controller through this visualization, what are the advantages? How are the controller parameter settings made?
The paper has too much self-citation, 15 papers identified.
minor issues:
- number the equations;
- a DC-DC converter is not a cyber physical system, which means that using it as a case study is not aligned with the text and title of the paper.
Author Response
Starting with the title, in my opinion it does not reflect the content of the paper. Why refer to cyber-physical systems when the methodology can be applied to any control plant.
***ANS***: Thanks for your remark. Since in this paper we focus on cyber-physical systems, and given that virtually all plants to be controlled can be modeled by cyber-physical systems, we think our title is appropriate.
The second point that must be clarified is what are the real contributions of the visualization component in the are of control systems. For example, focusing on Figure 3, what are the messages transmitted by this Figure? First, the axes need to be well identified, there are no units identified, second, what is the relevant meaning of colors? Left side, significant green area, what does that mean? a line at iL = 0, violet, what is it, and the variation of the gradient on the right, what does it mean? Questions like this needs to be clarified.
***ANS***: Thanks for your remark. We added the units to the figures and we better clarified the highlighted issues in the paper (see lines 521--529)
Another question is, how do I act on the controller through this visualization, what are the advantages? How are the controller parameter settings made?
***ANS***: Thanks for your remark. This is actually outside the focus of this paper. Here, we simply want to provide a tool for qualitative evaluation on a given controller.
The paper has too much self-citation, 15 papers identified.
***ANS***: Thanks for your remark. Actually, there are only 2 papers with exactly the same authors (one being the paper we are presenting an extended version of), and 8 if we consider papers containing all the authors of the current manuscript. Given that there are more than 70 citations in the current version of the paper, we think that the share should be fair.
minor issues:
number the equations;
***ANS***: Thanks for your remark. We have now numbered all equations.
a DC-DC converter is not a cyber physical system, which means that using it as a case study is not aligned with the text and title of the paper.
***ANS***: Thanks for the remark. We have better defined cyber-physical systems in the introduction (also referring to very well-known books), and better explained in the description of the case study why the DC-DC converter is actually a cyber-physical system (lines 16--20 and 450--454).
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have improved their paper and hence I suggest to accept it.
Author Response
Thanks for your evaluation
Reviewer 4 Report
Thanks for the answers to my questions. They say that acting on a controller is outside the scope of the paper, however, it will be important to have an idea of how visualization can be an improvement in the performance of the controllers. Without closing the loop, it is difficult to understand the real benefits of your contribution.
Author Response
Thanks for your remark. The importance of being able to provide a graphical representation of a controller (especially of an automatically generated one) is widely accepted in the literature for automatic control. In fact, virtually all papers in the subject (many of whom are cited in our paper) try to provide a visualization of the generated controller. This allows researchers to qualitatively compare their work with the methodology of other researchers. Our paper describes a widely applicable procedure to generate such pictures.
