CCN-Based Inter-Vehicle Communication for Efficient Collection of Road and Traffic Information
Round 1
Reviewer 1 Report
This paper presented the CNN-based approach for efficient road and traffic data collection. Simulation is conducted to show the effectiveness of the proposed approach. The paper is not well organized and some concepts are not explained clearly. My comments can be found below:
Figure 1 is not clear to the reader. CNN should appear in this figure, as it is the key contribution in the paper; The compare of CV, ECV, and ECV+ should be shown, which one is the author proposed, and the advantage. Algorithm 1 should be shown in a correct way; Simulation setup should be explained; Error in line 248; Line 305, (Figure 6d) -> Figure 6(d).Author Response
Response to Reviewer 1 Comments
Point 1: Figure 1 is not clear to the reader. CCN should appear in this figure, as it is the key contribution in the paper;
Response 1: Thanks for this comment.
Naming and caching, as important features of CCN, were included in Figure 1 of the previous manuscript. Now we make this clearer by explicitly mentioning them as components of CCN in Figure 1. CCN usually is implemented as application overlay. Therefore, routing is an independent module in the proposed method.
Point 2: The comparison of CV, ECV, and ECV+ should be shown, which one is the author proposed, and the advantage.
Response 2: Thanks for this comment.
We have followed this comment, and added a table (Table 3) and its description in Sec.3.4, to compare main targets of CV, ECV, and ECV + and their strengths and weaknesses.
Point 3: Algorithm 1 should be shown in a correct way;
Response 3: Thanks for this comment.
We now follow the algorithm format of LaTex and updated the description of algorithms. Because we added Algorithm I for CV, Algorithm II for ECV, the original algorithm for ECV+ now becomes Algorithm III.
Point 4: Simulation setup should be explained;
Response 4: Thanks for this comment.
Simulation setup (condition) is presented in Sec.4.2 and Table 4, and simulation scenario is shown in Figure 5. We further explain how Interest packets are generated there.
Point 5: Error in line 248; Line 305, (Figure 6d) -> Figure 6(d).
Response 5: Thanks for the comment.
This typo is corrected now. Figure 6 in the previous manuscript becomes Figure 7 in this revised manuscript.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Please find below my comments.
Related work section is not enough. Extend it to add more recent papers. Discuss this recent paper as well. DCS: Distributed caching strategy at the edge of vehicular sensor networks in information centric networking”, MDPI Sensors, vol. 19(20), pp. 1-20, October 2019 .
Add the comparative table of Related section by clearly highlighting aim, pros and cons.
Add the pseudo code of ECV and ECV+ parallelly for better understanding of the readers.
Add operational example of proposed schemes to clearly explain how it works.
Are the cache buffer capacities realistic? 10,000 GB? Defend it.
References are not enough. This area is very well explored and these references doesn't justify it.
Author Response
Response to Reviewer 2 Comments
Point 1: Related work section is not enough. Extend it to add more recent papers. Discuss this recent paper as well. DCS: Distributed caching strategy at the edge of vehicular sensor networks in information centric networking”, MDPI Sensors, vol. 19(20), pp. 1-20, October 2019 .
Response 1: Thanks for this comment.
The aforementioned paper is added as reference [16] and is reviewed in Sec.2.1. In addition, we added two other references as [17] and [18].
Point 2: Add the comparative table of Related section by clearly highlighting aim, pros and cons.
Response 2: Thanks for this comment.
A comparative table of related works has been added to Sec. 2.1 as Table 1.
Point 3: Add the pseudo code of ECV and ECV+ parallelly for better understanding of the readers.
Response 3: Thanks for this comment.
We have added pseudo code of CV (in Algorithm I) and ECV (in Algorithm II). ECV+ shares almost the same pseudo code as ECV and differs in how to control the cache probability, which is presented in Algorithm III.
Point 4: Add operational example of proposed schemes to clearly explain how it works.
Response 4: Thanks for this comment.
We have added detailed description to explain how proposed schemes work in Fig.2 and Fig.4.
In addition, we now give detailed explanation of Algorithm I for CV, Algorithm II for ECV, and Algorithm III for ECV+.
Point 5: Are the cache buffer capacities realistic? 10,000 GB? Defend it.
Response 5: Thanks for the comment.
The cache buffer capacity of 10,000GB is not realistic at this stage. But advances in storage technology will reduce the cost of cache memory, and will improve the performance of the proposed method when it becomes practical.
In fact, we mainly consider cache buffer of 100GB and consider 10,000GB as a potential in the future.
We clarified this in Sec.4.4.
Point 6: References are not enough. This area is very well explored and these references doesn't justify it.
Response 6: Thanks for this comment.
We have added new references [16], [17], [18], reviewed them in Sec.2.1 and compared them in Table 1.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper addresses a field in full expansion and has results obtained through simulation, which validates the proposed methods.
However, there are some issues that should be clarified and some small corrections could be made (the order is as they appear in manuscript):
Algorithm description (line 234-235) suggestion: reduce line spacing.
Algorithm (last line): “sleep for a period”: this period is the same as information generation interval investigated in 4.3.2? If so, a clearer expression would be needed. If not, it might be interesting (may be in future work) to analyze the effect of this sleep period value.
Algorithm: You use a Δ – Adjustment of cache probability. How this quantity affects the results?
4.2. Simulation conditions line 247 and line 392 ref. [17]: you use the network simulator Scenargie [17]. Please give a more detailed reference for [17].
Line 248: simulation is the Ginza area in Tokyo (Error! Reference source not found.). Please solve this error message.
Pag. 8 – 9 suggestion: try to obtain a best page fit to avoid almost blank page 8
4.3.1 Results Under Different Cache Buffer Capacity, line 264-265: “cache buffer capacity of each vehicle is changed from 1GB, 10GB, 100GB, 1000GB, to 10000GB”. Cache buffer of 100GB may be realistic, but 1000GB or 10000GB appears to be very large, at least from a practical point of view. A short comment on this buffer size might be useful.
Author Response
Response to Reviewer 3 Comments
Point 1: Algorithm description (line 234-235) suggestion: reduce line spacing.
Response 1: Thanks for this comment.
We have reduced the line spacing of the algorithm.
Point 2: Algorithm (last line): “sleep for a period”: this period is the same as information generation interval investigated in 4.3.2? If so, a clearer expression would be needed. If not, it might be interesting (may be in future work) to analyze the effect of this sleep period value.
Response 2: Thanks for this comment.
The period of sleep (in Algorithm III of the revised manuscript) is not the same as the information generation interval investigated in Sec.4.3.2.
It is expected that the period is long enough so that channel usage rate already becomes stable under the new cache probability and also short enough so that the algorithm can track the change in network traffic.
By some initial experiments, this period is set to 1second.
We have explained this in the 3rd paragraph of section 3.3. We will continue to investigate the impact of the period in the future.
Point 3: Algorithm: You use a Δ – Adjustment of cache probability. How this quantity affects the results?
Response 3: Thanks for this comment.
As for the adjustment value Δ, too large a value makes the system unstable while too small a value makes it difficult to track traffic change. The adjustment value Δ is set to 0.1, by initial experiment. We have explained this in the 3rd paragraph of section 3.3.
Point 4: 4.2. Simulation conditions line 247 and line 392 ref. [17]: you use the network simulator Scenargie [17]. Please give a more detailed reference for [17].
Response 4: Thanks for this comment.
This reference becomes [20] in the revised manuscript, and is modified as follows.
Mineo Takai, Jay Martin, Shigeru Kaneda and Taka Maeno, “Scenargie as a Network Simulator and Beyond,” Journal of Information Processing, vol.27, pp. 2-9, January 2019.
Point 5: Line 248: simulation is the Ginza area in Tokyo (Error! Reference source not found.). Please solve this error message.
Response 5: Thanks for the comment.
We have corrected this mistake.
Point 6: Pag. 8 – 9 suggestion: try to obtain a best page fit to avoid almost blank page 8
Response 6: Thanks for this comment.
We have refined the page layout.
Point 7: 4.3.1 Results Under Different Cache Buffer Capacity, line 264-265: “cache buffer capacity of each vehicle is changed from 1GB, 10GB, 100GB, 1000GB, to 10000GB”. Cache buffer of 100GB may be realistic, but 1000GB or 10000GB appears to be very large, at least from a practical point of view. A short comment on this buffer size might be useful.
Response 7: Thanks for this comment.
Currently, we mainly consider cache buffer of 100GB. The cache buffer capacity of 10,000GB is not realistic at this stage yet. But advances in storage technology will reduce the cost of cache memory, and will improve the performance of the proposed method when it becomes practical.
We have added the above description to Sec. 4.4.
Author Response File: Author Response.pdf
Reviewer 4 Report
The paper presents several techniques to improve cache availability, which ultimately reduces the response time for vehicular communications. The techniques are studied under different circumstances and with different parameters using simulations. The results are very interesting and informative.
Nevertheless, the introduction and the description of the different methods is not very clear. Figures 2 and 3 do not help much, since they lack descriptions of the elements and processes. It would help that all the description was complemented with specific examples of communication that illustrate how each method would handle the situation. Figures 2 and 3 could be more informative, showing the example that was mentioned in the text and describing the elements and showing the path of the information.
With the current status, the paper is confusing until the simulation and results are shown. At that point everything becomes more concise and clear.
From the scientific point of view, I miss some discussions (not necessarily new simulations) on the scenario where the density of vehicles is very low, such as suburban areas. This could be derived from one of the observations in the discussion, where low densities provoke many cache misses. What would happen at even lower densities? What would that mean for drivers? Would auxiliary communication systems be required? Would the missed information be relevant, considering that in low traffic there is much less going on? etc.
Author Response
Response to Reviewer 4 Comments
Point 1: Nevertheless, the introduction and the description of the different methods is not very clear. Figures 2 and 3 do not help much, since they lack descriptions of the elements and processes. It would help that all the description was complemented with specific examples of communication that illustrate how each method would handle the situation. Figures 2 and 3 could be more informative, showing the example that was mentioned in the text and describing the elements and showing the path of the information.
With the current status, the paper is confusing until the simulation and results are shown. At that point everything becomes more concise and clear.
Response 1: Thanks for this comment.
We have added detailed explanation for Fig.2 (CV) and Fig.4 (ECV), and a new figure (Fig.3) explaining when cache miss occurs. In addition, we have added Algorithm I for CV and Algorithm II for ECV.
Point 2: From the scientific point of view, I miss some discussions (not necessarily new simulations) on the scenario where the density of vehicles is very low, such as suburban areas. This could be derived from one of the observations in the discussion, where low densities provoke many cache misses. What would happen at even lower densities? What would that mean for drivers? Would auxiliary communication systems be required? Would the missed information be relevant, considering that in low traffic there is much less going on? etc.
Response 2: Thanks for this comment.
The cache miss problem often occurs in the CV method, but these problems have been solved in ECV and ECV +. Following the trend in Fig.8(d), when the vehicle density is low, the cache probability approaches 1. At that time, the difference in performance between the proposed method and the existing method is reduced.
At very low vehicle density, vehicles are not always connected, which greatly affects system performance. We have explained this in the end of section 4.4. In the future, we will further evaluate the performance of the proposed method under low vehicle density.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors made some corrections based on the reviewers comments. But I think the paper still needs to be improved before any publication.
Algorithms are not shown in a correct way; Figures need to be re-drawn, the font size is not consistent with the papers; English needs to be correctedAuthor Response
Response to Reviewer 1 Comments
Point 1: Algorithms are not shown in a correct way.
Response 1: Thanks for this comment.
All algorithms have been redrawn in the correct way, following the example obtained from the assistant editor.
Point 2: Figures need to be re-drawn, the font size is not consistent with the papers.
Response 2: Thanks for this comment.
Figures are redrawn and the font size of all figures is unified with that of main text.
Point 3: English needs to be corrected.
Response 3: Thanks for this comment.
We have refined the English writing.
Reviewer 2 Report
I am satisfied with the revised manuscript.
Author Response
Response to Reviewer 2 Comments
Point 1: English language and style are fine/minor spell check required.
Response 1: Thanks for this comment.
We have refined English writing.
Reviewer 4 Report
The paper is much improved with respect to the initial version. The figures help to understand the methods better. Some minor english style changes can be done to make the explanations clearer.
Author Response
Response to Reviewer 4 Comments
Point 1: Some minor English style changes can be done to make the explanations clearer.
Response 1: Thanks for this comment.
We have refined English writing.
Round 3
Reviewer 1 Report
no more comments.
