Trust Management Scheme of IoV Based on Dynamic Sharding Blockchain^†

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In the context of trust manageent in an Internet of Vehicles environment, the paper proposes a new approach to partitioning the transaction graph in order to more evenly distribute transaction parties among the underlying blockchain shards. The idea is that the partitioning should more explicitly reflect the transaction connectivity of the involved vehicle accounts. This is a natural heuristic leading toward less inter-shard traffic and so better throughput and computational performance compared to existing approaches, which indeed is documented in the simulations in section 6. On the scientific content side, the paper is of moderate interest, since the novelty amounts to the correct idea above, and the integration of several standard graph theory algorithms (like vertex cut, HDRF, Louvain, greedy partition). Despite the demonstrated numerical improvement, I do not consider the paper publishable in the present form. The reason is that the improvement only pertains to graph- and throughput-related metrics, and nothing is said about the trust management evaluation. It is not clear for what purpose the trust management material is included in the first place, there is no mention of it in Section 6. There is no justification of the applied DST framework - why should it cause any difference w.r.t. probability or other trust calculation paradigms, what is the impact of the proposed approach upon trust accuracy, how does the so-calculated trust impact other performance metrics in connection with the proposed graph partitioning., and finally, why does trust make it to the paper title at all? In my opinion, either the above questions should be addressed, or the paper should be remodeled to only contain the graph partitioning contribution, and suitably retitled; otherwise the motivation is insufficinet and the overall message a little confusing.

Minor remarks:

- figure1, algorith1 etc. - pls introduce spaces

- pls explain all used abbreviations, e.g., PBFT, HDRF

- in standard graph notation, G(V,E), V and E are vertice and edge sets, not particular instances thereof, cf. Table 1

-  pls italicize symbils within the body text, e.g., v, G

- the two algorithms should be completed with more elaborate comments

- "each evidence A in ... based on the data..." - probably \theta is missing

- in (4) the summation should be taken over intersecting A_j, A_i

Comments on the Quality of English Language

Minor faults, punctuation errors, and typos need to be corrected (in particular, pls correct "vertexs" to "vertices", "in the figure4" to "in figure 4", "the closer the number of edges in each partition is")

Author Response

1. Response to comment: There is no justification of the applied DST framework - why should it cause any difference w.r.t.

Response:We apologize for the inaccuracies in the description. As a method widely used in the field of autonomous driving for multi-sensor information fusion, we believe that the DST theory can also yield good results in trust management. This is achieved through the aggregation of reputational evidence from multiple other vehicles towards the target vehicle. We have made revisions to the pertinent descriptions in the paper.

 

2. 2. Response to comment: What is the impact of the proposed approach upon trust accuracy.

Response: We apologize for the oversight in accurately describing the lack of precision in trust management. We have supplemented relevant experiments in the updated version of the paper, found in lines 392-401.

 

3. 3. Response to comment: How does the so-calculated trust impact other performance metrics in connection with the proposed graph partitioning.

Response: We are very sorry for our negligence of the explanation. The performance metrics related to graph partitioning describe the performance of the blockchain itself, while trust computation is carried out through the DST theory. We have made the necessary corrections to the relevant description in our paper.We have also made additions and modifications to the effects of the proposed scheme on the performance of the blockchain in lines 4-9-421 of the manuscript.

 

4. 4. Response to comment: Why does trust make it to the paper title at all?

Response: In our conception, blockchain provides consistency, decentralization, and reliable data security support for trust management, while the DST theory serves as a supplement to the existing vehicle reputation recognition system. Together, they constitute our solution.

 

5. 5. Response to comment: The two algorithms should be completed with more elaborate comments

Response: We have added more detailed descriptions to the two algorithms according to the Reviewer’s comments.

 

6. 6. Response to comment: Pls explain all used abbreviations

Response: We sincerely apologize for the oversight regarding the abbreviations and have included relevant explanations.

 

7. 7. Response to comment: in standard graph notation, G(V,E), V and E are vertice and edge sets, not particular instances thereof, cf. Table 1

Response: We have made correction according to the Reviewer’s comments.

 

8. 8. Response to comment: Minor faults, punctuation errors, and typos need to be corrected

Response: We are very sorry for our incorrect writing and negligence, and once again, thank you for bringing it to our attention.

 

Special thanks to you for your good comments.

Reviewer 2 Report

Comments and Suggestions for Authors

The method is interesting. Yet two things are not clear:

1. the limitations of such a method ( for instance technological, computational resources, delay , the maximum number of users that could be supported...) ;

2. the extra cost associated to the proposed method (we all know that an improvement comes with benefits and drawbacks).

The authors are invited to process minor style issues as identified (marked in yellow together with the corresponding line number for easier identification, including repetitions, long phrases) in the given file. Regarding Figure 7, the issue here is that the figure is inserted before being mentioned/discussed in text.

Comments for author File: Comments.pdf

Author Response

1. Response to comment: The limitations of such a method

Response: Similar to other blockchain-based projects, blockchain itself still requires a significant amount of computing, storage, and network communication overhead.

 

2. 2. Response to comment: The extra cost associated to the proposed method

Response: Both PBFT and Raft consensus algorithms incur significant additional overhead in both intra-shard and inter-shard consensus.

 

3. 3. Response to comment: The authors are invited to process minor style issues as identified

Response: we will address the minor style issues as per the feedback provided by the reviewer. Thank you for the guidance.

 

Special thanks to you for your good comments.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper addresses the existing security and privacy risks associated with the concept of Internet of Vehicles (IoV). The authors emphasize very well the reduced scalability of blockchain solutions in securing IoV communication in the case of large-scale networks. Consequently, they provide a sharding scheme based on Graph Partitioning to improve the scalability of the blockchain and also apply the Dempster-Shafer theory. The evaluation of the proposed approach shows a significant improvement in terms of scalability compared to other baseline schemes such as Monoxide, HDRF, and Brokerchain. 

The article is very well-written, but there are some points that need to be addressed to improve the quality of the paper. Please see below my points regarding each section of the article.

1. The scientific contribution within this paper is very well explained in the Introduction.

2. The Related Work section is very well-organized and is based on relevant and recent research; most of the references are from the last 5 years.

3. The methodology divided between Sections 4 and 5 is enough detailed to ensure that the proposed scheme can be reproduced.

4. I appreciate that the validation used real-world datasets (lines 348-352). However, I suggest adding details about how these datasets can be accessed. This will enable future developments in the field and the reproduction of the experiment described in this paper.

5. The results obtained show that the proposed scheme achieves good performance. Are there any limitations related to the proposed approach? Please discuss them.

6. Provide a more comprehensive discussion of the recommendation for future research. Emphasize better how the scientific contribution within this paper can be used as a foundation for developing autonomous vehicles or other intelligent transportation systems (e.g., traffic lights control systems, local transportation scheduling, etc.).

Author Response

1. Response to comment: Adding details about how these datasets can be accessed.

Response: The Ethereum dataset can be obtained through various means such as relevant block explorers, command-line tools, and Ethereum API.

 

2. 2. Response to comment: Are there any limitations related to the proposed approach?

Response: The PBFT consensus algorithm demonstrates inefficiency when operating in large-scale networks, resulting in a limitation for each shard to accommodate additional nodes.

 

3. 3. Response to comment: Provide a more comprehensive discussion of the recommendation for future research.

Response: As suggested by the reviewer, we have included the relevant description.

 

Special thanks to you for your good comments.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The Authors have satisfactorily addressed all my comments regarding minor issues (except that G(v,e) in Table 1 still refers to a specific vertex and edge, and in Algorithm 1, lines 5, 7, and 19 still call for some explanation or reformulation).

Yet the most important issue I pointed out remains: Is this paper about transaction graph partitioning, where the Authors' contribution is unquestionable, or about trust management, where in my view there is no contribution? Moreover, there is no indication how/if these two topics affect each other. It is not enough to state that they together make the solution, since in fact the latter topic only "consumes" the solution of the former. I fear the reader can feel quite confused, not understanding what message the paper carries. In response to my comments, the Authors provided Fig. 8 to show the advantage of DST over a Dirichlet distribution-based DBTM trust scheme. But this advantage arises because DST better handles evidence conflicts compared to classical probability, which is a well-known fact that is not adequately shown to have anything to do with transaction graphs. Perhaps the figure should compare DST and DBTM under various graph partitioning methods, e.g. to show that the novel method favors DST even more (if this is indeed the case); a one-paragraph description and discussion of DBTM, in addition to just citing ref. [24], would then be in order to appreciate exactly why. In the present shape, I would sustain my advice that the trust aspects be left out, so that the sharding contribution is emphasized and there is no confusion what the paper is about.

Author Response

Responds to the reviewer’s comments:

1. Response to comment: Aboutminor issues

Response: We apologize for any inaccuracies in the description, and have made revisions to the representation in our table 1, as well as added more detailed descriptions to the algorithm 1.

 

2. 2. Response to comment: About important issue

Response: Thank you very much for your suggestions on our paper, which has provided us with a deeper understanding of the theme of the paper. We agree with your perspective on downplaying trust and emphasizing shard contribution. Therefore, we have optimized the expression of relevant content in the paper and adjusted the sequence of related chapters, with a focus on the optimization of transaction storage in blockchain sharding protocols, while treating trust calculation in trust management as a supplement to blockchain transaction storage. Additionally, we have removed the trust experiment data of DST and DBTM and provided a more reasoned explanation for our use of DST, these main changes are located in lines 214-218 and 270-275. Furthermore, we have also made minor changes to the expressions in lines 7-10, 60-61, 71-75, 176-177 and 418-422. We hope that the optimized expressions in our paper will not cause confusion to the readers. Finally, we would like to express our sincere gratitude for your valuable suggestions, which are of great importance to us.

 

Special thanks to you for your good comments.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the Authors' effort to address my comments. Though our discussion could continue, I do not see much point in going through more rounds of revisions. I insist that there is a contribution regarding graph partitioning that does improve the communiction aspects within the blockchain, but the link to trust management, in particular based on DST, is rather superficial. There is no direct numerical evidence in the present version that the improved graph partitioning makes for better trust assessment or management. I suppose, the removal of the previous Fig. 7 is a step backwards, but if the Authors' intention was to focus on the graph partitioning, the DST related theoretical introduction and further considerations could have been removed as well. I suggest renaming the paper accordingly.

Comments on the Quality of English Language

minor faults, can be improved easily

Author Response

1. Response to comment: DST related theoretical introduction and further considerations could have been removed as well.

Response: We appreciate your suggestions and agree with your viewpoints. Modifications and replacements have been made to the theoretical introduction and further considerations of the DST, which are located in lines 226-248 and lines 434-438 of the manuscript. Additionally, the deleted figure 7 and its corresponding description have been restored, appearing in lines 347-373 and 395-404 of the manuscript.

 

2. Response to comment: Suggestions for Renaming Papers

Response: We agree with the reviewer's suggestion and have made minor revisions to the title of the paper. The revised title of the paper is “Trust Management scheme of IoV based on dynamic sharding blockchain”.

Author Response File: Author Response.pdf