Parallel Multi-Level Simulation for Large-Scale Detailed Intelligent Transportation System Modeling

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposes a parallel multi-level simulation framework in the field of traffic simulation and verifies its feasibility through the coupling of CARLA and SUMO. The topic is timely and relevant, the methodological framework is clear, and the preliminary results provide valuable insights into the collaborative application of heterogeneous traffic simulation platforms. However, the work still shows several weaknesses, including insufficient emphasis on innovation, limited experimental design, a lack of depth in the literature review, and inadequate discussion of broader implications. With further improvements in methodology, result presentation, and scholarly connection, the article could make a more substantial academic and practical contribution.

1. Research Innovation and Methodological Refinement

The proposed multi-level parallel simulation framework presents certain novelty in intelligent transportation modeling, yet the articulation of its innovative contributions remains insufficient. The authors are encouraged to clarify how this approach differs from existing multi-layer simulation platforms, particularly regarding system architecture, parallel computing efficiency, and adaptability to diverse application scenarios. While the concept of the Buffer Zone is interesting, its validation currently relies heavily on single-vehicle experiments. It is recommended to extend the testing to multi-vehicle traffic flow environments in order to strengthen the robustness and persuasiveness of the proposed method.

2. Experimental Design and Result Presentation

The current experimental scale is rather limited, focusing primarily on the interaction between two simulators (CARLA and SUMO), which restricts the demonstration of the framework’s scalability. The authors are advised to extend the experiments by increasing the number of vehicles and testing in more complex traffic scenarios, while also reporting system performance in terms of computational cost, latency control, and synchronization accuracy. In addition, several figures and tables lack sufficient explanatory details, with some key terms left undefined. It is suggested to improve figure captions and provide clearer explanations in the text to ensure that readers can independently interpret the experimental results.

3. Literature Review and Scholarly Context

Although the paper reviews mainstream traffic simulation platforms, the proportion of self-citations is relatively high, and recent advances in parallel simulation and digital twin traffic systems are underrepresented. To strengthen the academic breadth and objectivity of the paper, it is recommended to include relevant studies from the past three years. Moreover, the discussion should better connect the proposed framework with emerging domains such as digital twins and smart city traffic management, thereby highlighting its academic significance and potential applications.

4. Depth of Discussion and Future Directions

The conclusion section briefly acknowledges certain limitations, but the overall depth of discussion remains insufficient. A more systematic reflection on potential challenges—such as map conversion consistency, cross-platform portability, and scalability in large-scale distributed environments—would enhance the study’s rigor. Additionally, the future research directions should be elaborated in more detail, for example: automated map partitioning, incorporation of multimodal traffic participants (pedestrians, bicycles, etc.), and integration with V2X communication modeling. Such extensions would significantly improve the versatility and practical relevance of the framework.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper presents a parallel multi-level simulation architecture for large-scale intelligent transportation systems (ITS) using CARLA and SUMO simulators. The authors propose an innovative approach to overcome computational limitations by partitioning the simulation into nanoscopic scenes with Buffer Zones, enabling seamless interaction between adjacent scenes. The work is highly relevant to the field of ITS and digital twin technology, and the integration of multiple simulation levels is a promising contribution. There are  several issues that should be addressed.
1. The introduction provides a good overview of ITS challenges but could better contextualize the gap in existing tools. A more detailed comparison with recent works in parallel and multi-scale simulation would strengthen the motivation for this work.
2. The paper conceptually introduces the Buffer Zone mechanism but does not quantify its performance impact. It is recommended to include an analysis of the computational and communication overhead introduced by the duplication and synchronization of vehicles in Buffer Zones.
3. To better highlight the advantages of the proposed parallel approach, a comparative analysis should be provided. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

1).  the code's GitHub link is provided. However the description of the implementation in section 5.2 is high level. It is suggested that it be simplified if someone wants to recreate the experiment. 

2). A bit of proof reading is needed. (for example "useful were features determined" should be changed to "useful features were identified"). Moreover some sentences are complex and long. they should be divided into smaller sentences. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This paper deals with nanoscopic/microscopic distributed/parallel road traffic simulation using CARLA nanoscopic and SUMO microscopic simulation.

The theme of the paper is interesting. It is an evolution of long-existing research on distributed road traffic simulation with the new elements being usage of digital-twins terminology and the nanoscopic simulation considering modern ITS elements (more sensors, vehicle-to-vehicle interactions, etc.). While this is a fair theme of the paper, the introduction part (which also contains related work) fails to deliver the "evolution part" of the story. Similar introductory sections with "lack of power of single standard computer" were used by distributed traffic simulation experts for decades. As such, it sounds quite strange in 2025, even if actually true. To help to set this research in a correct context, the authors should rephrase the introduction. It should contain infomation that the spatially decomposed distributed road traffic simulation is around for some time (some examples are listed below), but the power of distributed computer may still be needed, if the simulation is detailed enough. That would remove the (not good) first impression that the spatially distributed distributed simulation is a new idea.
- T. Potuzak, "Distributed-Parallel Road Traffic Simulator for Clusters of Multi-core Computers," 2012 IEEE/ACM 16th International Symposium on Distributed Simulation and Real Time Applications, October 2012, https://doi.org/10.1109/DS-RT.2012.36

Similarly, the usage of buffer zones, though maybe less elaborated, are used for decades (some examples bellow), and some of their form is necessary to make a spatially decomposed distributed road traffic simulation to work correctly. So, it should be also rephrased and set in a correct context to ensure it is clear, what exactly is the novelty of the buffer zones in this paper.
- A. Acosta, J. Espinosa, and J. Espinoza, "Distributed Simulation in SUMO Revisited: Strategies for Network Partitioning and Border Edges Management," in SUMO User Conference 2016, vol. 30, 2016.

Some of the statements in the paper are not correct. For example "Additionally, in the case of implementing ITSs in the real world, the nature of DT technology (i.e., its need to update the information about the transportation system from distributed sources, such as remote sensors, in real-time) necessitates that the real system is emulated in a decentralized way." - The simulation/emulation of a distributed system does not have to be distributed itself. In fact, the distribution makes any system more complex, including the simulation. So, this is not a valid point for justifying using of a distributed approach, unlike the high computational requirements mentioned earlier in the paper, which is a fair point supporting distributed architecture.

The structure of the paper is acceptable.

The review of the existing (and potentially convenient) simulators is Section 3 focused on three simulator, which were pre-selected by the authors to be bets match or their needs. Since this is not a systematic review paper, this is a fair approach.

The described features of the parallel/distributed simulation solution are quite well described and seems to be sound.

However, the paper does not contain any verification of the described solutions. Not counting the issues mentioned above, it is a quite nice proposal. However, for a journal paper, I would expect verification with a real-life or at least artificial, but sufficiently large, scenarios, measurement of the speed of the simulation depending on the number of utilized computers, the size of the simulated road traffic network, number of vehicles, pedestrians and other traffic elements, and so on. This is a major problem. Without the verification, the paper is more suitable for a conference rather than for a journal.

The figures are of sufficient quality.

The references are relevant and up-to-date, but the references, which are needed to better set this paper into context of current state-of-the-art are missing. See above for examples (just examples, the references in the review are by no means exhaustive).

 

Comments on the Quality of English Language

The English is acceptable, it is not so difficult to read the paper. However, there are some typos and errors. Proofreading by a grammar-skilled native speaker can improve the paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The author has revised the paper as required.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have satisfactorily addressed the required revisions, and the manuscript is now recommended for publication.

Reviewer 4 Report

Comments and Suggestions for Authors

This paper deals with nanoscopic/microscopic distributed/parallel road traffic simulation using CARLA nanoscopic and SUMO microscopic simulation.

The comments of the reviewers were quite thoroughly addressed including the rephrasing of the introduction and other comments. The verification comment, which was the main issue of the previous version of the paper could be addressed more extensively, but the authors did add important informations and justifications where these informations could no be provided. Overall, current version of the paper is acceptable as it is, although more and bigger experiments would further improve the quality of the paper. 

The structure of the paper is still acceptable.

The figures are of sufficient quality.

The references are relevant and up-to-date.