2024年3月18日将自己认为比较满意的工作投稿到领域内顶刊，不过这次比预期的早，5月27日就返回审稿意见，结果当然不出一样，被Reject。作为一个菜鸟科研人员，论文投稿顶刊，被拒稿实属正常。

在认真阅读审稿意见以后，发现编辑找了4个审稿人，当然有正面的意见，也有负面的意见。

其中的三个审稿人基本上意见都算中肯，总的结论就是稿件属于增量创新，还有假设的条件过多，限制论文的应用。

额外指出一下论文的细节问题，说明这些审稿人都是比较负责的，至少在读了论文以后，才给出评审意见。

剩余的一个审稿人，写了如下的很多审稿意见：

Questions/Comments:

• The paper describes the proposed theoretical framework, including the task models, schedulability analysis techniques, and energy optimization methods. However, the reproducibility and replicability of the experimental results could be improved by providing the following:

Detailed information about the experimental setup, including hardware and software configurations, parameter settings, and any relevant tools or libraries used.

Access to the data sets or workload generators used for the experiments, along with clear descriptions of their characteristics and generation processes.

Publicly available code or scripts used for implementing the proposed algorithms, schedulability analysis, and energy optimization techniques.

Step-by-step instructions or guidelines for reproducing the experiments, including any necessary preprocessing steps, parameter tuning, or post-processing of results.

Clarification on any assumptions, simplifications, or limitations that may affect the reproducibility or generalizability of the results.

• How will the proposed approach adapt to evolving hardware and software architectures, such as the increasing prevalence of multi-core processors and heterogeneous computing platforms?

• How can the proposed approach be integrated into existing real-time operating systems and middleware platforms, considering the potential need for modifications to scheduling algorithms, resource management, and power management mechanisms?

• While the proposed approach aims to optimize energy efficiency in mixed-criticality systems, how would it handle scenarios where energy consumption is not the primary concern, and other factors such as timing predictability or fault tolerance take precedence? What modifications or extensions to the model would be required to address these competing objectives effectively?

• Considering the increasing complexity of modern mixed-criticality systems, which often involve distributed architectures, heterogeneous platforms, and dynamic workloads, how scalable and adaptable is the proposed approach? There are challenges that might arise in extending the model to handle inter-task dependencies, shared resources, or complex task activation patterns. How would the proposed approach need to be modified to account for communication overheads and synchronization requirements in distributed systems?

• The paper assumes a specific task model and scheduling algorithm (Earliest Deadline First). However, real-world mixed-criticality systems may employ different task models, scheduling policies, or resource management strategies. How generalizable is the proposed approach to alternative task models (e.g., sporadic tasks, multi-frame tasks, task chains) or scheduling algorithms (e.g., rate-monotonic,fixed-priority preemptive)?

• Many modern mixed-criticality systems involve complex hardware architectures, including multicore processors, heterogeneous computing elements, and specialized accelerators. How would the proposed approach handle scenarios where tasks have specific affinities or preferences for certain hardware resources (e.g., GPU acceleration, FPGA offloading), and how would energy optimization strategies need to be adapted to account for these hardware-specific considerations? In heterogeneous computing environments, how would the proposed approach manage energy consumption and scheduling decisions across different types of processing elements, each with potentially different power and performance characteristics?

• The paper assumes a specific task model and scheduling algorithm (Earliest Deadline First). However, real-world mixed-criticality systems may employ different task models, scheduling policies, or resource management strategies. How generalizable is the proposed approach to alternative task models (e.g., sporadic tasks, multi-frame tasks, task chains) or scheduling algorithms (e.g., rate-monotonic,fixed-priority preemptive)?

• Several related works are missed. The authors can see the papers of XX.

Writing tips:

Some minor grammatical errors or typos may be present, but they do not detract from the overall quality of the paper.

To enhance the clarity and presentation of the work, the authors could consider the following improvements:

Include additional figures or visual aids to illustrate better the proposed approach, mathematical models, and key results, as the current paper relies heavily on text-based explanations.

Improve the consistency and formatting of mathematical notations, equations, and symbols throughout the paper to enhance readability and adherence to established standards or best practices.

Provide a more detailed explanation or examples of how the proposed approach can be integrated or extended to handle more complex scenarios, such as multi-core processors, heterogeneous architectures, or systems with more than two criticality levels.

Consider restructuring or reorganizing certain sections of the paper to improve the flow and logical progression of the content, particularly in the methodology and results sections.

刚开始以为这个审稿人非常认真负责，花了很多精力写这么多审稿意见，但仔细读下来发现，他说的都是正确的废话，所提的意见，换在领域内的任何文章基本都是适用。

此外，还不忘末尾要求引起他的文章。

这个主题的文章，不能说100%读过，但95%以上还是可以保证的，没有看过这位审稿人发过相似的文章，居然还让人引用。

至此可以大概率断定，该审稿人基本没有认真读文章，直接用论文的摘要放到AI软件，然后让其审稿，这样多省事。

这也是投稿十几年来遇到的新情况，所以不负责的审稿人哪里都有。

作为科研民工，面对拒稿不可怕，重要的是做好自己的研究工作，提升论文的组织与写作能力，在汲取有益的意见的情况下，不断改进与完善论文，论文总是能够发表的。

更多精彩请关注微信公众号“文忆天下”或搜索ZYWHQU