从读研究生开始，接触研究，至今已经十多年了；在这十多年中，做了点自己喜欢的研究，从一无所知的门外汉，不断的努力、探索；逐渐掌握了一些研究的方法。期间，也发表了一些论文；不过在论文投稿过程中一直有一个难题萦绕在心头。遇到这个难题，每次总是胆战心惊，并没有找到好的办法，尤其是一劳永逸的方法来解决。

不过这次的论文投稿经历，倒是给解决这个问题提供了比较好的思路。2021年7月14日将今年做的比较满意的研究工作投稿的国外期刊；让我意外不到的是这次的审稿周期出乎意料的短，8月30日就收到该稿件的评审意见，结果还是不错的，没有直接拒稿，而是给了修改的机会。从评审意见上来看，总共有4个审稿人，而且这些审稿人都是小同行，每个审稿人都提了比较中肯的意见和建议。

然而，这些意见了包含了我以前遇到的难题，即论文模型的假设不可接受以及论文所提出的新算法没有和现有的算法进行比较。第一个问题以前有成功的经历，所以还好解决；第二个问题就非常的麻烦。之所以没有将所提的新方法与现有算法进行比较，当然不是刻意回避，而是所做的工作比较新颖，找不到相同模型，相同条件的方法进行比较。以前遇到这种问题，这样答复结果挂了，所以这次是慎之又慎。以下是针对这两个问题的回答。

1. Also, the paper assumes, all the low criticality leveled tasks are dropped at the time of transition from low mode to high mode, this is also, not an acceptable assumption.

Response: This assumption is pessimistic, but it is acceptable. Because most studies about MC real-time systems such as [1-7], [9], [11-14], [23-27], [29-35], drop all LO-level tasks in HI-mode. First, the system rarely switches to HI-mode. Second, when the system is in HI-mode, the HI-level tasks must be executed as soon as possible. Third, LO-level tasks are mission-critical tasks, which will not cause fatal effects on the system. In recent years, some studies do not give up LO-level tasks in HI-mode, but provide degraded services for LO-level task. However, this task model is more complicated, we will study in future.

2. The paper compares the result of the proposed EANPS with NP-EDFVD, Is the NP-EDFVD uses the same assumption of dropping low critical task after the system migrate to high mode? Also, it is recommended to compare the energy consumption result of pre-emptive approaches with the same assumption.

Response: NP-EDFVD is used the same assumptions of dropping low critical task after the system migrate to high mode. In fact EANPS is based on NP-EDFVD. NP-EDFVD schedules tasks with the maximum processor speed.

Thank you for your suggestion. The real-time scheduling algorithm can be divided into non-preemptive scheduling and preemptive scheduling. Non-preemptive scheduling and preemptive scheduling are suitable for different applications. They have advantages and disadvantages. Non-preemptive scheduling has the following advantages:

•I/O scheduling cannot be preempted or the overhead of preemption is prohibitively expensive

•Non-preemptive scheduling is easier to implement, lower run-time overhead, require less memory etc.

•Non-preemptive scheduling can keep exclusive access to shared resources and reduce the overhead of synchronization.

•Non-preemptive scheduling is more general shared resources task.

However, the disadvantage of non-preemptive scheduling is low system utilization and not flexible enough.

The advantages of preemptive scheduling are high system utilization and more flexible. However, the disadvantages of preemptive scheduling are high system overhead and unpredictability.

Therefore, it is unreasonable to compare preemptive scheduling algorithms with non-preemptive scheduling algorithms while other assumptions are the same.

经过一周的修改，9月7日提交了修改稿；让人没有想到的是，10月10日就收到稿件的录用通知；这说明当时的给审稿人的答复已经完全得到了认可。这次投稿经历算是非常顺利的，中间过程几乎没有什么波澜，一切顺风顺水；而且稿件从投稿到录用不到3个月的时间，这速度也让人感到惊讶。更重要的是，这次投稿经历找到了解决问题的思路，所以更加坚定了信心。