Modeling Dew Computing in DISSECT-CF-Fog
Round 1
Reviewer 1 Report
This is a strong paper that I enjoyed reading. I like the problem that the authors/team was trying to address/solve.
A couple of suggestions that may make the paper better:
Table 2 could have a delta column, highlighting differences.
The introduction could add some clear examples of dew computing where intermittent internet connectivity occurs and is restored. One context that I can think of is airlines and cuiseships. Perhaps I am wrong with these examples, but some real-world examples would help.
Author Response
Thank you very much for your time and the valuable comments on the manuscript. We have made a revision of the paper regarding the changes requested, and we tried to take all your comments into account. The detailed response to the comments follows below.
- Table 2 could have a delta column, highlighting differences.
We have modified Table 2, accordingly.
- The introduction could add some clear examples of dew computing where intermittent internet connectivity occurs and is restored. One context that I can think of is airlines and cuiseships. Perhaps I am wrong with these examples, but some real-world examples would help.
We agree with the reviewer, we have extended Section 1 to respond to this comment.
Reviewer 2 Report
Nowadays, fog computing is extremely in demand. However, the management technologies of fog computing are still a challenge due to time and network instability factors in comparison with the quite understandable and sufficiently developed methods and tools of scheduling in cloud computing. In this regard, the paper addresses solving the problem of a fog computing simulation taking into account еру required QoS. The authors present an extension of the well-known DISSECT-CF-Fog simulator with a Dew Computing model. The aim of the study to enable the simulation of IoT-Dew-Fog systems in a cost-effective manner. The study results show that applying the proposed approach can lead to decreasing in processing time of IoT data exploiting an IoT-Dew-Fog environment.
I believe that highly professional work is presented. The paper is well structured and illustrated. A brief review adequately draws on the current state of related work.
Remarks.
Section 4 is brief. It would be useful to expand it by discussing the accuracy of the simulation results obtained based on the proposed approach in comparison with the practical results of operating a IoT-Dew-Fog system. In addition, the simulation results largely depend on the reliability of such a system. Will the fault tolerance of IoT-Dew-Fog systems be further simulated?
Author Response
Thank you very much for your time and the valuable comments on the manuscript. We have made a revision of the paper regarding the changes requested, and we tried to take all your comments into account. The detailed response to the comments follows below.
- Section 4 is brief. It would be useful to expand it by discussing the accuracy of the simulation results obtained based on the proposed approach in comparison with the practical results of operating a IoT-Dew-Fog system.
The evaluation in this paper is an artificial scenario, indeed. Nevertheless, real world systems can also be modelled by setting up their concrete parameters in the simulation. So far, we have not found detailed enough utilization logs of real, production IoT-Dew-Fog systems to be used in such a way.
- In addition, the simulation results largely depend on the reliability of such a system. Will the fault tolerance of IoT-Dew-Fog systems be further simulated?
There is an option in the simulator to allow and set scheduled shutdowns or down times for IoT and dew devices, for which the system can react automatically (see [1] for details). Similarly to this method, fault and system errors could be generated and the reactions analysed in the simulated environment. We plan to examine such scenarios in the future.
[1] A. Markus, G. Kecskemeti and A. Kertesz. Flexible Representation of IoT Sensors for Cloud Simulators. 25th Euromicro International Conference on Parallel, Distributed and Network-based Processing, 199-203, 2017. DOI: 10.1109/PDP.2017.87
Reviewer 3 Report
1. Line no 261-271, the authors are talking about smart devices, but there is no mention of the mobile technology of these smart devices (dominated by smartphones of 4G and 5G). though 5G is mentioned in the initial stage of introduction.
2. Fig 2, devicec is in the fringe zone of the mobile environment. comment on its position and connectivity when it reaches within the area defined by nodei and nodej.
4. in the simulation approach (table 1), what parameter of battery is considered? How the discharging is related to movement?
3. Ref 9-14 and 17-19, etal is used, so instead of it it would be advisable to give the all names.
Author Response
Thank you very much for your time and the valuable comments on the manuscript. We have made a revision of the paper regarding the changes requested, and we tried to take all your comments into account. The detailed response to the comments follows below.
- 1. Line no 261-271, the authors are talking about smart devices, but there is no mention of the mobile technology of these smart devices (dominated by smartphones of 4G and 5G). though 5G is mentioned in the initial stage of introduction.
We provided a detailed description in Section 3.2 and 4 to react to this comment. Nevertheless, the simulator is not dependent on the actual communication protocol or technology a system uses; any of them could be set up in a simulation by defining its latency, bandwidth, computing and storing capacity.
- 2. Fig 2, devicec is in the fringe zone of the mobile environment. comment on its position and connectivity when it reaches within the area defined by nodei and nodej.
In such cases, when device_c reaches that area, it will have similar abilities than device_a, it starts communicating with the preferred node. We responded to this comment in the paper as well, and strengthened Section 3.2 in this regard.
- 3. Ref 9-14 and 17-19, etal is used, so instead of it it would be advisable to give the all names.
We have revised the references, accordingly.
- 4. in the simulation approach (table 1), what parameter of battery is considered? How the discharging is related to movement?
In the third column of Table 1 we denoted that all considered simulators take into account the energy consumption of the devices. However, we did not differentiate, if a simulator can take into account battery drainage or not. For example, IoTSim-Edge supports battery discharging scenarios, but in DISSECT-CT-Fog we plan to introduce this in a future work. Nevertheless, DISSECT-CF-Fog can calculate energy consumption based on CPU usage.
