A Fog Computing Architecture with Multi-Layer for Computing-Intensive IoT Applications
Round 1
Reviewer 1 Report
In this paper, the authors propose a multi-layer fog computing platform for the enhancement of data analysis systems based on IoT devices to reduce latency. However, there are some parts that need to be improved.
- The differences between the proposed fog computing architecture and other similar architectures haven’t been claimed clearly. For instance，what is the difference between this work and the following work?
Ref1： Latency Minimization for Task Offloading in Hierarchical Fog-computing C-RAN Networks
- In the simulations, the comparison seems conducted within this work, and only cloud computing is considered. Comparison with other’s work is encouraged. Also, some key simulation parameters are missing.
- The process of data analytics and the offloading algorithm hasn’t been described clearly. Please revise this part.
Author Response
A1) In this paper, there is a difference in the proposed technology for transmitting the part that burdens the computation in the fog computing node to the cloud through offloading. The methodology of offloading was implemented by modifying the contents proposed in the eference [17] to suit this paper. We are the developers of reference [17].
A2) Key parameters required in the simulation are utilized in the offloading process. The key parameters are added in Figure 3. In addition, this paper focuses on the reduction of energy consumption and latency-delay time of devices rather than the cloud-only method of multi-layer based our fog computing through use-case study rather than relative comparison with other research studies. This was explained in detail in section 4.
A3) In Figure 3, the detailed offloading process is summarized and expressed in pictures. In this paper, offloading algorithm development does not fall under the core content. Therefore, the offloading algorithm was applied with reference to [17] developed by this author.
Reviewer 2 Report
The authors target the problem of data reduction at the edge/fog layer to reduce bandwidth-usage in IoT systems. By means of data analytics, irrelevant data shall be identified at a low layer such that only relevant data is sent through the network. To this end, the authors propose an architecture that makes use of a fog gateway to analyze and process collected data. Their work is of high relevance as the growing amount of data generated on the edge needs to be handled and preprocessed to not overload the network capacities.
Major language issues and a lack of structure make it difficult to understand what problem the researchers exactly address and where the main contribution of the article lies. Especially in the introduction, the authors mention and discuss a variety of relevant questions and research areas, making the introduction rather broad. How, for instance, does Covid-19, which is mentioned in the introduction, relate to the work presented in the article? It would be advisable to provide a more rigorous line of argumentation and a more explicit formulation of the pursued research goal.
As I understand it, the authors goal is to develop an IoT system architecture that is capable of filtering and preprocessing data (data analytics?) from the edge nodes before sending it to the cloud. This is to reduce the amount of transfered data and to reduce bandwidth usage. The vision itself is highly relevant, however, there exists a number of similar works that pursue similar approaches. The authors should provide a more comprehensive discussion of relevant related works, discuss their shortcomings, and argue how their address these shortcomings with their approach. (e.g. https://ieeexplore.ieee.org/abstract/document/9139356)
The fog module, which seems to be the central component of the proposed architecture, is taken from another research work [25]. The authors do not outline how their fog module works and why it is adequate for the task that it should carry out. Also, it is surprising to see that the main contribution of the work seems to be taken from another work. It should me made clear where the contribution lies, in case it is not the development of this fog gateway. Also, information should be provided on how the data processing is carried out by this gateway.
A validation and comparison is provided by simulating two scenarios, one with and one without the mentioned fog gateway with data analytics. Yet, the scenario is not described in detail and it is unclear what the four described configurations refer to (more than the number of surveilled areas). How big are these areas? What is surveilled in these areas? When visualizing the results, the authors choose to use a line chart, which is misleading, especially due to the smoothing applied. It seems that the delay decreases between config 2 and 3, which does not seem logical. Instead, a barchart should be used.
Finally, the conclusions section should summarize the finding of the work and discuss future work.
Author Response
We deeply appreciate your kind and important comments and suggestions. In addition, we made efforts to faithfully supplement and reflect it in the revised manuscript.
We agree with you.
[1] The content vaguely described in the introduction has been modified to convey more accurate meaning. For example, it is an expression for Covid-19, an expression to find a task that is processed by a fog computing node through filtering a large amount of data collected from various personal medical equipment in the health-care domain.
[2] We confirmed through references and experiments that at least 30% to 70% of the data collected by IoT devices is not related to the data analysis intended by the user. Communication setting, device configuration, etc. are the contents for information transmission and have nothing to do with the intended data analysis.
[3] We appreciate the introduction of the reference, https://ieeexplore.ieee.org/abstract/document/9139356. This is a very valuable comment that we could not confirm at the time we submitted the thesis. We are also currently conducting similar research for collaboration between fog computing nodes based on deep learning. However, the scope and experimental contents to be dealt with in this paper have clear limitations. The definition and function of the Fog module were explained by adding Figure 3.
[4] Four intelligent cameras are located at the vertex positions of the square. One camera can recognize and track an object within a 180 degree range, and by combining four cameras, capture data can be obtained from 8 locations within the shooting range.
[5] The conclusion section was modified to include research to add content that was not completed in the thesis and future research to enable collaboration between fog computing nodes based on deep learning.
Round 2
Reviewer 1 Report
I don't see a one-to-one response. The current version of manuscript can be accepted. But I'm not sure what the answer to my question is
Author Response
[1] We corrected and revised the overall repetition sentences.
[2] In abstract, We added the purpose of the study. "The main goal of this research is to use this multi-layer fog computing platform for enhancement of data analysis system based on IoT devices in realtime."
Reviewer 2 Report
The authors have improved the article, however, some crucial issues are still not addressed, which need to be solved before publishing. In particular, this concerns the clear statement of the research goal/contribution in the introduction and the graph in figure 6. The lines cannot be connected and you cannot apply a smoothening function here.
Author Response
1] We corrected and revised the overall repetition sentences.
[2] In abstract, We added the purpose of the study. "The main goal of this research is to use this multi-layer fog computing platform for enhancement of data analysis system based on IoT devices in realtime."
[3] We modified the Figure 6 graph based on your comments.
