A Function as a Service Based Fog Robotic System for Cognitive Robots

Round 1

Reviewer 1 Report

The topic are related with Robots and Wireless Communications.

In this paper, the author assume that there is no delay or any communication problems between Robots and Cloud  Layer. This problem is how do HPC and HTC  make available on site. I hope that you can discuss this issues.

Author Response

Point 1: In this paper, the author assume that there is no delay or any communication problems between Robots and Cloud  Layer. This problem is how do HPC and HTC  make available on site. I hope that you can discuss this issues.

Response 1: The comment of the reviewer on the communication delay is reasonable. Therefore, the author defined “service time” which includes computing time of cloud and networking delays, revised the part of results and discussion, and added some discussion on the topic of delays (Line 344 to Line 353). And the captions and the label of X axis of Figure 10 and 11 were changed to “service time”.

Reviewer 2 Report

This paper shows a distributed IIoT inspired system for cognitive robots, using the concepts of both fog and cloud computing. My main concern is that there is a lack of details regarding the different algorithms used in the experiment for speech recognition, image segmentation and so on. Please, expand this part of the paper.   Other comments: Line 106: I would expand this paragraph a little bit, explicitly stating the differences of [18], [25] and [27] with the authors' approach in this paper. Line 195: "computations is offloaded" => "computations are offloaded" Line 211: "security as well computing power" => "security as well as computing power" Line 224: "shows that the" => "shows the" Line 334: "average computing". How many executions? Details, please... Line 353: Figure 8's caption is in a different page than figure itself.

Author Response

Point 1: My main concern is that there is a lack of details regarding the different algorithms used in the experiment for speech recognition, image segmentation and so on. Please, expand this part of the paper.  

Response 1: According to the comment of reviewer, the author added details on speech recognition, image segmentation and syntactic parsing with references.

Speech recognition using Google Cloud was tagged two references of a web site and a paper for theoretic background. (References: 29, 30)

The part of image segmentation was enhanced by details of 3D segmentation with normal vector orientation.

Link parser also had a reference paper with a web site for the cloud service. (Line 310 to Line 313)

Point 2: Line 106: I would expand this paragraph a little bit, explicitly stating the differences of [18], [25] and [27] with the authors' approach in this paper.

Response 2: The previous fog robotic models were detailed with one more reference paper and the strong point of the comprehensive characteristic of proposed model was stressed. (Line 106 to Line 127)

 

Point 3:

Line 195: "computations is offloaded" => "computations are offloaded"  

Response 3: Corrected in Line 203

 

Point 4:

Line 211: "security as well computing power" => "security as well as computing power"  

Response 4: Corrected in Line 219

 

Point 5:

Line 224: "shows that the" => "shows the"  

Response 5: Corrected in Line 232

 

Point 6:

Line 334: "average computing". How many executions? Details, please...  

Response 6: Figure 10 shows the average service times of 20 times of trial with two types of FaaS-FR models to utilize the Google Cloud for speech recognition.

 

Point 7:

Line 353: Figure 8's caption is in a different page than figure itself. 

Response 7: Corrected