Bounding the Upper Delays of the Tactile Internet Using Deterministic Network Calculus
Round 1
Reviewer 1 Report
This paper investigates the delaying issues in tactile Internet using Deterministic Network Calculus. The authors have performed numerical analysis and some experiments on NS3 to show the impact of relevant parameters on a given network.
Generally, the paper is well-written and organized and follow a typical pattern. It is easy to follow the paper but I have following major reservations:
1- The tactile Internet has been investigated, the DNC has its own features. What is problem that this paper has addressed. It is a common understanding that these types of network will have delays. I am unable to understand the research gap that has been filled by this research.
2- Figure 1 is a way too typical example, Things are working fine and are in practice for years, How have the proposed research addressed this issue. Is there any issue at all?
3- The results presented in Figure 3 to 6. What are the values for comparison. what is the benchmark. Are the obtained results good or bad. How will we compare its performance. is there any performance comparison.
4- The IEEE P1918 is already there solving most of the issues. What addition is the proposed search making.,
5- I would propose to focus more on simulation experiments in these studies.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
In this paper, the authors proposed an analytical method for modeling and evaluating the end-to-end delay upper bound for Tactile Internet, which is a timely and interesting problem. Deterministic network calculus is employed as the primary tool in the proposed analytical method, which is agnostic to system implementations and thus applicable to the various networking domains in a Tactile Internet.
On the other hand, the reviewer has concerns about the following aspects of the manuscript that need to be addressed in a revision.
For the cases of multiple flows with single/multiple switches, the proposed model assumes interference among the flows; therefore, it obtains a delay bound for each flow that is impacted by the aggregated traffic of all other (interfering) flows. This implies that no delay bound can be guaranteed for any individual flow unless the traffic of all other flows sharing the same set of switches is strictly constrained, which is not realistic in many practical networking scenarios. SDN/NFV-based network slicing is expected to be a key attribute in future Internet (including Tactile Internet) that offers isolation (with resource reservation) among flows thus removing inter-flow interference. The authors are encouraged to refer to the following work on network calculus-based delay analysis for SDN/NFV-based networks: "Modeling and Performance Analysis for Service Function Chaining in the SDN/NFV Architecture" published in IEEE NetSoft 2018, and include related discussion in their revision.
Another issue of the proposed modeling method is the assumption of no data transformation (caused by computation functions) in the entire end-to-end system, which is required by the Tandem Theorem. However, a convergence of networking and edge-cloud computing is expected in the future Internet, especially in the master-network-operator scenario shown in Figure 1 where end-to-end delay actually comprises transmission latency in the network domain and process latency in the salve domain. The authors might find the delay analysis for composite network-compute system reported in the following paper useful as a reference: "Modeling and Performance Analysis for Composite Network-Compute Service Provisioning in Software-Defined Cloud Environments" published in Elsevier Digital Communications and Networks, vol. 1, no. 3, 2015.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Please add more refrences avoiding self citations if possible. Moderate english editing is required. Please elaborate on the benefits of you paper compaered to other work , and why we should consider it.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have tried to address my comments but I am not convinced.
In responding to my comment 1, you have explained what DNC is and have described network performance analysis. What you have provided is a common understanding. My question remains the same, "What is the problem that this paper has solved and which has not been solved in the existing literature". Moreover, Is there a problem at all. In other words what research gap has been filled and what is the observation that this is actually a research gap.
Figure 1 is a generic figure and does not actually reflect a tactile network. Simply writing a word " Bi-Directional Haptic Control" wont make it a tactile network figure.
In comparison, you can not compare your own results. You must have some base line model, some existing results, some standards that you have improved. There must be some performance improvement. This will take you back to my first comment.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
The revisions have addressed the reviewer's concerns.
Author Response
Thanks again for your comments.
Round 3
Reviewer 1 Report
Your claim is that there are no papers that use DNC for the analysis of tactile networks, and this study does the job. My concern remains the same there is no need to combine DNC and tactile networks because the findings and results can be two obvious.
This is the reason, the conclusions and results are too obvious in the proposed study
I am also not convinced that the authors have taken enough time to add a new comparison with [39].
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
