kumoh national institute of technology
Networked Systems Lab.

Review Comment

NSL > Education> Review Comment
Muhammad Royyan, Dong-Seong Kim, "Bio-inspired Congestion Control in Large-Scale Wireless Sensors Networks ", IEEE Communication Letter (IF 1.98)
By :
Date : 2018-02-19
Views : 55

Dear Author(s):

The review of the referenced manuscript, CL2018-0120, entitled "Bio-Inspired Congestion Control in Large-Scale Wireless Sensor Networks", is now complete.

One referee was negative about the letter and other two referees, although they have expressed interest, raised critical issues. The raised concerns include weak contribution and lack of novelty,
and incomplete performance evaluation with incorrect/unjustified wireless standard assumptions and simplistic network topology.

Additionally, some critical assumptions (such as constant traffic flow $K_i = K$, as well as $\alpa_ii = \beta$) and are not justified or referenced. Although deeply simplifying the stability analysis, they should be motivated and discussed in terms of WSN properties.

I am afraid that I fully agree with the raised comments and thus I am unable to accept the manuscript. Due to the fundamental nature of the flaws, your paper may not be resubmitted for review.

For the detailed list of comments and issues, please refer to the reviewers' comments at the end of this email.

Thank you for submitting your work to the IEEE Communications Letters.

Regards,

Dr. Marcello Caleffi
Associate Editor
IEEE Communications Letters

------------------------------------------------------------------------------------------------

Reviewer: 1

Comments to the Author
The paper introduces a new Bio-Inspired Congestion Control in Large-Scale
Wireless Sensor Networks. I like the idea of competitive Lotka-Volterra (C-LV) model to avoid congestion and further apply Particle Swarm Optimization (PSO) over C-LV to show fairness.

I believe the paper has potential to be published after some specific modifications, as mentioned below:

(1) The author introduced C-LV. But, there is no competitive analysis of C-LV shown. How much competitive is it?

(2) The PSO is a optimization tool. But there is no analysis showing how much close to optimal solution could the authors obtain? Generally these algorithm do not provide global optimal solution, but some near-optimal solution. Then how much close to global optimality could be achieved?

(3) Simulation results are good, but lack any comparison with existing results. Equation(3) ~ Equation(6) do not say anything about it.

------------------------------------------------------------------------------------------------

Reviewer: 2

Comments to the Author
The paper addresses the congestion control and the dynamic rate control in wireless sensor networks.

The problem is faced through an hybrid approach that uses two well-known algorithms, that are L-CV and PSO. The literature widely discusses these algorithms and many works already presents the same analytical models reported in Section II and Section III.

Simulations were conducted though ns-3. But, a simple and single-hop network based on the IEEE 802.11 (i.e., WiFi) is taken into account. WSNs, instead, use different MAC protocols, different physical settings and transmission rates, as well as different topologies (including multi-hop).

Figure 5 reports the most important results. Many works in the literature that uses L-CV and PSO in WSN report similar curves. Thus, it is not clear the novelty level of the submitted manuscript.

------------------------------------------------------------------------------------------------

Reviewer: 3

Comments to the Author
A hybrid bio-inspired algorithm was proposed for congestion control in large-scale WSNs in this paper. First, the C-LV model-based scheme is applied to avoid congestion in WSNs and to maintain fairness among sensor nodes. In addition, the PSO algorithm is applied to enhances the C-LV scheme by optimizing the parameter for minimizing end-to-end delay and to help the system adapt to changes in the number of sensor nodes. This paper is somehow interesting, detailed comments:
1. The complexity of the proposed method is not very clear, which shouldbe discussed.
2. The proposed method should be compared to the existing works to demonstrate the novelty of this paper, for example, this paper can compare to the following works in introduction or simulation results:
"Aggressive Congestion Control Mechanism for Space Systems," IEEE Aerospace and Electronic Systems Magazine, vol. 31, no. 3, pp. 28-33, June 2016.
3. The writing of this paper should be polished.

------------------------------------------------------------------------------------------------

Reviewer: 4

Comments to the Author
The paper presents a Bio-inspired algorithm that could be used to optimize the fairness of resources for wireless sensor networks. The algorithm is based on a combination of Lotka-voltera competition algorithm, as well as particle swarm optimization. Since the Lotka Voltera has parameters that reflects on the rate of resource consumption, determining this parameter is highly dependent on the scenario (e.g size of the species, number of species, etc.). In order to determine the optimum level for this, the authors used the PSO algorithm. The simulation was evaluated on a star topology.

The work is interesting, and there has been many works in the past that has looked at applying LV algorithms for managing resources in computer networks, which the authors has avoided to reference. This work gives an added flavour of using PSO on top to determine the optimal parameters. However, there was a number of disappointing aspects to this paper. Firstly, the authors should have done a comparison of just pure LV that does not have an optimum value of the parameter, and compare two extremes (when one species is highly competitive, compared to an even competition amongst all) - so in other words static parameters. Then we can see how the optimum value fairs against these two extreme cases. Secondly, it would have been nice if the authors also did tests for other random topologies (just one random topology would be sufficient), which already had predefined paths, and competition was conducted on these paths. This would be very interesting, and a nice contribution, because as the rate of transmission between different nodes changes, there are chances that the resource consumption that overlaps between paths will lead to dynamic changes of the parameters that is only a segment of that path (rather than the entire path). This would definitely make the paper more interesting, and will add value to the need for the algorithms. One thing I am also confused with, is that in the star topology, the nodes all send packets straight to a sink. Does that mean the competition is for the resources at the sink itself, because there want be any competition along each path? It would be good why the authors also only had 10 as a maximum particle search bound for the particles in the PSO. If they authors could fix all these issues, this would be a strong paper and should be considered for publication.