Energy-Efficient Wireless Sensor Networks Through PUMA-Based Clustering and Grid Routing
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript identifies the gap it aims to fill integrating PUMA with grid-based routing. However, the novelty statement needs stronger differentiation from existing hybrid metaheuristic and multi-hop works
The paper should explicitly clarify what unique technical mechanism PUMA-GRID introduces beyond combining existing components.
Consider adding a comparative conceptual diagram contrasting PUMA-GRID with prior AEO-GRID or PUMA-SH versions. I would high suggest to compare the work with following novel works. https://doi.org/10.3390/electronics13224388
The binary conversion process of PUMA’s continuous update functions is not defined.
The fitness function is central but needs deeper justification. The penalty coefficient α = 10 appears arbitrary.
Sensitivity analysis would strengthen credibility. The negative sign before the residual energy term should be clarified
The manuscript should justify normalizing or scaling the three terms; currently, distance and energy have very different magnitudes.
Some simulation parameters require justification or consistency checking. PUMA-GRID uses multi-hop, but LEACH in baseline comparison is single-hop. This inherently disadvantages LEACH unless LEACH-M is also included.
The manuscript should state clearly whether All protocols operate under the same clustering period, All use CH rotation each round or periodically, The number of CHs is fixed or adaptive in baselines?
Some figures lacks Axis labels, Units, Clear legends or color coding.
The CFI metric should include an explanation for why fairness is evaluated at the grid-cell level and not by coverage radius.
Section 3.6 (PO pseudocode) updates the global best only in exploitation, but the manuscript claims PUMA employs adaptive switching that reacts to improvements. This structural mismatch should be clarified.
Several symbols are reused in different contexts (e.g., N used for population size and number of nodes).
Acronyms should be defined at first mention (e.g., CMD).
The manuscript repeatedly discusses “machine-learning-inspired” routing, yet no ML components exist. Consider removing or revising this phrase.
Typographical issues:
Missing spaces (e.g., “CHtoBS”).
Equation numbering inconsistency (Equation 2 referenced where Equation 4 is intended).
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsPlease see the attachment.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper proposes PUMA-GRID, a hybrid WSN routing protocol combining metaheuristic CH selection with grid-based routing to improve energy use and network lifetime. The idea is relevant and experiments are extensive. However, the paper has several areas where clarity, depth, and rigor can be significantly improved.
- The introduction needs a clearer transition toward the proposed PUMA-GRID protocol and an explicit statement linking the problem to the solution. It should also more clearly contrast PUMA-GRID with existing protocols early on—highlighting, for example, that methods like LEACH and AEO lack energy-aware CH rotation, whereas PUMA-GRID incorporates it directly.
- The related-work section should be more critical and show clearly how PUMA-GRID addresses the specific weaknesses of previous protocols. It also needs a final synthesis that explicitly links insights from prior studies to the design choices of PUMA-GRID.
- Algorithm 4 is inconsistent with the earlier PO description: the adaptive exploration–exploitation switching mechanism from Algorithm 1 is missing and must be integrated. The grid-based routing algorithm also lacks clarity on grid construction and adjacency definitions, requiring brief explanation or citation. Additionally, the penalty coefficient α = 10 is used without justification and needs a short rationale to strengthen methodological rigor.
- The choice of a 40 m grid size is not justified, despite testing a 10–40 m range, and requires explanation. Figure clarity also needs improvement, as many graph labels are difficult to interpret from the text alone. Finally, the discussion of control overhead should explicitly link PUMA-GRID’s higher overhead to its improved performance, emphasizing it as an expected trade-off.
- The unexpectedly superior performance of AEO-GRID with an external base station requires deeper discussion—potentially examining whether AEO’s fitness function favors long-range transmissions or whether PO suffers from premature convergence in this scenario. Additionally, the conclusion would benefit from a qualitative comparison table summarizing the strengths and weaknesses of all evaluated protocols, including PUMA-GRID.
- The most critical issue is that Algorithm 4 is flawed and must incorporate the adaptive scoring mechanism from Algorithm 1. For Algorithm 5, clarification is needed on whether the argmin is computed over all cluster heads in adjacent grids or only a subset.
- Ensure that all references are consistently and correctly formatted according to the journal’s guidelines.
- The conclusion should more explicitly restate the paper’s contributions relative to the gaps highlighted earlier. Future work suggestions are appropriate but could be more specific, e.g., proposing “reinforcement learning for dynamic weight adaptation” instead of general “adaptive methods.”
- The manuscript contains minor grammatical errors and occasional awkward phrasing, warranting careful proofreading.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper introduces a hybrid clustering and grid-based routing protocol. The conceptual comparison in Fig. 2 is helpful, but the distinction from PUMA-SH and AEO-GRID needs quantitative evidence, not only descriptive differences.
The fitness function mixes quantities with different magnitudes distance in meters vs. energy in Joules. Although the authors argue that the weights implicitly serve as a scaler, a formal normalization or a sensitivity experiment would strengthen the justification.
The negative sign before the residual energy term (–w₃ · Energy) is correctly explained, but the impact of this term on convergence behaviour is not shown.
Provide plots of convergence curves for different energy-weight settings.
The penalty coefficient α = 10 appears to be chosen heuristically. A systematic sensitivity analysis (α ∊ {1, 5, 10, 20, 50}) would support the chosen value.
Algorithm 4 mixes continuous update rules (PUMA) with binary conversion, but the timing of binarization is unclear. Please specify explicitly:
Is binarization applied after every update?
Does binarization occur before or after bound-checking?
The exploration/exploitation counters (ScoreExplore, ScoreExploit) are mentioned but not precisely defined. Provide exact update conditions and thresholds.
In some steps the notation changes between leader, best solution, and best puma. These terms should be standardized.
The routing algorithm resembles an A*-inspired heuristic, but it lacks key elements such as heuristic distance evaluation or priority queuing. As a result, the algorithm is closer to greedy geographic forwarding.
Please clarify whether this approach truly includes A* components.
The grid relay decision uses
Dist(CHáµ¢, CHâ±¼) + Dist(CHâ±¼, BS).
This is equivalent to minimizing a 2-hop cost but does not generalize to longer paths, so the claim of “multi-hop corridors” is biased toward local decisions. Consider evaluating whether this produces suboptimal long-range paths.
The algorithm does not address routing loops. Provide explicit loop-avoidance conditions.
The simulation relies on 100 nodes in a 100×100 m field, but no justification is given for typical WSN density. Consider testing sparse (50 nodes) and dense (200 nodes) deployments to evaluate scalability.
Litterature review is not up to mark I would highly suggest to read following papers and write the litterature review based on these researches. https://doi.org/10.1049/pel2.70101
https://doi.org/10.3389/fphy.2024.1374138
The description mentions MATLAB simulations, but no runtime or hardware details are provided. Metaheuristics with P population × N nodes × I iterations can be computationally heavy; a runtime comparison with AEO and ABC would be informative.
The network models only static nodes; realistic WSNs experience node failures, mobility, and environment variability. Discuss limitations of your assumptions.
Results frequently state “improved performance” but several plots (FND, HND, LND) are highly sensitive to the weight combination. The authors should:
Provide confidence intervals over multiple runs (≥30 trials).
Report overall statistical significance (ANOVA or Wilcoxon test).
The coverage fairness index (CFI) is defined per grid, which is logical, but the authors should clarify:
What grid size was used to compute CFI?
Are CFI values comparable when grid size changes?
In multi-hop routing, the energy consumed by relay CHs may create bottlenecks (energy holes). The authors should examine relay-load distribution.
The manuscript is generally well-structured but overly long in the related-work section. Consider condensing repetitive descriptions.
Some equations lack numbering, and some variables (e.g., learning rates, movement coefficients) are not defined in tables.
Figures need higher resolution and consistent labeling conventions.
To convincingly argue superiority, consider adding comparisons with:
PSO-based CH selection with multi-hop routing
GWO-based clustering
At least one reinforcement-learning-based routing protocol
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsNo further questions.
Author Response
There are no further questions from the reviewer.
Reviewer 3 Report
Comments and Suggestions for AuthorsWe sincerely thank the authors for their comprehensive revisions in response to the comments. The manuscript has been improved, with all major comments addressed efficiently.
As final version, we recommend addressing the following minor points:
- Improve visual clarity by simplifying the figure and enlarging axis labels and legends
- Explicitly mention the main competing protocol, AEO-GRID, in the abstract for greater clarity.
- Add a brief bridging sentence in Section 2 to clearly link identified gaps to how PUMA-GRID addresses them.
- In Section 4.3, briefly clarify why the 40 m grid size was chosen over the similarly performing 30 m size.
- Perform a final proofreading pass to fix minor grammatical issues and improve phrasing.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
