{"id":"b821607e-3781-468e-b753-a153a325590a","arxiv_id":"2412.05949","paper_version":1,"verdict":"REJECT","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"A dual UAV cluster with collaborative beamforming is proposed for secure maritime links, with an improved mayfly optimizer, but the SINR model is dimensionally inconsistent.","lead":"Two UAV clusters use collaborative beamforming to relay data to a legitimate ship and jam an eavesdropper, while an improved mayfly optimizer balances signal strength, security, and energy. The read-worthy idea is secure long-range maritime relay, but the paper's central signal-to-noise model contains a dimensional error that invalidates the simulated comparisons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"SINR model in Eqs. (7)-(8) multiplies dB path losses (Eqs. (3),(6)) directly into linear power ratios, and double-counts array gain, so f1 and f2 are dimensionally invalid and every simulation comparison collapses.","rationale":"The reader's weakest_assumption targets exactly this dimensional inconsistency, and my independent check of Eqs. (3), (6), (7), (8) confirms it. This is the single most load-bearing concern because every quantitative result, including the Pareto fronts, convergence curves, and comparisons to non-CB/single CB/multi-hop, evaluates f1 and f2 through these equations. If the path-loss dB values were intended as linear factors, the notation '[dB]' and the 20log10 terms would be wrong; if they were intended as dB, the direct multiplication in Eq. (7) is invalid. In either reading, the reported SINR values do not represent physical signal-to-interference-plus-noise ratios. The additional multiplication by N_UR and N_UJ after normalizing array gain in Eqs. (2) and (5) compounds the error. I therefore agree with the REJECT verdict. The other issues raised by the reader (non-proof of NP-hardness, path-dependent energy model, under-specified hyperparameters) are secondary; even if corrected, they would not rescue the central claim while the SINR model remains invalid. A revised paper would need to correct the SINR formulation and rerun all simulations before any comparative claim can be assessed.","tokens_in":27535,"tokens_out":3740,"duration_ms":34913,"concrete_test":"Re-implement Eqs. (1)-(8) with PL_B and PL'_B converted to linear attenuation factors (e.g., PL_lin = 10^{-PL_dB/10}) and with the redundant N_UR and N_UJ multipliers removed (since G_v already accounts for array gain). Then recompute the SINR values for the optimized positions reported in Table V and Fig. 5. If the corrected SINRs are qualitatively different (sign change, ordering change, or more than a few dB shift), the paper's central claim that CB-based transmission is 'significantly better' is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Eq. (3) defines PL(Pr)[dB] as a sum including 20log10(d) and 20log10(4πfc/300), and Eq. (6) similarly defines PL'(Pj)[dB]. Yet Eqs. (7) and (8) insert these dB quantities as multiplicative factors in a linear SINR expression: γ_Bob = P_UR N_UR G_B PL_B / (P_UJ N_UJ G'_B PL'_B + σ^2). In linear radio modeling, the path-loss attenuation factor is 10^{-PL_dB/10}, not the dB value itself; multiplying a linear power by a dB number (which can be negative) is dimensionally meaningless. This is not a cosmetic issue: f1 and f2 are defined directly as these SINRs in Eqs. (11)-(12), and the paper's central claim (CB outperforms non-CB, single CB, and multi-hop; IMOMA outperforms competitors) is supported only by simulation results computed from this invalid objective. A second, independent flaw strengthens the concern: G_v in Eq. (2) already contains |AF_r|^2 (the beamforming array gain), so the extra factor N_UR in Eq. (7) counts the relay elements twice; the same holds for N_UJ in the denominator. Consequently, the reported SINR values in Fig. 5 and Table V (e.g., f1=15.5, f2=-27.9 for CB) are not physically interpretable, and the comparative conclusions are unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a dual-UAV-cluster maritime secure communication system in which one cluster forms a virtual antenna array relay and another forms a virtual antenna array jammer. It formulates a three-objective problem (maximize Bob SINR, minimize Willie SINR, minimize UAV flight energy), proposes an improved multi-objective mayfly algorithm with chaotic initialization and hybrid update strategies, and reports simulations claiming that the CB-based approach outperforms non-CB, single-CB, and multi-hop baselines and that IMOMA outperforms several metaheuristics. The system concept is relevant, but the stated SINR metric is dimensionally invalid and the energy objective is underspecified; the paper's central claims are therefore not supported.","tokens_in":27859,"tokens_out":7011,"duration_ms":68856,"significance":"If the system model were correct, the paper would address a timely problem: long-range maritime secure communications using UAV clusters with collaborative beamforming. The paper also has positive elements: it considers conflicting objectives, compares several baselines, and provides convergence diagnostics (IGD, ACR, solution distributions). However, the central quantitative claims rest on SINR expressions that mix dB and linear quantities and on an energy model that is not connected to any specified trajectory. The claimed CB advantage and the reported numerical comparisons are therefore not physically interpretable, and the significance of the findings cannot be evaluated.","major_comments":[{"comment":"The SINR expressions insert the dB path-loss values PL_B and PL'_B directly as multiplicative factors in linear power ratios. Since Eqs. (3) and (6) define path loss in dB (including 20log10 terms), the correct linear attenuation factor is 10^{-PL_dB/10} (or the equations must be rewritten in linear units before insertion). The same issue affects the noise term: Table IV lists σ² as -150 dBm, a logarithmic power, while Eq. (7) uses σ² as a linear noise power. Because f1 and f2 are defined directly as these SINRs in Eqs. (11)–(12), every objective value and every comparison in Section VI is computed from a dimensionally invalid metric. This is a central, load-bearing error.","section":"§III-B, Eqs. (7)–(8) and (3), (6)"},{"comment":"The factor N_UR in Eq. (7) appears to double-count the array size: G_B in Eq. (2) already contains |AF_r|^2, which for in-phase excitations is proportional to N_UR^2 (and the unnormalized array factor in Eq. (1) already sums over all relay UAVs). Unless the authors define P_UR as total power and G_B as a per-element gain, the extra N_UR multiplies the array gain a second time; the identical issue holds for N_UJ in the denominator. No such clarification is given, so the SINR values (e.g., f1=15.5, f2=-27.9 in Table V) are not physically interpretable. This extra factor also inflates the apparent advantage of CB over the non-CB baseline, making the conclusion in Section VIII an artifact of the formula rather than a demonstrated system result.","section":"§III-B, Eqs. (2) and (7)–(8)"},{"comment":"The NP-hardness claim is not established. The argument discretizes the continuous f3 and then states that the transformed problem 'can be regarded as a combinatorial optimization problem which is NP-hard [54]', without giving a reduction from a known NP-hard problem or specifying how the constraints in Eqs. (14d)–(14g) encode such a problem. The same holds for f1/f2, which are asserted to be 'usually NP-hard' by citation. The algorithm itself is a heuristic and does not require a rigorous NP-hardness proof, but the paper's motivation for a metaheuristic is weakened by the unsupported claim.","section":"§IV-C, Problem Analysis"},{"comment":"The energy objective f3 is underdetermined. Eq. (10) defines energy as an integral over a trajectory v(t), yet the optimization variables in SEMCMOP are only final UAV positions and excitation weights, and no trajectory is specified in the constraints or in the simulation setup. The values of f3 in Table V and the convergence experiments therefore cannot be reproduced or physically interpreted. If the authors intend straight-line constant-speed flight between initial and final positions, that assumption must be stated explicitly and included in Eq. (10) and in the constraint set.","section":"§III-C and §IV-B, Eq. (10) and Eq. (13)"}],"minor_comments":[{"comment":"In the large-scale dimension count, the jammer-set variables are written as (Xr, Yr, Zr, Ir) twice; the second should be (Xj, Yj, Zj, Ij).","section":"§IV-C"},{"comment":"The entries '6.6×1046.6×1046.6×104' and '1.4×1051.4×1051.4×105' appear corrupted by duplicated typesetting and should be corrected.","section":"Table V"},{"comment":"The comment '#Exploration phase' after Eq. (21) is inside the exploitation branch; it should read '#Exploitation phase'.","section":"Algorithm 3"},{"comment":"The claims that data-sharing overhead is 10–20 seconds and that the method saves 50–90% of time are not derived in this paper; they need a supporting calculation or a clearer reference to [74].","section":"Section VII"},{"comment":"The paper acknowledges that CB has limitations such as communication overhead and limited multi-user support, but these costs are not reflected in f1–f3; the Discussion in Section VII only addresses data-sharing overhead qualitatively.","section":"§II-B, last paragraph"}],"recommendation":"reject","confidential_remarks":"The manuscript self-reports that part of this work appeared in IEEE CSCWD 2023 [1]; I did not have that paper to compare, so the novelty of the extension beyond [1] should be checked by the editor. Also, the content is primarily a wireless-communications systems paper; the journal should confirm fit with its cs.DC scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the system concept is reasonable, but Eqs. (7) and (8) are fatal as written. The reader's stress-test is correct. PL_B and PL'_B are defined in dB by Eqs. (3) and (6), and the SINR formulas multiply them directly into a linear power ratio. A dB value is logarithmic; the correct path-loss factor is 10^{-PL/10}. On top of that, G_v in Eq. (2) already contains |AF_r|^2, which is the array gain, and Eq. (7) multiplies by N_UR again. Same for the jammer branch. So f1 and f2 are not physically interpretable, and every Pareto plot, comparison table, and curve in Section VI inherits the error. The paper's central claim—CB beats non-CB, single-CB, and multi-hop—is unsupported as stated.\n\nWhat is actually new is modest: applying CB-based UAV clusters to maritime relay with a jammer cluster, packaged as a three-objective problem (Bob SINR, Willie SINR, total flight energy). The IMOMA algorithm adds Tent-map initialization and hybrid WOA/AOA updates to a mayfly optimizer. That is a real engineering contribution, and the algorithm is described in enough detail to re-implement. The simulation setup is careful, and the overhead discussion in Section VII is a genuine attempt to address practicality.\n\nOther soft spots: the NP-hardness argument is a hand-wave with citations, not a proof; Eq. (10) evaluates energy from initial and final positions only, so f3 is trajectory-independent and cannot capture actual flight costs. On novelty, this overlaps the authors' CSCWD 2023 and ICC 2024 papers; the genuinely new piece is the optimizer, and there is no benchmark against those specific predecessors.\n\nCredit where due: the paper is clearly organized, the limitations of CB (overhead, limited multi-user support) are acknowledged, and the synchronization and data-sharing discussion shows the authors thought about implementation.\n\nVerdict: as submitted, the load-bearing metric is broken. The fix is straightforward—convert the dB values to linear and remove the extra element-count factors—but until then the simulation results are meaningless. A serious editor should desk-reject rather than spend referee cycles on a paper whose central comparisons are determined by a units error.","headline":"The dual-cluster maritime relay-and-jammer concept is coherent, but the SINR equations that carry the whole evaluation are dimensionally broken, so the reported results do not support the claims.","tokens_in":28467,"tokens_out":2142,"would_cite":false,"duration_ms":20834,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Twin UAV swarms, one relaying and one jamming, can give maritime links both long range and physical-layer secrecy at lower flight energy than multi-hop relays.","keywords":["maritime communications","UAV relay","collaborative beamforming","physical layer security","virtual antenna array","multi-objective optimization","improved mayfly algorithm","friendly jamming"],"falsifier":"Re-run the simulations with the dB path losses in Eqs. (3) and (6) explicitly converted to linear factors, $PL_{\\text{lin}}=10^{-PL_{\\text{dB}}/10}$, before they enter Eqs. (7) and (8). If the reported SINR values, for example about 20.8 dB for Bob and -39.9 dB for Willie in the larger network, change materially, then the dimensional slip in the objective functions is what produced the headline separation, and the comparison to multi-hop, non-CB, and single-CB baselines needs to be recomputed.","tokens_in":27308,"feed_emoji":"📡","tokens_out":6134,"duration_ms":56837,"temperature":0.7,"pith_summary":"This paper tries to establish that maritime wireless links can be made both longer and secret by dispatching two clusters of UAVs, one cluster forming a virtual antenna array to relay data to a legitimate ship and the other forming a virtual antenna array to beam jamming noise at an eavesdropping ship. The mechanism is collaborative beamforming: synchronizing the phases of many UAV-mounted antennas creates a strong directional signal at the intended receiver without any UAV flying the full distance. The paper formulates the problem as a three-objective optimization, namely maximize the legitimate ship's SINR, minimize the eavesdropper's SINR, and minimize total UAV flight energy, and solves it with an improved mayfly swarm algorithm. Simulation results claim the dual-cluster collaborative-beamforming approach beats non-CB, single-CB, and multi-hop baselines, and that the improved algorithm finds better frontier solutions than four comparison algorithms.","feed_headline":"Two UAV clusters beam secure data to ships at long range","feed_subtitle":"One cluster relays the signal, the second jams eavesdroppers; the design cuts flight energy versus multi-hop relays.","key_machinery":"The central object is the maritime UAV-enabled virtual antenna array (MUVAA), a swarm of synchronized UAVs whose collective array factor $AF_r(\\theta,\\phi)$ (relay) or $AF_j(\\theta',\\phi')$ (jammer), together with the antenna gain normalization and path-loss models, determines the SINR at Bob and Willie. The array factor and gain equations convert UAV positions and excitation current weights into directional gain toward each vessel; the SINR expressions in Eqs. (7) and (8) then combine the relay gain, jammer gain, and path losses into the objectives $f_1$ and $f_2$, while the propulsion energy model in Eqs. (9)-(10) gives $f_3$. The proposed IMOMA carries the optimization: Tent-chaotic initialization spreads the initial solution population, and hybrid WOA/AOA update rules move the relay and jammer sets with different step sizes and boundary handling.","core_discovery":"On its own terms, the paper's discovery is that two UAV clusters working as beamforming arrays can serve as a long-range relay and a long-range jammer at the same time, and that the joint placement and current-weight design of both arrays can be posed as a Pareto optimization that meaningfully separates the legitimate and eavesdropping vessels. The key outcome is a clean separation in SINR: in the larger-scale simulation the legitimate vessel's SINR is positive (about 20.8 dB) while the eavesdropper's is deeply negative (about -39.9 dB), and the relay-and-jammer arrangement achieves this with total flight energy about an order of magnitude below the multi-hop baseline. If the model holds, this means friendly jamming does not have to be close to the eavesdropper: collaborative beamforming lets a distant UAV swarm concentrate jamming power on the eavesdropper while a second swarm concentrates data power on the legitimate ship.","pith_inferences":["An untested but direct extension is to track vessels in motion rather than fixed positions; the paper itself lists this as future work, and a dynamic version would need the energy model to include trajectory, which the current start/end formula cannot capture.","The same two-array geometry should transfer to non-maritime long-range settings, such as rural or disaster-area links, where one swarm relays data and a second swarm protects against eavesdroppers; nothing in the SINR math is specific to sea-surface propagation except the path-loss constants.","A fair numerical test would recompute the SINR objectives after converting the dB path-loss values in Eqs. (3) and (6) into linear factors; if the reported results survive that conversion, the comparison claims are credible, and if not, the quantitative comparisons would need to be redone."],"forward_implications":["If the central claim is correct, a shore station can talk to a distant vessel through one tightly packed UAV swarm, and another swarm can shield that link from a known eavesdropper location without either swarm flying close to the vessels.","The reported SINR separation means physical-layer security can be achieved as a by-product of array geometry and current-weight design, rather than requiring cryptography or high-power jamming near the target.","The large energy gap versus multi-hop relaying would make the CB-based system the preferred architecture when UAV endurance is the constraint, provided the synchronization overhead is as small as the paper states.","The IMOMA's improvement on the eavesdropper-side objective (up to 43.20%) indicates that most of the algorithm's gain is in shaping the jamming array, which could be the deciding criterion in selecting an optimizer for this problem class."],"supporting_citations":[{"why":"Supplies the collaborative-beamforming relay method and the secure-and-energy-efficient UAV relay design that this dual-cluster system extends.","marker":"[7]"},{"why":"Provides the virtual-antenna-array formulation and multi-objective optimization approach for UAV swarms that underpins the MUVAA model.","marker":"[40]"},{"why":"Supplies the maritime air-to-sea path-loss model used in Eqs. (3) and (6).","marker":"[51]"},{"why":"Provides the closed-form UAV propulsion energy model used in Eq. (10) for the flight-energy objective.","marker":"[53]"},{"why":"Defines the mayfly optimization algorithm that IMOMA modifies with chaotic initialization and hybrid update rules.","marker":"[60]"},{"why":"Supplies the whale-optimization update strategy that IMOMA uses for the jammer-set positions and weights.","marker":"[64]"},{"why":"Supplies the arithmetic-optimization update strategy that IMOMA uses for the relay-set positions and weights.","marker":"[65]"}],"fun_headline_variants":["UAV pairs: relay and jammer clusters secure sea comms","Beamforming UAV swarms jam eavesdroppers, relay ship data","Two UAV arrays make secure long-range ship links","UAV cluster relay and jammer cut energy, boost SINR","Collaborative beamforming from UAV swarms shields ships"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the signal-loss numbers the paper computes in decibels are converted to ordinary multiplying factors before being fed into the signal-to-interference-plus-noise ratio formulas; if they are not, those formulas have no meaningful units.","fun_headline_variants_meta":{"raw":{"variants":["UAV pairs: relay and jammer clusters secure sea comms","Beamforming UAV swarms jam eavesdroppers, relay ship data","Two UAV arrays make secure long-range ship links","UAV cluster relay and jammer cut energy, boost SINR","Collaborative beamforming from UAV swarms shields ships"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000362,"raw_usage":{"total_tokens":1975,"prompt_tokens":991,"completion_tokens":984,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":607,"completion_tokens_details":{"reasoning_tokens":899}},"tokens_in":607,"tokens_out":984,"duration_ms":9361,"temperature":1.0,"reasoning_tokens":899,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T20:10:27.134071+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the simulations with the dB path losses in Eqs. (3) and (6) explicitly converted to linear factors, $PL_{\\text{lin}}=10^{-PL_{\\text{dB}}/10}$, before they enter Eqs. (7) and (8). If the reported SINR values, for example about 20.8 dB for Bob and -39.9 dB for Willie in the larger network, change materially, then the dimensional slip in the objective functions is what produced the headline separation, and the comparison to multi-hop, non-CB, and single-CB baselines needs to be recomputed.","supporting_citations":[{"cited_title":"Hybrid satellite-UA V- terrestrial networks for 6G ubiquitous coverage: A maritime communi- cations perspective,","cited_arxiv_id":null,"evidence_quote":"Supplies the maritime air-to-sea path-loss model used in Eqs. (3) and (6)."},{"cited_title":"A mayfly optimization algorithm,","cited_arxiv_id":null,"evidence_quote":"Defines the mayfly optimization algorithm that IMOMA modifies with chaotic initialization and hybrid update rules."},{"cited_title":"The whale optimization algorithm,","cited_arxiv_id":null,"evidence_quote":"Supplies the whale-optimization update strategy that IMOMA uses for the jammer-set positions and weights."},{"cited_title":"The arithmetic optimization algorithm,","cited_arxiv_id":null,"evidence_quote":"Supplies the arithmetic-optimization update strategy that IMOMA uses for the relay-set positions and weights."}],"review_version":1}