{"id":"ac06e7cb-2488-4763-8c80-a4cae0309ff2","arxiv_id":"2412.17934","paper_version":5,"verdict":"REJECT","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A simple ns-3 simulation shows that an obstacle between a UAV and a ground user reduces throughput and increases delay, and repositioning the UAV to restore line of sight recovers performance.","lead":"This paper simulates a UAV carrying a Wi-Fi access point serving a ground user in ns-3, with and without a building blocking the line of sight. It reports that throughput drops from about 51.8 to 41.5 Mbit/s and delay rises when the building blocks the link, and that moving the UAV to a clear position restores performance.","discovery_kind":"incremental","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The obstacle effect is confounded: scenario 1 uses Friis loss, scenario 2 uses HybridBuildings loss, so the throughput drop may reflect the model change rather than the building.","rationale":"The reader's weakest assumption flagged simulator fidelity and misconfiguration as the risk to the central claim. My stress-test narrows this to a concrete internal-confounding flaw: the paper never compares like with like. Because the two scenarios differ in both the presence of the building and the choice of propagation loss model, the simulation cannot support the causal statement that the building caused the QoS degradation. This concern is not about disagreement with the broader literature; it is about the internal validity of the presented evidence. The reader's verdict of REJECT is therefore not changed. I note in good faith that the qualitative direction of the result is expected and consistent with prior work, but the paper's own measurement approach, as described, does not demonstrate it. A single control run with HybridBuildingsPropagationLossModel and no building would settle the issue and would be inexpensive to perform.","tokens_in":7177,"tokens_out":2331,"duration_ms":23791,"concrete_test":"Re-run the obstacle-free scenario (UAV at (30.0, 0.0, 10.0), UE at (0.0, 0.0, 0.0), 5 GHz, same traffic and settings as Table II) while replacing Friis with HybridBuildingsPropagationLossModel and leaving the scene empty of buildings. If the average throughput falls to approximately 41.5 Mbit/s or the end-to-end delay rises to approximately 10.5 ms, then the reported obstacle effect is an artifact of the propagation-model switch rather than of the building itself.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is that placing an obstacle between UAV and UE degrades QoS and that repositioning to LoS restores it. The reported evidence, however, does not isolate the obstacle as the cause. In Section IV-A, the obstacle-free scenario uses the Friis propagation loss model (Table I), while the obstacle scenario uses the HybridBuildingsPropagationLossModel (Table II). The comparison therefore changes two variables at once: building presence and propagation model. The measured drop from 51.819 to 41.5377 Mbit/s and the delay increase from 6.61881 to 10.5386 ms could be produced by the different default attenuation and shadowing behavior of HybridBuildingsPropagationLossModel even with no building present. No control run with HybridBuildingsPropagationLossModel and an empty scene is reported, so the building's marginal effect is unidentified. This is load-bearing because the entire conclusion rests on attributing the QoS change to the obstacle and to the UAV repositioning. The physically implausible 100 ms guard interval in both scenarios further weakens the link to realistic 802.11ac operation, but the model confound alone is sufficient to undermine the central claim as stated.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a simulation-based study of an air-to-ground Wi-Fi link between a UAV-mounted access point and a ground user in ns-3. It compares an obstacle-free scenario using the Friis propagation loss model with an obstacle scenario using the HybridBuildingsPropagationLossModel, reporting average throughput, PDR, packet loss, and end-to-end delay over 10 runs. The authors claim that placing a building between the UAV and the user degrades QoS and that moving the UAV to a line-of-sight position restores performance. They position the work as an obstacle-aware channel characterization and as a step toward obstacle-aware UAV positioning.","tokens_in":7393,"tokens_out":4804,"duration_ms":46974,"significance":"If the claims were established, the paper would provide a modest simulation-based confirmation that 5 GHz air-to-ground links degrade when a building blocks line of sight and that repositioning recovers performance. The paper has some strengths: the simulation parameters are listed in tables, and the use of 10 repeated runs is better than a single run. However, the central comparison conflates the propagation model with the presence of the obstacle, the reported statistics lack any uncertainty measure, and a listed Wi-Fi guard interval of 100 ms is physically implausible for 802.11ac. As written, the result is largely a demonstration of ns-3's built-in building attenuation rather than an independent characterization of air-to-ground channels, so its significance for the UAV placement literature is currently limited.","major_comments":[{"comment":"The obstacle-free and obstacle scenarios differ in two variables at once: the presence of the building and the propagation loss model. Scenario 1 uses the Friis path loss model, while scenario 2 uses HybridBuildingsPropagationLossModel. No control run with HybridBuildingsPropagationLossModel and an empty scene is reported. The observed drop in average throughput from 51.819 to 41.5377 Mbit/s and the increase in delay from 6.61881 to 10.5386 ms are therefore not causally attributable to the obstacle. This is load-bearing because the paper's conclusion rests on attributing the QoS change to the building. A control experiment with the same propagation model and the building present/absent, or an otherwise identical setup using one propagation model, is needed before the central claim can be evaluated.","section":"Section IV-A and IV-C"},{"comment":"The paper reports averages over 10 simulation runs but provides no standard deviations, confidence intervals, per-run values, or distribution plots for the main metrics. Without a measure of run-to-run variability, the reported differences in throughput and delay cannot be distinguished from simulation noise. The authors should report variance or confidence intervals, and ideally show per-run data or box plots, to support the quantitative comparison.","section":"Section IV-C and Tables I-II"},{"comment":"The guard interval is listed as 100 ms. In 802.11ac OFDM, guard intervals are on the order of 0.4 to 3.2 microseconds, so 100 ms is not a physically valid parameter and suggests a unit or configuration error. This parameter undermines the fidelity of the reported results. The authors must correct the guard interval to a standard 802.11ac value and rerun the simulations, or explicitly justify why 100 ms was used.","section":"Tables I and II"},{"comment":"The central claim is substantially overstated relative to the evidence. A single static scenario with one building, one ground user, and one frequency band is used to support a broad conclusion about UAV positioning with obstacles improving QoS and coverage. Moreover, since HybridBuildingsPropagationLossModel is specifically designed to add building attenuation, the simulation outcome largely restates the model's assumptions rather than providing an independent empirical characterization. The paper should be reframed as a demonstration of an ns-3 simulation workflow, with the scope clearly limited to this scenario, or it should include additional scenarios and a validation discussion.","section":"Sections I, V, and VI"}],"minor_comments":[{"comment":"The figure captions contain repeated typos and are internally inconsistent: 'Throghput' appears in several captions, Fig. 8 is captioned as a probability distribution function but appears to show a throughput time series, and Section VI refers to a 'Fig. 9' that does not exist.","section":"Figures 4-8"},{"comment":"Table II omits the packet size row that is present in Table I, so the two listed configurations are not fully comparable as presented. The table headers also misspell 'SCERARIO'.","section":"Tables I and II"},{"comment":"The statement that obstacle-aware UAV repositioning 'has not been investigated so far' is contradicted by the paper's own references, including [1], [2], and [22]. The novelty claim should be softened and positioned relative to those works.","section":"Section I"},{"comment":"There are typographical issues such as 'Th e' in the abstract and 'Tthroughput' in the metrics list. These should be corrected for readability.","section":"Abstract and Section IV-B"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is quite close to the minimum bar for a publishable simulation study. The core qualitative message is plausible and likely true, but the current evidence is not clean: the missing HybridBuildings control run is a serious confound, and the 100 ms guard interval suggests a configuration error. If the authors add the control experiment, correct the guard interval, and reframe the contribution as a demonstration of an ns-3 obstacle-aware simulation workflow, the paper could become acceptable for a venue that values reproducible simulation studies. As submitted, however, the central quantitative comparison is not yet supported."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nShort take: the paper's conclusion that an obstacle degrades UAV-to-ground throughput and delay is physically true, but this simulation does not establish it. The obstacle-free scene uses FriisPropagationLossModel and the obstacle scene uses HybridBuildingsPropagationLossModel, so the comparison changes two variables at once. No control run with HybridBuildings and an empty scene is reported. That confound is load-bearing and, as reported, invalidates the headline comparison.\n\nWhat is actually there: a minimal ns-3 example of placing a building and applying the HybridBuildings model. That could be a useful tutorial snippet. The qualitative direction—LoS outperforms NLoS—is correct, and the within-scenario repositioning comparison (same propagation model) is less vulnerable.\n\nSoft spots, aside from the confound: the table lists a 100 ms guard interval for 802.11ac, which is physically impossible (guard intervals are nanoseconds to microseconds). The figures are mislabeled, the typo in 'throughput' appears in captions and text, and no variance or confidence intervals are given for the 10 runs. No code or config is shipped. The references include 2025 preprints from a different author surname in a 2024 paper, which makes the related-work review unreliable. The paper claims 'AG channel characterization' but measures only throughput, PDR, and delay—link-layer QoS, not channel parameters like path loss or fading.\n\nNet: this is a tutorial-grade simulation, not a research contribution. The topic is relevant and the basic idea is sound, but the execution and reporting are below the bar. The confound alone is sufficient for rejection as a research preprint. I would not cite it, and I wouldn't spend a reading group session on it. If the author adds a control run, fixes the guard interval, and supplies code, the paper might serve a workshop audience as a reproducible example. As it stands, a serious referee would reject it.\n\nRecommendation: desk reject, or send back with a clear request for the control run and corrected parameters.\n\nBest.","headline":"Confounded propagation-model comparison and an impossible guard interval undermine an otherwise unremarkable ns-3 demonstration.","tokens_in":7910,"tokens_out":6063,"would_cite":false,"duration_ms":55325,"reading_group":"no","serious_thinker":"unclear","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that inserting a building between a UAV-mounted Wi-Fi access point and a ground user cuts throughput from 51.8 to 41.5 Mbit/s and raises delay from 6.6 to 10.5 ms, and that moving the UAV to an unobstructed spot restores…","keywords":["Unmanned Aerial Vehicles","Obstacle-aware Communication","UAV Positioning","Channel Characterization","Air-to-ground channel","ns-3 simulation","Quality of Service","Line of Sight"],"falsifier":"Run the same scenario in the simulator with the building present but with its attenuation set to zero while keeping all other parameters; if throughput and delay do not return to the unobstructed values, the reported degradation comes from the obstacle attenuation model. Conversely, a field test with a small UAV at 10 m altitude, 30 m horizontal distance, and a building between it and a 5 GHz ground client would show whether a roughly 20% throughput drop and a 4 ms delay increase actually occur.","tokens_in":6950,"feed_emoji":"📡","tokens_out":8041,"duration_ms":70023,"temperature":0.7,"pith_summary":"The paper is trying to establish that obstacle-aware UAV positioning matters for air-to-ground channel quality, not just coverage geometry. It simulates a 5 GHz Wi-Fi link between a stationary UAV and a ground user in ns-3, first with clear line of sight and then with a building between them. The building lowers average throughput from 51.819 to 41.5377 Mbit/s and raises end-to-end delay from 6.61881 to 10.5386 ms, while packet delivery ratio stays at 99%. Moving the UAV back to a position where the building is no longer in the way restores the original throughput and delay. If true, this supports building obstacle awareness into UAV placement algorithms, since higher-frequency future links will be increasingly blockage-sensitive.","feed_headline":"One building blocks a fifth of a UAV link's throughput","feed_subtitle":"Moving the drone to restore line of sight brings throughput and delay back to baseline.","key_machinery":"The central machinery is the pair of propagation-loss models in the ns-3 simulator: the Friis free-space path-loss model for the obstacle-free scenario and the HybridBuildingsPropagationLossModel for the obstacle scenario, the latter adding building-dependent attenuation and shadowing to the same 5 GHz Wi-Fi link. The building block is a rectangular obstacle with coordinates x_min=10, x_max=20, y_min=0, y_max=50, z_min=-30, z_max=30, and the UAV moves between position 1 (30,0,10) and position 2 behind the building. These models carry the argument by producing the throughput, delay, and delivery metrics that are compared across the obstructed and unobstructed configurations.","core_discovery":"In this paper's terms, the central discovery is that a single static obstacle placed in an air-to-ground Wi-Fi link measurably degrades channel quality, and that repositioning the UAV to re-establish line of sight recovers that quality. The claim is supported by ns-3 simulations at 5 GHz: with the Friis propagation loss model and no obstacle, average throughput is 51.819 Mbit/s and end-to-end delay is 6.61881 ms; with HybridBuildingsPropagationLossModel and a building spanning x=10–20 m, y=0–50 m, z=±30 m, throughput falls to 41.5377 Mbit/s and delay rises to 10.5386 ms, while PDR remains 99% and packet loss remains 1%. The paper further reports that raising the frequency to 10 GHz under the same configuration decreases throughput further, and that relocating the access point to the original unobstructed position restores performance. The intended conclusion is that channel characterization for UAV links should include obstacles, and UAV placement should treat line-of-sight as a primary optimization criterion.","pith_inferences":["Going beyond the paper, the roughly 20% throughput penalty could be used as a simple scalar objective for obstacle-aware placement: maximize the fraction of users with clear line of sight to the UAV, and the QoS metrics should follow.","A testable extension is to vary the building's height, width, and horizontal offset; one would predict that throughput degradation grows as the building increasingly blocks the direct path, with a threshold above which packet delivery ratio begins to fall.","The setup could also be extended to multiple UAVs or moving blockers; a natural prediction is that blockage recovery time, rather than steady-state throughput, becomes the dominant QoS factor when obstacles move."],"forward_implications":["If the claim is correct, UAV placement algorithms should treat line of sight as a first-class optimization target, not just coverage radius or user density.","Obstacle-aware repositioning is a workable recovery mechanism: in this scenario, moving the access point from the blocked position back to the unobstructed position restores the original throughput and delay values.","Carrier frequency strengthens the effect: the paper's 10 GHz run shows throughput falling further below the 5 GHz obstructed case, so obstacle awareness should matter more for higher-frequency air-to-ground links.","Channel characterization for aerial links should routinely report both line-of-sight and non-line-of-sight configurations, because throughput and delay are strongly configuration-dependent."],"supporting_citations":[{"why":"Supplies prior experimental air-to-ground and ground-to-air channel characterization that this simulation-based measurement approach builds on.","marker":"[5]"},{"why":"Frames the opportunities and challenges of UAV wireless communications, motivating why line of sight and positioning are core concerns.","marker":"[6]"},{"why":"Represents obstacle-free UAV placement optimization that this paper contrasts with its obstacle-aware measurement approach.","marker":"[10]"},{"why":"Provides the concept that buildings determine line-of-sight probability, which the obstacle scenario instantiates.","marker":"[21]"},{"why":"Closest prior work on traffic- and obstacle-aware UAV positioning, which this paper complements with a direct channel measurement.","marker":"[22]"}],"fun_headline_variants":["Obstacle blocks a fifth of UAV link throughput","Repositioning drone restores lost throughput from obstacle","UAV link suffers 20% throughput hit from single obstacle","Clear line of sight is crucial: obstacle test shows UAV QoS drop","Obstacles hurt UAV links, but moving drone fixes it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the simulator's building attenuation model, as configured here, captures how a real building blocks a 5 GHz signal; if the configuration is unrealistic, the measured throughput drop may be a simulation artifact rather than a real obstacle effect.","fun_headline_variants_meta":{"raw":{"variants":["Obstacle blocks a fifth of UAV link throughput","Repositioning drone restores lost throughput from obstacle","UAV link suffers 20% throughput hit from single obstacle","Clear line of sight is crucial: obstacle test shows UAV QoS drop","Obstacles hurt UAV links, but moving drone fixes it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000304,"raw_usage":{"total_tokens":1742,"prompt_tokens":938,"completion_tokens":804,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":554,"completion_tokens_details":{"reasoning_tokens":721}},"tokens_in":554,"tokens_out":804,"duration_ms":9024,"temperature":1.0,"reasoning_tokens":721,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:07:45.934869+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same scenario in the simulator with the building present but with its attenuation set to zero while keeping all other parameters; if throughput and delay do not return to the unobstructed values, the reported degradation comes from the obstacle attenuation model. Conversely, a field test with a small UAV at 10 m altitude, 30 m horizontal distance, and a building between it and a 5 GHz ground client would show whether a roughly 20% throughput drop and a 4 ms delay increase actually occur.","supporting_citations":[{"cited_title":"UAV -assisted wireless communications: An experimental analysis of A2G and G2A channels","cited_arxiv_id":null,"evidence_quote":"Supplies prior experimental air-to-ground and ground-to-air channel characterization that this simulation-based measurement approach builds on."},{"cited_title":"Line-of- Sight Probability for mmWave -Based UAV Communications in 3D Urban Grid Deployments,","cited_arxiv_id":null,"evidence_quote":"Provides the concept that buildings determine line-of-sight probability, which the obstacle scenario instantiates."},{"cited_title":"Traffic and Obstacle -Aware UAV Positioning in Urban Environments Using Reinforcement Learning,","cited_arxiv_id":null,"evidence_quote":"Closest prior work on traffic- and obstacle-aware UAV positioning, which this paper complements with a direct channel measurement."}],"review_version":1}