{"id":"265dae60-94b2-4076-9c25-6572ced06298","arxiv_id":"1908.01486","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A 2x2 optical MIMO link using repetition coding and maximum ratio combining reduces bubble-induced outage in an underwater laser link from 34.6% to below 1% at 780 Mbit/s.","lead":"Air bubbles in water can block or bend underwater laser signals; this experiment shows that using two transmitters and two receivers cuts link failures from more than one in three to under one in one hundred. A generalist might read it because underwater robots, divers, and seafloor sensors need reliable optical links in bubbly water.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline 'outage <1%' is statistically unsupported: Table 2 censors all MIMO losses below 5 out of 500, and exact counts/confidence intervals are not reported.","rationale":"The reader's weakest assumption identified the unmeasured spatial correlation between MIMO branches as the main gap. That is a real limitation for generalization, but the empirical MIMO PLR itself already provides indirect evidence against strong correlation: if the two branches faded together, MIMO's PLR would track SISO's 13-34%, rather than dropping below 1%. The more immediate load-bearing issue is the statistical support for the exact headline number. The paper explicitly censors every MIMO PLR to '<1%' whenever fewer than five of 500 packets are lost, and it reports no exact counts, confidence intervals, or raw data. The central claim is a quantitative threshold claim ('from 34.6% to less than 1%'), and the reported numbers cannot establish that threshold at conventional confidence unless the hidden counts are very small. This is not an accusation of error; it is a missing-support problem that can be settled by releasing the exact counts. The paper otherwise contains no fitted parameters, no circular reasoning, and the BER/PLR tables are plausible and detailed. The verdict should remain CONDITIONAL: the experimental demonstration is credible, but the precise outage reduction should be verified against raw counts or a larger sample before the '<1%' figure is treated as a robust design rule.","tokens_in":6852,"tokens_out":15029,"duration_ms":155847,"concrete_test":"Request from the authors the raw per-packet BER records (or at least the exact lost-packet counts) for the nine MIMO rows of Table 2, and compute exact Clopper-Pearson 95% confidence intervals for the PLR in each scenario. If every interval's upper bound is below 1%, the headline claim is statistically supported; if any interval's upper bound exceeds 1%, the claim 'outage probability less than 1%' should be revised to the observed point estimate with its uncertainty, or the experiment should be rerun with substantially more packets (e.g., 5,000 per scenario) to resolve the sub-1% region.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—that 2x2 MIMO reduces outage probability from 34.6% to less than 1%—rests entirely on Table 2, in which every MIMO entry is reported as '<1%'. The text immediately above the table states: 'for the loss packet number less than five, the PLR is denoted as <1%.' With n=500 packets per scenario, this censoring means the observed PLR is anywhere between 0% and 0.8%, with the exact count undisclosed. This is too coarse to support the '<1%' probability claim: under a binomial model, 4 losses out of 500 (point estimate 0.8%) has a 95% Clopper-Pearson upper bound near 1.8%; even 2 losses yields an upper bound near 1.1%. Unless every MIMO scenario had at most one lost packet, or the sample size were larger, the data do not establish that the true outage probability is below 1% at conventional confidence. The qualitative robustness conclusion may survive, but the precise headline number is not supported by the reported statistics. This concern applies to the measured result itself, independent of the path-correlation question.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper experimentally investigates the effect of air bubbles on underwater wireless optical communication (UOWC) links and evaluates the mitigation provided by spatial diversity. A testbed with two laser-diode transmitters and two photodiode receivers is used to compare SISO, SIMO, MISO, and 2x2 MIMO with repetition coding and maximum ratio combining, at a data rate of 780 Mbit/s over 16-QAM OFDM. Bubble-induced impairment is varied by three bubble sizes and three positions along an 80-cm water channel. The central claim is that the 2x2 MIMO scheme reduces outage probability from 34.6% (worst-case SISO) to less than 1% in all tested scenarios, while also improving average BER from 1.31e-2 to 1.42e-3. The paper reports BER distributions, BER variances, and packet loss rates for the four schemes.","tokens_in":7067,"tokens_out":2990,"duration_ms":31533,"significance":"If the core claims hold, the paper provides a practically useful result: a relatively simple 2x2 spatial-diversity configuration can make an OFDM UOWC link robust against bubble-induced blockage, a problem that is identified as a key challenge in underwater optical communications. The experimental methodology is a strength: the testbed is real-time, the bubble size and position are varied systematically, and a direct comparison across SISO, SIMO, MISO, and MIMO is provided. The paper also offers a clear demonstration that MRC-based receive diversity combined with RC-based transmit diversity improves both BER and packet-level reliability. However, the headline quantitative claim ('outage < 1%') currently rests on censored packet-loss data, and the independence of the two diversity paths is asserted from geometry rather than measured, so the statistical and physical support for the headline is incomplete.","major_comments":[{"comment":"The central quantitative claim that 2x2 MIMO reduces outage probability below 1% is not supported by the reported statistics. The text states 'for the loss packet number less than five, the PLR is denoted as <1%', so with n=500 packets each MIMO entry only indicates an observed count between 0 and 4. Under a binomial model, 4 losses out of 500 give a 95% Clopper-Pearson upper bound near 1.8%, and even 2 losses give an upper bound near 1.1%. The data therefore do not establish that the true outage probability is below 1% at conventional confidence unless every MIMO scenario had at most one lost packet. Please report the exact packet-loss counts for each MIMO scenario, or increase the sample size, or reframe the claim as a qualitative robustness result. The qualitative conclusion that diversity reduces packet loss is likely unaffected, but the specific '<1%' claim is load-bearing in the abstract and needs proper statistical support.","section":"Table 2"},{"comment":"The diversity gain is attributed to 'multiple independent light paths', but the manuscript provides no measurement of the spatial correlation between the two transmitter-receiver links. The only evidence is the geometric separation of transmitters (10 cm) and receivers (8 cm). If a bubble cluster blocks both paths simultaneously, or if the optical beams are not actually separated at the bubble location, maximum ratio combining cannot reduce the outage to below 1% in general. Please provide simultaneous per-link SNR or BER time series, or a fading correlation coefficient across the two links, and discuss the bubble cluster size relative to the TX/RX separation to justify the independence assumption.","section":"Fig. 2 and 'At last' (diversity mechanism)"},{"comment":"The term 'outage probability' is used as a synonym for packet loss rate, but the exact criterion is not specified. Is a packet counted as lost when its BER exceeds the SD-FEC limit of 2e-2? Does the BER variance in Table 1 include or exclude lost packets? How are partial packet losses handled? These definitions matter because the main numerical claim (34.6% to <1%) is expressed in terms of outage probability, yet the manuscript never defines the mapping from BER to outage. Please state the outage criterion explicitly and clarify whether the average BER in Fig. 7 is computed over all packets or only over successfully received packets.","section":"Section 'At last' (outage definition)"}],"minor_comments":[{"comment":"There are several typographical errors: 'Almouti-coded' should be 'Alamouti-coded', 'Hamutuas' should be 'Hamamatsu', 'vairance' should be 'variance', 'flcution' should be 'fluctuation', 'repsectivley' should be 'respectively', and 'cruve' should be 'curve'. These should be corrected in revision.","section":"General"},{"comment":"The red MIMO curve is described in the text but may be difficult to distinguish in grayscale print; consider using different line styles or adding a legend in addition to color.","section":"Fig. 5"},{"comment":"The PDF analysis is presented qualitatively; the claim that distributions 'remain relatively constant' with diversity would be strengthened by a quantitative distance or variance measure across the PDFs, rather than only visual inspection.","section":"Fig. 6"},{"comment":"Several references are incomplete or inconsistently formatted (e.g., [2], [3] lack full page ranges and publisher information), and the manuscript still contains placeholder text such as 'Received Month X, XXXX'. Please complete the reference list and remove placeholders.","section":"References"},{"comment":"The description of received optical power ('with one LD' vs. 'two-LD case') is confusing; clarify whether the MISO case transmits the same signal from both LDs and how the received power is measured in each configuration.","section":"Fig. 2 (received power)"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and addresses a timely problem. The main concern is that the headline '<1% outage' claim depends on censored data and an unmeasured independence assumption; both are fixable in revision with additional experiments or a more careful statistical statement. I would encourage the editor to request those revisions rather than reject, given the practical value of the testbed and the systematic comparison."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First thing you should know: this paper is worth reading if you care about underwater optical wireless links, but the headline number in the abstract is statistically overclaimed. The '<1%' outage probability is not actually supported by the table as reported.\n\nWhat's new: prior work on bubble effects ([6], [7]) looked at received power statistics and fading models. This paper is the first (as far as I can tell) to experimentally compare SISO/SIMO/MISO/MIMO under controlled bubble sizes and positions at link level, with BER, BER variance, PDFs, and packet-loss rates. The testbed is clean: 780-Mbit/s OFDM over 80 cm, two transmitters separated by 10 cm, two receivers separated by 8 cm, three bubble size regimes, three bubble positions. The qualitative result is unmistakable: MIMO is dramatically more robust than SISO. The worst SISO packet-loss rate is 34.6%; every MIMO entry in Table 2 is below 1% as reported. The variance and PDF comparisons support the same story.\n\nThe soft spots are real, and the stress-test note hits the main one. Table 2 censors every MIMO value as '<1%' whenever fewer than five packets out of 500 are lost. With n=500, that means the reported number is anywhere from 0% to 0.8%, and the exact count is not given. The stress-test is correct that this does not establish the true outage probability is below 1%: four losses out of 500 has a Clopper-Pearson upper bound near 1.8%, and even two losses has an upper bound above 1%. So the abstract's 'outage probability ... less than 1%' is an overstatement. The sample PLR might be below 1%, but the paper doesn't report enough to know, and it never gives confidence intervals. This is fixable by reporting exact packet-loss counts, running more packets, and using the word 'sample PLR' instead of 'outage probability.' I'd treat the qualitative claim—MIMO essentially eliminates the bubble-induced outage in this testbed—as credible; the precise numerical claim is not.\n\nThe other soft spot is the path-independence assumption. The diversity gain depends on the two links fading independently. The geometry (10 cm TX spacing, 8 cm RX spacing) makes independence plausible, but it's not measured, and the paper doesn't report any correlation statistic. That limits generalization but doesn't undermine the observed results for this testbed. Minor notes: only one batch of 500 packets per scenario, no repeated trials; bubble sizes are rough area estimates from photos; some copyediting issues. The citation pattern looks fine; [6] and [7] are the right comparison points, and the self-citations are not load-bearing.\n\nWho is this for: researchers working on UOWC or optical diversity who want an experimental data point on bubble impairment and MIMO mitigation. It's not a breakthrough, but it's a solid, useful study.\n\nRecommendation: it deserves a serious referee. The experimental work is genuine, and the main issue is statistical reporting, not the physics. I'd send it to review with a clear request: either report exact counts and intervals, or soften the abstract claim. No desk reject.","headline":"Useful experimental study of diversity against bubble blockage in UOWC, but the abstract's '<1%' outage claim is statistically overreported.","tokens_in":7603,"tokens_out":4700,"would_cite":false,"duration_ms":47477,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A 2×2 MIMO underwater optical link with repetition coding and maximum ratio combining cuts bubble-induced outage from 34.6% to under 1%.","keywords":["underwater wireless optical communication","spatial diversity","MIMO","bubble-induced fading","maximum ratio combining","repetition coding","OFDM","outage probability"],"falsifier":"Put a single bubble cluster wide enough to cover both transmitter apertures at the position nearest the transmitter, and measure the packet loss rate of the 2×2 MIMO link; if the outage rises above roughly 1% while the single-link outage is high, the spatial-independence assumption, not the combining scheme, is the limiting factor.","tokens_in":6643,"feed_emoji":"🌊","tokens_out":8008,"duration_ms":80921,"temperature":0.7,"pith_summary":"This paper reports an experimental underwater wireless optical communication (UOWC) system in which a 2 by 2 MIMO link with repetition coding at the transmitter and maximum ratio combining at the receiver keeps a 780-Mbit/s OFDM link operational when air bubbles pass through the water channel. The authors systematically vary bubble size (roughly 2 to 15 square millimeters) and position along an 80-centimeter tank, and they show that the single-link version suffers packet loss up to 34.6 percent, while the MIMO version keeps outage below 1 percent in every tested scenario. They also show that intermediate diversity schemes—one transmitter with two receivers or two transmitters with one receiver—reduce but do not eliminate outage. The point of the study is that spatial diversity, not more optical power or more complex modulation, is what makes an underwater optical link robust to bubble-induced blockage.","feed_headline":"Spatial diversity cuts underwater bubble outage from 34.6% to under 1%","feed_subtitle":"Two lasers and two detectors keep a 780-Mbit/s underwater optical link working when bubbles block single paths.","key_machinery":"The load-bearing mechanism is spatial diversity realized as a two-by-two MIMO link: two transmitters send the same signal (repetition coding), and two receivers combine the two detected signals by maximum ratio combining. In MRC, each received signal is weighted by the complex conjugate of its channel coefficient, so the combined SNR is proportional to the sum of the per-path SNRs, $S\\sum_i |h_i|^2/N_0$, which is optimal among linear combiners when the noise on each branch is independent and equal-power. The physical setup matters too: the transmitters are spaced 10 cm apart and the receivers 8 cm apart in an approximately 80-cm water tank, giving four partially distinct light paths through the bubble field. Repetition coding is chosen over Alamouti-style space-time coding because intensity-modulated direct-detection systems add optical powers constructively, and the paper cites this as a low-complexity effective transmit-diversity format.","core_discovery":"The central claim is that transmit and receive spatial diversity can make an underwater optical link robust against the dynamic, partial blocking caused by air bubbles. Concretely, the paper demonstrates a 780-Mbit/s 16-QAM OFDM link over roughly 80 cm of tank water in which two 450-nm laser diodes transmit the same signal (repetition coding) and two photoreceivers combine the received signals with maximum ratio combining. Across nine scenarios—three bubble sizes at three positions—the two-by-two link keeps the packet loss rate below 1 percent, whereas the no-diversity single link reaches 34.6 percent in the worst case (small bubbles near the receiver). The paper also reports that diversity shrinks BER variance by about three orders of magnitude and lowers the average BER from 1.31×10⁻² to 1.42×10⁻³ in the no-bubble reference. The explanation offered is that MIMO provides multiple light paths that are blocked independently, so when one path is obscured another still carries signal.","pith_inferences":["The authors do not measure the spatial correlation of the two links; if a bubble cluster is wide enough to block both transmitter beams or both receiver apertures simultaneously, the under-1% outage claim would not generalize, so varying TX/RX spacing and bubble-cluster width is the natural next experiment.","Treating bubble blockage as spatially selective fading suggests standard diversity-outage scaling should apply; if so, a 3×3 or 4×4 system could tolerate denser bubble fields or reach even lower outage, but the paper does not test this.","Because MRC relies on channel estimates from the received OFDM symbols, the results implicitly assume the bubble field is quasi-static over an OFDM symbol; faster-moving bubbles or stronger turbulence would stress the channel-estimation update rate, an untested regime.","The testbed uses a short 80-cm tank without lenses, so the link has a large optical spot at the receiver; in a longer, lensed link the bubble-induced fading statistics and the diversity gain may differ, so field validation in open water remains an open step."],"forward_implications":["A 2×2 MIMO with repetition coding and maximum ratio combining holds outage below 1% for all nine bubble size and position scenarios tested, making link reliability essentially independent of bubble condition within this range.","SIMO and MISO reduce outage but not everywhere: packet loss rates of 1.2% to 9.6% remain in several scenarios, showing that two diversity branches are not always enough but four are.","BER variance drops by roughly three orders of magnitude with the 2×2 link, meaning real-time stability improves, not just average throughput.","Because the 780-Mbit/s link uses standard 16-QAM OFDM and no lenses, the robustness gain is attributable to diversity rather than to extra link margin or higher transmit power."],"supporting_citations":[{"why":"Prior single-link experiment emulating bubble-induced blocking; establishes the impairment this paper mitigates.","marker":"[6]"},{"why":"Statistical fading analysis for UOWC with air bubbles; motivates the missing link-level BER and outage study.","marker":"[7]"},{"why":"RF spatial diversity theory; supplies the diversity principle transferred to the underwater optical channel.","marker":"[13]"},{"why":"Comparison of space-time coding for direct-detection optical links; supports repetition coding over Alamouti coding.","marker":"[15]"},{"why":"Defines maximum ratio combining as the optimal linear diversity combiner; the receiver algorithm used here.","marker":"[17]"}],"fun_headline_variants":["Bubble-proof underwater link: 2x2 MIMO slashes outage from 34.6% to <1%","Underwater optical link survives bubbles: 2x2 diversity cuts outage to <1%","MIMO tricks keep 780-Mbit/s underwater link bubble-proof: outage <1%","Two lasers, two detectors: underwater link shrugs off bubbles, outage <1%"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole diversity gain depends on the four light paths being blocked independently enough by the bubbles; if one bubble cluster can obscure both transmitters or both receivers at the same time, the claimed under-1% outage will not hold.","fun_headline_variants_meta":{"raw":{"variants":["Bubble-proof underwater link: 2x2 MIMO slashes outage from 34.6% to <1%","Underwater optical link survives bubbles: 2x2 diversity cuts outage to <1%","MIMO tricks keep 780-Mbit/s underwater link bubble-proof: outage <1%","Two lasers, two detectors: underwater link shrugs off bubbles, outage <1%"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000787,"raw_usage":{"total_tokens":3425,"prompt_tokens":854,"completion_tokens":2571,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":470,"completion_tokens_details":{"reasoning_tokens":2470}},"tokens_in":470,"tokens_out":2571,"duration_ms":16356,"temperature":1.0,"reasoning_tokens":2470,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:11:44.584294+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Put a single bubble cluster wide enough to cover both transmitter apertures at the position nearest the transmitter, and measure the packet loss rate of the 2×2 MIMO link; if the outage rises above roughly 1% while the single-link outage is high, the spatial-independence assumption, not the combining scheme, is the limiting factor.","supporting_citations":[{"cited_title":"Performance Evaluation of Underwater Wireless Optical Communications Links in the Presence of Different Air Bu bble Populations","cited_arxiv_id":null,"evidence_quote":"Prior single-link experiment emulating bubble-induced blocking; establishes the impairment this paper mitigates."},{"cited_title":"Statistical Studies of Fading in Underwater Wireless Optical Channels in the Presence of Air Bubble, Temperature, and Salinity Random Variations","cited_arxiv_id":null,"evidence_quote":"Statistical fading analysis for UOWC with air bubbles; motivates the missing link-level BER and outage study."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"RF spatial diversity theory; supplies the diversity principle transferred to the underwater optical channel."},{"cited_title":"Do We Really Need OSTBCs for Free -Space Optical Communication with Direct Detection?","cited_arxiv_id":null,"evidence_quote":"Comparison of space-time coding for direct-detection optical links; supports repetition coding over Alamouti coding."},{"cited_title":"Linear diversity combining techniques","cited_arxiv_id":null,"evidence_quote":"Defines maximum ratio combining as the optimal linear diversity combiner; the receiver algorithm used here."}],"review_version":1}