{"id":"325ec999-6a9e-455c-844a-a1e3a05c0019","arxiv_id":"2506.02656","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":2.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"A 10 bit/s polarization-encoded key distribution run over 5 km of 1550 nm fiber is reported, using bright multiphoton pulses in a single basis with QBER below 1%.","lead":"This paper demonstrates a slow, lab-scale version of quantum key distribution over 5 km of optical fiber at 10 bits per second, encoding each bit as horizontal or vertical polarization. It is framed as a training guide rather than a secure system, and a general reader should see it as a hardware walkthrough with no new security or speed record.","discovery_kind":"incremental","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed <1% QBER is not an independent measurement: Bob's H/V decision uses a count threshold chosen from the same data, and Fig. 5 vs Fig. 7 disagree about the V-pulse count level. A fixed, pre-registered threshold applied to raw timetags would settle it.","rationale":"Read in good faith, the paper is a laboratory training guide: it builds a 10-Hz, one-basis polarization channel over 5 km and reports 3000 bits with low measured error. For that claim, the decisive evidence is the bit recovery rule. The paper states the receiver 'duly separates' two count levels and estimates QBER from the same record; this is circular unless a threshold is fixed by calibration. The contradiction between Fig. 5 (V ≈7.5e3 cps) and Fig. 7 (V at dark-count level) makes the classification rule under-specified. This is not merely an external concern: Sec. II explicitly acknowledges polarization drift but provides no time-resolved calibration of the two count distributions. A pre-registered-threshold re-analysis of the raw timetags would settle whether the reported <1% QBER is real. This concern is close to the reader's weakest_assumption, and I agree with it; I would sharpen it by emphasizing that the two count distributions are never characterized and that the manuscript's own figures disagree about the V level. The security/name issue (multiphoton, single basis, no sifting) is also real and independently disqualifying for 'QKD', but the experimental bit-error claim already fails the threshold test. Verdict: unchanged (REJECT).","tokens_in":7455,"tokens_out":4587,"duration_ms":47126,"concrete_test":"Request the raw TTU timetag file and Alice's stored QRNG key. Before analyzing the run, fix a decision rule from a separate calibration: e.g., define H as MCSS-gated counts above the midpoint between the 1-m calibration H and V levels, and V otherwise. Apply this fixed rule to each MCSS cycle for the full 5 minutes and compare against Alice's key. Report the per-minute bit error rate and the threshold's margin. If no fixed threshold gives <1% error over the full run, or if the threshold that zero-errors the first 2.5 minutes misclassifies >1% of the final 2.5 minutes, then the reported QBER is an artifact of post-hoc thresholding and the central transmission claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Bob's bit decision is described in Sec. II as separating 'two levels of counts above the dark counts' and associating ≈2×10^4 counts with H and ≈7.5×10^3 with V (Fig. 5), while Fig. 7 says V pulses sit at the dark-count floor. The receiver has a single H-arm detector with no active polarization stabilization; the separation threshold is therefore not a calibrated instrument response but a data-dependent classification boundary. Because the same 5-minute record is used to choose the threshold and to estimate the QBER (Figs. 7-8), the reported <1% error rate is not an independent characterization of channel quality. The claim 'no bit flipped for 2.5 minutes' is exactly what one expects from any threshold placed between two well-separated count modes; it does not establish that the V state was correctly transmitted, nor that the later drift was correctly tracked. The load-bearing assumption is that a single count-rate threshold remains a valid discriminator for the entire 5 km run without birefringence compensation. No calibration data over time are provided to support that assumption.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a low-speed (10 Hz) polarization-encoding QKD setup over 5 km of fiber at 1550 nm, using weak coherent pulses, a single H-arm SPAD, and count-rate thresholding to recover bits. The authors claim that 3000 binary bits were distributed in 5 minutes, with zero QBER for the first 2.5 minutes and QBER below 1% overall. The stated purpose is to serve as an initial experimental guide for laboratory-scale demonstrations of polarization-encoded QKD.","tokens_in":7680,"tokens_out":4123,"duration_ms":40521,"significance":"If the central claims were reproducible and the QBER were independently measured, this would be a modest training demonstration rather than a novel scientific result. The system uses a single polarization basis, has no sifting, no decoy states, no privacy amplification, and explicitly operates with multiphoton pulses; as such, it does not demonstrate QKD in any cryptographically meaningful sense. The manuscript does provide a useful component-level checklist and a clear schematic that could help a student assemble a polarization-encoding testbed, and it honestly acknowledges several of its own limitations. However, the experimental evidence is internally inconsistent, and the reported error rate is not an independent metric because the same data are used both to choose the H/V discrimination threshold and to estimate the QBER.","major_comments":[{"comment":"The phase-to-voltage mapping is stated inconsistently. The text first says 'at φ = 0 or V = 0 V (φ = π or V = 4 V), we get H-polarized (V-polarized) light,' but later says 'In Fig. 4, it is marked that the voltage level 0 V (4V) modulates the refractive index ... inducing the phase shift φ = π (φ = 0).' These two sentences assign opposite phase shifts to the same voltages, and the reader cannot tell whether a calibration error or a typographical slip is present. Since the entire bit-recovery scheme rests on the 0 V / 4 V assignments, this contradiction must be resolved.","section":"Section II"},{"comment":"The reported count rates are mutually inconsistent by orders of magnitude. Fig. 2 states that after 5 km the 'peak photonic counts are ≈ 1.2 × 10^6 per second,' Fig. 4 states '≈ 4.5 × 10^4 counts/sec' for the H-polarized case, and Fig. 5 reports '≈ 2 × 10^4 counts' for H and '≈ 7.5 × 10^3 counts' for V. Fig. 7 then states that vertically polarized pulses sit at the dark-count level. These numbers cannot all describe the same system under the same operating conditions. Because the threshold method depends on the relative heights of these count distributions, the inconsistency directly affects the reliability of the bit-recovery and QBER claims.","section":"Section II, Figs. 2, 4, 5, 7"},{"comment":"The reported QBER is not an independent measurement. Bob's decision rule is described as separating 'two levels of counts above the dark counts' and associating them with H or V, with the threshold effectively chosen from the observed data (e.g., the ≈2×10^4 vs ≈7.5×10^3 levels in Fig. 5 and the 'dark count level' in Fig. 7). The same 5-minute record is then used to estimate the QBER in Fig. 8. A threshold selected post hoc will always 'explain' the data on which it was fit; the claimed <1% error rate therefore does not characterize channel quality. The authors should pre-register a fixed threshold from a separate calibration run and apply it to the full raw timetag stream, or report a per-instance decision rule with confidence intervals.","section":"Section II, Figs. 6–8"},{"comment":"The claim of 'key distribution' is not supported by any security analysis, and the paper's own acknowledgments of multiphoton pulses and the PNS attack make this a central limitation, not a small caveat. With an average photon number not specified, a single basis, no decoy states, and no privacy amplification, the system is a classical polarization channel with quantum-relevant components. The title and framing as a QKD implementation are therefore misleading; at best this is a proof-of-principle testbed for polarization encoding. The authors should either provide a full security analysis under explicit assumptions or explicitly relabel the demonstration as a classical channel test, in which case the 'error rate less than unity' claim has no cryptographic meaning.","section":"Abstract and Section II"}],"minor_comments":[{"comment":"The abstract states that '3000 binary bits were distributed ... with an error rate less than unity,' while Section III claims 'QBER remained below 1% throughout.' 'Less than unity' conventionally means less than 1, i.e., less than 100%, which is trivially true for any useful channel; this ambiguity should be fixed.","section":"Abstract and Section III"},{"comment":"The abstract and Section II say 3000 bits were distributed in 5 minutes, which is consistent with the 10 Hz rate, but Fig. 7 is labeled 'first 5000 bits/pulses.' At 10 Hz for 5 minutes, only 3000 pulses occur, so the axis and the claim need to be reconciled.","section":"Fig. 7 and Section II"},{"comment":"There are numerous typographical and grammatical errors that impede readability: 'aLiN bO3', 'comprized', 'eleminated', 'Pockel cell' (should be 'Pockels cell'), 'SPDAD' (likely 'SPAD'), 'Oscilloscop' in Fig. 1, and inconsistent capitalization of 'Horizontal' and 'Vertical.' A careful language edit is needed.","section":"Throughout"},{"comment":"The description of the TTU sampling is unclear: 'The temporal resolution of TTU is kept 1 × 10^-2 s that takes 100 data counts samples after each second.' At 10 Hz, this is not obviously consistent with 'one sample per cycle of the MCSS.' Please state the actual binning and selection procedure with a timing diagram or pseudocode.","section":"Section II"}],"recommendation":"reject","confidential_remarks":"The manuscript reads more like an undergraduate laboratory report than a research paper in quantum information. The internal count-rate inconsistencies and the circular QBER estimation are not local blemishes; they undermine the central demonstration. I would not encourage resubmission without a fundamentally redesigned experiment and a pre-registered analysis protocol."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a lab-training writeup, not a research contribution. It claims QKD but demonstrates a 10 bit/s, single-basis, multiphoton polarization channel over 5 km, and its headline error rate is not an independent measurement.\n\nWhat it does well: the component chain is clearly described, and the synchronization requirements between the clock, FPGA, intensity modulator, and detector form a useful checklist for someone building a first setup. The author honestly says this is a simplified, laboratory-scale demonstration and lists the upgrades needed for real QKD. That self-aware framing is decent.\n\nWhere it falls down: the QBER estimate is circular. Bob chooses a count-rate threshold after seeing the data, then uses that same threshold to classify bits and compute the error rate. That tells you a threshold can separate two count modes; it does not tell you the polarization states were transmitted correctly, and it does not characterize drift over time. The quantitative statements also contradict each other: one passage assigns phi=0 to 0 V and phi=pi to 4 V, the next says the opposite; count rates jump between 1.2e6, 3.1e4, and 2e4 counts/s across figures; and the abstract's \"error rate less than unity\" is a trivial bound, even though the conclusion claims <1%. No raw data or error bars are provided.\n\nThe architecture itself is already present in the cited refs [7-9], which do polarization-coded QKD over fiber with active compensation and far higher rates. So there is no new technical result here. The paper's value, if any, is pedagogical, and that value would be much higher if the threshold were calibrated on a separate dataset and the raw timetags were released.\n\nVerdict: desk reject. I would not send this to referees as a research paper. A heavily revised teaching note could be considered by a lab-oriented venue, but this version is not there.","headline":"A teaching-lab writeup that overclaims QKD; the QBER is not independently measured, the numbers are internally inconsistent, and the architecture is already in the cited literature.","tokens_in":8240,"tokens_out":3612,"would_cite":false,"duration_ms":35276,"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":"A 5 km fiber link carried 3000 polarization-encoded QKD bits at 10 bits per second with reported error under 1%.","keywords":["quantum key distribution","polarization encoding","weak coherent pulses","1550 nm telecom wavelength","phase modulator","birefringence","single-photon avalanche detector","quantum bit error rate"],"falsifier":"Send a known alternating H/V sequence over the same 5 km fiber with the count-rate threshold fixed before the run begins, decode the bits from the recorded counts, and compare the decoded sequence to the sent one for the full five minutes. If errors appear before the 2.5-minute mark, or if the measured H and V count distributions overlap substantially at any point, the claimed zero-error interval and sub-1% QBER do not survive.","tokens_in":7192,"feed_emoji":"🔐","tokens_out":7039,"duration_ms":63284,"temperature":0.7,"pith_summary":"This paper reports a laboratory-scale quantum key distribution experiment that encodes bits as horizontal or vertical polarization of weak coherent pulses at 1550 nm. Alice sends 10 pulses per second through a 5 km fiber, and Bob separates the two polarizations with a polarizing beam splitter while detecting only the horizontal arm. In a five-minute run the authors report exchanging 3000 bits, with zero bit-flip errors in the first 2.5 minutes and a total quantum bit error rate below one percent. The goal is to establish that the simplest single-basis polarization setup can work on a telecom fiber, and to document the synchronization and alignment steps needed to build it.","feed_headline":"Polarization QKD test sends 3000 bits over 5 km fiber","feed_subtitle":"Single-basis, 10-bit-per-second setup stays error-free for 2.5 minutes, then drifts under 1 percent.","key_machinery":"The mechanism that carries the argument is the Pockel-cell phase modulator: an FPGA applies a random 0 V or 4 V on each cycle of a 10 Hz master clock, and the lithium-niobate crystal converts those voltages into phase shifts $\\varphi=0$ or $\\varphi=\\pi$ between horizontal and vertical field components, preparing each weak coherent pulse as $|\\psi\\rangle=\\cos(\\varphi/2)|H\\rangle+\\sin(\\varphi/2)|V\\rangle$. The same master clock synchronizes the intensity modulator, the phase modulator, and the time-tagging unit, so each pulse is aligned with the voltage that encoded it. At Bob, a polarizing beam splitter sends H light to the single-photon detector and V light away from it; the two count levels are separated by a fixed threshold, the data-processing step that actually converts optical counts into bits.","core_discovery":"On its own terms, the paper's discovery is that a polarization-encoded channel with one basis, one detector arm, and 10 Hz clocking can carry key material over 5 km of standard fiber. Horizontal and vertical states are prepared by applying 0 V or 4 V to a lithium-niobate phase modulator, and Bob reconstructs the bits from the count rate measured by a single InGaAs single-photon detector behind a polarizing beam splitter: roughly $2\\times 10^4$ counts per second marks H, and roughly $7.5\\times 10^3$ counts per second marks V. The authors report that no bit was flipped for the first 2.5 minutes and that the QBER stayed below one percent across the five-minute observation period, after which polarization drift lowered the contrast between the two states.","pith_inferences":["The QBER figure is not fully independent: the same threshold used to classify bits was selected after viewing the data, so a pre-registered threshold test would be a stricter check of the scheme.","The 2.5-minute stable window is a drift timescale, not a physical limit; active birefringence compensation or polarization-maintaining fiber should extend it, and the setup's own references suggest those routes.","A two-basis extension, adding a diagonal basis, would restore sifting and close the single-basis security gap at the cost of a second detection arm or faster phase modulation.","At 10 bit/s and only 3000 bits, finite-key effects and photon-number-splitting vulnerabilities dominate, so the result is best read as a feasibility and training demonstration of the optical channel, not as a secure key source."],"forward_implications":["A single polarization basis and a single detector arm are enough to move 3000 bits over 5 km at 10 bit/s with reported error below one percent.","Because the key rate is fixed by the 10 Hz master clock, increasing the clock frequency is the paper's stated route to higher bit rates.","The slow drift that appears after 2.5 minutes stays small enough that ordinary error correction can clean the channel without fast active compensation.","Removing sifting simplifies Bob's receiver, but it also means the demonstration operates outside the full BB84 security model."],"supporting_citations":[{"why":"Supplies the original BB84 protocol that this single-basis polarization implementation is built around.","marker":"[1]"},{"why":"Shows active polarization control as the standard remedy for drift, which this experiment deliberately omits.","marker":"[7]"},{"why":"Demonstrates real-time birefringence compensation on fiber, the technique this setup leaves for later improvement.","marker":"[8]"},{"why":"Provides the weak-coherent-pulse security context that motivates the faint-pulse choice and the PNS caveat.","marker":"[12]"},{"why":"Defines the photon-number-splitting attack, the vulnerability the paper accepts by using multiphoton pulses.","marker":"[15]"},{"why":"Supplies the quantum random number generator that creates the key loaded into the FPGA.","marker":"[29]"},{"why":"Provides the error-correction method the authors say can handle the sub-1% drift-induced errors.","marker":"[30]"}],"fun_headline_variants":["Single-basis QKD delivers 3000 bits over 5 km","Slow QKD: 3000 bits, 5 km, one polarization basis","10-bit/s QKD test: 3000 bits over 5 km fiber","One-basis QKD: 3000 bits on 5-km link"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The scheme assumes that a fixed count-rate threshold, chosen after the data were already recorded, will keep cleanly separating horizontal from vertical pulses over 5 km of ordinary fiber for at least 2.5 minutes, even though the fiber's birefringence is drifting and no active polarization compensation is running.","fun_headline_variants_meta":{"raw":{"variants":["Single-basis QKD delivers 3000 bits over 5 km","Slow QKD: 3000 bits, 5 km, one polarization basis","10-bit/s QKD test: 3000 bits over 5 km fiber","One-basis QKD: 3000 bits on 5-km link"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000621,"raw_usage":{"total_tokens":2844,"prompt_tokens":878,"completion_tokens":1966,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":494,"completion_tokens_details":{"reasoning_tokens":1881}},"tokens_in":494,"tokens_out":1966,"duration_ms":13259,"temperature":1.0,"reasoning_tokens":1881,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:19:21.661503+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Send a known alternating H/V sequence over the same 5 km fiber with the count-rate threshold fixed before the run begins, decode the bits from the recorded counts, and compare the decoded sequence to the sent one for the full five minutes. If errors appear before the 2.5-minute mark, or if the measured H and V count distributions overlap substantially at any point, the claimed zero-error interval and sub-1% QBER do not survive.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows active polarization control as the standard remedy for drift, which this experiment deliberately omits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates real-time birefringence compensation on fiber, the technique this setup leaves for later improvement."},{"cited_title":"Zhao, C.-H","cited_arxiv_id":null,"evidence_quote":"Provides the weak-coherent-pulse security context that motivates the faint-pulse choice and the PNS caveat."},{"cited_title":"Zapatero, ´A","cited_arxiv_id":null,"evidence_quote":"Defines the photon-number-splitting attack, the vulnerability the paper accepts by using multiphoton pulses."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the quantum random number generator that creates the key loaded into the FPGA."},{"cited_title":"Haider, M","cited_arxiv_id":null,"evidence_quote":"Provides the error-correction method the authors say can handle the sub-1% drift-induced errors."}],"review_version":1}