Automated Polarization Basis Adjustment and Security Monitoring in Quantum Communication via Coincidence Entropies

Carlos Guerra-Y\'anez; Josef Bla\v{z}ej; Josef Vojt\v{e}ch; Mat\v{e}j Holubi\v{c}ka; Tom\'a\v{s} Nov\'ak

REVIEW 4 major objections 4 minor 5 references

Automated Polarization Basis Adjustment and Security Monitoring in Quantum Communication via Coincidence Entropies

T0 review · 4 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A scalar coincidence-entropy metric can automatically re-align polarization bases in an all-fiber quantum receiver.

desk verdict Novel cost function, one convincing run, but the invariance under local bit flips means it may align to the wrong Bell state; security monitoring is not demonstrated. read the letter →

arxiv 2505.24071 v1 pith:ZGHGAYXD submitted 2025-05-29 quant-ph

Tomáš Novák , Carlos Guerra-Yánez , Matěj Holubička , Josef Vojtěch , Josef Blažej This is my paper

classification quant-ph

keywords quantumkeydistributionpolarizationentanglementbirefringencecompensationcoincidenceentropybasisalignmentBBM92protocolfiber-basedsingle-photondetectioneavesdropping

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to solve the birefringence problem in fiber-based quantum communication: single-mode optical fibers randomly rotate photon polarization, so an all-fiber receiver cannot assume its polarization bases stay aligned. It introduces coincidence entropies, defined from the coincidence counts already measured in a quantum key distribution setup, which quantify how random the joint measurement outcomes are. These entropies are constructed to be independent of which Bell state was transmitted, while still being sensitive to how well the two receivers' polarization bases are aligned. Summing the eight same-basis and cross-basis entropy terms over both receivers gives a single cost function $H_{\mathrm{total}}$, and a gradient-descent routine over fiber polarization-controller paddles maximizes it. In a polarization-entangled quantum key distribution testbed, the routine converged to a quantum bit error rate (QBER) of about 4.2% and $H_{\mathrm{total}}\approx 13.9$ in about 200 minutes.

What carries the argument

The load-bearing object is the coincidence-entropy cost function $H_{\mathrm{total}} = H_A + H_B$, built from the sixteen polarization coincidence counts $C_{jk}$. For each of the four polarizations $j$, the same-basis term $H_j^{\mathrm{same}}$ is the binary Shannon entropy of the normalized counts $C_{jj}$ and $C_{jk}$, so it is large when the correlation expected for the shared basis is strong; the cross-basis term $H_j^{\mathrm{diff}}$ is the binary entropy of the two normalized counts in the other basis, so it is large when those outcomes are uniformly random, realizing the mutual-unbiasedness condition. The gradient routine estimates the change in $H_{\mathrm{total}}$ after small shifts of the three paddle angles of each of the four electrically driven polarization controllers, then steps the paddles in the direction of increasing entropy; detector-efficiency weights correct the counts before the entropies are computed. This machinery reduces the optical alignment problem to a scalar optimization that needs only the coincidence data the protocol already collects.

What would settle it

Start the gradient routine from many random paddle settings and record the endpoint values. If any run converges to $H_{\mathrm{total}} \geq 13.9$ with QBER substantially above 4.2%, then high coincidence entropy does not by itself imply aligned bases, and the central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that coincidence entropy alone is a sufficient objective for aligning two polarization-sensitive receivers connected by single-mode fiber, using no auxiliary laser and no prior knowledge of the transmitted Bell state. For each polarization $j \in \{H,V,D,A\}$, the metric combines $H_j^{\mathrm{same}}$, a binary entropy that approaches 1 only when the expected same-basis coincidence channel dominates, with $H_j^{\mathrm{diff}}$, a binary entropy that approaches 1 only when the two cross-basis outcomes are equally likely; summing over all four polarizations at both receivers bounds $H_{\mathrm{total}}$ by 16. The authors argue that these two conditions together enforce both strong quantum correlations in a shared basis and the requirement that the two cross-basis outcomes be equally likely (mutual unbiasedness between the $Z$ and $X$ bases), and demonstrate that a stochastic gradient routine on this scalar reaches $\mathrm{QBER}_{\lvert\phi^+\rangle} \approx 4.2\%$ and $H_{\mathrm{total}} \approx 13.9$. They attribute the residual gap to the minimal achievable QBER of about 1%, set by their source visibility near 98%, plus wavelength-dependent birefringence across the source's roughly 60 nm bandwidth.

Load-bearing premise

The argument assumes the entropy landscape has no deceptive maxima: every configuration with near-maximal coincidence entropy also has low quantum bit error rate, so high entropy genuinely means the polarization bases are aligned.

Editorial extensions

If this is right

All-fiber QKD receivers can be re-aligned automatically during operation, using coincidence counts alone rather than a bright reference laser.
Because the entropy construction is the same for every Bell state, the same routine can align receivers without knowing which entangled state the source emits.
Coincidence counts normally discarded in sifting can be repurposed to monitor cross-basis correlations, giving a security check that catches intercept-resend attacks; a drop in $H_{\mathrm{total}}$ is the warning signal.
The residual QBER of about 4.2% is attributed to wavelength-dependent fiber transformations across the broadband spectrum, so compensating the source's 60 nm bandwidth should bring the system closer to the ~1% visibility limit.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the metric relies only on counts that a QKD receiver already measures, it could be embedded as a continuous health monitor in deployed networks, flagging channel drift or tampering before the key rate collapses.
The paper reports one successful convergence; whether the $H_{\mathrm{total}}$ landscape contains misleading maxima is untested. Running the routine from many random paddle configurations and checking that every high-entropy endpoint also has low QBER would settle this.
The adaptive trial-count formula $m(H_{\mathrm{total}})$ is derived but not implemented; using it to set integration time dynamically could cut the roughly 200-minute convergence substantially.
Adding specific QBER terms to the cost function, as the authors suggest, would allow the same machinery not just to align bases but to select which Bell state the receivers measure, effectively turning the receiver into a tunable Bell-state analyzer.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

First, the core idea: defining coincidence entropies from the 16 coincidence counts and using H_total as a scalar cost function for gradient-descent polarization alignment is genuinely new. The definitions are simple and the single experimental run converging to QBER 4.2% in ~200 minutes is a real demonstration, not a simulation. The paper also does a decent job of explaining why this is useful for all-fiber receivers.

The soft spot is the Bell-state ambiguity. As you'd expect from the entropy symmetry, H_total is invariant under swapping the two outputs in a basis at one receiver, so a local Pauli-X at Bob leaves H_total unchanged while rotating the inferred Bell state. The paper even acknowledges in the conclusion that a composite cost function would be needed to align to an arbitrarily chosen Bell state; the current H_total only guarantees alignment to some Bell state. That is a material caveat. In practice, if the starting point is in the basin of the intended Bell state, the algorithm works, but nothing in the cost function protects the system from converging to a different basin. The single run reported is therefore evidence of feasibility, not of robust alignment.

The security-monitoring claim in the abstract is also overreaching. The paper shows H_total tracks entanglement quality in principle, but there is no demonstration that it detects intercept-resend attacks; the conclusion wisely downgrades to 'may serve as a tool.' The missing error bars on the main convergence curves and the lack of multiple independent trials are real but minor concerns; the linear fit of sigma_H vs H_total is a start.

The math is correct, and the citation pattern looks reasonable, though the paper leans on four references. The experiment is reproducible in principle, but no data or code are shipped, which is a weak point for a systems paper.

Who is this for? Researchers building all-fiber QKD receivers and looking for practical alignment metrics. It is not a fundamental result, and the Bell-state ambiguity needs to be fixed before widespread use. With a revision that either adds QBER terms to the cost function or explicitly runs multiple trials and reports the resulting Bell states, this could be a solid engineering contribution.

Recommendation: send it to peer review. The specific concerns are addressable, and the idea deserves scrutiny and replication. I would not cite it in my own work until the ambiguity is resolved, but I would bring it to a reading group to discuss the symmetry problem.

Referee Report

4 major / 4 minor

Summary. The paper proposes a method for automated polarization-basis alignment in an all-fiber entangled-photon detection setup. Coincidence entropies H_same and H_diff are defined from coincidence counts in the H/V and D/A bases, and their sum H_total is used as a scalar cost function for a stochastic gradient-descent algorithm that adjusts fiber paddle polarization controllers. The authors report a single experimental run converging to QBER_Φ+ ≈ 4.2% and H_total ≈ 13.9 in about 200 minutes, and they suggest that coincidence entropies can serve as a security-monitoring tool to detect eavesdropping, e.g., intercept-resend attacks.

Significance. If the central claims are sound, the method offers a useful, alignment-free alternative to auxiliary-laser calibration for entanglement-based QKD receivers, using only coincidence data. The paper provides a concrete algorithm, an experimental demonstration, and a noise model for H_total. However, the claimed Bell-state independence of the cost function, while convenient for basis alignment, creates an unresolved selectivity problem: the algorithm does not discriminate between different Bell states, and the security-monitoring claim is not backed by data. These issues materially affect the scope of the conclusions, but they are addressable in a revision.

major comments (4)

[Section 2, Eq. (6)] The cost function H_total is invariant under a local bit flip at one receiver. Since binary entropy satisfies H(p,1-p)=H(1-p,p), swapping C_jj and C_jk for all j (equivalently applying Pauli-X on Bob's qubit) leaves every H_same and H_diff term unchanged, hence H_total is unchanged. For the ideal states, this maps Φ+ to Ψ+. Therefore a configuration with H_total≈13.9 and QBER_Φ+≈4.2% is indistinguishable, by H_total alone, from one with H_total≈13.9 and QBER_Φ+≈95.8%. Because Section 3 states that no information about the target Bell state is used in the optimization, the single reported convergence does not demonstrate that the algorithm reliably finds the intended Bell-state sector; it only shows that the chosen starting point lay in the basin of the Φ+ sector. The paper should either acknowledge and discuss this limitation explicitly or modify the algorithm to include a state-selective term (as hinted in the future-work paragraph).
[Abstract and Section 4] The security-monitoring claim is not supported by the presented evidence. No attack scenario, simulation, or measurement is reported that shows how H_total responds to an intercept-resend attack or any other eavesdropping strategy. Moreover, because of the bit-flip invariance of H_total noted above, a channel that unitarily maps the shared state to a different Bell state would leave H_total maximal while changing the error rate with respect to the expected state; such an effect is precisely relevant to QKD key sifting. The conclusion's cautious phrasing ("could be utilized", "may serve") is appropriate, but the abstract's assertion that coincidence entropies "can be employed to monitor the quality of entanglement transmission... enhancing the system's ability to detect potential eavesdropping attempts" overstates what is demonstrated.
[Section 3, Figs. 3-4] The experimental results are reported as a single run without error bars on H_total or QBER. The converged value H_total≈13.9 is presented without an uncertainty estimate, even though Eq. (8) provides σ_H(H_total) from which a standard error of the mean could be computed. Without this statistical information, it is difficult to assess the stability of the convergence or the significance of the difference between the m=1 and m=5 trials. The authors should report the mean and SEM of the final quantities, or at least show the scatter of the steady-state time series.
[Section 3, gradient-descent procedure] The optimization landscape of H_total over the 12 paddle angles is not analyzed, and only one starting point is tested. The authors state that the step sizes were varied over 10°, 5°, and 1°, but the figures appear to label steps as "10 step, 3 step, 1 step," and no multiple independent trials with random initial paddle settings are reported. The claim "we are able to align the polarization bases" is therefore stronger than the evidence supports; the possibility of converging to a deceptive local maximum (or to an undesired Bell-state sector) is not addressed. This is acceptable for a proof-of-principle, but the language should be tempered accordingly.

minor comments (4)

[Section 2, Eq. (6)] The symbol H is used both for the binary entropy function H(p1,p2) and for the summed coincidence entropy H; this creates confusion. A different symbol for the binary entropy (e.g., S) would improve readability.
[Section 2, after Eq. (6)] The text says that H_same_j≈1 in the ideal case of C_jj≫C_jk, but for a perfectly anti-correlated state (C_jj≪C_jk) the same conclusion holds due to entropy symmetry. Clarifying this would also make the limitation discussed in major comment 1 more transparent to the reader.
[Section 3, Figs. 3-4] There is an inconsistency between the text, which lists step sizes 10°, 5°, and 1°, and the figure captions, which mention "10 step, 3 step, 1 step" (presumably 10°, 3°, and 1°). Please correct this discrepancy.
[Section 2, QBER definitions] The denominator β in the QBER expressions is defined as CHV+CVH+CDA+CAD+Σ C_jj, but it is not immediately obvious why the eight cross-basis coincidence terms (e.g., CHD) are excluded from β; a short explanation of the sifting process would help the reader follow the definition.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: coincidence entropy optimization is a closed-loop control procedure externally validated by QBER.

full rationale

The paper's derivation chain is self-contained and non-circular. The coincidence entropies in Eq. (6) are defined from measured coincidence counts, and the cost function H_total in Eq. (7) is a sum of these measured quantities. The gradient descent algorithm adjusts EPC paddles based only on H_total measurements; the reported QBER|phi+> = 4.2% is computed independently from the same coincidence counts and is not used as a fitting target. Thus the 'prediction' of successful alignment is externally validated rather than forced by construction. There are no load-bearing self-citations: the cited references are prior external works on polarization compensation, not the authors' own unverified results. The Bell-state independence of H_total, correctly stated in the paper, means that the cost function does not discriminate between different Bell-state sectors, which could cause a run to converge to a high-H_total configuration with high QBER relative to a particular target Bell state. This is a robustness or correctness limitation, not circular reasoning, because the paper's central claim is basis alignment and the reported QBER provides an independent check for the demonstrated run. The auxiliary fit of sigma_H in Eq. (8) is descriptive and not used as an input to the alignment optimization. Therefore no circularity is present.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The alignment algorithm rests on standard quantum measurement assumptions (Eq. 1), the stationarity of the entangled source, and the slow variation of fiber birefringence. The only numeric fit in the paper is the peripheral sigma_H calibration in Eq. (8), which is not used in the demonstrated alignment result. No new physical entities are introduced.

free parameters (2)

sigma_H vs H_total linear fit: intercept = 0.06
Intercept of the linear fit in Eq. (8) to the standard deviation of H_total measured in 5-second intervals (Fig. 2). Used to propose adaptive trial counting, but not implemented.
sigma_H vs H_total linear fit: slope = 0.01
Slope of the same fit; this relationship is used to compute the required number of trials m in Eq. (9), which is not part of the central alignment result.

assumptions (4)

domain assumption Measured coincidence counts are proportional to projective probabilities |<jk|Psi>|^2 (Eq. 1), with negligible dark counts and accidental coincidences.
Standard single-photon detection assumption; Eq. (1) is the basis for the QBER and entropy definitions. Dark counts and accidentals are not subtracted, which may inflate QBER.
domain assumption The entangled photon source is stationary and produces near-perfect copies of a Bell state (reported visibility of about 98%).
The optimization assumes a fixed, known state structure; the 2% admixture of noise is small but not modeled.
domain assumption Fiber birefringence is a unitary, slowly varying transformation that can be compensated by paddle-based polarization controllers on the timescale of the 5-second measurement intervals.
The algorithm correctness depends on the fiber transformation being quasi-static during the gradient estimation; this is stated qualitatively in Section 1 and not quantified.
domain assumption Satisfying the mutual unbiasedness conditions in Eqs. (3) and (4) is sufficient to guarantee QKD security.
The paper asserts that these conditions 'impose quantum correlations at the receivers that are sufficient for QKD applications' but provides no proof or citation for sufficiency.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Automated Polarization Basis Adjustment and Security Monitoring in Quantum Communication via Coincidence Entropies." pith.science (2026). https://pith.science/paper/ZGHGAYXD

@misc{pith2026250524071,
  author       = {Pith},
  title        = {Pith review of: Automated Polarization Basis Adjustment and Security Monitoring in Quantum Communication via Coincidence Entropies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZGHGAYXD}},
  note         = {Machine review of arXiv:2505.24071}
}

read the original abstract

Polarization-sensitive receivers for single photons are of crucial importance in various applications within the fields of quantum communication and quantum sensing, and are more commonly implemented in free-space optics rather than in optical fibers. This is primarily due to the unpredictable and varying birefringence in single-mode optical fibers. We present a method for birefringence compensation in an all-fiber detection setup that relies solely on coincidence measurements of a polarization-entangled state, or known correlations in a prepare-and-measure scenario. We define coincidence entropies as functions of the measured coincidence counts. These quantify the randomness of measurement outcomes, remain independent of the transmitted Bell state, and serve as indicators of the degree of entanglement. By leveraging coincidence entropy as a cost function in a gradient descent algorithm, we are able to align the polarization bases between two distinct polarization-sensitive receivers. Additionally, coincidence entropies can be employed to monitor the quality of entanglement transmission, thereby enhancing the system's ability to detect potential eavesdropping attempts, such as intercept-resend quantum attacks.

Figures

Figures reproduced from arXiv: 2505.24071 by the authors.

**Figure 2.** Dependance of standard deviation on the total coincidence entropy. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

[1]

A study of polarization compensation for quantum networks,

Perani´ c, M., Clark, M., Wang, R., Bahrani, S., Alia, O., Wengerowsky, S., Radman, A., Lonˇ cari´ c, M., Stipˇ cevi´ c, M., Rarity, J., Nejabati, R., and Joshi, S. K., “A study of polarization compensation for quantum networks,”EPJ Quantum Technology10(30) (2023)

work page 2023
[2]

Ai-assisted polarization basis alignment for quan- tum key distribution system receivers,

Mantey, S. T., Silva, N. A., Pinto, A. N., and Muga, N. J., “Ai-assisted polarization basis alignment for quan- tum key distribution system receivers,” in [23rd International Conference on Transparent Optical Networks (ICTON)], Tu.C2.5, IEEE (2023)

work page 2023
[3]

Fibre polarization state compensation in entanglement- based quantum key distribution,

Shi, Y., Poh, H. S., Ling, A., and Kurtsiefer, C., “Fibre polarization state compensation in entanglement- based quantum key distribution,”Optics Express29(20), 31741–31752 (2021)

work page 2021
[4]

Real-time polarization compensation method in quantum communi- cation based on channel muller parameters detection,

Tan, Y., Wang, J., Wu, J., and He, Z., “Real-time polarization compensation method in quantum communi- cation based on channel muller parameters detection,”Communications Engineering3(57) (2024)

work page 2024
[5]

Analytical solution to the problem of polarization drift compensation in an all-fiber qkd system,

Mayboroda, V., Rudavin, N., Kupriyanov, P., Fat’yanov, O., and Shakhovoy, R., “Analytical solution to the problem of polarization drift compensation in an all-fiber qkd system,”Optics Express32(26), 45421–45435 (2024)

work page 2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

A study of polarization compensation for quantum networks,

Perani´ c, M., Clark, M., Wang, R., Bahrani, S., Alia, O., Wengerowsky, S., Radman, A., Lonˇ cari´ c, M., Stipˇ cevi´ c, M., Rarity, J., Nejabati, R., and Joshi, S. K., “A study of polarization compensation for quantum networks,”EPJ Quantum Technology10(30) (2023)

work page 2023

[2] [2]

Ai-assisted polarization basis alignment for quan- tum key distribution system receivers,

Mantey, S. T., Silva, N. A., Pinto, A. N., and Muga, N. J., “Ai-assisted polarization basis alignment for quan- tum key distribution system receivers,” in [23rd International Conference on Transparent Optical Networks (ICTON)], Tu.C2.5, IEEE (2023)

work page 2023

[3] [3]

Fibre polarization state compensation in entanglement- based quantum key distribution,

Shi, Y., Poh, H. S., Ling, A., and Kurtsiefer, C., “Fibre polarization state compensation in entanglement- based quantum key distribution,”Optics Express29(20), 31741–31752 (2021)

work page 2021

[4] [4]

Real-time polarization compensation method in quantum communi- cation based on channel muller parameters detection,

Tan, Y., Wang, J., Wu, J., and He, Z., “Real-time polarization compensation method in quantum communi- cation based on channel muller parameters detection,”Communications Engineering3(57) (2024)

work page 2024

[5] [5]

Analytical solution to the problem of polarization drift compensation in an all-fiber qkd system,

Mayboroda, V., Rudavin, N., Kupriyanov, P., Fat’yanov, O., and Shakhovoy, R., “Analytical solution to the problem of polarization drift compensation in an all-fiber qkd system,”Optics Express32(26), 45421–45435 (2024)

work page 2024