pith. sign in

arxiv: 2604.25194 · v1 · submitted 2026-04-28 · 🪐 quant-ph · cs.NI

Quantum-enhanced Network Tomography

Yufei Zheng , Zihao Gong , Saikat Guha , Don Towsley This is my paper

Pith reviewed 2026-05-07 16:42 UTC · model grok-4.3

classification 🪐 quant-ph cs.NI
keywords network tomographyquantum probesoptical networksFisher information matrixlink transmissivitycontinuous-variable quantum opticslink identifiability
0
0 comments X

The pith

Quantum probes consisting of squeezed or entangled light pulses improve the estimation of link transmissivities in general optical networks.

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

Network tomography infers the loss on each internal link by sending probes from accessible nodes and measuring the output. This work replaces ordinary laser pulses with blocks of coherent states that carry squeezing or weak temporal-mode entanglement, which homodyne receivers can use to extract more information about channel transmissivity. An algorithm selects probe routes that make every link distinguishable while packing as many independent information sets as possible into the measurements. The resulting Fisher information matrix supplies two performance numbers, its determinant and the trace of its inverse, that directly compare quantum and classical accuracy on any network topology. If the approach holds, operators can monitor the same set of links with either fewer probes or tighter uncertainty bounds.

Core claim

Assuming a subset of nodes can generate and detect the quantum probes and that intermediate nodes support all-optical switching, blocks of n coherent-state pulses augmented with continuous-variable squeezing (n=1) or weak temporal-mode entanglement (n>1) carry information about the transmissivity of every traversed link. A probe-construction algorithm selects routes that guarantee link identifiability while maximizing the number of information-orthogonal sets of transmissivities. The Fisher information matrix induced by these probes is evaluated through its determinant and the trace of its inverse, yielding a concrete characterization of the quantum improvement over classical probes for any

What carries the argument

The Fisher information matrix induced by the chosen set of quantum probes, which encodes how much each link transmissivity can be estimated from the homodyne measurement outcomes.

If this is right

  • Every link transmissivity becomes uniquely recoverable from the end-to-end quantum measurements.
  • The determinant of the Fisher information matrix grows with the use of squeezing or entanglement, indicating greater total information.
  • The trace of the inverse matrix shrinks, corresponding to lower average estimation variance across all links.
  • The same performance metrics apply unchanged to any network topology once the routing algorithm is executed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Operators could monitor larger networks with the same number of measurements or maintain the same accuracy with fewer access points.
  • The routing algorithm might be adapted to estimate additional parameters such as phase or dispersion alongside loss.
  • Networks already equipped with all-optical switches gain extra monitoring capability without new hardware beyond the quantum sources and detectors.

Load-bearing premise

A subset of nodes must be able to prepare and detect the squeezed or entangled pulses and to route them through all-optical switching at every intermediate node.

What would settle it

Running the routing algorithm on a small test network, sending the quantum probes, and finding that the observed estimation variance for the transmissivities is no smaller than the classical shot-noise bound would show that the claimed quantum improvement does not occur.

Figures

Figures reproduced from arXiv: 2604.25194 by Don Towsley, Saikat Guha, Yufei Zheng, Zihao Gong.

Figure 1
Figure 1. Figure 1: Tr(I −1 e,d ) − Tr(I −1 s,d ) with respect to η1 and η2. When N is large enough, 2Nn(n − c1Sη) dominates nc2 1Sη, we drop the nc2 1Sη term in the denominator to get a cleaner upper bound, Tr(I −1 s,d ) ≤ Sη(n − c1Sη) Nn . (16) Similarly for Tr(I −1 e,d ), the intuition is best understood by focusing on the dominating terms, even though this does not yield a lower bound: Tr(I −1 e,d ) ∼ 2(n − cnSη) · 4Nn(n … view at source ↗
Figure 2
Figure 2. Figure 2: A network and two sets of probes that guarantees identifiability. view at source ↗
Figure 3
Figure 3. Figure 3: Differences in the two metrics for the two-channel setup. view at source ↗
read the original abstract

Network tomography refers to the use of inference techniques for inferring internal network states from end-to-end probes. Quantum probes, implemented by sending blocks of $n$ coherent-state pulses augmented with continuous-variable (CV) squeezing ($n=1$) or weak temporal-mode entanglement ($n>1$) over a lossy channel to a receiver with homodyne detection capabilities, are known to carry information about the channel transmissivity. Assuming a subset of nodes in an optical network is capable of sending and receiving such probes through intermediate nodes with all-optical switching capabilities, we leverage these quantum probes to estimate link transmissivities. To determine how to route the probes in a network, we propose a probe construction algorithm that guarantees link identifiability, while maximizing the number of information orthogonal sets of transmissivities. A set of probes induces a Fisher information matrix (FIM). We then derive two metrics, the determinant of the FIM and the trace of its inverse, to evaluate the performance of the probes. In particular, our results can be used to characterize the quantum improvement in estimating link transmissivities in a general optical network.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes a quantum-enhanced network tomography framework for estimating link transmissivities in optical networks. Quantum probes consist of blocks of n coherent-state pulses augmented with CV squeezing (n=1) or weak temporal-mode entanglement (n>1), routed via all-optical switches at intermediate nodes. A probe construction algorithm is introduced to guarantee link identifiability while maximizing the number of information-orthogonal sets of transmissivities. Performance is quantified using two FIM-derived metrics—the determinant of the Fisher information matrix and the trace of its inverse—to characterize quantum improvements relative to classical coherent-state probes.

Significance. If the derivations and assumptions hold, the work could provide a systematic method for designing quantum probes that yield measurable advantages in network parameter estimation, with potential relevance to quantum communication infrastructure. The choice of standard FIM metrics (det(FIM) and tr(FIM^{-1})) is appropriate for comparing quantum and classical performance and for assessing identifiability.

major comments (3)
  1. [Abstract] Abstract: The FIM is stated to be induced by the set of probes, yet no explicit expression is supplied for the FIM in terms of the quantum-state covariance matrix, squeezing parameter, entanglement strength, or the vector of link transmissivities. Without this, it is impossible to verify how the quantum resources increase det(FIM) or decrease tr(FIM^{-1}) relative to the classical case.
  2. [Abstract] Abstract: The probe construction algorithm is claimed to guarantee link identifiability and to maximize the number of information-orthogonal sets, but the manuscript provides neither pseudocode, a formal statement of the algorithm, nor a proof of these guarantees. This omission renders the central methodological contribution unverifiable.
  3. [Abstract] Abstract: The load-bearing assumption that all-optical switching at intermediate nodes preserves squeezing or entanglement without additional loss or decoherence (beyond the modeled link transmissivities) is stated but not incorporated into the FIM derivation or the probe-design procedure. Any switch-induced mixing or extra loss would alter the effective covariance matrix, directly lowering det(FIM) and raising tr(FIM^{-1}), thereby invalidating the claimed quantum improvement.
minor comments (2)
  1. [Abstract] The abstract refers to 'our results' that can be used to characterize quantum improvement but does not summarize what those concrete results are (e.g., analytic bounds, scaling with n, or example network calculations).
  2. [Abstract] Notation for the block size n and its relation to the squeezing/entanglement parameters should be introduced more explicitly to clarify how these enter the FIM.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. We address each major point below and will incorporate revisions to enhance the clarity, verifiability, and rigor of the manuscript.

read point-by-point responses

  1. Referee: The FIM is stated to be induced by the set of probes, yet no explicit expression is supplied for the FIM in terms of the quantum-state covariance matrix, squeezing parameter, entanglement strength, or the vector of link transmissivities. Without this, it is impossible to verify how the quantum resources increase det(FIM) or decrease tr(FIM^{-1}) relative to the classical case.

    Authors: We agree that an explicit expression is necessary for full verification. Although the manuscript derives the FIM from the probe states, we will add a dedicated subsection in the revised version that provides the closed-form FIM in terms of the covariance matrix, squeezing parameter, entanglement strength, and link transmissivities, explicitly showing the quantum improvement over the classical case. revision: yes

  2. Referee: The probe construction algorithm is claimed to guarantee link identifiability and to maximize the number of information-orthogonal sets, but the manuscript provides neither pseudocode, a formal statement of the algorithm, nor a proof of these guarantees. This omission renders the central methodological contribution unverifiable.

    Authors: This is a valid observation. We will include a formal statement of the algorithm, pseudocode, and a proof of the identifiability guarantee together with the maximization of information-orthogonal sets. These will be added to the main text or a new appendix in the revision. revision: yes

  3. Referee: The load-bearing assumption that all-optical switching at intermediate nodes preserves squeezing or entanglement without additional loss or decoherence (beyond the modeled link transmissivities) is stated but not incorporated into the FIM derivation or the probe-design procedure. Any switch-induced mixing or extra loss would alter the effective covariance matrix, directly lowering det(FIM) and raising tr(FIM^{-1}), thereby invalidating the claimed quantum improvement.

    Authors: We acknowledge this modeling assumption requires explicit treatment. The current derivation assumes ideal switching with losses only from the links. In revision we will incorporate this assumption directly into the FIM derivation, add a discussion of potential switch-induced effects, and clarify the conditions under which the reported quantum improvements remain valid. revision: partial

Circularity Check

0 steps flagged

No circularity: standard FIM metrics applied to independently defined quantum probes and routing algorithm

full rationale

The derivation begins with quantum probes (coherent pulses plus CV squeezing or weak entanglement) whose effect on transmissivity is modeled via standard quantum optics loss channels. A probe construction algorithm is then proposed to select routes that guarantee identifiability while maximizing the count of information-orthogonal transmissivity sets; these criteria are defined directly from the network graph and the linear mapping of link transmissivities to probe outcomes, without reference to the FIM. The FIM itself is induced by the chosen probes exactly as in classical estimation theory, after which det(FIM) and tr(FIM^{-1}) are computed from the standard definitions. No equation reduces to its own input by construction, no parameter is fitted on a subset and then relabeled a prediction, and no load-bearing step rests on a self-citation. The quantum-improvement claim follows from comparing the resulting FIM eigenvalues to the classical coherent-state case under the stated all-optical switching assumption, which is an external modeling choice rather than a derived identity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Review performed on abstract only; full derivations, assumptions, and any fitted parameters are not visible. The approach rests on standard quantum optics and classical estimation assumptions without introducing new free parameters or entities in the visible text.

axioms (2)
  • domain assumption Quantum probes (squeezed coherent states or weakly entangled temporal modes) carry information about channel transmissivity under homodyne detection
    Stated as known in the abstract; no derivation supplied.
  • domain assumption All-optical switching at intermediate nodes allows lossless routing of probes between sender and receiver nodes
    Invoked to enable the probe routing scheme.

pith-pipeline@v0.9.0 · 5491 in / 1462 out tokens · 51512 ms · 2026-05-07T16:42:44.646116+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

41 extracted references · 41 canonical work pages

  1. [1]

    Srlg failure localization in optical networks.IEEE/ACM Transactions on Networking, 19(4):989–999, 2011

    Satyajeet S Ahuja, Srinivasan Ramasubramanian, and Marwan Krunz. Srlg failure localization in optical networks.IEEE/ACM Transactions on Networking, 19(4):989–999, 2011

    work page 2011

  2. [2]

    OUP Oxford, 2007

    Anthony Atkinson, Alexander Donev, and Randall Tobias.Optimum experimental designs, with SAS, volume 34. OUP Oxford, 2007

    work page 2007

  3. [3]

    Algorithms for network topology discovery using end-to-end measurements

    Laurent Bobelin and Traian Muntean. Algorithms for network topology discovery using end-to-end measurements. In2008 International Symposium on Parallel and Distributed Computing, pages 267–

    work page

  4. [4]

    The convergence of a class of double-rank minimization algorithms 1

    Charles George Broyden. The convergence of a class of double-rank minimization algorithms 1. general considerations.IMA Journal of Applied Mathematics, 6(1):76–90, 1970

    work page 1970

  5. [5]

    Network tomography on general topolo- gies.ACM SIGMETRICS Performance Evaluation Review, 30(1):21–30, 2002

    Tian Bu, Nick Duffield, Francesco Lo Presti, and Don Towsley. Network tomography on general topolo- gies.ACM SIGMETRICS Performance Evaluation Review, 30(1):21–30, 2002

    work page 2002

  6. [6]

    Simple network management protocol (SNMP)

    Jeffrey D Case, Mark Fedor, Martin L Schoffstall, and James Davin. Simple network management protocol (SNMP). Technical report, 1989

    work page 1989

  7. [7]

    Network Tomography: Recent Developments.Statistical Science, 19(3):499 – 517, 2004

    Rui Castro, Mark Coates, Gang Liang, Robert Nowak, and Bin Yu. Network Tomography: Recent Developments.Statistical Science, 19(3):499 – 517, 2004

    work page 2004

  8. [8]

    Network health and e-science in commercial clouds

    Ryan Chard, Kris Bubendorfer, and Bryan Ng. Network health and e-science in commercial clouds. Future Generation Computer Systems, 56:595–604, 2016

    work page 2016

  9. [9]

    Network tomography for internal delay estimation

    Mark J Coates and Robert D Nowak. Network tomography for internal delay estimation. In2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No. 01CH37221), volume 6, pages 3409–3412. IEEE, 2001

    work page 2001

  10. [10]

    MIT press, 2022

    Thomas H Cormen, Charles E Leiserson, Ronald L Rivest, and Clifford Stein.Introduction to algorithms. MIT press, 2022

    work page 2022

  11. [11]

    Parameter orthogonality and approximate conditional inference

    David Roxbee Cox and Nancy Reid. Parameter orthogonality and approximate conditional inference. Journal of the Royal Statistical Society: Series B (Methodological), 49(1):1–18, 1987

    work page 1987

  12. [12]

    Quantum network tomography.IEEE Network, 38(5):114–122, 2024

    Matheus Guedes De Andrade, Jake Navas, Saikat Guha, In` es Monta˜ no, Michael Raymer, Brian Smith, and Don Towsley. Quantum network tomography.IEEE Network, 38(5):114–122, 2024

    work page 2024

  13. [13]

    Efficient and reliable network tomography in hetero- geneous networks using bittorrent broadcasts and clustering algorithms

    Kiril Dichev, Fergal Reid, and Alexey Lastovetsky. Efficient and reliable network tomography in hetero- geneous networks using bittorrent broadcasts and clustering algorithms. InSC’12: Proceedings of the International Conference on High Performance Computing, Networking, Storage and Analysis, pages 1–11. IEEE, 2012

    work page 2012

  14. [14]

    Network tomography from measured end-to-end delay covariance

    Nick G Duffield and F Lo Presti. Network tomography from measured end-to-end delay covariance. IEEE/ACM Transactions On Networking, 12(6):978–992, 2004

    work page 2004

  15. [15]

    A new approach to variable metric algorithms.The computer journal, 13(3):317–322, 1970

    Roger Fletcher. A new approach to variable metric algorithms.The computer journal, 13(3):317–322, 1970

    work page 1970

  16. [16]

    Resource allocation via graph neural networks in free space optical fronthaul networks

    Zhan Gao, Mark Eisen, and Alejandro Ribeiro. Resource allocation via graph neural networks in free space optical fronthaul networks. InGLOBECOM 2020-2020 IEEE Global Communications Conference, pages 1–6. IEEE, 2020

    work page 2020

  17. [17]

    Multiparameter squeezing for optimal quantum enhancements in sensor networks.Nature communications, 11(1):3817, 2020

    Manuel Gessner, Augusto Smerzi, and Luca Pezz` e. Multiparameter squeezing for optimal quantum enhancements in sensor networks.Nature communications, 11(1):3817, 2020. 18

    work page 2020

  18. [18]

    Netscope: Prac- tical network loss tomography

    Denisa Ghita, Hung Nguyen, Maciej Kurant, Katerina Argyraki, and Patrick Thiran. Netscope: Prac- tical network loss tomography. In2010 Proceedings IEEE INFOCOM, pages 1–9. IEEE, 2010

    work page 2010

  19. [19]

    A family of variable-metric methods derived by variational means.Mathematics of computation, 24(109):23–26, 1970

    Donald Goldfarb. A family of variable-metric methods derived by variational means.Mathematics of computation, 24(109):23–26, 1970

    work page 1970

  20. [20]

    Quantum-enhanced quickest change detection of transmission loss.Physical Review Letters, 135(21):210801, 2025

    Saikat Guha, Tiju Cherian John, Zihao Gong, and Prithwish Basu. Quantum-enhanced quickest change detection of transmission loss.Physical Review Letters, 135(21):210801, 2025

    work page 2025

  21. [21]

    Non- adaptive fault diagnosis for all-optical networks via combinatorial group testing on graphs

    Nicholas JA Harvey, Mihai Patrascu, Yonggang Wen, Sergey Yekhanin, and Vincent WS Chan. Non- adaptive fault diagnosis for all-optical networks via combinatorial group testing on graphs. InIEEE INFOCOM 2007-26th IEEE International Conference on Computer Communications, pages 697–705. IEEE, 2007

    work page 2007

  22. [22]

    Fisher information-based experiment design for network tomography.ACM SIGMETRICS Performance Evaluation Review, 43(1):389–402, 2015

    Ting He, Chang Liu, Ananthram Swami, Don Towsley, Theodoros Salonidis, Andrei Iu Bejan, and Paul Yu. Fisher information-based experiment design for network tomography.ACM SIGMETRICS Performance Evaluation Review, 43(1):389–402, 2015

    work page 2015

  23. [23]

    Cambridge University Press, 2021

    Ting He, Liang Ma, Ananthram Swami, and Don Towsley.Network tomography: identifiability, mea- surement design, and network state inference. Cambridge University Press, 2021

    work page 2021

  24. [24]

    Growing AWS internet peering with 400 GbE, 2023.https://aws.amazon.com/blogs/networking-and-content-delivery/ growing-aws-internet-peering-with-400-gbe/

    Fredrik Korsb¨ ack, Lincoln Dale, and Dave McGaugh. Growing AWS internet peering with 400 GbE, 2023.https://aws.amazon.com/blogs/networking-and-content-delivery/ growing-aws-internet-peering-with-400-gbe/

    work page 2023

  25. [25]

    Link budget analysis for free-space optical satellite networks

    Jintao Liang, Aizaz U Chaudhry, Eylem Erdogan, and Halim Yanikomeroglu. Link budget analysis for free-space optical satellite networks. In2022 IEEE 23rd International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM), pages 471–476. IEEE, 2022

    work page 2022

  26. [26]

    Hybrid classical-quantum communication networks

    Joseph M Lukens, Nicholas A Peters, and Bing Qi. Hybrid classical-quantum communication networks. Progress in Quantum Electronics, page 100586, 2025

    work page 2025

  27. [27]

    Efficient identification of addi- tive link metrics via network tomography

    Liang Ma, Ting He, Kin K Leung, Don Towsley, and Ananthram Swami. Efficient identification of addi- tive link metrics via network tomography. In2013 IEEE 33rd International Conference on Distributed Computing Systems, pages 581–590. IEEE, 2013

    work page 2013

  28. [28]

    Stephen Nellis. Cisco and qunnect build quantum network using new york fiber optic cables.https://www.reuters.com/business/media-telecom/ cisco-qunnect-build-quantum-network-using-new-york-fiber-optic-cables-2026-02-18/, February 2026. Accessed: 2026-04-17

    work page 2026

  29. [29]

    Multiparameter gaussian quantum metrology.Physical Review A, 98(1):012114, 2018

    Rosanna Nichols, Pietro Liuzzo-Scorpo, Paul A Knott, and Gerardo Adesso. Multiparameter gaussian quantum metrology.Physical Review A, 98(1):012114, 2018

    work page 2018

  30. [30]

    Springer, 2006

    Jorge Nocedal and Stephen J Wright.Numerical optimization. Springer, 2006

    work page 2006

  31. [31]

    Jupiter evolving: transforming google’s datacenter network via optical circuit switches and software-defined networking

    Leon Poutievski, Omid Mashayekhi, Joon Ong, Arjun Singh, Mukarram Tariq, Rui Wang, Jianan Zhang, Virginia Beauregard, Patrick Conner, Steve Gribble, et al. Jupiter evolving: transforming google’s datacenter network via optical circuit switches and software-defined networking. InProceedings of the ACM SIGCOMM 2022 Conference, pages 66–85, 2022

    work page 2022

  32. [32]

    Conditioning of quasi-newton methods for function minimization.Mathematics of computation, 24(111):647–656, 1970

    David F Shanno. Conditioning of quasi-newton methods for function minimization.Mathematics of computation, 24(111):647–656, 1970

    work page 1970

  33. [33]

    Adjustment of an inverse matrix corresponding to a change in one element of a given matrix.The Annals of Mathematical Statistics, 21(1):124–127, 1950

    Jack Sherman and Winifred J Morrison. Adjustment of an inverse matrix corresponding to a change in one element of a given matrix.The Annals of Mathematical Statistics, 21(1):124–127, 1950. 19

    work page 1950

  34. [34]

    Passive network tomography using em algorithms

    Yolanda Tsang, Mark Coates, and Robert Nowak. Passive network tomography using em algorithms. In2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No. 01CH37221), volume 3, pages 1469–1472. IEEE, 2001

    work page 2001

  35. [35]

    Network delay tomography.IEEE Transactions on Signal Processing, 51(8):2125–2136, 2003

    Yolanda Tsang, Mark Coates, and Robert D Nowak. Network delay tomography.IEEE Transactions on Signal Processing, 51(8):2125–2136, 2003

    work page 2003

  36. [36]

    Network tomography: Estimating source-destination traffic intensities from link data

    Yehuda Vardi. Network tomography: Estimating source-destination traffic intensities from link data. Journal of the American statistical association, 91(433):365–377, 1996

    work page 1996

  37. [37]

    Lukas Velush. Boosting our connectivity with our own next-generation optical network, 2023.https://www.microsoft.com/insidetrack/blog/ boosting-our-connectivity-with-our-own-next-generation-optical-network/

    work page 2023

  38. [38]

    Quantum network tomography for general topology with spam errors.arXiv preprint arXiv:2511.01074, 2025

    Xuchuang Wang, Matheus Guedes De Andrade, Guus Avis, Yu-Zhen Janice Chen, Mohammad Hajies- maili, and Don Towsley. Quantum network tomography for general topology with spam errors.arXiv preprint arXiv:2511.01074, 2025

    work page arXiv 2025

  39. [39]

    Estimating network loss rates using active tomog- raphy.Journal of the American Statistical Association, 101(476):1430–1448, 2006

    Bowei Xi, George Michailidis, and Vijayan N Nair. Estimating network loss rates using active tomog- raphy.Journal of the American Statistical Association, 101(476):1430–1448, 2006

    work page 2006

  40. [40]

    Towards unbiased end-to-end network diagnosis.IEEE/ACM Transactions on Networking, 17(6):1724–1737, 2009

    Yao Zhao, Yan Chen, and David Bindel. Towards unbiased end-to-end network diagnosis.IEEE/ACM Transactions on Networking, 17(6):1724–1737, 2009

    work page 2009

  41. [41]

    A quantum speedup in localizing transmission loss change in optical networks

    Yufei Zheng, Yu-Zhen Janice Chen, Prithwish Basu, and Don Towsley. A quantum speedup in localizing transmission loss change in optical networks. In2025 IEEE International Conference on Quantum Computing and Engineering (QCE), volume 1, pages 958–968. IEEE, 2025. 20 0.0 0.2 0.4 0.6 0.8 1.0 1 0.0 0.2 0.4 0.6 0.8 1.0 2 0 1e5 2e5 3e5 4e5 det(FIM s) det(FIM e)...

    work page 2025