PaperlayerView exact revision on arXiv

Source-linked AI summary

Distributed Quantum Sensing

Zheshen Zhang, Quntao Zhuang

arXiv:2010.14744v2quant-phphysics.optics

TL;DR

Distributed quantum sensing asks how entangled states shared by multiple sensors can improve measurement tasks beyond the standard quantum limit. This Review develops a quantum-information formulation, surveys protocols and experiments, and reports optimality results in specified settings alongside demonstrated advantages in RF sensing and data classification. It also identifies mixed-state Fisher-information analysis as a practical limitation.

  • Problem

    Distributed sensing needs a framework for assessing how shared entanglement across multiple sensors can improve estimation of global properties beyond classical sensing limits.

  • Method

    The Review formulates DQS from a quantum-information perspective, compares it with DCS, analyzes continuous- and discrete-variable protocols, and examines applications and nonidealities.

  • Results

    In lossless conditions, the original DQS protocol is optimal among all DQS protocols, while experiments report advantages over DCS in RF sensing and reduced error probability in data classification.

  • Takeaways & Limitations

    DQS provides a framework and demonstrated route for using multipartite entanglement in distributed sensing and quantum-enhanced data processing.

  • Takeaways & Limitations

    Experimental imperfections produce mixed output states, making exact Fisher-information optimization more challenging and motivating upper bounds or tractable restricted input-state classes.

Abstract

from arXiv · show

A plethora of applications hinge on a network or an array of sensors to undertake measurement tasks. A rule of thumb for sensing is that a collective measurement taken by $M$ independent sensors can improve the sensitivity by $1/\sqrt{M}$, known as the standard quantum limit (SQL). Quantum resources such as entanglement and squeezed light can be harnessed to surpass the SQL. Distributed quantum sensing is an emerging subject dedicated to investigating the performance gain enabled by entangled states shared by multiple sensors in tackling different measurement problems. This Review formulates distributed quantum sensing from a quantum-information perspective and describes distributed quantum sensing protocols and their experimental demonstrations. The applications enabled by distributed quantum sensing and an outlook for future opportunities will also be discussed.

1. Introduction

Quantum sensing offers near-term opportunities to outperform classical sensing, while distributed quantum sensing extends quantum-metrology gains from single sensors to networks sharing entangled states. This Review formulates DQS, surveys its protocols and demonstrations, examines applications, and discusses future prospects.

  • Motivation: Quantum metrology uses nonclassical resources to enhance measurements across applications including gravitational-wave detection, target detection, microscopy, biological sensing, and phase tracking.LIGO surpasses the standard quantum limit by injecting squeezed light into a Michelson interferometer.
  • Motivation: Earlier quantum-metrology experiments focused on improving performance at a single sensor.Recent theoretical work instead proposed distributed protocols using shared entangled states across multiple sensors.
  • Motivation: Distributed quantum sensing targets performance gains from shared entanglement when multiple sensors probe global properties of an object.This extends quantum sensing from individual measurement devices to distributed sensing scenarios.
  • Review scope: The Review compares DQS with distributed classical sensing, describes continuous-variable and discrete-variable protocols, and analyzes quantum advantages under practical nonidealities.It also quantifies performance and accounts for noise and other experimental imperfections.
  • Review scope: Two recent experiments illustrate DQS applications in optical phase sensing and radio-frequency sensing, followed by an outlook for future prospects.The applications are reviewed in the context of experimental demonstrations.

2. Overview

DQS uses entangled probes distributed across multiple sensors to infer a global property, whereas DCS uses separable sensor states. Fair comparison requires explicit resource accounting, typically by constraining total sensor power.

  • Distributed sensing model: DQS estimates a global property from measurement data collected by multiple sensors jointly interrogating an object with an entangled probe state.A quantum circuit prepares the shared probe before the sensors interrogate the object and postprocess their data.
  • Distributed sensing model: DCS uses separable sensor states, such as a product state, while applying the same distributed-data strategy to infer a global property.Figure 1 contrasts entangled inputs for DQS with separable inputs for DCS.
  • Fair comparison: Resource counting is essential because DCS could otherwise outperform DQS simply by using more power at the sensors.One widely used scheme compares protocols with the same total sensor power and defines the SQL from the constrained DCS sensitivity.
  • Fair comparison: The SQL is defined through a distributed classical-sensing protocol subject to the chosen resource constraint.The comparison baseline therefore depends on how resources are counted.

3. Distributed Quantum Sensing Protocols

Distributed quantum sensing estimates global properties of multiple parameters using entangled probes shared by multiple sensors, extending single-parameter quantum sensing beyond the SQL setting. The Review develops continuous-variable and discrete-variable protocols, analyzes their performance under imperfections, and identifies conditions for optimality and unresolved loss-dependent cases.

  • Foundations: Single-parameter sensing prepares a probe, applies a parameter-dependent unitary, measures the output, and constructs an estimator whose precision is quantified by rms error.Independent probes yield SQL scaling δα_M = δα/√M.
  • Foundations: DQS uses an entangled probe state shared by M sensors to estimate a global property of multiple parameters, unlike conventional multiparameter estimation.The global property can be reduced to estimating a weighted average of the parameters.
  • Displacement sensing: For identical displacements and weights, a squeezed-vacuum state distributed through a balanced beam-splitter array supports homodyne-based displacement sensing.Equivalent reductions transform the multimode protocol into a single-mode displacement-estimation problem.
  • Displacement sensing: The identical-displacement DQS protocol is optimal among Gaussian DQS protocols and, without loss, among all DQS protocols.The corresponding separable Gaussian protocol is SQL-limited, while the DQS performance can beat a non-tight separable lower bound near η = 1.
  • General protocols: For lossless general displacement sensing, the separable protocol is optimal within Gaussian states and among all DCS protocols, whereas the entangled protocol can saturate Heisenberg scaling.The entanglement benefit arises when photons are concentrated in a few beam-splitter input modes rather than evenly distributed.
  • Phase sensing: With loss, adding a nonzero displacement can improve phase-sensing performance, but the optimal entangled probe remains unknown even when restricted to Gaussian states.This uncertainty is explicitly reported for the lossy phase-sensing setting.
  • Variable encodings: The CV DQS protocol has better reported performance than the generalized twin-Fock DV protocol, although both achieve Heisenberg scaling.The reported comparison is δα_E = 2/√[2N_S(N_S + 2)] for the DV protocol versus the cited CV results.

4. Performance Limits

This section reviews quantum-sensing performance limits, beginning with quantum Cramér–Rao, standard quantum, and Heisenberg bounds before applying them to DQS and DCS protocols.

  • The section first reviews quantum Cramér–Rao, standard quantum, and Heisenberg limits for quantum sensing.It then uses the quantum Cramér–Rao bound to analyze DQS and DCS performance and assess protocol optimality.

4.1. Quantum Cramér-Rao bound and the standard quantum limit

The quantum Cramér–Rao framework quantifies estimation precision through Fisher information, with product-state strategies subject to SQL scaling and entangled strategies outside that restriction.

  • The quantum Cramér–Rao bound gives the asymptotic precision limit for unbiased estimation of a single parameter from a quantum state.
  • Fisher information can be defined using Uhlmann fidelity or the symmetric logarithmic derivative.
  • For pure states generated unitarily, Fisher information is obtained from the variance of the generator.This variance expression does not generally hold for mixed states, although the fidelity and symmetric-logarithmic-derivative formulations remain valid.
  • Adaptive strategies can attain the asymptotic Cramér–Rao bound for single-parameter estimation while retaining product-form input states.
  • Allowing arbitrary separable inputs and joint measurements still preserves the SQL scaling constraint.
  • The SQL corresponds to 1/√M scaling, where M is the number of sensing attempts.

4.2. Beating the standard quantum limit

Entanglement can surpass the SQL by producing superadditive Fisher information, with Heisenberg scaling in finite-dimensional settings and specialized advantages for constrained continuous-variable sensing.

  • 4.2. Beating the standard quantum limit: Entanglement across M measurements can approach Heisenberg precision scaling, δα_M ∝1/M, whereas separable states obey additive Fisher information.
  • 4.2.1. Finite-dimensional probes: For finite-dimensional probes, separable inputs are bounded by additive Fisher information, while entangled inputs can achieve Fisher information proportional to M^2.The resulting optimal precision follows δα_M ∝1/M.
  • 4.2.2. Infinite-dimensional probes: For infinite-dimensional probes, a mean occupation constraint is required because unbounded generator spectra can otherwise make Fisher information unbounded.
  • 4.2.2. Infinite-dimensional probes: In displacement sensing under a total energy constraint, separable probes use iid squeezed-vacuum states and their asymptotic Fisher information is independent of M.
  • 4.2.2. Infinite-dimensional probes: For continuous-variable sensing, concentrating energy into a global mode through a beam-splitter transform is optimal, with entanglement generated from a single-mode squeezed vacuum.

4.3. Precision limits in the presence of noise

In the presence of loss and other imperfections, the Review derives Fisher-information upper bounds and identifies optimal Gaussian probes in several DQS settings, while leaving some bounds’ achievability unresolved.

  • General bounds: Experimental imperfections produce mixed probe states, making Fisher-information evaluation more challenging than for pure states.The Review notes that exact optimization is difficult and considers upper bounds or Gaussian-state restrictions as tractable alternatives.
  • General bounds: At η = 1, the Fisher-information upper bound recovers the lossless result and is achievable; at η < 1, its achievability remains unclear.Purification yields the same upper bound as the direct analysis.
  • Optimal Gaussian states: A single-mode squeezed-vacuum state is the optimal Gaussian probe for single-mode sensing.The Gaussian-state optimization maximizes Fisher information at θ = n = 0, with zero displacement optimal under the stated photon-number constraint.
  • Optimal Gaussian states: For separable Gaussian inputs, an iid product of single-mode squeezed-vacuum states achieves the optimized bound, and the lossless case confirms Gaussian optimality.At η = 1, the separable-input result reduces to the earlier lossless expression.
  • Optimal Gaussian states: For entangled inputs, reducing the M-probe problem to one probing mode yields an achievable bound, and η = 1 again confirms Gaussian optimality.The entangled probe is constructed using the protocol introduced earlier in the Review.

4.4. Multiparameter quantum Cramér-Rao bound

Distributed sensing of a global scalar function generally requires a multiparameter quantum Cramér–Rao analysis. The Review formulates the Fisher-information matrix and derives the corresponding weighted precision limit.

  • Problem formulation: Although one scalar function is estimated, distributed sensing generally involves multiple unknown parameters and therefore requires a multiparameter quantum Cramér–Rao bound.The Review considers commuting generators for the distributed unitary model.
  • Fisher-information matrix: The estimator’s precision is characterized by a covariance matrix, while each parameter contributes a symmetric-logarithmic-derivative element to the Fisher-information matrix.The matrix formulation treats the parameters jointly rather than estimating only a single local quantity.
  • Precision bound: For an arbitrary weight matrix G, the Review derives the multiparameter Cramér–Rao bound.Choosing G = w^T w yields the ultimate limit for estimating the weighted global parameter.
  • Practical evaluation: Evaluating the Fisher-information matrix is generally challenging but tractable for Gaussian states.Alternative bounds based on the right logarithmic derivative are also noted.

5. Applications of Distributed Quantum Sensing

DQS has been demonstrated for optical phase and RF sensing and extended to supervised learning with entangled sensor networks. These applications use multipartite entanglement to improve global-parameter estimation or classification error.

  • Experimental demonstrations: Two CV-based experiments demonstrate DQS for optical phase sensing and radio-frequency sensing.The optical experiment uses multipartite entanglement, while the RF experiment uses an optical transducer to encode the field.
  • Optical phase sensing: In optical phase sensing, averaging measurements across sensors increases signal-to-noise ratio through entanglement-enabled cancellation of measurement noise.Figure 8 compares the DQS and DCS measurement sensitivities against the SQL.
  • RF sensing: The RF experiment evaluates average amplitude and central- and edge-node phase differences, with DQS outperforming DCS for edge-node phase-difference estimation.The comparison uses DCS with the entangled sideband state turned off as the SQL benchmark.
  • RF sensing: The entangled sensor network is reconfigurable: tuning the quantum circuit tailors multipartite entanglement to minimize uncertainty for a specified RF-sensing problem.Theory and experiments compare different settings for three RF-sensing tasks.
  • Machine-learning applications: SLAEN combines a classically trained variational quantum circuit with an entangled sensor network for supervised classification.Its training data configure both the classical algorithms and variational quantum circuits.
  • Machine-learning applications: SLAEN achieves lower error probability than classical supervised learning across simulations with different source brightness and system efficiencies.Figure 12 plots the training-step error probabilities for both approaches.

6. Outlook

The Review identifies unresolved questions about optimal DQS protocols, non-Gaussian resources, and discrete-variable implementations, while pointing to scalable hardware and broader applications as future directions.

  • Open theory: A general optimal DQS scheme remains unknown, including the utility of non-Gaussian probe states and measurements under loss.The Review specifically notes GKP states as a resource that can compensate for loss but requires further study.
  • Open theory: Discrete-variable DQS protocols are less explored than continuous-variable protocols, especially under practical nonidealities and scalable entangled-state generation.The Review highlights magnetic and stress sensing as relevant application areas.
  • Applications: DQS applications may extend to optical gyroscopes, optomechanical transducer arrays, and atomic force microscopes.These proposals target navigation, inertial sensing, and readout signal-to-noise ratio.
  • Scalability: Current demonstrations use table-top quantum-optics platforms, while on-chip squeezed-light, entangled-photon, and transducer technologies offer a route toward scalable deployment.The Review describes these developments as promising more scalable and cost-effective DQS implementations.
Loading 2010.14744v2…