PaperlayerView exact revision on arXiv

Source-linked AI summary

Demonstration of a Reconfigurable Entangled Radiofrequency-Photonic Sensor Network

Yi Xia, Wei Li, William Clark, Darlene Hart, Quntao Zhuang, Zheshen Zhang

arXiv:1910.08825v3quant-phphysics.optics

TL;DR

Distributed sensing applications need multiple sensors to work jointly, while most entanglement-enhanced sensing demonstrations use a single sensor. This work experimentally demonstrates a reconfigurable, three-sensor RF-photonic network using continuous-variable multipartite entanglement, with estimation variances below the SQL across distributed sensing tasks.

  • Problem

    Most entanglement-enhanced sensing demonstrations operate at a single sensor, although many applications require arrays of sensors to work collectively.

    Source: p. 1

  • Method

    The experiment uses three RF-photonic sensors with electro-optic transducers and tailors shared continuous-variable multipartite entanglement to distributed sensing tasks.

    Source: pp. 2–3

  • Results

    The entangled network’s normalized estimation variance beats the SQL by 3.2 ± 0.1 dB across the three distributed sensing tasks, while classical data remain at the SQL.

    Source: p. 4

  • Takeaways & Limitations

    The experiment establishes a connection between entanglement structure and quantum advantage in distributed sensing and points to potential positioning, navigation, timing, astronomy, and imaging applications.

    Source: p. 5

    The operational range can be further enlarged using noiseless linear amplifiers or continuous-variable error correction.

    Source: p. 5

Abstract

from arXiv · show

Quantum metrology takes advantage of nonclassical resources such as entanglement to achieve a sensitivity level below the standard quantum limit. To date, almost all quantum-metrology demonstrations are restricted to improving the measurement performance at a single sensor, but a plethora of applications require multiple sensors that work jointly to tackle distributed sensing problems. Here, we propose and experimentally demonstrate a reconfigurable sensor network empowered by continuous-variable (CV) multipartite entanglement. Our experiment establishes a connection between the entanglement structure and the achievable quantum advantage in different distributed sensing problems. The demonstrated entangled sensor network is composed of three sensor nodes each equipped with an electro-optic transducer for the detection of radiofrequency (RF) signals. By properly tailoring the CV multipartite entangled states, the entangled sensor network can be reconfigured to maximize the quantum advantage in distributed RF sensing problems such as measuring the angle of arrival of an RF field. The rich physics of CV multipartite entanglement unveiled by our work would open a new avenue for distributed quantum sensing and would lead to applications in ultrasensitive positioning, navigation, and timing.

A. Single RF-photonic sensor enhanced by squeezed light

The theoretical framework maps RF fields onto optical quadrature displacements and uses homodyne measurements to estimate field parameters. A squeezed effective mode is distributed among sensors, with its entanglement configuration matched to the target distributed measurement.

A. Single RF-photonic sensor enhanced by squeezed light

  • RF modulation transfers field information to optical sidebands and, in the weak-field regime, to a displacement on the effective mode’s phase quadrature.A phase-squeezed effective mode is required for quantum-enhanced sensitivity.Source: A. Single RF-photonic sensor enhanced… · pp. 7–8

I. THEORETICAL FRAMEWORK

B. Entangled RF-photonic sensor network

  • A two-mode squeezed-vacuum source produces a squeezed effective mode that a beam-splitter array distributes across sensors with weights chosen for the target sensing task.The minimum-variance configuration requires the entanglement-distribution weights to match the postprocessing weights; loss is modeled through transmissivity η.Source: B. Entangled RF-photonic sensor network · pp. 8–9

C. Performance analysis

The performance analysis compares entangled and separable sensor networks under matched sensing resources. It reports an estimation-variance advantage for the entangled network while noting the resource overhead of the optimum separable benchmark.

C. Performance analysis

  • The comparison constrains central-mode optical energy, while matching the squeezing photon number between entangled and separable schemes.The analysis notes that fixing total photon number is appropriate when probe power is constrained, but RF-photonic sensing ideally uses as much central-mode optical power as the device accommodates.Source: C. Performance analysis · pp. 9–10
  • A ∼10% estimation-variance advantage over the optimum separable sensor network is reported for the entangled network under matched total squeezing photon number.The optimum separable scheme uses ∼7.9 dB of squeezing and achieves 2.7 dB noise reduction, compared with the measured ∼3.2 dB squeezing in the three-sensor network.Source: C. Performance analysis · p. 10
  • The optimum separable benchmark requires a separate squeezed-light source at each sensor, creating substantial resource overhead.Source: C. Performance analysis · p. 10

D. Finite difference method for the estimation of AoA of the RF field

Angle-of-arrival estimation is reduced to estimating phase differences between RF sensors, with wavelength and spacing linking phase to angle. Finite-difference weights determine the estimated phase gradient and can be chosen to reduce both variance and discretization error.

D. Finite difference method for the estimation of AoA of the RF field

  • Known wavelength and sensor spacing convert the measured inter-sensor phase difference into an angle-of-arrival estimate, with integer ambiguity removable when spacing is much less than a wavelength.The method is not generally restricted to that small-spacing assumption.Source: D. Finite difference method for the… · pp. 10–11
  • For a central-node phase gradient, weights (−1/2, 0, 1/2) minimize both estimation variance and discretization error.This choice gives a discretization error of O(∆x^3).Source: D. Finite difference method for the… · p. 11
  • At an edge node, weights (−3/2, 2, −1/2) yield a phase-gradient estimate with discretization error O(∆x^3).The weights are one solution to the constraints on the estimator's expectation value.Source: D. Finite difference method for the… · p. 11
  • For a second-order derivative, central-node weights (1/2, −1, 1/2) give error O(∆x^4), while edge-node weights (1, −2, 1) give O(∆x^3).The source states there is only one possible parameter set for each case.Source: D. Finite difference method for the… · p. 11

II. EXPERIMENTAL DETAILS

The experimental-details section covers apparatus and phase locking, noise and squeezing calibration, and individual-sensor performance across input powers.

II. EXPERIMENTAL DETAILS

  • The section introduces the experimental components and phase-locking mechanisms.The passage labels this as Sec. II A.Source: §II · p. 11
  • It reports calibration of shot-noise, the electronic noise floor, and squeezing across RF frequencies.These calibrations are also assigned to Sec. II A in the passage.Source: §II · p. 11
  • It presents individual-sensor performance at different input powers and reports agreement between experimental data and a theoretical model.This material is identified as Sec. II C.Source: §II · p. 11

A. Detailed description of experimental setup

A phase-squeezed optical source is distributed across three RF-photonic sensors, where electro-optic modulation encodes RF signals and balanced homodyne detection supplies data for estimation.

A. Detailed description of experimental setup

  • A semiconductor laser, filtered by a mode-cleaning cavity, supplies light for second-harmonic generation and the homodyne local oscillator.The source describes a 1550-nm laser and a 775-nm pump used for the optical parametric amplifier.Source: A. Detailed description of… · p. 11, Figure S3 · p. 12
  • The optical parametric amplifier generates phase-squeezed light that two variable beam splitters distribute among three RF-photonic sensors.The variable-beam-splitter ratios determine the continuous-variable multipartite entangled state.Source: A. Detailed description of… · p. 12, Figure S3 · p. 12
  • At each sensor, an electro-optic modulator transfers a portion of the coherent central mode into quadrature displacement in the squeezed 11-MHz sidebands.The displacement depends on the amplitude and phase of the probed RF field.Source: A. Detailed description of… · p. 12, Figure S3 · p. 12
  • Balanced homodyne detection addresses the squeezed phase quadrature, and demodulated sensor signals are recorded for postprocessed estimation and noise-variance analysis.The setup reports classical interference visibility above 97% at each sensor and photodiode quantum efficiency of about 88%.Source: A. Detailed description of… · p. 12, Figure S3 · p. 12

B. Calibration of shot-noise and squeezing levels

Calibration establishes a shot-noise reference separated from the detector's electronic noise floor and measures frequency-dependent squeezing, including at the RF carrier.

B. Calibration of shot-noise and squeezing levels

  • 13 dB separates the measured shot-noise level from the electronic noise floor, allowing sub-shot-noise behavior to be observed.The shot-noise spectrum was flat beyond 1.9 MHz, and the measured level matched a model accounting for local-oscillator power, detector efficiency, amplifier gains, and analyzer bandwidth.Source: B. Calibration of shot-noise and… · pp. 12–13, Figure S4 · p. 13
  • Approximately 4 dB of squeezing was observed at the 11-MHz RF carrier, with sub-shot-noise behavior above approximately 5.5 MHz.No squeezing was observed at lower frequencies, likely due to residual amplified spontaneous-emission and Brillouin-scattering noise; narrower-linewidth mode-cleaning cavities may permit lower-frequency observation.Source: B. Calibration of shot-noise and… · p. 13, Figure S4 · p. 13

C. Performance of individual sensors

The study varies VBS splitting ratios to redistribute squeezed light among the three sensors and measures each sensor’s noise power. The measured noise powers agree well with the theoretical model.

C. Performance of individual sensors

  • Measured noise powers across three VBS-tuning procedures agree well with the theoretical predictions.The procedures vary VBS 1 or VBS 2 while holding the other splitting ratio fixed.Source: C. Performance of individual sensors · p. 13
  • With VBS 2 fixed at 50:50, varying VBS 1 reallocates squeezed light from Sensor 1 while Sensors 2 and 3 share the remainder equally.Sensor 1’s share ranges from 0% to 100%.Source: C. Performance of individual sensors · p. 13
  • With VBS 1 fixed at 50:50, varying VBS 2 keeps Sensor 1 at 50% while Sensors 2 and 3 divide the rest.A separate procedure fixes VBS 2 at 0:100 and varies VBS 1 to split light between Sensors 1 and 3, leaving Sensor 2 unused.Source: C. Performance of individual sensors · p. 13

1. Phase-difference estimation at a central node

The experiment estimates the RF phase difference at a central node using a tailored three-sensor entangled state. The entangled network reduces estimation uncertainty relative to the classical separable scheme.

1. Phase-difference estimation at a central node

  • Weights [1/2, 0, −1/2] are implemented with VBS settings 50:50 and 0:100, with a π-phase delay at Sensor 3.The applied phases at Sensors 1 and 2 are swept in opposite directions while RF-field amplitudes are identical.Source: §1 · p. 13
  • Homodyne measurements from the three sensors are weighted to form an unbiased estimator of the central-node phase difference.The phase difference is emulated by an RF-field phase difference across the sensors.Source: §1 · p. 13
  • 3.5 dB reduction in estimation uncertainty is observed for the entangled network versus the classical separable scheme.The experiment estimates phase difference at the central node served by Sensor 2.Source: §1 · p. 14

2. Optimization of multipartite entangled state

The experiment tunes the VBS settings to optimize multipartite entanglement for estimating average RF-field amplitude and central-node phase difference. The phase-difference curves are asymmetric, consistent with correlations between sensor measurement noise.

2. Optimization of multipartite entangled state

  • The largest entangled-network advantage for average RF-field amplitude occurs at VBS 1 transmissivity 0.33, when all sensors receive equal optical power.VBS 2 is fixed at 50:50 during this optimization.Source: §2 · p. 14
  • For central-node phase-difference estimation, VBS 2 is fixed at 0:100 while VBS 1 is tuned across its range.A negative transmissivity represents a sign flip on homodyne data during postprocessing.Source: §2 · p. 14
  • The asymmetric entangled-network behavior in the central-node phase-difference comparison is a signature of quantum correlations between measurement noise at different sensors.The experimental data agree with the theoretical model.Source: §2 · p. 14

E. Theoretical model for the experiment

The model expresses distributed RF sensing as estimation of weighted phase or amplitude combinations, then derives estimation variances for specified sensor weights and circuit settings. It compares these variances with the classical SQL and models the reported experiments.

E. Theoretical model for the experiment

  • For average RF-field amplitude, the model gives a minimum estimation variance at VBS 1 splitting ratio r = 1/3, with theory and experiment in close agreement.The model uses equal sensor weights and keeps VBS 2 at 50:50.Source: Figure S7 · p. 15, E. Theoretical model for the experiment · p. 16
  • For central-node phase-difference estimation, the model’s variance is minimized at r = 1/2 for r > 0 and r = −1/2 for r < 0.For r < 0, the model states that entangled and classical separable estimation variances are identical.Source: E. Theoretical model for the experiment · p. 16
  • The model defines the SQL as the minimum estimation variance of a classical separable sensor network and normalizes reported variances to that unit.The model describes distributed estimation as a weighted combination of sensor phase or amplitude parameters.Source: E. Theoretical model for the experiment · pp. 14, 16
  • For edge-node phase-difference estimation, the minimum estimation variance is achieved at r = −1/4.The model also reports the positive-r variance condition and its plotted experimental comparison.Source: E. Theoretical model for the experiment · p. 16

III. ENTANGLEMENT DISTRIBUTION

The authors demonstrated a tabletop entangled sensor network and simulated distributing its entanglement over low-loss optical fibers. The simulation shows a distance-dependent but persistent estimation-variance reduction.

III. ENTANGLEMENT DISTRIBUTION

  • The simulated entanglement-enhanced sensitivity degrades with distance but never diminishes.The fiber-distributed network is simulated using low-loss optical fibers with 0.14 dB/km loss.Source: §III · p. 16, Figure S8 · p. 16
  • The experimental network was demonstrated on a free-space tabletop platform.Source: §III · p. 16
  • 2-dB (1-dB) estimation variance reduction is simulated for sensors 10 km (30 km) from the entanglement source.The simulation uses experimental parameters for squeezing, detector efficiency, and additional loss.Source: Figure S8 · p. 16, §III · p. 16
Loading 1910.08825v3…