Source-linked AI summary
Advances in Photonic Quantum Sensing
Stefano Pirandola, Bhaskar Roy Bardhan, Tobias Gehring, Christian Weedbrook, Seth Lloyd
TL;DR
Quantum photonic sensing faces unresolved limits in optimal estimation, discrimination, and practical measurement design. This paper develops a framework for quantum parameter estimation and hypothesis testing, then reviews photonic sensing applications and their reported advantages. It highlights quantum reading and sub-wavelength optical resolution while identifying open theoretical and experimental challenges.
Problem
Optimal estimation and discrimination of bosonic loss remain open, despite their importance for quantum optical communication and sensing.
Method
The paper formulates general quantum estimation and hypothesis-testing tools, classifies channel protocols, and reviews theoretical and experimental photonic-sensing applications.
Results
Quantum resources can enable quantum-reading advantages at low energies and constant-accuracy resolution of sub-wavelength-separated point sources.
Takeaways & Limitations
Entanglement-based sensing may support faster optical readers, denser memories, and resolution beyond the Rayleigh limit.
Takeaways & Limitations
Optimal measurements and input states are not generally known, and adaptive strategies remain open for bosonic-loss estimation and discrimination.
Abstract
from arXiv · showhide
Quantum sensing has become a mature and broad field. It is generally related with the idea of using quantum resources to boost the performance of a number of practical tasks, including the radar-like detection of faint objects, the readout of information from optical memories or fragile physical systems, and the optical resolution of extremely close point-like sources. Here we first focus on the basic tools behind quantum sensing, discussing the most recent and general formulations for the problems of quantum parameter estimation and hypothesis testing. With this basic background in our hands, we then review emerging applications of quantum sensing in the photonic regime both from a theoretical and experimental point of view. Besides the state-of-the-art, we also discuss open problems and potential next steps.
Estimation and discrimination protocols
Quantum sensing is framed through quantum channel estimation and discrimination: estimating a continuous parameter or distinguishing discrete channel-encoded values. Protocols range from memoryless block strategies to general quantum-comb schemes incorporating entanglement, adaptive operations, and sequential channel uses.
- Problem formulations: Quantum channel estimation estimates a continuous parameter encoded in a channel, whereas quantum channel discrimination distinguishes discrete parameter values with prior probabilities.The basic discrimination case is binary symmetric discrimination, equivalent to retrieving a classical bit encoded in the parameter.
- Block protocols: In a block-unassisted protocol, Alice sends n copies of an input state through a memoryless channel and Bob jointly measures the n-copy output.Estimation uses a continuous-outcome measurement to construct an unbiased estimator with error variance δθ2, while discrimination uses a dichotomic measurement with mean error probability perr.
- Quantum-comb protocols: The most general protocol uses unlimited entanglement and adaptive quantum operations, represented as a quantum comb whose slots contain the unknown channel.An optimal quantum measurement produces an outcome that is classically processed.
- Protocol classification: Quantum combs encompass block-unassisted, block-assisted, and sequential protocols.Block-assisted schemes retain an idler reference during output measurement, while sequential schemes transmit an input through n successive channel instances.
Performance of channel estimation
Channel-estimation performance is bounded by the quantum Cramér–Rao bound, with SQL and Heisenberg scaling determined by resources and protocol structure. Teleportation covariance reduces many adaptive problems to block-assisted estimation, while bosonic-loss estimation remains unresolved.
- Quantum Cramér–Rao bound: The quantum Cramér–Rao bound sets the ultimate performance for optimized quantum-comb channel estimation.The bound is expressed through the quantum Fisher information and Bures fidelity.
- Scaling limits: SQL gives δθ2 ≳ n^-1, whereas the Heisenberg limit gives δθ2 ≳ n^-2; with photon number N, the corresponding scalings are δθ2 ≳ N^-1 and δθ2 ≳ N^-2.The Heisenberg limit is achievable for unitary phase estimation using sequential protocols or N00N states in a single block-assisted use.
- Teleportation covariance: Teleportation covariance implies SQL-limited channel estimation and allows adaptive protocols to be reduced to block-assisted protocols with n maximally entangled probes.This reduction follows by teleportation simulation, stretching the comb, and monotonicity and multiplicativity of quantum Fisher information.
- Teleportation-covariant channels: δp2 ≥ p(1−p)n^-1 for depolarizing, dephasing, and erasure channels, while thermal-noise estimation satisfies δn̄2 ≥ n̄(n̄ + 1)n^-1.These SQL limits apply to probability estimation and excess-noise estimation, respectively.
- Bosonic loss: The optimal estimation limit for bosonic transmissivity η is unknown, making bosonic-loss estimation an open problem of central importance for quantum optical communications.For pure loss, the best-known performance is δη2 ≥ γN^-1, while coherent states achieve δη2 ≥ ηN^-1 and squeezing can improve the prefactor.
Performance of channel discrimination
Binary channel discrimination retrieves a classical bit by distinguishing two equally likely channels, with optimal minimum-error performance determined by the Helstrom bound. For jointly teleportation-covariant channels, teleportation simulation reduces adaptive discrimination to Choi-state discrimination, but this reduction fails for bosonic loss with fixed noise.
- Binary channel discrimination: Binary discrimination distinguishes two equally likely channels by retrieving the corresponding classical bit, with minimum error optimized by the Helstrom POVM.The optimal error depends on the trace distance between the two possible output states.
- Teleportation-covariant channels: Joint teleportation covariance enables channel simulation from Choi matrices and converts the adaptive comb output into a global channel acting on multiple Choi states.This follows from teleportation simulation and channel stretching.
- Teleportation-covariant channels: For jointly teleportation-covariant channels, maximally entangled inputs achieve the bound, and finite-dimensional diamond distance equals the trace norm distance between Choi matrices.Thus adaptive discrimination can be characterized directly through the corresponding Choi states.
- Error bounds: Fidelity-based bounds and the quantum Chernoff bound provide lower and upper bounds on discrimination error, including adaptive protocols for jointly teleportation-covariant channels.For multimode Gaussian states, both fidelity and quantum Chernoff quantities have closed-form formulas, yielding asymptotic functionals of bosonic Choi matrices.
- Scope and limitation: The reduction applies to Pauli channels, erasure channels, and thermal-noise parameters in bosonic Gaussian channels with fixed transmissivity, but not to bosonic loss with fixed noise.Loss channels with different transmissivities are not jointly teleportation-covariant.
Quantum reading of classical data
Quantum reading models classical optical-memory readout as quantum channel discrimination between cells with different reflectivities, comparing classical and EPR transmitters under matched energy constraints. Entanglement can provide substantial advantages at low photon numbers, including near-complete single-cell readout and operation at very low total energies, while related work extends the framework to capacities, alternative states, and experimentally realized error-free schemes.
- Quantum-channel model: Classical optical-memory readout is modeled as discriminating cells with reflectivities η0 and η1, using signal modes and reference modes to probe and detect the stored bit.The cell stores u = 0, 1 in two equiprobable reflectivities, with η1 > η0.
- Transmitter comparison: The comparison evaluates coherent-state, classical, and EPR transmitters with an optimal Helstrom measurement, defining quantum advantage as ∆:= JEPR −Jclass under fixed probing energy.Positive ∆ indicates that the EPR transmitter retrieves more information than the optimal classical transmitter.
- Quantum advantage: ∆→1 bit per cell can occur for η1 →1, meaning the EPR transmitter can fully read an ideal cell in suitable parameter regimes.Positive advantage is typical at low signal photon numbers and high reflectivities, and it can already occur with a single probe per cell, n = 1.
- Global energy constraint: At total energies NT ≲10 photons, even a monochromatic EPR transmitter with nEPR = 1 can outperform arbitrary classical transmitters across extremely large bandwidths.This low-energy regime could enable faster optical readers and denser memories.
- Extensions and related work: Subsequent studies introduced quantum reading capacity and error exponents, identified optimal or advantageous Fock, entangled coherent, and non-Gaussian states, and developed alternative unitary models.These extensions include multi-cell error-correction coding, superadditive capacity, and settings where both beamsplitter inputs and outputs are accessible.
- Experimental realization: An experimentally implemented Mach–Zehnder scheme achieved perfect discrimination between a beamsplitter with η1 = 1 and one with η0 < 1 using phase shifts and photon-counting outcomes.The perfect-reflectivity case directs the photon to one detector, whereas the lower-reflectivity case directs it to either of two others.
Quantum illumination of targets
Quantum illumination uses entanglement to improve standoff detection of low-reflectivity targets in bright thermal noise. The approach has yielded theoretical performance advantages, explicit receiver proposals, microwave extensions, and experimental demonstrations.
- Quantum illumination of targets: Entangled quantum illumination enhances detection of low-reflectivity targets compared with non-entangled transmitters.The original qubit-based protocol established entanglement as a resource for standoff target detection.
- Quantum illumination of targets: Gaussian quantum illumination discriminates target absence from low-reflectivity detection in bright thermal noise using continuous-variable bosonic modes.The target reflectivity satisfies η ≃0, while the environment has ¯n ≫1 mean thermal photons.
- Quantum illumination of targets: The EPR transmitter achieves a 6dB advantage in the error-probability exponent over the coherent-state transmitter.The comparison assumes low-energy signals and many transmitted modes under a local energy constraint.
- Quantum illumination of targets: An explicit quantum receiver based on feed-forward sum-frequency generation achieves the EPR performance and demonstrates an advantage in detection probability versus false-alarm probability.This provides a receiver-level realization of the predicted quantum illumination advantage.
- Quantum illumination of targets: Quantum-metrology analysis shows a 3dB-enhancement of the signal-to-noise ratio over local measurements, while experiments demonstrate illumination advantages with entangled Gaussian states.Experimental platforms include parametric down-conversion sources, photon-counting CCD detection, and stored idlers for joint measurements.
Optical resolution beyond the Rayleigh limit
Quantum sensing addresses the Rayleigh-limit loss of optical resolution, where classical imaging and photon shot noise severely constrain the separation of point-like sources. The section develops quantum-resolution theory for diverse optical sources and measurements, and reviews proof-of-principle demonstrations beyond the classical limit.
- Optical resolution beyond the Rayleigh limit: The classical Rayleigh length is approximately λ/a, and photon shot noise can severely limit resolving closer point-like sources.Here λ is the emitted-light wavelength and a is the observing lens’s numerical aperture.
- Optical resolution beyond the Rayleigh limit: Optical resolution can be formulated as estimating the loss parameters of two lossy bosonic channels for sources ranging from attenuated classical light to bright or entangled states.The ultimate resolution depends on the sources’ optical properties and separation, and can be enhanced by entangled or quantum-correlated emission.
- Optical resolution beyond the Rayleigh limit: Subsequent theory established information-optimal image-plane sinc-Bessel modes for hard apertures, generalized results to arbitrary point-spread functions, and studied alternative optimal measurements.These measurements include homodyne, heterodyne, and quantum-optimal detection strategies.
- Optical resolution beyond the Rayleigh limit: Proof-of-principle experiments demonstrated super-resolution using image-inversion interferometry, heterodyne detection, phase-shift projection, and spatial-mode projections.Experiments used simulated incoherent sources, double slits, single-mode fibers, spatial light modulators, and electron-multiplying CCD detection.
Discussion and outlook
Quantum sensing has advanced substantially, but optimal measurements and several experimental implementations remain unresolved. Key challenges span quantum reading, illumination receivers and converters, radar memory and bandwidth requirements, and alignment for super-resolution.
- Open theoretical and experimental challenges: Quantum metrology and hypothesis testing can often determine ultimate quantum-mechanical performance, but optimal measurements are not generally known to be implementable.The remaining gap is between theoretical limits and practical measurement procedures.
- Open theoretical and experimental challenges: Quantum reading lacks a fully demonstrated experiment accessing a single cell output within a classically coded one- or two-dimensional array.A proof-of-principle experiment based on unitary beamsplitter discrimination has been reported, but not the complete readout scenario.
- Quantum illumination and radar: Quantum illumination still needs practical receivers approaching the Helstrom bound, while microwave implementations require highly efficient microwave-optical quantum converters.A possible alternative is fully microwave illumination using superconducting Josephson parametric amplifiers, phase conjugation, idlers, and transmon single-photon detection.
- Quantum illumination and radar: Quantum radar requires low-loss idler storage, microwave pulses with a time-bandwidth product of 10^6 or more, multi-bin interrogation, and solutions for random-amplitude targets and clutter.Unlike classical radar, the quantum counterpart must increase bandwidth at constant signal brightness and may currently query only one polarization, azimuth, elevation, range, or Doppler bin at a time.
- Super-resolution: Super-resolution schemes generally require accurate centroid knowledge, making maximum alignment before separation estimation essential for realistic quantum-optimal performance.Current approaches include spatial-mode demultiplexing, image inversion, and heterodyne-based super-localization.