Quantum Illumination Radar

• Quantum illumination radar is a target-detection method using entangled signal–idler pairs that exploit nonclassical correlations to reduce detection errors in high-noise, low-brightness regimes.
• It employs Gaussian-state models with two-mode squeezed vacuum states to outperform classical coherent-state transmitters under severe thermal and loss conditions.
• Advanced microwave implementations, including electro-optomechanical converters and JPAs, showcase measurable improvements despite challenges in idler storage, transduction efficiency, and joint measurement.

Searching arXiv for recent and foundational papers on quantum illumination radar, microwave QI, receivers, and practical implementations. Quantum illumination radar is a target-detection architecture in which an entangled source prepares signal–idler pairs, the signal interrogates a weakly reflecting region immersed in bright thermal noise, and the idler is retained for a joint or correlation-based measurement with the return. In the canonical formulation, the propagation channel is so lossy and noisy that return–idler entanglement is destroyed, yet the transmitter’s initial entanglement imprints correlations that can lower target-detection error probability relative to any classical transmitter of equal energy, especially in the low-brightness, high-background regime (Shapiro, 2019, Barzanjeh et al., 2015).

1. Historical development and conceptual scope

Quantum illumination emerged from quantum channel-discrimination theory and was first formulated as a target-presence test rather than as a ranging or imaging protocol. Early discrete-variable proposals emphasized entangled single-photon illumination, but the decisive theoretical framework became the continuous-variable Gaussian-state model, in which two-mode squeezed vacuum states are transmitted in the regime $N_S \ll 1$, $N_B \gg 1$, and $\kappa \ll 1$. Within that formulation, the relevant comparison is not against unentangled single photons but against the optimal classical coherent-state transmitter at the same energy (Shapiro, 2019).

Within the broader quantum-radar literature, quantum illumination is distinct from interferometric quantum radar and from purely metrological localization protocols. Interferometric proposals emphasize phase estimation and super-resolution, whereas quantum illumination is fundamentally a binary hypothesis-testing problem: $H_0$ corresponds to background only, and $H_1$ corresponds to a weak reflection embedded in the same bright background. This distinction is central because the celebrated QI advantage is tied to detection in entanglement-breaking noise, not to the survival of coherent interference through the channel (Mathews, 2022, Karsa et al., 2023).

Microwave frequencies are the natural regime for radar-oriented quantum illumination because the thermal background is intrinsically bright there. Optical QI experiments therefore demonstrated the principle under artificially added noise, whereas microwave QI aims at the environmental conditions encountered by classical radar itself. This is the motivation behind the hybrid microwave–optical architectures of electro-optomechanical converters, superconducting Josephson devices, and related transducers (Barzanjeh et al., 2015, Barzanjeh et al., 2014).

2. Signal–idler model and the origin of the quantum advantage

In the Gaussian formulation, each signal–idler pair is typically a two-mode squeezed vacuum state. One standard representation is

$$|\psi_m\rangle_{SI} = \sum_{n=0}^\infty \sqrt{\frac{N_S^n}{(N_S+1)^{n+1}}}\,|n\rangle_{S_m}|n\rangle_{I_m},$$

with mean photon number $N_S$ per mode. Its crucial nonclassical feature is the phase-sensitive cross-correlation

$$\langle \hat a_{S_m}\hat a_{I_m}\rangle = \sqrt{N_S(N_S+1)},$$

which exceeds the classical bound $N_S$ in the low-brightness regime (Shapiro, 2019).

Target interaction is modeled mode by mode. Under $H_0$, the receiver sees only thermal background,

$N_B \gg 1$0

whereas under $N_B \gg 1$1,

$N_B \gg 1$2

Here $N_B \gg 1$3 is the round-trip transmissivity, and $N_B \gg 1$4 is the mean background photon number per mode. In the regime $N_B \gg 1$5, $N_B \gg 1$6, and $N_B \gg 1$7, the return–idler state is separable, but the residual phase-sensitive return–idler correlation under $N_B \gg 1$8 remains stronger than any classical transmitter of equal energy can induce (Shapiro, 2019, Barzanjeh et al., 2015).

The standard asymptotic comparison follows from the quantum Chernoff analysis. For $N_B \gg 1$9 iid mode pairs, Gaussian-state QI obeys

$\kappa \ll 1$0

while the optimal coherent-state benchmark satisfies

$\kappa \ll 1$1

in the same limit. The factor-of-4 improvement in the error exponent is the canonical $\kappa \ll 1$2 quantum-illumination advantage (Shapiro, 2019).

The essential point is therefore not “entanglement survives noise,” but rather that entanglement at the transmitter produces a return–reference correlation structure that remains operationally useful even after the channel has become entanglement breaking. This counterintuitive feature is the defining theoretical result of quantum illumination (Shapiro, 2019, Karsa et al., 2023).

3. Microwave architectures and receiver realizations

The first explicit microwave-QI radar architecture used an electro-optomechanical converter to generate an entangled microwave signal and optical idler. After linearization, the effective interaction Hamiltonian is

$\kappa \ll 1$3

where the optical–mechanical term acts as a parametric down-conversion interaction and the microwave–mechanical term acts as a beam-splitter interaction. The transmitter emits the microwave mode toward the target, retains the optical idler, and a second converter phase conjugates and upconverts the microwave return for joint optical detection with the idler (Barzanjeh et al., 2015, Barzanjeh et al., 2014).

A major practical deviation from canonical QI is quantum-enhanced noise radar. In that protocol, a Josephson parametric amplifier generates a two-mode squeezed microwave state, but the idler is measured immediately and stored as a classical record; the returned signal is measured separately, and the two records are correlated digitally. The central engineering consequence is that the protocol does not require joint quantum measurement or quantum memory, while still preserving a measurable advantage over a classical source saturating the classical correlation bound (Chang et al., 2018).

A further receiver simplification is the hetero-homodyne cascaded POVM. It first heterodynes the return, then uses that classical record to modulate a homodyne measurement on the stored idler. Because it does not require a quantum interaction between the return and the idler, it is substantially more compatible with microwave hardware than OPA-like or SFG-like joint receivers. In the non-sequential setting, its performance matches the $\kappa \ll 1$4 quantum advantage of phase-conjugate and parametric-amplifier receivers; with sequential detection it yields a $\kappa \ll 1$5 advantage over a conventional coherent-state radar and a $\kappa \ll 1$6 advantage over coherent-state radar with sequential detection (Reichert et al., 2023).

Architecture	Core feature	Stated consequence
Electro-optomechanical microwave QI	Microwave signal and optical idler generated by an EOM converter; return is phase conjugated and upconverted	Error probability superior to any classical microwave radar of equal transmitted energy (Barzanjeh et al., 2015)
Quantum-enhanced noise radar	JPA-generated two-mode squeezed microwave source with immediate idler measurement and digital correlation	Eliminates quantum memory and outperforms a classical bound-saturating noise source by as much as an order of magnitude (Chang et al., 2018)
JTWPA-based microwave quantum radar	Traveling-wave parametric amplifier used as broadband entangled source	Ultrawide $\kappa \ll 1$7 bandwidth at X-band addresses JPA bandwidth limits (Livreri et al., 2021)
Cavity-magnonics microwave QI	YIG magnon mode mediates microwave–optical entanglement	Orders-of-magnitude lower detecting error probability than the electro-optomechanical prototype and any classical microwave radar of equal transmitted energy (Cai et al., 2020)

These architectures reveal a recurring tradeoff. The closer one stays to the Tan–Shapiro optimality proofs, the stronger the asymptotic advantage, but the heavier the requirements on idler storage, phase coherence, transduction efficiency, and genuinely joint measurements. The more one adapts the scheme to existing microwave electronics, the more one sacrifices theoretical optimality in exchange for implementability (Barzanjeh et al., 2015, Chang et al., 2018).

4. Performance bounds, SNR gains, and proof-of-principle experiments

The ideal Gaussian-state QI advantage is a $\kappa \ll 1$8 improvement in error exponent over the optimal coherent-state benchmark, but practical receivers typically realize $\kappa \ll 1$9. OPA and phase-conjugate receivers attain the latter limit in the canonical regime, and the hetero-homodyne receiver reproduces that $H_0$0 gain without direct return–idler quantum interaction (Shapiro, 2019, Reichert et al., 2023).

In the microwave domain, the clearest proof-of-principle departure from purely classical noise radar came from the quantum-enhanced noise-radar experiment of Sandbo Chang et al. The comparison was against an “ideal” classical source engineered to saturate the classical correlation bound for two sidebands. The total system noise temperature was calibrated as $H_0$1 K, corresponding to roughly $H_0$2 photons/Hz of added noise, while the signal power was in the range $H_0$3–$H_0$4 photons/Hz. Even in that regime, the detected covariance from the quantum source was systematically higher than the classical reference, and the quantum enhancement factor exceeded a factor of ten at low detected source power, i.e. an order-of-magnitude improvement in the covariance-based detection statistic (Chang et al., 2018).

The same work also made explicit what practical QI can mean at the receiver. The idler beam was measured immediately, converted into a classical record, and digitally correlated with the later return over arbitrary time delays. That architecture relinquishes the Helstrom-optimal joint measurement but removes the need for quantum memory and exact prior knowledge of the target range, thereby aligning QI more closely with classical noise-radar workflows (Chang et al., 2018).

The gap between theoretical and experimental gains remains large. Optical and microwave demonstrations have established modest but nonzero advantages over relevant classical baselines, whereas the full $H_0$5 bound still depends on collective or strongly non-Gaussian receivers that are not yet practical. The field’s central empirical result is therefore not that operational radar systems already realize the full asymptotic theory, but that quantum-correlated transmitters can outperform carefully engineered classical competitors under realistic noise and loss (Karsa et al., 2023, Chang et al., 2018).

5. Variants, extensions, and adjacent quantum-radar modalities

Quantum illumination has been extended beyond binary target-presence testing. In quantum Doppler radar, the parameter of interest is the Doppler scale $H_0$6, equivalently radial velocity, rather than a binary reflectivity hypothesis. In the high-thermal-noise, low-transmissivity, few-signal-photon regime, a multimode SPDC-based quantum protocol yields a quantum Fisher information up to a factor of about $H_0$7 larger than the coherent-state benchmark, i.e. a $H_0$8 advantage in the asymptotic precision of Doppler estimation (Wei et al., 2024).

Other proposals grouped under “quantum radar” are conceptually adjacent but not identical to standard QI. A localization protocol based on $H_0$9 entangled photons uses all spatial degrees of freedom to obtain an uncertainty in localization that is $H_1$0 times smaller for each spatial direction than what could be achieved by $H_1$1 independent photons. That protocol is a quantum-metrology scheme for three-dimensional target localization, not an idler-assisted QI protocol, and its advantage is correspondingly more fragile to photon loss (Maccone et al., 2019).

Discrete-variable QI has also re-emerged in high-dimensional Bell-state formulations. Finite-dimensional analysis shows a threshold dimensionality below which there is no quantum advantage and above which the Nair–Gu lower bound is approached asymptotically. However, when both systems operate with error-probability exponents $H_1$2 lower than the Nair–Gu bound, the required entangled-state dimensionality is much larger than for Gaussian-state QI, and neither system has appreciable quantum advantage in low-brightness noise (Shapiro, 2024).

These extensions clarify that “quantum radar” is not a single protocol family. Standard Gaussian-state QI remains the most developed framework for noisy target detection; Doppler QI extends the same logic to velocity estimation; high-dimensional Bell-state QI revisits discrete-variable optimality at extreme dimensionality; and entanglement-based localization protocols address a different, metrological target-estimation problem (Wei et al., 2024, Shapiro, 2024, Maccone et al., 2019).

6. Limitations, controversies, and outlook

The practical constraints are severe and recurrent. Idler storage loss directly reduces the QI exponent; bandwidth and time-bandwidth product constrain how rapidly the necessary mode number $H_1$3 can be accumulated; and realistic radar operation requires interrogation of multiple range–angle–Doppler bins rather than a single known resolution cell. These limitations led Shapiro’s retrospective assessment to conclude that the practical utility of QI for radar is “severely limited,” even though the underlying information-theoretic lesson remains important (Shapiro, 2019).

Benchmarking is also contentious. Many microwave demonstrations have compared quantum transmitters against classical references that were technologically realistic but not always the ideal coherent-state benchmark. A careful distinction is therefore needed between relative advantage over an implemented classical system and absolute advantage over the optimal classical transmitter allowed by theory. Recent reviews emphasize that the current state of the field is best described as proof-of-principle rather than deployable radar (Karsa et al., 2023).

Quantum illumination radar also does not automatically imply stealth. A target-side “quantum radar warning receiver” analysis found that detectability depends strongly on the realization of the protocol: a single-photon implementation can be easier for the target to detect than for the radar to exploit, whereas a Gaussian-state implementation with a homodyne-type receiver can remain comparatively stealthy in the same simplified model (Krelina, 2019). A common misconception is therefore that any use of entanglement produces an inherent low-probability-of-intercept benefit; the literature does not support that as a general statement.

Current research directions target specific bottlenecks rather than a wholesale escape from radar constraints. Broadband Josephson traveling-wave parametric amplifiers address the narrow bandwidth of JPAs (Livreri et al., 2021). Cavity-magnonics sources increase quantum correlations per emitted microwave photon and thereby reduce equal-energy error probabilities (Cai et al., 2020). At a more theoretical level, non-Markovian reservoir engineering has been proposed to restore noisy-QI resolution asymptotically toward its ideal form when a bound state exists in the system–bath spectrum (Wu et al., 23 Oct 2025). The likely trajectory is therefore incremental: more practical receivers, better transducers, broader-band sources, and niche short-range sensing applications before any claim to a general-purpose operational quantum radar becomes credible (Karsa et al., 2023).