Quantum Integrated Sensing and Communication

• Quantum Integrated Sensing and Communication (QISAC) is a paradigm that utilizes shared quantum resources, such as entangled states and coherent hardware, to perform simultaneous sensing and communication.
• Recent studies demonstrate its implementation through protocol-level integration, fiber-optic networks, and variational measurement circuits, offering improvements in security and precision.
• The field addresses practical trade-offs between communication throughput and sensing accuracy using techniques like Fisher information analysis and adaptive receiver control.

Quantum integrated sensing and communication (QISAC) is used in the recent literature for a family of architectures in which sensing and communication are co-executed through shared quantum states, shared quantum-optical measurements, or shared quantum-enabled infrastructure. In the narrowest sense, the same quantum carrier, entangled resource, or receiver observation is used simultaneously to convey information and estimate an unknown physical parameter; in a broader systems sense, the same fiber, optical front-end, base station, or network control plane supports both quantum communication and sensing functions (Liu et al., 2024). The resulting landscape is heterogeneous: some works study direct protocol-level integration over entangled qubits or qudits, some study coherent-state optical links with receiver-side sensing–communication tradeoffs, some demonstrate fiber networks that combine QKD with vibration or disturbance sensing, and some provide hybrid or architectural precursors in which quantum sensing is integrated with classical communications or with quantum-network orchestration (Krikidis, 11 Jan 2026).

1. Scope, definitions, and major forms of integration

A consistent pattern across the literature is that “integration” can occur at different layers. At the protocol level, integration means that the same transmitted quantum system supports both message recovery and parameter estimation. At the receiver-design level, the same measurement outcomes are reused for detection and sensing. At the infrastructure level, the same deployed fiber or shared coherent optical hardware supports QKD and environmental monitoring. At the architecture level, communication and sensing are coupled through entanglement distribution, synchronization, storage, and control rather than through a single joint waveform or a single estimation-theoretic objective (Nikoloska et al., 20 Nov 2025).

Form of integration	Representative papers	Shared resource or task
Protocol-level direct QISAC	(Liu et al., 2024, Nikoloska et al., 20 Nov 2025, Krikidis, 11 Jan 2026)	Same entangled state, qudit, or homodyne observation used for communication and sensing
Infrastructure-integrated QKD and sensing	(Xu et al., 2024, Liu et al., 2024, Zhao et al., 22 Dec 2025)	Same fiber network, coherent hardware, pilot structure, or Sagnac loop
Hybrid or enabling architectures	(Alsaui et al., 7 Sep 2025, Popovski et al., 18 Sep 2025, Jebraeilli et al., 25 Feb 2025, Koike-Akino et al., 2022)	Quantum sensing front-end, entanglement-aware networking, distributed sensing support, or quantum inference layer

The literature is also explicit that not every “quantum” contribution to sensing-and-communications is equally central to QISAC. The Wi-Fi AutoQML study treats the quantum element as a downstream inference engine operating on ISAC-derived data, not as a quantum-native sensing or communication mechanism, and is therefore best described as adjacent to QISAC rather than constitutive of it (Koike-Akino et al., 2022). By contrast, the homodyne BPSK optical link, entanglement-based secure sensing-and-communication protocol, and variational qudit design all place the quantum resource inside the joint sensing-and-communication mechanism itself (Krikidis, 11 Jan 2026).

2. Direct protocol-level QISAC with entanglement and variational measurement

One direct formulation uses Bell pairs to perform secure direct communication and remote sensing with the same entangled resource. In the entanglement-based QISAC protocol, Alice prepares Bell pairs initially in $|\psi^-\rangle^{\otimes m}$, keeps one photon from each pair, sends the partners to Bob, uses $U_0=I$ and $U_1=\sigma_x$ to encode one classical bit, and then probes an unknown phase through repeated application of

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$

After $N$ passes, the payload states become

$\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$

and

$\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$

Bob measures

$$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$

so that $\langle \hat{O_1}\rangle=-\cos N\theta$ and $\langle \hat{O_2}\rangle=-\sin N\theta$, enabling simultaneous bit decoding and phase estimation. The paper gives $U_0=I$0 and, for the payload fraction $U_0=I$1, obtains

$U_0=I$2

which it interprets as Heisenberg-type $U_0=I$3 scaling in the number of sensing passes. Its communication security is characterized by

$U_0=I$4

with a reported secrecy threshold of $U_0=I$5 QBER, while sensing security is expressed through an asymmetric Fisher-information comparison with an $U_0=I$6 QBER threshold (Liu et al., 2024).

A second direct formulation uses maximally entangled qudits and treats QISAC as a joint message-decoding and parameter-classification problem. Charlie prepares

$U_0=I$7

Alice encodes a message by

$U_0=I$8

restricting the message alphabet to $U_0=I$9, and Bob estimates both the classical message and an unknown discrete parameter $U_1=\sigma_x$0 after a trainable pre-measurement unitary $U_1=\sigma_x$1. The communication back-off is

$U_1=\sigma_x$2

with throughput defined as

$U_1=\sigma_x$3

The sensing metric is the correct classification probability $U_1=\sigma_x$4, not Fisher information or mean-squared error. In the reported simulations for $U_1=\sigma_x$5 and $U_1=\sigma_x$6 with $U_1=\sigma_x$7 discrete values in $U_1=\sigma_x$8, Bob uses a variational measurement circuit together with two fully connected neural networks, each with two hidden layers of 1024 neurons and ReLU activations, trained with Adam at $U_1=\sigma_x$9 and a quantum update step $$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$0. The main result is a numerically demonstrated throughput–estimation-accuracy trade-off, with improved sensing performance relative to the fixed superdense-coding measurement, especially at intermediate communication back-off (Nikoloska et al., 20 Nov 2025).

These two lines of work use different sensing metrics and different communication abstractions, but they share a core methodological claim: a communication-optimal receiver or encoding rule is not generally QISAC-optimal. In one case this is expressed analytically through Fisher information and secrecy capacity; in the other it is learned variationally as a task-dependent measurement.

3. Receiver-level tradeoffs in quantum-optical links

A more receiver-centric formulation appears in the quantum optical homodyne BPSK setting. Here the transmitted coherent states are

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$1

the channel is a one-mode phase-insensitive Gaussian channel with transmissivity $$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$2, thermal noise $$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$3, and an unknown deterministic phase rotation $$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$4, and the receiver uses a homodyne measurement with local-oscillator phase $$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$5. The scalar observation model is Gaussian with means

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$6

and variance

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$7

The design problem is

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$8

where

$$U^\theta=e^{-iJ_z \theta}=| 0 \rangle \langle 0 | +e^{i\theta } | 1 \rangle \langle 1 |.$$9

and $N$0 is block Fisher information for the phase estimate. In high SNR,

$N$1

so the communication-optimal and sensing-optimal LO phases differ by exactly $N$2:

$N$3

The paper then proposes a two-loop algorithm combining EM-based joint symbol detection and phase estimation with adaptive LO retuning, thereby making the sensing–communication trade-off a receiver-control problem rather than a separate synchronization step. It is explicit that the sensing metric here is classical Fisher information under homodyne detection, not quantum Fisher information (Krikidis, 11 Jan 2026).

A hybrid extension appears in the integration of quantum illumination radar into an otherwise classical full-duplex base station. In this IQSCC formulation, downlink and uplink communication remain classical, while sensing is performed through a TMSV-based QI radar with signal–idler correlation detection. The integrated optimization maximizes

$N$4

subject to a radar sensing constraint

$N$5

and solves the resulting non-convex joint power-and-beamforming design via successive convex approximation. The paper’s quantitative system-level point is that the quantum radar advantage relaxes the required sensing SINR: for $N$6 and $N$7, the required sensing SINR drops from $N$8 dB in conventional ISAC to $N$9 dB in the proposed IQSCC system, and the steady-state sum-rate reaches about $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$0, about $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$1 above the conventional ISAC baseline (Alsaui et al., 7 Sep 2025).

Together these works establish a central QISAC theme: sensing and communication are coupled by a shared receiver degree of freedom—LO phase in one case, radar SINR budget in another—and the quantum contribution can be operationalized as a constrained resource-allocation problem rather than only as a standalone sensing gain.

4. Fiber-integrated quantum communication and distributed sensing networks

A major experimental strand realizes QISAC through shared optical-fiber infrastructure. In the integrated sensing and quantum network scheme, a star network combines Gaussian-modulated coherent-state CV-QKD with a distributed sensing protocol called spectrum phase monitoring. The center node provides the laser and coherent detector, the child nodes modulate return signals on different RF carriers, and the same network supports secure key distribution and vibration sensing. The communication subsystem is a round-trip multi-band CV-QKD network with

$\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$2

where $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$3 is network capacity, while the sensing subsystem monitors the spectrum to identify which node is vibrating and uses phase recovery plus unwrapping to reconstruct the vibration waveform. Experimentally, the network capacity is 8, the per-user secret key rate is approximately $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$4 under $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$5 standard fiber transmission, and the sensing subsystem reaches a vibration response bandwidth from $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$6 to $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$7, a strain resolution of $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$8, and a spatial resolution of $\lvert \psi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert01\rangle-e^{iN\theta }\lvert10\rangle)$9 under shot-noise-limited detection (Xu et al., 2024).

A related downstream quantum access network integrates discrete-modulation CV-QKD and vibration sensing on the same fiber by encoding eight users simultaneously on sidemode quantum states of a single laser source and separating them through a filter network. The same coherent DSP pipeline recovers QKD signals and extracts forward/backward phase perturbations for localization. Over an $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$0 single-mode fiber, the average key rate is $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$1 bits per second, while vibration localization is demonstrated with spatial resolution of $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$2 at $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$3, $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$4 at $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$5, and $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$6 at $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$7. The integration is structurally tight: pilot tones, coherent detection, and DSP are reused, and only one additional heterodyne detector at the server is added for localization (Liu et al., 2024).

A third platform uses a 30 km Sagnac loop to combine ring-based phase-encoded BB84, weak-measurement-enhanced quasi-static sensing, and null-frequencies localization for dynamic disturbances. In QKD mode the loop delivers a raw key generation rate of $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$8 with QBER of $\lvert \phi^{-}_{\theta}\rangle=\frac{1}{\sqrt{2}(\lvert00\rangle-e^{iN\theta }\lvert11\rangle).$9 for 0 phase difference and $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$0 for $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$1 phase difference over 20 minutes without active phase stabilization. When QBER exceeds a threshold, the same loop is reconfigured into a perception system: weak measurement resolves quasi-static disturbances as small as $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$2, corresponding to a time-delay variation of $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$3 through

$$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$4

and null-frequency localization estimates the position of dynamic disturbances with a reported mean localization error of $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$5 (Zhao et al., 22 Dec 2025).

These fiber-network works exemplify a systems interpretation of QISAC. The quantum communication layer is not sacrificed; instead, the same channel becomes an object of sensing, monitoring, and self-diagnosis. They also show that integration can be simultaneous, partially simultaneous, or mode-switched, depending on whether pilot-assisted sensing remains compatible with the communication operating point.

5. Quantum-enabled sensing front ends and architectural precursors

Some influential works are not direct QISAC protocols but expand the architectural design space. One example is the Rydberg atomic receiver paradigm, where a quantum atomic medium rather than a classical electronic front end performs the field transduction for both communications and radar-like sensing. The receiver output after lock-in processing is modeled as

$$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$6

with a noise-inclusive form

$$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$7

The same atomic front end is used as both communication receiver and radar receiver. Proof-of-concept experiments report radar ranging with $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$8 RMSE and about $$\hat{O_1}=\sigma_x\otimes \sigma_x,\qquad \hat{O_2}=\sigma_y\otimes \sigma_x,$$9 resolution over $\langle \hat{O_1}\rangle=-\cos N\theta$0–$\langle \hat{O_1}\rangle=-\cos N\theta$1, and communication using 4-FSK at $\langle \hat{O_1}\rangle=-\cos N\theta$2 with BER below $\langle \hat{O_1}\rangle=-\cos N\theta$3, while maintaining communication under wideband radar interference with ISR from $\langle \hat{O_1}\rangle=-\cos N\theta$4 dB to $\langle \hat{O_1}\rangle=-\cos N\theta$5 dB (Chen et al., 16 Jun 2025).

At the network-architecture level, the 1Q framework proposes first-generation wireless systems integrating classical and quantum communication through quantum base stations, quantum cells, quantum user equipment, and hybrid resource allocation across classical time-frequency resources and quantum entanglement resources. Distributed quantum sensing is treated as an explicit service class, and the paper states a key architectural rule: if a device has quantum coverage, it must also have classical coverage, but not vice versa. It further identifies resource descriptors such as entanglement type, generate-and-store versus generate-and-measure mode, maximum tolerable entanglement-generation latency, minimum acceptable fidelity, and the dual timing constraints $\langle \hat{O_1}\rangle=-\cos N\theta$6 and $\langle \hat{O_1}\rangle=-\cos N\theta$7 (Popovski et al., 18 Sep 2025).

A closely related sensing-centric architecture is STQS, which interlaces sensing, memory, communication, and computation through Quantum Sensing Chips and a Quantum Sensing Processing Unit. Its workflow is explicitly organized as probe preparation, sensing, storage, memory, retrieval, and processing, with spatially distributed GHZ-type sensing states and temporal buffering. The paper models the temporal memory update as

$\langle \hat{O_1}\rangle=-\cos N\theta$8

uses swap-test overlap as an intermediate comparison metric, and validates components on IBM Marrakesh and IonQ Forte devices (Jebraeilli et al., 25 Feb 2025).

These papers broaden QISAC beyond direct waveform co-design. They show that communication can enter as entanglement distribution, synchronization, memory transport, or quantum-enabled front-end transduction. A plausible implication is that future QISAC systems will require both physical-layer co-design and network-level orchestration of fidelity, latency, coherence, and state-comparison workflows.

6. Adjacent layers, misconceptions, and open problems

A recurring misconception is that any application of quantum methods to sensing data inside a communication system is already a full QISAC realization. The Wi-Fi AutoQML study makes clear why that interpretation is too broad. Its contribution is a proof-of-concept application of automated quantum machine learning to Wi-Fi ISAC-derived human pose recognition, with a search space over ansatz type, embedding type, qubit count, layer count, and learning rate. The best AutoAnsatz result is a 3-layer MPS with 10 qubits, angle embedding, and 54 variational parameters; the paper reports greater than 80% accuracy in the limited-data regime, above 90% for DNN, QNN, and SVM when sufficient labeled data are available, and 89% pruning over 2,000 trials. Yet it also states explicitly that the experiments are not performed on real quantum hardware and that the quantum element lies in the downstream inference model rather than the sensing waveform, communication protocol, or joint communication–sensing resource design (Koike-Akino et al., 2022).

Another misconception is that all QISAC works claim a quantum advantage of the same type. The reviewed literature uses several distinct standards. Some works target Heisenberg-type scaling in a sensing-pass resource, as in the Bell-pair protocol with $\langle \hat{O_1}\rangle=-\cos N\theta$9 (Liu et al., 2024). Some emphasize a practical receiver trade-off between BER and Fisher information rather than ultimate bounds (Krikidis, 11 Jan 2026). Some show communication-throughput gains made possible by a relaxed sensing constraint in hybrid quantum-radar systems (Alsaui et al., 7 Sep 2025). Experimental fiber platforms, by contrast, primarily establish coexistence, stability, and multifunctionality rather than a metrological limit beyond classical sensing (Xu et al., 2024).

The literature is also candid about limitations. The variational qudit QISAC protocol is simulated in an idealized setting with no decoherence, gate noise, channel loss, or measurement imperfections (Nikoloska et al., 20 Nov 2025). The homodyne BPSK link is currently restricted to BPSK, quasi-static deterministic phase rotation, and classical Fisher information under homodyne detection (Krikidis, 11 Jan 2026). The 1Q and STQS architectures sacrifice mathematical rigor or formal sensing-theoretic optimization in favor of system design vocabulary and component integration (Popovski et al., 18 Sep 2025). Fiber-network platforms typically offer incomplete integrated security analyses, modest network scale, or mode switching instead of fully simultaneous operation (Zhao et al., 22 Dec 2025).

Across the surveyed works, the main open problems are consistent. End-to-end QISAC requires tighter coupling between physical-layer state design, receiver measurement, security analysis, and network control. Several papers explicitly name future directions: validation on real quantum processors for AutoQML (Koike-Akino et al., 2022), non-BPSK modulation and nonclassical light sources in quantum optical QISAC (Krikidis, 11 Jan 2026), continuous-valued and multi-parameter sensing together with more expressive learned measurements in variational QISAC (Nikoloska et al., 20 Nov 2025), and wavelength-multiplexed or more simultaneous communication-and-sensing operation in integrated fiber systems (Zhao et al., 22 Dec 2025). This suggests that QISAC is converging not toward a single canonical model but toward a layered field spanning direct joint protocols, quantum-optical receiver co-design, multifunctional network infrastructure, and quantum-enabled sensing front ends.