Integrated Sensing & Communication (ISAC) Networks

• Integrated Sensing and Communication (ISAC) networks are transformative wireless systems that merge environmental sensing with data transmission using unified signal designs.
• They leverage shared hardware and spectrum, employing advanced waveform design and cooperative architectures to balance throughput with sensing accuracy.
• ISAC is pivotal for applications like smart mobility, airspace management, and industrial automation, facilitating context-aware, resource-efficient networks.

Integrated Sensing and Communication (ISAC) networks constitute a transformative paradigm in wireless systems, embedding environmental perception natively into the physical infrastructure of data communications. Unlike conventional architectures optimized for either pure data transfer or standalone sensing, ISAC networks pursue unified signal models, waveforms, and resource management to jointly provide high-rate/reliable communications and accurate, real-time environmental sensing, all over shared hardware, spectrum, and infrastructure. ISAC networks are at the core of emerging 6G systems, promising digital twins, context-aware services, efficient spectrum use, and new applications in smart mobility, airspace management, industrial automation, and security.

1. Mathematical Foundations and System Models

The canonical ISAC network consists of multi-antenna base stations (BSs) transmitting a common waveform for both user communications and environmental interrogation. The joint signal model at time $t$ is

$\mathbf{x}(t) = \mathbf{W}\mathbf{s}(t),$

where $\mathbf{s}(t) \in \mathbb{C}^{d \times 1}$ carries $d$ communication symbols and $\mathbf{W} \in \mathbb{C}^{N_t \times d}$ is the shared precoder. Communications and sensing receive branches observe: $$\begin{aligned} \mathbf{y}_c(t) &= \mathbf{H}_c\mathbf{x}(t) + \mathbf{n}_c(t), \ \mathbf{y}_s(t) &= \mathbf{H}_s\mathbf{x}(t) + \mathbf{n}_s(t), \end{aligned}$$ with $\mathbf{H}_c$ the comm channel, $\mathbf{H}_s$ the composite target (multi-path) response, and $\mathbf{n}_c, \mathbf{n}_s$ independent noises.

Optimal ISAC design is formulated as a trade-off between communication rate

$$R(\mathbf{W}) = \log_2\det\left( \mathbf{I} + \frac{1}{\sigma_c^2} \mathbf{H}_c\mathbf{W}\mathbf{W}^H\mathbf{H}_c^H \right),$$

and sensing accuracy, typically via the vector CRLB for parameter vector $\boldsymbol{\eta}$ (target delay, AoA, etc.): $$\mathrm{MSE}(\mathbf{W}) = \mathbb{E}\| \hat{\boldsymbol{\eta}} - \boldsymbol{\eta} \|^2.$$ Weighted-sum and Pareto frontier optimization constitutes the prevailing formulation: $$\max_{\mathbf{W}}\,\, \alpha R(\mathbf{W}) - (1-\alpha)\mathrm{MSE}(\mathbf{W}), \quad \|\mathbf{W}\|^2_F \leq P,$$ with $\alpha \in [0,1]$ and $P$ the power constraint (Cui et al., 16 Oct 2025).

Network-level extensions introduce resource-sharing across time, frequency, and spatial domains, multiplexing users, sensing tasks, and network nodes.

2. ISAC Network Architectures and Cooperative Operation

ISAC network design spans multiple integration levels, from spectrum cohabitation to full resource, waveform, and hardware sharing. At the network scale, ISAC leverages multi-cell and distributed architectures:

• CoMP and Multi-static Radar: Cooperative BS clusters simultaneously serve communication users and act as distributed radar apertures, fusing multi-static echoes for high-resolution, wide-area sensing (Meng et al., 2024, Xu et al., 2023, Thomä et al., 17 Nov 2025).
• Stochastic Geometry Models: BS/UE/target placements are modeled via spatial point processes (PPP, Poisson cluster, Poisson hole), enabling tractable analysis of coverage, SINR, detection, and the identification of optimal deployment/topology parameters (Wang et al., 2024, Gan et al., 2024).
• Resource and Degree-of-Freedom Allocation: Network-level spatial DoF are partitioned for user multiplexing, interference nulling, and sensing beams, subject to per-cluster and backhaul constraints (Meng et al., 2024). Optimization variables include beamforming vectors, time/frequency split ratios, and cluster composition.

Distributed ISAC protocols implement task-, data-, or waveform-level cooperation—with performance increasing from mere task handover (low backhaul) to full joint waveform/beamformer design (high backhaul). Data-fusion and Over-the-Air Computation expedite joint target estimation across the network (Thomä et al., 17 Nov 2025, Meng et al., 2024).

3. Integrated Waveform Design and Signal Processing

ISAC networks implement various multiplexing and waveform approaches:

• TDM and FDM ISAC: Isolate sensing and communication into orthogonal time or frequency resources. Fully isolates tasks but underutilizes resources (Cui et al., 16 Oct 2025).
• Spatial-Domain Superposition: Concurrently transmit comm and sensing beams with joint optimization over directions and power split (Cui et al., 16 Oct 2025).
• Unified Waveform Designs: OFDM, OTFS, and custom dual-domain waveforms (OFDM + DD-sinusoid) maximize both data capacity and delay-Doppler ambiguity resolution within regulatory constraints. Minimal rate penalty ($<0.1$ bps/Hz at $\alpha\sim 10^{-3}$), but up to 20 dB CRLB improvement in range estimation (Tagliaferri et al., 2022).
• Model-driven Deep Learning: Deep unfolded networks jointly estimate comm symbols and sensing parameters from received blocks, exploiting mutual correction and outperforming classical (2D-DFT) and black-box methods at low complexity (Jiang et al., 2023).

Optimization metrics involve joint maximization of mutual information, SINR, and Fisher information, subject to regulatory power and OOB constraints.

4. Network-Level Metrics, Trade-offs, and Scaling Laws

ISAC networks are characterized by:

• Coverage Probability: Probability of SINR (comm) and CRLB (sensing) below thresholds, often derived in closed form for PPP deployments (Gan et al., 2024, Wang et al., 2024).
• Unified Spectral Efficiency: Bits/s/Hz/km$^2$ combining area sum-rate and effective "spectral efficiency" of localization accuracy (Meng et al., 2024).
• Pareto Fronts: The achievable Sensing-vs-Throughput region is piecewise linear and convex—each Pareto segment corresponding to successive bottlenecks in resource-constrained links (Andrews et al., 15 Jan 2026).
• Scaling Laws: For distributed MIMO sensing clusters of $N$ nodes, localization CRLB scales as $1/(\ln^2N)$; communication rate saturates with dense deployments, while sensing gains persist, suggesting hybrid clusters of $N \sim 10$ optimally balance performance and backhaul (Meng et al., 2024, Gan et al., 2024).
• Conditional Ergodic Rates: Tightening sensing (CRLB) constraints improves the ergodic rate; the converse is weaker due to interference-limited regimes (Gan et al., 2024). In dense networks, ~30x increase in joint coverage is observed as BS density increases 1 to 10/km$^2$.

5. RIS-Assisted and Hybrid ISAC Networks

Reconfigurable Intelligent Surfaces (RIS), including hybrid STAR-RIS (simultaneous transmission and reflection), are central in extending ISAC functionality to challenging environments (NLoS, mmWave/THz) (Chepuri et al., 2022, Yigit et al., 2024, Illi et al., 29 Apr 2025):

• Channel Coupling: Joint comm-sensing designs are most effective when channels are coupled (overlapping spatial subspaces); RIS phase configuration can align comm and sensing channel subspaces, balancing performance (Chepuri et al., 2022).
• RIS Optimization: Alternating optimization (AO) and SDR-based methods jointly tune beamformers, RIS phase shifts, and artificial noise for maximizing min-sensing power, enforcing secrecy constraints and backhaul constraints (Illi et al., 29 Apr 2025, Yigit et al., 2024).
• Hybrid Active/Passive STAR-RIS: Full-space coverage for both comm and sensing via active transmissive (remote) and passive reflective (local) RIS elements; optimization formulations maximize sum-SINR and minimize Cramér–Rao bound for all targets/users.

6. Applications, Prototypes, and Emerging Use Cases

ISAC networks support:

• Airspace and Vehicular Networks: Real-time UAV/eVTOL tracking, low-earth airspace management, and cooperative vehicular SLAM integrating comm and multipath sensing (Cui et al., 16 Oct 2025, Ren, 20 Jan 2026, Luo et al., 2024).
• Smart Homes/Industries: Device-free vital sign/gesture detection, digital twins, sub-meter vehicle/robot localization, electromagnetic environment/material mapping (Luo et al., 2024, Wymeersch et al., 16 May 2025).
• Security and Trustworthy Networks: Dynamic trust authentication, physical-layer fingerprinting, and intent inference for low-altitude security (Ren, 20 Jan 2026).
• Prototyping: FPGA-based ISAC platforms demonstrate ≥1.5 Gbps downlink, 3° angular, and 0.18 m range resolution under joint comm/sensing operation (Luo et al., 2024).

7. Open Challenges and Future Directions

Substantive research directions include:

• Waveform and Protocol Innovation: Autoencoder-based waveform synthesis, semantic-layer stacks treating both data and "echo" streams, and multi-agent, adaptive resource allocation (Cui et al., 16 Oct 2025, Lu et al., 2023).
• Multi-node Synchronization and Data Fusion: Sub-ns sync for distributed clusters, robust fusion (edge/cloud hierarchy), AI-based mission control (Thomä et al., 17 Nov 2025).
• Cross-layer Optimization: Quantitative frameworks mapping design DoF (antennas, bandwidth, pilot overhead) to KPIs (resolution, coverage, reliability, latency) (Wymeersch et al., 16 May 2025).
• RIS/STAR Hardware Realism: Wideband modeling, mobility/tracking, finite-resolution phase shifters, multi-function metasurfaces (Chepuri et al., 2022, Yigit et al., 2024).
• Security and Privacy: Physical-layer authentication, joint radar-comm privacy, adversarial robustness, and trust models for sensing communication convergence (Ren, 20 Jan 2026, Lu et al., 2023).
• Self-Learning/AI: Distributed semantic compression, self-adaptive DoF allocation, edge/cloud ML for resource/policy management (Wang et al., 2024, Wymeersch et al., 16 May 2025).

Achieving robust, scalable, and efficient ISAC networks in 6G hinges on deep cross-layer integration, AI/edge-driven protocols, cooperative multi-node architectures, and precise characterization and optimization of the sensing-communication resource trade-space (Cui et al., 16 Oct 2025, Meng et al., 2024, Wymeersch et al., 16 May 2025).