Intelligent Sensing and Communications (ISAC)

• ISAC is a unified framework integrating radar sensing and wireless communication by sharing hardware, spectrum, and protocol resources.
• Advanced architectures like DFRC transceivers, RIS, and distributed networks enable joint optimization of data rate and sensing accuracy.
• AI-driven techniques and multi-objective optimization manage trade-offs to support real-time, context-aware applications in next-generation networks.

Intelligent Sensing and Communications (ISAC) is a technical paradigm that unifies radio-based sensing (e.g., radar, localization, mapping) and wireless data communication within a single hardware, spectrum, and protocol stack, increasingly driven by AI. ISAC is foundational for 6G and beyond, promising significant gains in spectral, energy, and hardware efficiency by eschewing the long-standing separation between communication and sensing resources. Architectures range from dual-functional transceivers integrating both operations at the waveform and beamformer levels to large-scale, distributed, semantic-aware networks encompassing heterogeneous sensor modalities. Cutting-edge designs combine advanced optimization and AI/ML to manage complex trade-offs, support self-organization, and enable distributed, context-aware operation.

1. Core ISAC Principles and System Models

ISAC abandons the traditional hardware and protocol decoupling of radar and communications, instead employing a unified transmit–receive chain and common spectral resources (Zhang et al., 29 Dec 2025). In practical terms:

• The transmit waveform $x(t)$ is engineered to be simultaneously information-bearing and probing.
• Both sensing and communications multiplex over time–frequency–space using multi-carrier, chirp-OFDM, or OTFS schemes (Zhang et al., 29 Dec 2025, Kaushik et al., 2023).
• Generic receive models for an ISAC node are:
  • Sensing: $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$, with $h_s(\theta)$ a reflection coefficient, $\tau$ delay, $w(t)$ noise.
  • Communication: $y_c(t) = h_c x(t) + z(t)$.

Key system-level objectives are data delivery at high reliability, low latency, and accurate environment mapping or target detection/localization, through hardware sharing and joint signal processing.

2. Architectures: From Dual-Function Transceivers to Distributed Intelligent Networks

ISAC deployment models range from monolithic base station (BS) architectures to distributed, goal-driven mesh networks:

• Dual-Function Radar–Communication (DFRC) Transceiver: The classical ISAC implementation for 5G-Advanced and 6G, where the BS shares antennas and RF chains for both communications and radar sensing (Magbool et al., 2024, Kaushik et al., 2023).
• Reconfigurable Intelligent Surfaces (RIS) and Intelligent Metasurfaces (IM): Programmable electromagnetic surfaces enhance both communication and sensing by controlling the channel spatial degrees of freedom—improving coverage, SNR, and spatial resolution (Wu et al., 14 Nov 2025, Meng et al., 2023, Li et al., 16 Jun 2025).
• Distributed ISAC (DISAC): 6G extensions emphasizing networks of spatially distributed nodes, heterogeneous sensors (RF, LiDAR, camera), semantic data fusion, and goal-oriented information flow replace classical centralized fusion (Strinati et al., 2024, Strinati et al., 2024, Stylianopoulos et al., 17 Apr 2025). Nodes compress and transmit semantic features (rather than raw data), and fusion centers orchestrate resources and perform high-level inference.

The move toward distributed architectures enables super-resolution sensing via aperture synthesis, reduction of energy/backhaul demand, and context-driven operation at scale.

3. Signal Processing, Optimization, and AI-Native ISAC

ISAC requires complex, joint optimization of transmit beamformers, waveforms, and programmable surfaces. Characteristic features include:

• Multi-objective Optimization: The fundamental problem is often cast as maximizing a weighted sum of communication rate and sensing accuracy,

$$\max_{x(t)}\ \alpha R_C(x) - (1-\alpha) \mathrm{MSE}_s(x)$$

or similar, with trade-off parameter $\alpha$ (Zhang et al., 29 Dec 2025, Meng et al., 2023, Vaezi et al., 17 Apr 2025). Constraints include power, constant-modulus requirements, and hardware limitations.

• Alternating and Manifold Optimization: Coupled, non-convex variables such as beamformers, waveforms, and RIS/IM phase shifts are addressed using block-coordinate descent, semidefinite relaxation (SDR), majorization-minimization (MM), and Riemannian manifold methods (Meng et al., 2023, Wu et al., 14 Nov 2025, Li et al., 16 Jun 2025, Liu et al., 2022).
• AI-Driven Design: Deep learning, reinforcement learning, and graph neural networks facilitate:
  • Data-driven waveform and beamformer synthesis under difficult channel models or partial observability (Vaezi et al., 17 Apr 2025);
  • Resource allocation policies generalizing over changing topologies (Zhang et al., 29 Dec 2025);
  • Semantic encoder/decoder design for distributed, goal-efficient communication in DISAC (Strinati et al., 2024, Strinati et al., 2024).

Papers document the use of unsupervised DNNs and neural network-based algorithm unrolling to achieve near-optimal trade-offs in real-time with drastically reduced computational cost (Vaezi et al., 17 Apr 2025).

4. Intelligent Surfaces and Metasurfaces in ISAC

Programmable surfaces such as RIS, IM, STAR-IRS, and IOS are central to high-performance ISAC:

• Operational Models: Passive RIS elements provide phase-only reflection; active and hybrid variants supply gain and/or direct echo measurement; STAR-IRS and IOS enable simultaneous transmit/reflect/refract for full-space joint coverage (Wu et al., 14 Nov 2025, Meng et al., 2023, Zhang et al., 2023, Wei et al., 2022).
• Performance Metrics: SNR scaling for sensing (e.g., $20 \log_{10} M$ dB as a function of RIS elements $M$), CRB for estimation, and spectral efficiency enhancements have been quantified in a variety of system models (Wei et al., 2022, Meng et al., 2023).
• Channel Coupling and Trade-off: The capacity to adjust the "coupling" between the communication and sensing subspaces by RIS/IM phase assignment is essential—optimizing for joint or orthogonal operation as required (Chepuri et al., 2022).
• Optimization Problems: Joint beamforming and RIS/IM phase optimization problems are standard, with solutions via AO, SDR, SCA, and learning-aided methods (Wu et al., 14 Nov 2025, Meng et al., 2023, Li et al., 16 Jun 2025).

Empirical and analytical results demonstrate quadratic SNR gains in sensing, 30–50% increases in achievable communication rate, and sharp CRB reductions as RIS/IM size grows (Liu et al., 2022, Wei et al., 2022, Li et al., 16 Jun 2025, Wu et al., 14 Nov 2025).

5. Distributed, Semantic, and Goal-Oriented ISAC

Emerging 6G research expands ISAC into the distributed and semantic paradigms:

• DISAC and Semantic RANs: DISAC abstracts from node-centric ISAC to a networked model in which multiple, heterogeneous sensors generate semantic descriptors of the environment, interpreted and fused by semantic managers (SeMF) for goal-driven action (Strinati et al., 2024, Strinati et al., 2024, Stylianopoulos et al., 17 Apr 2025).
• Mathematical Formalism: Consensus Kalman filtering, graph-based aggregation, and Bayesian fusion define the distributed estimation backbones. Semantic representations $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$0—low-dimensional, goal-relevant features—are fused to minimize overall semantic distortion rather than raw data reconstruction error.
• Resource Optimization: Task-oriented rate–distortion trade-offs, joint connect-compute allocation, and distributed convex programming are key strategies (Strinati et al., 2024).
• Performance Gains: DISAC demonstrates 30–50% improved localization accuracy, 80% latency reduction in multi-node V2X use-cases, and substantial energy/backhaul savings compared to monolithic ISAC (Strinati et al., 2024, Stylianopoulos et al., 17 Apr 2025).

Table: Architectural Features across ISAC Generations

Generation	Architecture	Objective
ISAC	Centralized DFRC	Joint rate/sensing
DISAC	Distributed, semantic	Goal/task optimization

6. Trade-offs, Metrics, and Evaluation

ISAC system design is fundamentally governed by the trade-off between communication throughput and sensing fidelity. Quantitative metrics include:

• Communication: Achievable rate $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$1, spectral efficiency, reliability (PER/BER).
• Sensing: CRB, MSE of parameter estimation, detection probability $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$2 vs. false alarm $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$3, tracking RMSE.
• Integrated Objectives: Weighted-sum or vectorized Pareto regions between these metrics, with trade-offs precisely rendered as functions of joint beamformer/surface/waveform parameters (Meng et al., 2023, Zhang et al., 29 Dec 2025, Vaezi et al., 17 Apr 2025).
• Resource Constraints: Power, bandwidth, latency, and computational budgets, with end-to-end latency evaluated from sensing-to-action cycle.

Empirical studies and simulation benchmarks validate the joint design's advantage: e.g., a RIS-enhanced ISAC with $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$4 elements can increase echo SNR by $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$5 dB over baseline, halve the lower CRB for angle estimation, and raise detection probability above $y_s(t) = h_s(\theta)x(t-\tau) + w(t)$6 in NLoS (Wei et al., 2022, Liu et al., 2022, Meng et al., 2023).

7. Open Challenges and Future Research Directions

Key open problems for ISAC and DISAC research include:

• Theoretical Limits: Derivation of sharp, closed-form Pareto frontiers for rate–sensing accuracy in large-scale, partially observed, and distributed networks (Zhang et al., 29 Dec 2025, Meng et al., 2023).
• Channel Modeling: Wideband, near-field, and mobility-aware models for realistic programmable surfaces; robust estimation and real-time beam training (Chepuri et al., 2022, Wu et al., 14 Nov 2025).
• Scalable AI, Trust, and Security: Robustness to adversarial/poisoned AI agents, federated/self-supervised learning under distribution drift, and privacy of semantic sensing (Zhang et al., 29 Dec 2025, Strinati et al., 2024).
• Multi-modal ISAC: Generalization from RF-only ISAC to SoM (synesthetic multi-modal) frameworks incorporating vision, LiDAR, and inertial sensors for rich, cross-domain fusion (Cheng et al., 2023).
• Standardization and Real-World Prototyping: 3GPP, IEEE, and ITU ongoing and planned work to define ISAC/DISAC interfaces, performance metrics, and testbeds for evaluation (Zhang et al., 9 Apr 2025, Magbool et al., 2024, Kaushik et al., 2023).

Emergent topics include cooperative multi-RIS/IM, semantic rate-distortion bounds, digital twin-aided scenario simulation, and real-time distributed goal-driven orchestration.

---

References:

• (Zhang et al., 29 Dec 2025) AI-Native Integrated Sensing and Communications for Self-Organizing Wireless Networks: Architectures, Learning Paradigms, and System-Level Design
• (Meng et al., 2023) Intelligent Surface Empowered Integrated Sensing and Communication: From Coexistence to Reciprocity
• (Wu et al., 14 Nov 2025) Intelligent Reflecting Surfaces for Integrated Sensing and Communications: A Survey
• (Vaezi et al., 17 Apr 2025) AI-Empowered Integrated Sensing and Communications
• (Strinati et al., 2024) Towards Distributed and Intelligent Integrated Sensing and Communications for 6G Networks
• (Strinati et al., 2024) Distributed Intelligent Integrated Sensing and Communications: The 6G-DISAC Approach
• (Chepuri et al., 2022) Integrated Sensing and Communications with Reconfigurable Intelligent Surfaces
• (Li et al., 16 Jun 2025) Intelligent Metasurface-Enabled Integrated Sensing and Communication: Unified Framework and Key Technologies
• (Kaushik et al., 2023) Integrated Sensing and Communications for IoT: Synergies with Key 6G Technology Enablers
• (Stylianopoulos et al., 17 Apr 2025) Distributed Intelligent Sensing and Communications for 6G: Architecture and Use Cases
• (Magbool et al., 2024) A Survey on Integrated Sensing and Communication with Intelligent Metasurfaces: Trends, Challenges, and Opportunities
• (Cheng et al., 2023) Intelligent Multi-Modal Sensing-Communication Integration: Synesthesia of Machines
• (Liu et al., 2022) Integrated Sensing and Communication with Reconfigurable Intelligent Surfaces: Opportunities, Applications, and Future Directions
• (Wei et al., 2022) Intelligent Reflecting Surface assisted Integrated Sensing and Communication System