Uplink Dual-Functional ISAC Overview

• Uplink Dual-Functional ISAC is an advanced framework where uplink transmissions are used both for communication and for sensing targets or environmental parameters.
• It leverages multi-antenna, full-duplex, and reconfigurable intelligent surface technologies along with techniques like SIC and deep learning for optimal resource allocation.
• This approach enhances spectral efficiency and joint performance compared to traditional FDSAC, serving as a key enabler for 6G networks and beyond.

Uplink Dual-Functional Integrated Sensing and Communication (ISAC) refers to systems and frameworks where uplink transmissions from user equipment (UE) serve both as information-bearing signals (enabling communications) and as probing waveforms for environmental or object sensing. In such architectures, the base station (BS) employs the received uplink signal to simultaneously detect targets or estimate environmental parameters and decode user data, leveraging the same physical channel and spectral resources. This dual use has been recognized as a cornerstone of 6G and future wireless systems, yielding significant improvements in spectral efficiency, hardware reuse, latency reduction, and joint communication–sensing performance compared to traditional frequency-division (FDSAC) or time-division multiplexing approaches.

1. Systems and Architectural Principles

1.1 Generic Uplink ISAC System Model

A canonical uplink dual-functional ISAC system consists of multiple single-antenna UEs transmitting modulated data and, optionally, explicit pilots. The BS is equipped with a multi-antenna (MIMO) array, potentially augmented by intelligent or reconfigurable surfaces (e.g., RDARS, RIS) or movable antenna elements. The received BS signal is thus a composite of:

• Superposed user data: $x_C(t) \in \mathbb{C}^{K\times1}$
• Optional local sensing/probing waveform: $x_S(t) \in \mathbb{C}^{M_t \times 1}$
• Environmental echo or target response: $G$ or equivalent operator for continuous arrays
• Additive white Gaussian noise and possible self-interference (in full-duplex, FD, setups)

A concrete instance is provided by the RDARS-aided system, where 256 passive antenna/reflector elements at the BS are individually switchable between reflection and direct-connection modes, controlling signal pathways for optimized communications and distributed sensing (Wang et al., 2023).

1.2 Integration Modalities

• Full-duplex (FD) ISAC: Both radar probing (for sensing) and uplink user communication occur simultaneously and in the same frequency band at the BS (He et al., 2023, Guo et al., 2023, Guo et al., 2023).
• Monostatic, Bistatic & Network Sensing: Prototype systems support monostatic (self-echo), bistatic (cross-node), and multi-node networked scenarios, exploiting shared hardware and waveforms for all (Yang et al., 2024).

2. Information-Theoretic Performance Metrics

2.1 Mutual Information (MI) Framework

ISAC system performance is naturally assessed via mutual information (MI) metrics for both communication and sensing links:

• Communication MI: $I_{\text{comm}} = I(x_C; y_U | H)$, representing channel capacity under Gaussian signaling, with or without interference cancellation.
• Sensing MI: $I_{\text{sense}} = I(G; y_U | x_S)$, quantifying the Fisher-information or mutual information of the physical/environmental parameters observable via uplink signals.

These metrics facilitate a unified analysis and trade-off of achievable rates, degrees of freedom (DoF), and Pareto boundaries compared to isolated FDSAC approaches (Ouyang et al., 2022, Piao et al., 2023).

2.2 High-SNR Slope and Diversity Order

High-SNR analyses reveal the asymptotic scaling laws:

• Communication slope: Achievable multiplexing gain; for K users and N antennas, it's typically $\min(N, K)$.
• Sensing slope: Related to sensing dimensionality, e.g., NM/L for radar with M transmit, N receive, and L pulses (Ouyang et al., 2022).

In NOMA-ISAC and MIMO/MIMO-OFDM frameworks, these slopes determine the DoF unlocked by ISAC protocols and how they outperform FDSAC in spectral efficiency (Zhao et al., 2023, Ouyang et al., 2022).

3. Receiver and Signal Processing Methodologies

3.1 Successive Interference Cancellation (SIC) and Projection-based Separation

Two principal approaches resolve the interference between the data streams and the sensing echo at the BS:

• SIC-based Receivers: Communications-centric (C–C) or Sensing-centric (S–C) decoding orders, where one functionality is treated as interference while the other is decoded, followed by subtraction and re-processing (Zhao et al., 10 Feb 2025, Zhao et al., 2023, Ouyang et al., 2022, He et al., 2023). The choice impacts achievable regions and diversity order.
• Flexible Projection (FP) Receivers: Dynamic trade-off factor $\alpha$ enables seamless transition between projection and SIC extremes, optimizing for channel condition or SIR. Dynamic FP (DFP) receivers adapt online, and PDFP/block-structured variants further reduce error propagation and exploit temporal coherence (Yu et al., 4 Mar 2025).

3.2 Covariance and Subspace Optimization

Covariance-based designs (e.g., MI-based with water-filling solutions (Ouyang et al., 2022, Piao et al., 2023)), subspace-based beamformers for CAPAs (Zhao et al., 10 Feb 2025), and iterative algorithms for joint filter/signal/positioning optimization (e.g., MM, PDD, ADMM) target feasible, closed-form, or analytic update rules for the non-convex ISAC optimization landscape.

3.3 Deep Learning-based Solutions

Supervised and unsupervised DNNs can approximate near-optimal resource and waveform allocations (e.g., power profiles, beamforming weights) by learning from the joint distribution of communication CSI and sensing statistics (Qi et al., 2024). Forward pass runtimes (<1 ms) enable real-time adaptation.

4. System Prototypes and Emerging Hardware

4.1 Reconfigurable Distributed and Intelligent Surfaces

RDARS-based prototypes implement dynamic mode switching (reflection vs connection) per element, providing macro-diversity, passive beamforming, and directly sampled echoes for user localization (Wang et al., 2023). RIS and metasurface-empowered BSs extend this paradigm, offering quartic-coupled channels and programmable spatial propagation environments for joint FD-ISAC (Guo et al., 2023, Guo et al., 2023, Illi et al., 29 Apr 2025).

4.2 Movable Antenna Arrays

Movable antenna (MA) systems optimize spatial DoF by physically repositioning receive elements to maximize both communication and sensing performance under tight CRB or SINR constraints. MM + PDD methods provide low-complexity updates for joint beamformer, power, and position control (Guo et al., 29 Oct 2025).

4.3 Continuous-Aperture Architectures

CAPA-based ISAC architectures replace classical discrete antenna arrays with operator-defined, spatially continuous transmit/receive surfaces, achieving higher CR and SR and a larger joint rate region compared to S-PDA arrays (Zhao et al., 10 Feb 2025).

4.4 Uplink-Centric Prototypes

Experimental networks leveraging mmWave arrays, SDRs, and TDD-5G NR physical layer have demonstrated uplink ISAC feasibility, reporting sub-meter range RMSE, 2.3° angle RMSE, and 500+ Mbps instantaneous UL rates under realistic loads and mobility (Yang et al., 2024).

Prototype/Concept	Unique Feature(s)	Measured UL ISAC Performance
RDARS (Wang et al., 2023)	256-element reconfigurable panel, reflection/connection modes	81.8 Mbps UL, AoA error 4°, range RMSE 0.5 m
FD-ISAC + RIS (Guo et al., 2023, Guo et al., 2023)	Full-duplex; RIS-assisted joint optimization	30-40% UL rate gain at large RIS; flexible radar/comm trade-off
CAPA (Zhao et al., 10 Feb 2025)	Continuous-aperture receive/transmit arrays	+25% CR and +27% SR vs SPDA; closed-form rate expressions
5G-NR mmWave (Yang et al., 2024)	Multi-domain, multi-user, networked	2.3° angle RMSE, 0.3 m range RMSE, 508 Mbps UL

5. Optimization and Trade-Off Characterization

5.1 Joint Communication–Sensing Optimization

The ISAC optimization is typically formulated as a weighted sum or multi-objective problem:

$$\max_{W, \{\mathbf{u}_k\}, \{q_k\}} \ \alpha R_{\text{sense}}(W, \mathbf{u}_k, q_k) + (1-\alpha) R_{\text{comm}}(W, \mathbf{u}_k, q_k)$$

subject to hardware, power, and SINR constraints. Water-filling, alternating maximization, or neural approximation are used to solve for jointly optimal resource allocations (Ouyang et al., 2022, Yu et al., 4 Mar 2025, Guo et al., 29 Oct 2025, Qi et al., 2024).

5.2 Pareto Region and DoF Gains

The ISAC achievable rate region with time-sharing between C–C and S–C (or their respective outer/inner SIC orders) strictly contains that of FDSAC. ISAC unlocks the full multiplexing gain in both functionalities rather than splitting them among orthogonal resources. For instance, at high SNR the DoF for communication and sensing under ISAC equal rank(H) and rank(R_G), respectively, whereas FDSAC's are scaled down by the bandwidth split factors (Ouyang et al., 2022, Zhao et al., 2023).

5.3 Effects of Channel, Power, and Resource Allocation

• Increasing the number of antennas, user groups (NOMA), or array aperture expands both communication and sensing rate regions.
• FD ISAC consistently saves transmit power relative to HD-ISAC at fixed performance, with RIS/MAs further boosting this gain (He et al., 2023, Guo et al., 29 Oct 2025).
• Optimal mode switching (reflection/connected) in RDARS enables balancing between macro-diversity (communication) and spatially distributed echoes (sensing/localization) (Wang et al., 2023).
• Artificial noise (AN) in RIS-ISAC can be harnessed for both secrecy and sensing, increasing beampattern power but requiring careful joint design under secrecy and total-power constraints (Illi et al., 29 Apr 2025).

6. Extensions, Implementation Issues, and Practical Insights

6.1 Implementation Complexity and Real-Time Feasibility

Efficient optimization algorithms—for example, MM+PDD instead of generic interior-point solvers—are critical for practical deployments, with per-iteration scaling as $O(N^3)$ for most matrix operations (Guo et al., 29 Oct 2025).

Deep learning predictors can offer sub-millisecond inference for online waveform/beamformer selection, robust to CSI estimation inaccuracies and environmental drift (Qi et al., 2024).

6.2 Sensing-Aided Communication

Sensing output, such as AOA or range, can be used to guide UL beam-tracking for reduced alignment overhead and lower outage rates, e.g., reducing exhaustive beam searches from 1.28 s to 0.32 s with minimal communication interruption at 4 m/s terminal velocity (Yang et al., 2024). SLAM applications leverage ISAC modalities for joint localization and mapping, blending radio, vision, and inertial cues.

6.3 Challenges and Open Problems

• Robust design under imperfect or partial CSI, and under multipath or MIMO radar/target conditions
• Cross-cell and multi-user coordination for joint ISAC resource allocation
• Efficient hardware implementation for moving arrays, massive RIS, and real-time signal processing
• Security–secrecy–sensing trade-offs in adversarial or dynamic environments

7. Comparative Summary

Class	Uplink Communication Metric	Uplink Sensing Metric	Notable Implementation(s)/Findings
MIMO/OFDM ISAC (Piao et al., 2023)	Weighted sum, MI (water-filling)	Sensing MI via mutual information	Smooth trade-off curve, efficient eigen-optimization
RDARS ISAC (Wang et al., 2023)	log(1+SNR)-like, with macro-diversity gains	Angle RMSE (4°), range RMSE (0.5 m)	Hardware-prototype with mode switching
NOMA-ISAC (Zhao et al., 2023, Ouyang et al., 2022)	Per-user/group rates and outage	Sensing MI (water-filling); high-SNR slope $M^2/L$	Enhanced rate-region via SIC ordering
CAPA-ISAC (Zhao et al., 10 Feb 2025)	Closed-form AWGN rate (CR)	Closed-form, per-symbol SR	CAPA strictly outperforms SPDA
Movable Antenna ISAC (Guo et al., 29 Oct 2025)	CRB-minimized (MM+PDD)	CRB for angle estimation	Up to 5× lower CRB than fixed arrays
DL-based ISAC (Qi et al., 2024)	Normalized, weighted comm. rate	Normalized, weighted sensing rate	Real-time DNN prediction
RIS-powered FD ISAC (Guo et al., 2023, Guo et al., 2023, Illi et al., 29 Apr 2025)	Sum-rate (under unit-modulus phase)	Radar output SINR, beampattern, secrecy	Up to 40% UL rate gain, joint secrecy-sensing optimization

• "Demo: Reconfigurable Distributed Antennas and Reflecting Surface (RDARS)-aided Integrated Sensing and Communication System" (Wang et al., 2023)
• "Integrated Sensing and Communications: A Mutual Information-Based Framework" (Ouyang et al., 2022)
• "Integrated sensing and full-duplex communication: Joint transceiver beamforming and power allocation" (He et al., 2023)
• "Performance of Downlink and Uplink Integrated Sensing and Communications (ISAC) Systems" (Ouyang et al., 2022)
• "Joint Beamforming and Power Allocation for RIS Aided Full-Duplex Integrated Sensing and Uplink Communication System" (Guo et al., 2023)
• "Mutual Information Metrics for Uplink MIMO-OFDM Integrated Sensing and Communication System" (Piao et al., 2023)
• "On the Secrecy-Sensing Optimization of RIS-assisted Full-Duplex Integrated Sensing and Communication Network" (Illi et al., 29 Apr 2025)
• "Uplink-Downlink Duality for Beamforming in Integrated Sensing and Communications" (Attiah et al., 17 Sep 2025)
• "A Framework for Uplink ISAC Receiver Designs: Performance Analysis and Algorithm Development" (Yu et al., 4 Mar 2025)
• "On the Performance of Uplink ISAC Systems" (Ouyang et al., 2022)
• "Joint Beamforming for RIS Aided Full-Duplex Integrated Sensing and Uplink Communication" (Guo et al., 2023)
• "ISAC Prototype System for Multi-Domain Cooperative Communication Networks" (Yang et al., 2024)
• "Cramér-Rao Bound Optimization for Movable Antenna-Empowered Integrated Sensing and Uplink Communication System" (Guo et al., 29 Oct 2025)
• "Downlink and Uplink NOMA-ISAC with Signal Alignment" (Zhao et al., 2023)
• "Deep Learning-based Design of Uplink Integrated Sensing and Communication" (Qi et al., 2024)
• "Downlink and Uplink ISAC in Continuous-Aperture Array (CAPA) Systems" (Zhao et al., 10 Feb 2025)