Future industrial 6G applications will impose stringent requirements on ultra-reliable low-latency communications (URLLC) and precision sensing enabled by integrated sensing and communication (ISAC) techniques, motivating a comprehensive study of the communication–sensing (C–S) trade-off under finite blocklength transmission. Therefore, this paper investigates the fundamental C–S performance limits in a network-assisted full-duplex (NAFD) cell-free ISAC system with URLLC users. To address the theoretical gap in the finite blocklength regime, closed-form upper-bound expressions are derived for key communication metrics, including transmission delay and decoding error probability (DEP), and a Cramér–Rao lower bound (CRLB) framework is established for multi-static sensing. Furthermore, to explicitly characterize the C–S trade-off, the ISAC network availability is evaluated and the Pareto frontier is obtained using the non-dominated sorting genetic algorithm II (NSGA-II). The results demonstrate that increasing the blocklength improves sensing accuracy at the cost of higher communication latency. To address this inherent conflict, this study proposes a DDQN-based finite blocklength optimization (FBLO) algorithm that performs blocklength selection under URLLC and sensing quality-of-service requirements to achieve a favorable C–S trade-off. Simulation results validate that the proposed algorithm achieves near-optimal performance with reduced computational overhead.

Cooperative cell-free (CF) integrated sensing and communication (ISAC) systems emerge as a promising architecture for supporting 6G dynamic Internet of Things (IoT) scenarios. However, the mobility of user equipment (UE) and the inherent non-scalability of CF networks pose critical challenges to the practical deployment of CF ISAC systems. This paper presents a comprehensive mobility management framework for cooperative CF ISAC systems to enhance their deployability and scalability. This framework not only establishes a foundational operation paradigm to obtain the mutual promotion of communication and sensing (C&S) performance in multi-static ISAC, but also employs the dynamic cooperative clustering method and dynamic management mechanism to ensure seamless service for mobile UEs. First, we establish the mathematical signal model of the proposed mobility management framework and conduct the analysis of mobility-aware C&S performance in CF ISAC systems. Subsequently, a distributed, low-complexity initial access scheme is designed to tackle the tightly coupled challenges of access point (AP) clustering and AP mode selection, which can ensure communication reliability and sensing accuracy in static scenarios. Furthermore, to achieve the trade-off between mobility-induced handover loss and per-slot C&S performance, a dynamic access scheme is introduced for dynamic scenarios, comprising the dynamic adaptive hysteresis handover strategy and the kinematic information-based dynamic clustering update algorithm. Theoretical and numerical analyses validate that the proposed access schemes significantly enhance the operability and practicality of CF ISAC systems through low complexity and flexible operations. Meanwhile, simulation results demonstrate that the framework empowers cooperative CF ISAC systems to achieve superior mobility-aware C&S performance, ensuring the stable and efficient system support in 6G dynamic IoT scenarios.

The deployment of cell-free massive multiple-input multiple-output (CF-mMIMO) systems in urban areas enriches wireless channel characteristics, making it a promising approach for fingerprint-based localization. However, the high cost of label collection and the unreliable generalization performance significantly limit its widespread application. Recently, model-agnostic meta-learning (MAML) has achieved remarkable success in few-shot learning by extracting common knowledge from existing tasks. But the use of a forcibly shared meta-parameter for model initialization often struggles with task heterogeneity in practical applications. To address these challenges, we propose a novel gradient-based task-aware meta-learning (GTML) framework for fingerprint-based localization. We first model the localization problem in a new environment with limited fingerprint data as a meta-learning problem for new task adaptation. Then, we propose an improved embedded task-aware method based on training gradients to reduce the overhead of task-specific feature extraction. The proposed GTML includes two paradigms using task-specific information to customize the global meta-learner: for a few historical tasks, a weighted paradigm is introduced to compensate for task heterogeneity; for a larger set of historical tasks, a clustered paradigm is used to capture the distribution of training tasks and learn group-specific meta-parameter. The simulation results using Wireless Insite software demonstrate that the proposed GTML enables rapid adaptation to a new environment with a few training samples. Moreover, compared to the vanilla MAML, GTML improves the localization accuracy by over 11% while maintaining low computational overhead.

Multi-static Integrated Sensing and Communication (ISAC) is a potential technology for future sixth-generation (6G) and cell-free (CF) network is a suitable architecture to integrate it. Current research on multi-static ISAC and full-duplex communication (MIFC) CF systems is scarce, and the adoption of full-duplex (FD) access points (APs) inevitably leads to significant self-interference (SI) and exorbitant deployment costs. Utilizing network-assisted full-duplex (NAFD) technology to implement MIFC CF systems can effectively avoid the above issues. However, in addition to the challenge posed by highly coupled cross-link interference (CLI) and multi-user interference, NAFD-based MIFC CF systems must also address the mutual interference between sensing signals and communication signals. This paper proposes a practical MIFC CF system based on NAFD technology and introduces a four-stage interference management mechanism, which integrates direct interference suppression with indirect interference suppression techniques. Within this mechanism, we initially derive the data transmission estimated channel state information (CSI), the maximum a posteriori ratio test (MAPRT) target detector and inter-AP estimated CSI. Then, we furnish the expressions for communication achievable rate and sensing signal-to-interference-plus-noise ratio (SINR) after direct interference cancellation based on the estimated CSI. Furthermore, a deep learning (DLN)-based joint communication and sensing precoding (JCSP) algorithm is devised for indirect interference suppression. Simulation results demonstrate the effectiveness of the direct interference suppression strategy and DLN-based JCSP algorithm in the proposed interference management mechanism, which can achieve the trade-off between communication and sensing performance.

Network-assisted full-duplex (NAFD) cell-free (CF) network emerges as a promising solution for enabling massive ultra-reliable and low-latency communications (mURLLC). In the massive Internet-of-Things (mIoT) scenarios where users’ active statuses change, the existing NAFD resource scheduling schemes require frequent invocation of optimization algorithms, causing significant energy loss and operational delays. So they are not conducive to mURLLC. This paper proposes a dual time-scale resource scheduling scheme, which combines the improved long-term AP duplex mode optimization method with the short-term power allocation optimization method to further enhance the mURLLC ability of NAFD CF networks. In the proposed long-term AP duplex mode optimization method, we first derive the closed-form expressions of active users’ time overflow (TO) probability as the service latency indicator. Operating on a superframe as a large time-scale unit, the improved long-term AP duplex mode optimization method initially employs a long-term active user prediction algorithm to forecast active users in an upcoming superframe and then leverages the long-term AP duplex mode optimization algorithm based on multi-agent deep reinforcement learning to achieve optimal long-term AP mode selection which minimizes the TO probability. In the proposed short-term power allocation optimization method, we design a heuristic algorithm to ensure active users in each coherence time can receive high-reliable and low-latency service. Simulation results demonstrate the effectiveness of the proposed scheme. Compared with the short-term AP mode and power joint optimization methods, the dual time-scale resource scheduling scheme achieves similar spectral efficiency and a much lower TO probability, while also avoiding the frequent AP mode optimization and switching, making it more suitable for mURLLC.

Fingerprint technique is a promising enabler for mobile terminals (MTs) localization in rich scattering environments, such as urban areas and indoor corridors. In this paper, we investigate fingerprint-based localization for massive multiple- input multiple-output (MIMO) orthogonal frequency- division multiplexing (OFDM) systems with deep convolutional neural networks (DCNNs). By taking full advantage of the high resolution in the angle domain and the delay domain in massive MIMO-OFDM systems, we first propose an efficient angle-delay channel amplitude matrix (ADCAM) fingerprint extraction method. Then a DCNN enabled localization method is proposed, in which the modeling error for fingerprint similarity calculation can be overcome. Both DCNN classification and DCNN regression are considered. For practical implementation, a hierarchical DCNN architecture is proposed. Numerical simulation results demonstrate that DCNN performs well in achieving high localization accuracy as well as reducing storage overhead and computational complexity.

Fingerprint techniques are promising localization strategies in rich scattering environments, such as urban areas and indoor corridors. However, most existing approaches rely on multiple base station (BS) cooperation and suffer from multipath propagation. In this paper, we propose a fingerprint based single-site localization method for massive multiple-input multiple-output (MIMO) orthogonal frequency-division multiplexing (OFDM) systems. The new angle delay channel power matrix (ADCPM) fingerprint is extracted from instantaneous channel state information (CSI) by taking full advantage of the high resolution in the angle and delay domains for massive MIMO-OFDM systems. The applicable fingerprint similarity criterion，, as well as location estimation method, are proposed to reduce measurement, storage, and matching overheads. Numerical results demonstrate the desirable performance of the proposed localization method.

Two-stage precoding is a promising transmission strategy for multi-user frequency division duplex (FDD) massive multiple-input multiple-output (MIMO) systems due to its large multiplexing gain with overhead reduction in both downlink channel estimation and channel state information (CSI) feedback. The performance of existing two-stage precoding schemes mainly depends on appropriate selection and clustering of the users, which is sometimes difficult to realize in a realistic scenario with limited number of users having different covariance eigenspace. In this paper, we propose a new agglomerative clustering method for user grouping which can be easily implemented in realistic scenario. We also develop an average signal-to-leakage-plus-noise ratio (SLNR) based downlink group scheduling method to achieve combination of user groups in a particular time-frequency slot. Numerical results validate the performance improvement of the proposed methods over existing methods.