Contact-free sensing based on WiFi channel state information (CSI) has shown considerable potential for human activity recognition and indoor localization. However, jointly addressing these two tasks remains challenging because raw CSI signals usually suffer from high-dimensional channel redundancy, task-irrelevant variations, and temporally entangled multi-scale fluctuations. To address these issues, this paper proposes a dual-task learning framework that emphasizes task-aligned subspace construction and structured temporal decomposition. Specifically, a Multi-task Aligned Re-ranking Subspace Principal Component Analysis (MARS-PCA) module is designed to re-rank principal components according to their discriminative relevance to both activity recognition and localization, thereby retaining a compact CSI representation that is more consistent with the dual-task objective. In addition, a multi-level wavelet decomposition front-end is introduced to separate CSI temporal responses into sub-band components, allowing transient activity-related dynamics and relatively stable location-related patterns to be represented more explicitly. The refined and decomposed features are then modeled by a lightweight temporal prediction module with channel-wise task regulation. Experiments on a public WiFi CSI dataset show that the proposed method achieves good performance in both activity recognition and indoor localization.

Wi-Fi sensing provides a privacy-preserving and device-free sensing modality for stationary crowd counting with a low deployment cost. However, labeled channel state information (CSI) data are difficult to obtain at scale, and CSI distributions vary significantly across deployment environments, leading to limited generalization. While semisupervised learning (SSL) has shown promise in Wi-Fi-sensing tasks, existing approaches primarily focus on human activity recognition (HAR) or localization and do not effectively address pseudolabel reliability or cross-domain robustness in crowd counting. To address these challenges, we propose a Wi-Fi-based crowd counting via domain-adversarial semisupervised learning (WiCount-DASL), a domain-adversarial SSL framework that jointly leverages limited labeled data and abundant unlabeled data while aligning feature representations across scenarios. The framework incorporates a classwise, adaptive pseudolabel thresholding mechanism and a targeted signal-level augmentation strategy to improve pseudolabel quality and robustness. Extensive experiments across multiple real-world deployment scenarios demonstrate that WiCount-DASL achieves competitive and robust counting accuracy under limited labeled data compared with representative baselines.

Deep Learning technology has emerged as an exciting innovation in the field of applications in wireless communications systems. In this proposed work, fully connected neural network based on deep learning (DL-FCNN) as well as convolutional neural network (CNN) models were employed to determine the channel state information by estimating channel coefficients using a minimal number of pilot symbols. The findings consider the length of the pilot symbol and the cross-cell interference as the main aspects in the estimation of channel coefficients. Results demonstrate that the proposed deep learning method outperforms traditional approaches such as least square (LS) and minimal mean square estimation (MMSE) techniques for measuring the normalized mean square errors (NMSE), which is thought to be a loss-function or channel estimation error. Additionally, the deep learning (DL) method provides better results in the presence of pilot contamination in multi-cell cellular networks and at various levels of cross-cell interference. In conclusion, deep learning models thereof can efficiently be utilized for channel estimation operations as well as physical layer signal processing applications in Massive MIMO Networks.

Low Earth orbit (LEO) satellite systems provide ubiquitous global connectivity for massive grant-free random access Internet of Things (IoT) applications. Full frequency reuse (FFR) improves spectrum efficiency in spectrum sharing scenarios but introduces severe adjacent beam and cross-system co-channel interference. Meanwhile, the high mobility of LEO satellites hinders accurate instantaneous channel state information (iCSI) acquisition, and random direction-of-arrival (DOA) estimation errors cause statistical CSI (sCSI) mismatch, which degrades beamforming performance and makes it difficult to balance transmission robustness, user fairness, and onboard computational complexity. To address these issues, we propose a low-complexity Hybrid Optimized Robust Beamforming (HORBA) algorithm. We first construct a robust joint optimization model to characterize the coupling effects of DOA errors, outdated CSI, and multi-dimensional interference, with constraints on per-user minimum SINR and cross-system interference temperature. Then, based on the block coordinate descent framework, we decouple the original non-convex problem into two convex subproblems, which are solved via generalized eigenvalue decomposition and first-order Taylor expansion, combined with an adaptive sampling mechanism that balances accuracy and complexity. Simulation results verify that our algorithm outperforms typical benchmarks in sum rate and robustness, maintains low onboard processing complexity, and effectively alleviates edge user rate polarization.

ABSTRACT Beamforming optimisation methods stand as a crucial component for providing higher robustness and efficiency of wireless communication systems in the rapidly evolving landscape of mobile telecommunication. At first, essential data for the validation is collected by varying the sampling points, locations and user counts, and it is selected optimally using an Updated Random Attribute-based Cock-hen-chicken Optimiser (URA-CO). Once the required data for the validation is collected, then the Channel State Information (CSI) is considered for performing beamforming. Here, the Adaptive Stacked Sparse Autoencoder with Spatial Channel Attention (ASSA-SCA) method is employed to execute the beamforming and antenna selection process in the energy-efficient Fifth-Generation (5G) networks. The parameters in the developed ASSA-SCA are tuned using URA-CO that enhances the beamforming vector values and selects the most suitable antenna to transfer the information in the 5G networks. Various analyses are performed in the developed model over the classical techniques to prove its effectiveness.

Purpose This study proposes an early-stage, non-invasive assistive framework as a proof-of-concept for American Sign Language recognition (ASLR) using radio-frequency (RF) sensing and time-series deep learning (DL) techniques. The framework aims to explore the technical feasibility of ASLR as a potential future support mechanism for communication between individuals with hearing impairments and the hearing community. Traditional approaches, such as using camera-based systems or wearable gloves, face limitations in privacy, environmental adaptability, and full-body motion capture. Methods We propose software-defined radio (SDR) sensing using Universal Software Radio Peripheral (USRP) kits to transmit RF signals and capture signal-induced variations in wireless channel state information (WCSI). These dynamic distortions generate unique time-series patterns in the RF data, which were processed and classified using a DL architecture optimised for time-series analysis. Data for 20 American signs were acquired via WCSI using an orthogonal frequency division multiplexing transceiver to measure human sign imprints. Results Comparative analysis of time-series DL algorithms achieved 98% recognition accuracy, without compromising privacy. Key strengths of the proposed conceptual framework include robustness in nonintrusive operation and low-light conditions, as well as interpretability of complex full-body movements. Conclusions The findings suggest potential applicability for individuals with hearing impairments in future communication-support technologies, with possible implications for fostering more inclusive environments. Integrating SDR frequency sensing with advanced DL techniques, this work can provide preliminary evidence to advance ASLR research and lay the groundwork for future scalable, privacy-preserving assistive systems.

Vehicle-to-everything (V2X) communication systems impose stringent latency and reliability requirements that are difficult to satisfy in highly dynamic wireless environments. Although reconfigurable intelligent surfaces (RISs) and unmanned aerial vehicles (UAVs) have independently demonstrated potential in enhancing wireless coverage, most existing RIS–UAV frameworks rely on idealized assumptions such as perfect channel state information (CSI) and static user scenarios. In this paper, a multi-RIS-assisted UAV-enabled V2X communication framework is proposed that explicitly accounts for vehicular mobility, latency constraints, and mobility-induced CSI aging. Multiple RIS panels are cooperatively deployed to eliminate coverage blind spots and ensure link continuity in realistic V2X environments. A joint UAV mobility and RIS phase optimization approach is proposed under outdated CSI to improve link reliability. Additionally, a time-varying performance analysis is carried out for understanding the dynamic behavior of signal-to-noise ratio (SNR) and average bit error rate (ABER) for mobility-aware CSI aging. Simulation results demonstrate that the proposed framework reduces the ABER by approximately 75% compared to a conventional single-RIS system under outdated CSI at 20 dB SNR ( vs. ), while substantially suppressing outage intervals in high-mobility V2X scenarios ( m/s, CSI delay ms), confirming the effectiveness of cooperative multi-RIS assistance for safety-critical vehicular communications.

In ubiquitous Wi-Fi sensing, human motion interval segmentation is crucial for applications ranging from basic intrusion detection to advanced activity understanding. Existing methods often treat the Channel State Information (CSI) primarily as time series, overlooking its rich information in the spatial and frequency domains. To address this, we propose a training-free motion segmentation method that exploits the spatiotemporal features of CSI. We first analyze the discriminative spatial distributions of the CSI Ratio on the complex plane and construct a spatiotemporally dual-constrained local density estimator to characterize motion-induced perturbations. To overcome subcarrier selection challenges, we introduce a packet-level asymmetric truncation-based fusion algorithm, which yields a feature representation with a pronounced bimodal histogram. This enables the automatic determination of the optimal segmentation threshold based on the distribution characteristics of the truncated density image. Experiments in typical indoor environments demonstrate that the proposed method achieves high accuracy in both motion event detection and interval localization.

Orthogonal time-frequency space modulation combined with multiple-input multiple-output transmission (MIMO-OTFS) has emerged as a strong waveform candidate for sixth-generation (6G) wireless networks because of its robustness against high mobility and doubly selective channels. However, reliable signal detection in large-scale MIMO-OTFS systems remains challenging owing to severe delay-Doppler coupling and channel state information (CSI), particularly under Rayleigh and Rician fading conditions. This paper proposes a residual-learning minimum mean square error neural detector (RL-MMSE-ND) to address these challenges under both perfect and imperfect CSI, including scenarios with up to a 20% channel estimation error. The proposed detector integrates a conventional minimum mean square error (MMSE) front-end with a lightweight residual-learning neural network that learns only the residual interference caused by CSI mismatch and delay-Doppler effects, thereby preserving MMSE stability while introducing minimal learning overhead. Extensive simulations were conducted for representative large-scale MIMO-OTFS configurations to evaluate the bit error rate (BER) versus signal-to-noise ratio (SNR), BER versus delay spread, power spectral density (PSD) characteristics, inference latency, and training convergence. Numerical results demonstrate that the proposed RL-MMSE-ND achieves 10-13dB SNR gain at a BER of 10-3 compared to zero-forcing equalization and MMSE detectors, while requiring 3-6dB lower SNR than the maximum likelihood, QR decomposition-based M-algorithm detection, and deep learning-based detectors under both Rayleigh and Rician fading. Moreover, the proposed method reduces inference latency by over 60% compared to long short-term memory and bidirectional long short-term memory detectors and achieves 5-8dB lower out-of-band emissions while adding only linear computational overhead beyond MMSE detection. These results confirmed the novelty and practical significance of the proposed approach for robust, low-latency, and low-complexity MIMO-OTFS detection in future 6G wireless systems.

In recent years, channel state information (CSI)-based sensing technology has gradually attracted widespread attention as a contactless and low-cost approach for robotic arm motion understanding. Despite continuous progress in CSI-based human sensing, existing methods of robotic motion sensing still face two key challenges when directly applied to robotic motion sensing: (1) CSI perturbations induced by robotic arm motion are weak and locally distributed, making fine-grained feature extraction difficult. (2) Discriminative information in long robotic arm motion sequences is sparsely concentrated in a few key intervals, and its adaptive temporal selection and enhancement remain challenging. To address the above challenges, this paper proposes an efficient multi-stage robotic arm motion recognition method (named MSPoolNet). The proposed method consists of three key modules: an adaptive temporal downsampling module, a temporal gating module, and a Transformer-based feature encoding module. Specifically, the adaptive temporal downsampling module processes the raw CSI signal at the input stage to achieve local pattern extraction. The temporal gating module adaptively reweights temporal features, dynamically highlighting key temporal segments while suppressing irrelevant information. The proposed Transformer-based feature encoding module replaces conventional self-attention with pooling operations, enabling global information interaction and fine-grained feature representation in a computationally efficient manner. Extensive experimental results demonstrate that the proposed method achieves state-of-the-art performance on two representative public datasets, maintaining a compact model size with an accuracy exceeding 99%.

Purpose Secure transmission in wireless communication systems has garnered significant attention recently. A key challenge is thwarting eavesdroppers while optimizing the performance of legitimate users and security. This paper aims to propose a novel scheme using index modulation (IM) for a space-time block code (STBC)-aided simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) system to ensure physical layer security (PLS). Design/methodology/approach STAR-RIS is designed as an access point (AP) to adjust array element phases for signal generation. By grouping the STAR-RIS and using IM, different antenna combinations are activated for transmission. The zero-forcing (ZF) precoding method generates artificial noise (AN) based on the channel state information (CSI) of the legitimate user, which is then added to the transmitted signals. Finally, Alamouti coding is integrated into the STAR-RIS for signal transmission. Findings Theoretical and simulation analyses of bit error rate (BER) and secrecy rate show the scheme prevents eavesdropping and enhances the secrecy rate. At a 30dB signal-to-noise ratio (SNR), the secrecy rate increases by approximately 8 bits/s/Hz compared to the no-AN scheme. Furthermore, compared to traditional STAR-RIS, at BER = 10−4, the SNR gain is approximately 2dB. Research limitations/implications The good results are based on CSI in the communication system. Practical implications It could be used for V2V communication networking and secured communications. Social implications It gives more ideas and research on the coding in the RIS. Originality/value Designing STAR-RIS as an AP simplifies channel links and, combined with Alamouti code, greatly boosts diversity gain via spatial-temporal coding. Moreover, this paper uses the STAR-RIS to implement IM, enabling the additional transmission of index bits and improving spectral efficiency. Meanwhile, adding AN to signals, a key design aspect, prevents eavesdropping and enhances security, showing a holistic approach to optimizing wireless security.

Wi-Fi Channel State Information (CSI)-based indoor localization enables high-precision positioning, but its deployment across multiple environments faces two major challenges: privacy concerns from centralizing CSI data, and severe statistical heterogeneity (non-IID) arising from the strong environment-dependency of CSI. This heterogeneity creates a stability–plasticity trade-off in federated learning—maintaining precision in known environments (stability) while adapting to unseen domains (plasticity). To address this trade-off, we propose AdaFed-LDR, which combines server-side Confidence-Weighted Adaptive Aggregation with client-side Layerwise Dynamics Regularization (LDR). The aggregation recalibrates client contributions based on feature covariance changes, while LDR imposes depth-dependent constraints—stronger constraints on shallow layers to preserve environment-agnostic features and weaker constraints on deeper layers to allow environment-specific adaptation. Evaluated across 8 indoor environments using Leave-One-Out Cross-Validation and 5 random seeds, AdaFed-LDR achieved a mean localization error (MLE) of 0.41 cm in known environments, corresponding to an 88.2% reduction compared with FedAvg. In domain generalization to unseen environments, AdaFed-LDR achieved an MLE of cm, demonstrating an improvement over FedPos ( cm). With one adaptation sample per reference point, MLE improved to 21 cm. Ablation experiments confirmed that combining the two proposed components achieved the highest improvement (83.9%) compared with applying them individually, supporting AdaFed-LDR as a reproducible approach to the stability–plasticity trade-off in federated CSI-based localization.

In next-generation wireless communication systems, spectrum efficiency can be realized through the integration of hybrid beamforming (HBF) and non-orthogonal multiple access (NOMA). To maximize the synergy between these two technologies, it is essential to accurately cluster users within beams. Most existing studies on clustering overlook practical constraints and assume perfect channel state information (CSI). However, obtaining full CSI is impractical in realistic environments due to high feedback overhead and potential CSI errors. To address these challenges, this paper adopts an opportunistic beamforming (OBF) framework based on a partial CSI environment. The OBF facilitates channel estimation and HBF precoder design using only signal-to-interference-plus-noise ratio (SINR) feedback. Subsequently, clustering and power allocation (PA) are performed utilizing the feedback SINR from OBF without requiring additional feedback information. While conventional NOMA focuses on maximizing either throughput or fairness, this paper proposes a scheme that selects users with high SINR to maximize system throughput while minimizing the throughput disparity among users to enhance fairness. Furthermore, a power allocation method that satisfies the minimum successive interference cancellation (SIC) power requirement is employed to ensure stable decoding. Simulation results demonstrate that the proposed clustering scheme enhances the sum-rate compared to conventional SINR-based clustering methods while maintaining fairness. Consequently, this study suggests a promising approach to improving NOMA performance in practical partial CSI environments.

Nowadays, farmers across the globe are gradually adopting intelligent farming, which is facilitated by a variety of cutting-edge technologies. The advancement of intelligent farming applications is greatly aided by the internet of farming things (IoFT). Massive IoFT devices generally possess constrained resources, making it challenging to meet the battery and computational requirements of intelligent farming applications through local computation. RF energy harvesting enabled mobile edge computing (RFE-MEC) addresses this issue by harvesting RF energy from an access point, offloading and computing tasks at the edge in a nearby access point. In the proposed scheme, multiuser nonorthogonal multiple access allows the IoFT devices to simultaneously offload computationally intensive tasks to the MEC server for processing. The delay outage probability closed-form expression is formulated for the RFE-NOMA-MEC intelligent farming system under a Rayleigh fading channel. The impact of imperfect channel state information on the RFE-NOMA-MEC is considered. Tunicate enhanced northern goshawk optimization algorithm (TNGO) has been proposed to discover the optimal parameter set to minimize delay outage probability. The results indicate that the system performance is enhanced using TNGO when the optimal time switching factor, power allocation coefficient and task allocation ratio are utilized.

TH-DRL is proposed for optimizing spectral and energy efficiency in 6G MIMO-MC-CDMA systems. Unlike existing flat DRL or optimization-based approaches, the proposed TH-DRL framework uniquely integrates a lightweight Transformer encoder with a hierarchical decision-making architecture to jointly optimize subcarrier allocation, power control, and SIC ordering. Architecturally, the structure integrates a Transformer encoder with a two-tier hierarchical DRL model to enhance adaptability in dynamic wireless conditions. The Transformer learns spatiotemporal dependencies from channel state information, interference patterns, and user dynamics to generate context-aware features that is used for making effective decisions. A high-level policy-gradient agent handles subcarrier allocation and user clustering, while a low-level DQN agent manages power control and successive interference cancellation order, jointly improving throughput and energy efficiency. Convergence, scalability, and detection performance are evaluated based on Rayleigh channels at 28/100GHz with bandwidths of 400MHz-1GHz serving a varying number of users (10-100) served by a base station equipped with 64-256 antennas. Training consists of a replay buffer of samples within a range of 104-10⁶ over 5000-10,000 episodes. The results showed that the convergence is stable around episode 600 with consistent gain over the baseline methods achieving 15-18% higher spectral efficiency, up to 22% energy savings, and peak performance at SE = 32.7 bits/s/Hz, EE = 14.8 bits/J, SINR ≈ 34dB, and BER ≈ 10⁻5. This confirms that Transformer-enhanced hierarchical DRL offers scalable, low-latency, energy-aware resource management for dense 6G networks.

Accurate indoor positioning is vital for applications such as augmented reality and autonomous robotics. Channel state information (CSI)-based methods, particularly when combined with beamforming, massive multiple input multiple output (mMIMO) techniques, and artificial intelligence (AI) algorithms, offer enhanced indoor user equipment (UE) positioning accuracy and robustness in complex indoor environments. In this paper, we present an AI-driven CSI-based indoor positioning method for mMIMO systems, where channel impulse, channel frequency, and angular response domain features are extracted from the CSI data and combined to form both uni-domain and multi-domain feature sets. We introduce a deep attention network (DAN), an AI algorithm that leverages attention mechanisms to effectively integrate and process multi-domain CSI data, thereby enhancing UE positioning performance. We evaluate DAN using a publicly available mMIMO dataset and compare its performance against the baseline and multi-domain convolutional neural network (CNN) models. Our results show that multi-domain DAN outperforms CNN approaches in positioning accuracy, though at the cost of increased inference complexity-highlighting a trade-off between performance and computational overhead. These findings demonstrate the potential of attention mechanisms and multi-domain CSI features for accurate indoor UE positioning systems.

The research paper evaluates how an Intelligent Reflecting Surface (IRS)-assisted Orthogonal Frequency Division Multiplexing (OFDM) system performs when transmitting over frequency-selective fading channels with adaptive per-subcarrier phase-shift optimization. The receiver uses an adaptive phase-shift optimization algorithm, enabling it to combine reflected and directly received signals. The system evaluation was conducted via Monte Carlo simulations that tested 64 subcarriers with 32 IRS elements, using QPSK modulation and 4-tap frequency-selective channels. The system performance metrics are measured through Bit Error Rate (BER) and spectral efficiency tests, which operate under imperfect Channel State Information (CSI) and Additive White Gaussian Noise (AWGN). The simulation results show that the proposed optimized IRS configuration achieves better performance than both conventional OFDM systems without IRS and IRS-assisted systems that use random phase shifts. At 20 dB SNR, the proposed optimized IRS configuration achieves a BER of 0.008, which is 16 times lower than the random IRS configuration and 21 times lower than conventional OFDM. At 25 dB SNR, the spectral efficiency reaches approximately 7.0 bps/Hz, representing a gain of 2.5 bps/Hz over the random IRS configuration and 2.8 bps/Hz over conventional OFDM. The implementation of adaptive per-subcarrier IRS phase optimization in broadband multicarrier communication systems enables the significant improvements in reliability and capacity enhancements for upcoming wireless networks.

This paper introduces a unified framework for activity detection in IRS-assisted IoT networks that simultaneously addresses three critical practical challenges: asynchronous transmissions, heterogeneous power levels used by devices to report their local observations, and signal blockage in dynamic environments. The system leverages an intelligent reflecting surface (IRS) to enhance detection reliability, with optional incorporation of a direct line-of-sight (LoS) path. Departing from conventional approaches that handle these challenges in isolation, we formulate a comprehensive detection problem as a binary hypothesis test and develop a structured hierarchy of four detectors: an optimal detector alongside three computationally efficient detectors designed for practical scenarios with different levels of prior knowledge about noise variance, channel state information, and device transmit powers. This hierarchical approach systematically bridges the gap between theoretical optimality and implementation practicality. For each detector, we derive closed-form expressions for both detection and false alarm probabilities, establishing theoretical performance benchmarks and revealing fundamental scaling laws. Extensive simulations validate our analytical results and extract actionable design guidelines by systematically evaluating the impact of key system parameters including the number of antennas, samples, users, and IRS elements on detection performance, providing valuable insights for 6G IoT system designers. The proposed framework effectively bridges theoretical optimality with implementation practicality, providing a scalable solution for IRS-assisted IoT networks in emerging 6G systems.

OT-ADG: Optimal-transport-driven adaptive data generation for WiFi sensing in airborne mobile networks

Massive Multiple-Input Multiple-Output (Massive MIMO) technology, a fundamental part of the fifth-generation (5G) wireless networks, and a component of next-generation (6G) networks, provides unprecedented spectral efficiency and space multiplexing advantages in large-scale antenna arrays. The achievement of such gains is basically conditional to the correct channel state information (CSI) measurements. The paper provides a systematical and exhaustive overview of channel estimation algorithms of Massive MIMO systems including Least Squares (LS), Minimum Mean Squared Error (MMSE), compressed sensing (CS) algorithms with channel sparsity, angle-domain and two-stage hybrid algorithms of frequency-division duplex (FDD) systems, and deep learning (DL)-based estimators such as convolutional neural networks (C The techniques are evaluated relative to normalized mean squared error (NMSE), computational complexity, pilot overhead, pilot contamination robustness and deployability. The open literature is used to consolidate quantitative comparisons. The problem of open research such as near-field estimation of XL-MIMO, RIS-assisted channel, grant-free access, and federated learning are outlined and critically examined.