ABSTRACT In this paper, the localisation of an unmanned aerial vehicle (UAV) in a cooperative manner in global positioning systems (GPS)‐denied environments using multiple base stations and multiple integrated reflecting surfaces (IRSs) is performed. A two‐stage scheme is proposed to localise the UAV. In the first stage, a phase gain matrix is used for IRSs, and a symbol is transmitted by the base transceiver station (BTS), and the first received symbol is observed by the UAV. In the second stage, the inverse phase gains are set for IRSs, and again the same symbol is transmitted by BTSs, and a second received symbol is observed by the UAV. Using only these two received symbols, the distances between the UAV and the BTSs and the distances between the UAV and the IRSs are estimated in the two different versions of the proposed algorithm, and then the location of the UAV is extracted via trilateration. Some distance estimation error analysis is also provided in the paper. Simulation results show the efficiency of the proposed scheme in terms of one order of magnitude faster run time of the BTS version of the proposed algorithm, whilst it is slightly less accurate than a state‐of‐the‐art algorithm in the literature.
ABSTRACT In multi‐base integrated sensing and communication (ISAC) scenarios, weak signals, multipath fading and interference severely degrade radar signal reconstruction and waveform design, posing significant challenges to simultaneously achieving reliable communication and high‐accuracy sensing. To address these challenges, we propose a multi‐objective genetic algorithm‐based ISAC waveform design that jointly optimizes the peak‐to‐average power ratio (PAPR) and the peak sidelobe ratio (PSLR). To enhance the robustness of ISAC signal reconstruction under adverse channel conditions, we exploit the error‐correction capability of forward error correction (FEC) in orthogonal frequency division multiplexing (OFDM) communication systems, which enables reliable recovery of corrupted subcarriers. Building upon this insight, we design an OFDM‐ISAC waveform employing reserved subcarriers and FEC coding, and formulate a joint PAPR‐PSLR optimization model under Bose–Chaudhuri–Hocquenghem coding constraints. Owing to the inherent nonlinearity and nonconvexity of the resulting optimization problem, we further develop a two‐stage genetic algorithm‐based optimization framework to jointly optimize PAPR and PSLR. The proposed approach preserves the QAM characteristics of OFDM waveforms, enabling accurate signal reconstruction at the receiver even in the presence of interference, thereby improving sensing performance. Extensive simulation results validate the effectiveness of the proposed method.
ABSTRACT Accurate channel estimation is essential for robust vehicular communication, particularly under high‐mobility conditions where traditional methods like least squares, minimum mean square error, data pilot‐aided and orthogonal matching pursuit often fall short due to error propagation and computational inefficiencies. This paper presents an enhanced gated recurrent unit (GRU)‐based deep learning framework that incorporates attention‐based feature selection and adaptive noise filtering to improve channel estimation performance in IEEE 802.11p‐based vehicular networks. The proposed model effectively captures temporal and spatial dependencies in dynamic wireless environments without requiring prior channel knowledge. Simulation results show an average normalized mean squared error (NMSE) reduction of approximately 80.6% across diverse scenarios compared to conventional methods, demonstrating its superiority in both accuracy and computational efficiency. Notably, the model achieves approximately dB NMSE during epoch‐based convergence and maintains high‐resolution prediction quality, as validated through visual heatmaps and binned confusion‐matrix analysis. These findings position the proposed GRU‐based estimator as a scalable, real‐time solution for next‐generation vehicular communication systems.
ABSTRACT Time‐evolving traffic flow forecasting is playing a vital role in intelligent transportation systems and smart cities. However, the dynamic traffic flow forecasting is a highly nonlinear problem with complex temporal‐spatial dependencies. Although the existing methods have provided great contributions to mine the temporal‐spatial patterns in the complex traffic networks, they fail to encode the globally temporal‐spatial patterns and are prone to overfitting on the pre‐defined geographical correlations, and thus hinder the model's robustness in the complex traffic environment. To tackle this issue, in this work, we proposed TSFusion , a multi‐grained temporal‐spatial graph learning framework to adaptively augment the globally temporal‐spatial patterns obtained from a crafted graph transformer encoder with the local patterns from the graph convolution by a crafted gated fusion unit with residual connection techniques. Under these circumstances, our proposed model can mine the hidden global temporal‐spatial relations between each monitor station and balance the relative importance of local and global temporal‐spatial patterns. Experiment results demonstrate the strong representation capability of our proposed method, and our model consistently outperforms (more than 11.5% for MAE) other strong baselines on various real‐world traffic networks.
In this paper, a robust direction-of-arrival (DOA) estimation method based on the swin transformer (DOA-SwinT) is proposed for estimating the DOAs of narrowband coherent sources using a uniform linear array under challenging conditions, such as low signal-to-noise ratio (SNR) and a limited number of snapshots. The swin transformer (SwinT) achieves high efficiency and accuracy in classification tasks owing to its low computational complexity and high performance, which is enabled by its window attention mechanism and hierarchical architecture. The proposed DOA-SwinT fuses both low-level and high-level features extracted from array signal data, thereby improving DOA estimation accuracy. Furthermore, DOA-SwinT can handle coherent signals and exhibit good DOA estimation performance. Simulation results demonstrate that the proposed method achieves substantially higher estimation accuracy than existing deep-learning-based methods and traditional high-resolution methods for coherent signals under equivalent SNR and snapshot conditions.
The Industrial Internet's expansion intensifies the need for low-latency and high-reliability end-to-end deterministic networks. While TSN-5G integration offers a promising path, complex delay-bound prediction and inherent stochastic fluctuations remain key bottlenecks, leading to total delay jitter and reliability degradation. To address this challenge, we propose an "SNC-DNC fusion framework" employing a stochastic-probabilistic identity deterministic cross-domain mapping (SPI-DCM) mechanism that unifies stochastic and deterministic network calculus for cross-domain latency modelling. Specifically, its core innovation-a priority- and burst-sensitive mechanism-adaptively converts SNC to DNC parameters for periodic, aperiodic, and bursty flows while maintaining probabilistic accuracy, complemented by a delay-exponential decay elastic service curve, capturing 5G's burst-load dynamics. Simulations show that SPI-DCM can tighten the delay bound by up to approximately 40% compared with the improved MGF-SNC and WCD methods under identical constraints, while preserving reliability and analytical scalability, demonstrating its effectiveness and practical value.
ABSTRACT Accurately identifying jammer types is critical for reliable communication, but current methods fail in real‐world environments due to signal coexistence and the open‐set recognition challenge. Existing approaches primarily focus on closed‐set scenarios, limiting robustness against unknown interference. We propose a solution centred on the novel IQSpectro‐TransRes Joint Network and a specialised ArcCider Loss. Our bi‐domain feature fusion employs a Transformer branch for I/Q sequence data and a ResNet branch for short‐time Fourier transform spectrograms, ensuring robust feature extraction. Crucially, the ArcCider Loss explicitly enforces feature compactness and dispersion, providing a mechanism for open‐set recognition. Experimental validation on a complex compound jamming dataset demonstrates the superiority of our approach. The Joint Network achieves a high closed‐set accuracy of 98.22%. More significantly, the ArcCider Loss scheme outperforms traditional methods and recent state‐of‐the‐art OSR frameworks, achieving both high closed‐set performance and effective identification of unknown interference. This provides a resilient and practical framework for jamming signal recognition in dynamic, open‐world environments.
ABSTRACT With the rapid expansion of cyberspace and the increasing complexity of cyber threats, massive volumes of heterogeneous security data are generated every day. Effectively extracting structured information from such data has become a critical challenge in cybersecurity and information extraction. Cybersecurity event extraction, as a critical task in cybersecurity and information extraction, aims to automatically identify and structure key elements of security events, thereby supporting threat intelligence analysis and enhancing proactive defence capabilities. However, despite the growing importance of this area, there is still a lack of systematic surveys of existing cybersecurity event extraction methods. To address this gap, this paper presents a thorough review of cybersecurity event extraction techniques, covering their evolution from rule‐based methods to traditional machine learning models, and further to recent advances in deep learning and large language model (LLM)‐based paradigms. Representative methods are categorised and analysed in terms of their underlying principles, strengths and limitations. Finally, this survey summarises the key challenges faced in cybersecurity event extraction and discusses promising future research directions, aiming to provide valuable insights for researchers and practitioners in this field.
ABSTRACT Integrated sensing and communication (ISAC) is a key enabling technology for 6G wireless networks. This paper presents a unified waveform allocation framework for downlink (DL) and uplink (UL) MIMO–OFDM ISAC systems. We derive complete signal models for frequency‐selective channels and formulate per‐subcarrier spectral efficiency (SE), sensing rate (SR), mutual information (MI) and minimum mean squared error (MMSE) expressions. Five waveform allocation schemes are investigated: optimal‐for‐communication (OPC), optimal‐for‐sensing (OPS), MMSE‐based joint allocation, equal allocation and random allocation. Analytical expressions for DL/UL SE, SR and weighted MI are provided. A projected‐gradient MMSE optimization algorithm achieves locally optimal waveform distributions under total power constraints. Simulation results with 3GPP channel models demonstrate that the MMSE scheme achieves superior sensing‐communication balance, OPC maximizes SE and OPS maximizes SR. Computational complexity and convergence analysis confirm the framework's practical applicability for 6G ISAC systems. The proposed framework provides a complete resource‐allocation toolkit for practical deployment.
ABSTRACT Millimetre Wave (mmWave) Cell‐free massive Multiple‐Input Multiple‐Output (CF‐mMIMO) is a key architecture for beyond 5G (B5G) networks; nevertheless, its energy efficiency (EE) remains constrained by dense access points (APs) deployments, rapidly varying channels, and the strict low‐latency requirements for real‐time decisions. Classical optimisation techniques and Machine Learning (ML) or existing Deep Reinforcement Learning (DRL)‐based solutions exhibit limited scalability, rely on centralised global CSI, and insufficiently capture dynamic User‐Centric (UC) topology and mmWave channel variations. To address these limitations, we propose a joint uplink power control and topology optimisation framework. It leverages a multi‐agent Deep Deterministic Twin Policy Gradient (MATD3) algorithm combined with an adaptive AP‐User Equipment (UE) association mechanism termed UC Dynamic Proximity and demand (UCDPD). Within a centralised training with decentralised execution (CTDE) paradigm to ensure fast convergence and stability, UCDPD dynamically forms adaptive UC clusters to reduce state dimensionality. Simulation results demonstrate that the proposed method outperforms standalone MATD3, baselines’ single and multi‐agent DRL and traditional techniques. It achieves up to 495 Mbit/joule in EE and average spectral efficiency (SE) above 17 bps/Hz. This represents significant improvement of about 10–20%, and 32–60% compared to reference multi‐agent and single‐agent schemes, respectively. The proposed framework remains robust across network densities, under transmit power levels (15–100 mW) and with different learning rates. These findings underscore the scalability and practicality of the proposed algorithm for next‐generation self‐optimising wireless communication s .
This paper proposes a novel weighted -nearest neighbour algorithm for indoor positioning of continuously moving users based on soft range limitation and dynamic threshold. By introducing a position penalty function, reference points closer to the previous position are given higher weights resulting in a soft range limit. Meanwhile, a method for global fingerprint library simplification is proposed, and a dynamic threshold is introduced to avoid local extreme value phenomena. In addition, an adaptive parameter is adopted to maximize the positioning capability of the algorithm. Through numerical simulations and physical experiments, the results demonstrate that the proposed algorithm significantly outperforms existing five baseline positioning algorithms in terms of accuracy. Compared with the traditional -nearest neighbour algorithm, its performance in simulation and experimental tests is improved by and , respectively.
Efficient placement of drone base stations (DBSs) in Internet of Things (IoT) networks plays a vital role in enhancing coverage and minimising signal degradation, particularly in complex urban environments. This paper presents a multi-objective optimisation framework based on the Strength Pareto Evolutionary Algorithm 2 (SPEA2) to simultaneously maximise user coverage and minimise average path loss. The proposed model integrates realistic 3D urban propagation characteristics and supports flexible deployment across multiple urban environments-urban, dense urban, high-rise, and suburban. Furthermore, this study leverages SPEA2 to derive optimal three-dimensional (3D) DBS placements by exploiting its strong non-dominated sorting and density estimation mechanisms, ensuring robust handling of the conflicting objectives of path-loss minimisation and coverage maximisation. A comprehensive simulation analysis is conducted, including sensitivity evaluation of key parameters such as user density and threshold signal levels. Results reveal a clear trade-off between coverage and path loss and demonstrate the model's capacity to generate Pareto-optimal DBS configurations that suit diverse application scenarios, including emergency response and energy-constrained deployments. Simulation results show that the optimised SPEA2-based configuration achieves a minimum average path loss of 86.9 dB with 100 search agents in urban scenarios and provides up to 100% coverage when deploying four or more DBSs at a threshold level of 110 dB. These quantitative results demonstrate the explicit contribution of SPEA2 in enhancing DBS deployment efficiency and resolving the trade-offs inherent in practical IoT environments.
Reconfigurable intelligent surface (RIS) assisted symbiotic radio communication (SRC) networks are considered a key solution to address the challenges of massive Internet of Things connectivity. There, the resource optimization problems are always NP-hard and formulated with time varying parameters. Learning via deep neural network has proven to be an effective way to encapsulate the optimized solution as a function of the optimization problem. However, the learning framework is challenged by meeting complex communication constraints involved. To handle this issue and accommodate the time varying SRC environment, this work proposes a novel deep reinforcement learning framework to solve the classical passive beamforming optimization problem. System sum-rate is maximized while ensuring strict communication requirements. The formulated constrained Markov decision process is transformed into a Lagrange dual problem. The deep deterministic policy gradient framework is adopted introducing a monitor module to approximate Lagrange multipliers. The actor and monitor networks are trained using the derived Lagrangian-related gradients, ensuring constraint satisfaction during the training. Simulation results demonstrate that the proposed method achieves the performance upper bound, significantly improving RIS transmission rate and system sum-rate compared to the counterparts. It also demonstrates adaptability to environmental dynamics, computational efficiency, reduced transmission resource usage, and lower probability of violating communication requirements.
Disasters often lead to severe disruptions of terrestrial backbone networks, resulting in information isolation in affected areas. In such situations, very small aperture terminals (VSATs) can be rapidly deployed to establish terrestrial-satellite communication links, enabling the transmission of large volumes of situational awareness data, such as high-definition videos and three-dimensional maps. However, the high mobility of low earth orbit (LEO) satellites necessitates frequent satellite switching to maintain service continuity. This challenge is further exacerbated in emergency scenarios, where the timely and reliable transmission of post-disaster data is of critical importance, while unpredictable disturbances can severely degrade the stability of ground-satellite backhaul links through abrupt channel variations and resource fluctuations. Consequently, robust and efficient satellite selection is crucial for sustaining reliable and high-throughput satellite backhaul. To address these challenges, this paper proposes a disturbance-aware deep reinforcement learning (DRL) framework for adaptive satellite selection, aiming to support robust and high-capacity terrestrial-satellite backhaul for VSATs in disaster scenarios. Specifically, the proposed approach jointly accounts for link quality, handover (HO) frequency, and connection failures in the decision-making process. Moreover, the robustness and stability of the DRL model is improved by incorporating the stochastic disturbances encountered in LEO satellite networks into the training stage. These disturbances comprise weather-induced carrier-to-noise ratio (CNR) degradation and reductions in available satellite bandwidth caused by user contention. Simulation results demonstrate that the proposed approach achieves robust and stable performance under dynamic conditions, outperforming existing methods.
In 5G-advanced and 6G networks, integrated sensing and communication (ISAC) is pivotal. However, in complex urban environments, non-line-of-sight (NLOS) propagation often degrades the sensing and communication performance of ISAC system. Accordingly, we propose a UAV-assisted ISAC system where the unmanned aerial vehicle (UAV) acts as both a sensing platform and a relay. This study derives closed-form expressions for outage probability, communication information rate, and sensing information rate, validated via Monte Carlo simulations. We analyse the impact of UAV deployment on the weighted sum information rate and introduce the chaotic particle swarm optimization (CPSO) algorithm to optimize the UAV's position for maximizing this metric. Simulations reveal that system performance is highly sensitive to UAV placement, and adjusting weight coefficients enables an effective trade-off between communication and sensing. The CPSO algorithm demonstrates superior stability and global search capability over standard PSO, avoiding premature convergence and enhancing overall efficiency.
The need for secure information exchange is increasingly felt with an exponential rise in the number of small devices often associated with critical processes, which frequently share sensitive data. Recently, 5G networks provided integrated platforms for the connectivity of an enormous number of small devices in the form of Internet of Things (IoT), across diversified and heterogeneous networks. The growth in connectivity is also spreading the threat surface exponentially, requiring tailored security approaches suited to the specific needs of these devices. Compressive sensing (CS) has proven to be an appropriate choice for the security needs of small devices. This research proposes the construction of a single combined measurement matrix built from two discrete chaotic sequences for enhanced randomness in the data. Additionally, a separate chaotic system is used in a novel way to construct orthogonal matrices. The initial values and other parameters of the chaotic systems are calibrated to build considerable diffusion and confusion in the data, thereby enhancing security. Besides, these parameters also serve as keys enabling layered security. The combination of multi-chaotic systems further reinforces the encryption process with enriched coding and heightening security. Both theoretical analysis and simulation suggest that the proposed scheme has significantly improved the overall security, which in particular makes it resilient against known statistical, brute-force and differential attacks. Moreover, the use of CS also significantly improves the transmission efficiency. The proposed scheme is effective in both security enhancement and improving efficiency, particularly for small resource-constrained devices (i.e., IoTs, D2D and sensor networks).
In this paper, an untrusted bidirectional amplify and forward relay system is considered through the use of dual IRSs to facilitate the secure bidirectional information transmission between users U1 and U2. The task of maximizing the overall secrecy rate for two-way communications system is formulated by jointly designing the phase shifts of both IRSs along with transmit power of U1/U2 and the untrusted relay R, which is generally challenging. To facilitate a better grasp of the problem, an effective alternating optimization and semi-definite relaxation-based iterative algorithm is designed to solve the resulting non-convex problem. Experimental results illustrate that the presented algorithm is both effective and flexible in improving the system security performance, and achieves significant performance gains over three baseline methods.
Federated learning (FL) is a prominent paradigm in contemporary machine learning, renowned for its capability to aggregate efficient and accurate models without directly collecting user data. However, FL faces numerous security challenges, including potential privacy disclosure risks as well as poisoning and free-rider attacks perpetrated by malicious users. To address these challenges, we propose a dual-encryption and reputation evaluation defense mechanism on FL called DRFL to defend against malicious attacks. Specifically, we first design a dual-encryption strategy to achieve secure aggregation while protecting user privacy. Subsequently, we construct a secure reputation evaluation mechanism within the ciphertext environment to evaluate users' local ciphertext gradients, thereby defending against poisoning and free-rider attacks. Simultaneously, blockchain technology is integrated to document the entire FL process, enhancing transparency and openness in the training procedure. We further provide security analysis to justify the privacy guarantees of DRFL, and conduct extensive experiments on three real-world datasets. Experimental results demonstrate that DRFL effectively defends against poisoning and free-rider attacks and improves model utility by 3%-4% over Median, Trimmed-Mean and Krum, while achieving accuracy comparable to the benign FedAvg baseline.
Millimeter-wave (mmWave) Internet-of-Vehicles (IoV) communications rely on large antenna arrays to combat severe path loss and to provide high beamforming gains. However, practical vehicular deployments often adopt analog or hybrid beamforming with constant-modulus constraints, leading to a highly non-convex joint beamforming design problem whose iterative solvers incur prohibitive latency under fast channel variations. To enable low-latency beamforming for dynamic IoV scenarios, we propose an end-to-end deep learning framework that directly maps instantaneous channel realizations to the transmit and receive beamformers. Three representative backbones are investigated, including a multilayer perceptron (MLP), a DCGAN-style convolutional encoder, and a U-Net encoder-decoder with multi-scale feature extraction. Moreover, beyond conventional supervised imitation learning, we emphasize rate-driven learning and develop training strategies that directly optimize the achievable-rate objective. Extensive simulations demonstrate that the proposed rate-driven learning substantially improves performance over the previous two-stage scheme, with the U-Net backbone consistently providing the best accuracy. Under the premise that the learned solutions achieve over 95% of the optimization-based benchmark on average, the trained networks provide an approximately 800- inference-time speedup compared with the iterative based method, highlighting their computational efficiency and practical potential for latency-sensitive mmWave IoV systems.
Efficient data-detection techniques for massive multiple-input multiple-output (mMIMO) systems with low complexity remain a major research focus, aiming to achieve high performance without excessive computational cost. Therefore, this article introduces signed quadrature integrated detection (SQUID), a framework that integrates spatial modulation (SM) with a compressive sensing (CS)-based approach. By relying on the sparse property of signed quadrature SM, CS algorithms enable the detection with lower computational cost than the optimum method, while maintaining strong performance. Numerical results demonstrate that SQUID achieves superior performance compared to traditional methods while further reducing computational complexity. The proposed method is thoroughly evaluated under both perfect and imperfect channel state information conditions, and in each case, the SQUID method delivers robust and accurate detection, confirming its suitability for realistic mMIMO deployments.
