This paper presents a multi-task spatiotemporal deep learning framework designed for zero-shot geographic generalization in predictive channel modeling for low Earth orbit (LEO) satellite communications at Q-band (39 GHz). Unlike many existing methods that rely on location-specific training, the coordinate-based long short-term memory (LSTM) architecture processes historical sequences of atmospheric and geographic data spanning 60 min to simultaneously predict weather conditions and excess path loss (EPL) for 5 h, incorporating inherent Gaussian uncertainty quantification. The proposed framework is designed to improve geographic generalization, which remains a challenge for existing machine learning approaches in satellite communications. A comprehensive evaluation conducted across various European climates demonstrates exceptional dual performance, achieving a root mean square error (RMSE) of 0.026 dB at the training locations and an average error of 0.213 dB across ten entirely unseen European cities, demonstrating strong generalization and representing a substantial improvement over existing methods. The framework maintains consistent prediction performance across all untrained European cities, while the uncertainty estimation mechanism provides calibrated confidence measures that may support operational decision-making. Furthermore, the coordinate-based inference strategy enables prediction at previously unseen locations without location-specific retraining, making the proposed framework a promising approach for large-scale predictive link adaptation in future LEO satellite networks. Although the proposed framework achieved promising results on ITU-R model-generated datasets, further validation using independent real-world satellite link measurements is needed to confirm its practical applicability under operational conditions.
The rapid growth of cognitive Internet of Vehicles (IoV) networks requires efficient spectrum sharing mechanisms to support dynamic vehicular communications under limited spectrum resources. UAV-assisted cognitive relaying provides flexible coverage enhancement for vehicular networks. However, highly dynamic air-to-ground channels, vehicle mobility, and primary user (PU) protection constraints make long-term resource scheduling highly challenging. Conventional optimization methods and reactive reinforcement learning approaches often suffer from limited adaptability under dynamic network conditions due to their short-term decision mechanisms. To address these challenges, this paper proposes a predictive resource scheduling framework based on latent-state world modeling for UAV-assisted cognitive IoV networks. Specifically, the high-dimensional channel and vehicular states are mapped into a compact latent representation to capture temporal dependencies between channel evolution and queue dynamics. Based on the learned latent states, a VoI-aware predictive scheduling strategy is developed to evaluate the long-term effects of scheduling actions on information freshness and buffer stability through multi-step latent-state prediction. The proposed framework improves scheduling robustness under incomplete channel state information (CSI) and dynamic wireless environments. Simulation results under different signal-to-noise ratio (SNR) conditions and traffic loads demonstrate that the proposed framework achieves improved VoI-oriented scheduling performance and spectrum utilization compared with conventional model-free reinforcement learning baselines such as TD3 and PPO. These results confirm the effectiveness of integrating latent-state prediction into proactive spectrum resource scheduling for cognitive vehicular networks.
In the current artificial intelligence era, mobile network channels play a crucial role in the development of networks with 4G, 5G, and next generation 6G. This research focuses on intellectual security architecture (ISA) to mitigate web-oriented attacks that use hybrid deep learning models to stall the operation of mobile networks. With the rapid evolution of mobile communication systems, especially in the context of edge-based content sharing, network security has become a crucial aspect of maintaining robust and reliable service. The integration of artificial intelligence (AI) into security frameworks has opened new frontiers in identifying and mitigating cyber threats such as web-based attacks. This research suggests a deep learning system that uses lightweight bidirectional long short-term memory (Bi-LSTM) networks and transformer-based multi-head attention (MHA) to smartly detect web attacks. Apart from dataset accuracy, the proposed ISA is assessed for edge feasibility through testing. The ISA architecture requires 30,087 parameters and 0.42 MB of storage with a medium inference latency under 90 ms, thereby supporting on device deployment on limited storage 6G edge node. The proposed model is evaluated on a labeled web traffic dataset and trained using an additional zero shot, cross-dataset evaluation on independent edge IIoT dataset. The proposed model preserves effective attack/normal separability AUC of 0.93
The rapid emergence of the power low-altitude economy, characterized by wide-area drone-assisted grid inspections and dynamic aerial dispatching, demands microsecond-level determinism and ultra-high reliability across complex air-ground communication networks. However, traditional time-sensitive networking (TSN) suffers from severe scalability limitations over extensive distances, while standard deterministic networking (DetNet) lacks the autonomous adaptability required to manage highly dynamic traffic and stochastic topological perturbations during low-altitude flight operations. To address these bottlenecks, this paper proposes an agentic artificial intelligence (AI)-driven, high-reliability deterministic Internet Protocol version 6 Plus (IPv6+) routing architecture tailored to power low-altitude-economy networks. The architecture combines explicit segment routing over IPv6 (SRv6) with agentic AI-enabled autonomous orchestration in an intelligent closed loop. Specifically, we formalize an intent translation function that maps high-level, multi-dimensional low-altitude dispatching requirements into segment identifier (SID) lists and uses periodic mapping to obtain an approximately linear hop count trend for the variable processing, scheduling, and queuing component of delay in the evaluated setting. To support continuous operation after unpredictable physical link disruptions, an agentic AI module performs real-time state prediction and autonomous path re-optimization alongside an intent-driven packet replication and elimination function (PREF). The evaluation compares intent-driven large-scale deterministic networking (Intent-LDN) with representative SRv6–DetNet hybrid and graph deep-reinforcement-learning approaches and examines the contributions of the AI controller, PREF, flexible ethernet (FlexE) isolation, and topology-independent loop-free alternate (TI-LFA) recovery. In the modeled 2000 km jitter scenario, Intent-LDN records a maximum end-to-end (E2E) jitter of 18.2 s, compared with 842.5 s for TSN. Pre-installed TI-LFA repair remains below 15 ms across the evaluated 20–200-node range, and the full agentic policy achieves a 99.9
As millimeter-wave systems are deployed in 5G cellular networks and expected to expand to new frequency ranges in 6G, massive MIMO is a key technology to overcome the challenges of millimeter waves. The use of complex radio front-ends increases power consumption, and reliable consumption models are of great interest for the community. However, existing power consumption models often lack either generality or validation with real hardware measurements. This work addresses this gap by proposing a generalized and parameterized power consumption model derived from real hardware measurements collected through extensive surveys of circuit implementations. The proposed model enables a more empirically grounded comparison between the different massive MIMO architectures.
Wireless body area networks play a crucial role in continuous and real-time health monitoring; however, their practical deployment is severely constrained by limited energy resources, event-driven traffic characteristics, and scalability requirements in hospital environments. To address these challenges, this paper proposes a hybrid lightweight medium access control protocol that combines IEEE 802.15.6 and IEEE 802.15.4 communication technologies for reliable and energy-efficient health monitoring. In the proposed architecture, low-power biomedical sensor nodes deployed on the patient’s body transmit physiological data to a hand-belt coordinator using IEEE 802.15.6 with an energy-aware TDMA (EATDMA) mechanism. The hand-belt acts as an aggregation and relay node, forwarding prioritized data to the central ICU-monitoring hub through a bitmap-assisted MAC over an IEEE 802.15.4 Zigbee transceiver (CC2420). A comprehensive mathematical energy consumption model is developed to analyze the impact of sensor density, number of monitored patients, event probability, packet size variation, and number of transmission sessions per superframe. The MATLAB-based simulations demonstrate that the proposed hybrid MAC significantly reduces energy consumption while maintaining scalability and reliability. Results show that the proposed scheme achieves energy savings of up to approximately 60
This paper investigates cross-modal task-oriented semantic communication for a low-altitude unmanned aerial vehicle (UAV) transmitting mission-relevant semantic packets to a ground station over a wireless uplink. Different from bit-level transmissions, the UAV first extracts task-relevant semantics from heterogeneous onboard sensors and then adaptively determines the sensing modality combination, semantic granularity, and transmit power according to channel variation and mission urgency. To capture the long-term coupling among semantic utility, energy consumption, and delay, the problem is formulated as a Markov decision process (MDP) that maximizes the long-term average semantic utility subject to average energy and delay constraints. A hierarchical deep reinforcement learning (DRL) framework is developed, where a high-level recurrent proximal policy optimization (PPO) policy selects the cross-modal semantic mode and semantic granularity, and a low-level soft actor–critic (SAC) policy determines the transmit power. In addition, a task-oriented semantic encoder is designed with cross-modal alignment loss, multi-granularity consistency loss, task loss, quantization loss, and rate regularization. The proposed framework enables semantic-level decision-making rather than merely physical-layer resource control. Extended simulation results show that the proposed hierarchical DRL-based scheme outperforms several baselines in terms of semantic utility, energy consumption, and delay.
The integration of Wireless Sensor and Actuator Networks (WSANs) is a cornerstone of modern IoT and Industrial IoT (IIoT) frameworks. These systems generally consist of resource-limited devices that rely on IEEE 802.15.4 connectivity and the Time Slotted Channel Hopping (TSCH) MAC protocol. Although earlier studies introduced the Modified Bidirectional IPv6 Multicast Protocol (MBMRF) to mitigate multicast challenges within TSCH-based WSANs, those evaluations relied primarily on random network topologies. This paper builds upon that foundation by introducing a new theoretical framework aimed at examining end-to-end multicast latency across the entire routing path. To achieve greater analytical precision, we shift our focus from random deployments to constrained topologies featuring deliberate node placement. Our cross-layer evaluation fills a significant gap in the literature by incorporating routing-layer latency, an area often overlooked in studies restricted to MAC layer performance. Additionally, we introduce a dedicated analytical framework for RPL-enabled WSANs that accounts for synchronous TSCH scheduling, representing a distinct improvement over traditional asynchronous Contiki/RDC models. The proposed protocol and model are rigorously evaluated using three key performance metrics: average multicast packet delivery delay, packet delivery ratio (PDR), and energy consumption. Through Cooja-based simulations utilizing Zolertia (Z1) hardware, we verify our theoretical model across the transport, network, MAC, and RDC layers. Our findings indicate that MBMRF reduces latency and improves PDR by employing link-layer (LL) unicast for smaller multicast groups, while concurrently improving energy efficiency through a mixed mode that switches to LL broadcast as node density increases. The high degree of alignment between our analytical findings and simulation data demonstrates the reliability of the model in representing the intricate nature of multicast traffic within industrial scale wireless environments.
High-altitude platform systems (HAPS) have recently attracted significant attention due to their unique characteristics and wide range of potential applications. In particular, cloud-based HAPS (C-HAPS) provides a framework for deploying and delivering cloud services directly from HAPS station data centers. In our previous work, we introduced a C-HAPS-based architecture for environmental and infrastructure monitoring that integrates a wireless sensor network (WSN) with a blockchain model. In this paper, we propose a novel energy-aware and cache-enhanced routing protocol tailored for C-HAPS. The proposed approach addresses the resource constraints of sensor nodes that relay data to the HAPS data center by jointly considering residual energy and geographical location as the primary metrics for routing decisions. Furthermore, we introduce a data caching mechanism based on the concepts of cache data and cache location to further improve network performance and efficiency. Simulation results demonstrate that the proposed protocol outperforms existing terrestrial–aerial routing schemes in terms of packet delivery ratio, data request delay, energy consumption, and cache hit ratio, confirming its effectiveness and robustness for next-generation HAPS-based IoT environments.
Managing the Peak-to-Average Power Ratio (PAPR) in OFDM remains challenging due to unpredictable peak reformation caused by Power Amplifier (PA) distortion, pulse shaping, and multipath fading. Existing PAPR reduction methods fail to adapt dynamically, leading to spectral regrowth and the reintroduction of peaks. Additionally, modifications to reduce PAPR disrupt subcarrier orthogonality, increasing inter-symbol interference (ISI). A balance between PAPR reduction and ISI mitigation remains an unresolved trade-off in real-world OFDM deployments. Hence, this paper proposes an SLM–Wavelet Volterra-Pre-Distortion Reinforced Equalization Network, implemented within a Reinforcement Recurrent Neural Network Learning (RRNNL) framework, to dynamically balance PAPR reduction and ISI mitigation in OFDM systems. The first hidden layer employs Wavelet-Selective Partial Mapping to optimize phase adjustments and subcarrier allocation, minimizing peak power while preserving orthogonality. The second layer applies Volterra-Polynomial Guarded Pre-Distortion Filtering to correct PA distortion and mitigate ISI dynamically. The third layer incorporates Lagrange Multiplier and Zero-Forcing Equalization (ZFE) to optimize resource allocation and eliminate residual interference without reintroducing peak amplification. Experimental results demonstrate that the proposed method effectively balances PAPR reduction and ISI mitigation, achieving superior signal integrity and improved spectral efficiency compared to existing models.
This paper proposes a cross-layer scheduling algorithm for joint flow control and resource allocation to optimize the time-average utility function for the dual-connectivity multi-queue multi-server (D-C MQMS) system with a stochastic arrival process. First, we derive the capacity region of the D-C MQMS system by a finite set of linear inequalities. The capacity region characterizes the maximum input rate supported by the system, which is useful for network management. Furthermore, the characterization of the distance between the input rate vector and the boundary of capacity region is exhibited when the channel model is an ON/OFF channel. Then we consider two cases: (I) input rates within the capacity region, and (II) input rates exceeding the capacity region. Two scheduling algorithms for joint flow control and resource allocation, FCRA-I and FCRA-II, are proposed by Lyapunov optimization method. Next, the performance on the utility function and queue delay is analyzed. Finally, by comparing the proposed method with the single-connectivity (S-C) system, we verify the algorithm’s effectiveness by evaluating the average system throughput, system user perceived throughput (SUPT), the throughput-fairness utility function, and the average system backlog.
Multi-unmanned aerial vehicle (Multi-UAV) low-altitude networks require efficient transmission of perception information under limited bandwidth and dynamic channels. Semantic communication is a promising solution. However, existing designs rarely address two critical challenges: Detector confidence is often miscalibrated, and multiple UAVs may transmit redundant semantics from overlapping views. To address these issues, we propose a confidence-calibrated visual semantic communication framework for multi-UAV networks. Each UAV extracts semantic tokens and local detection hypotheses from onboard images. It then estimates calibrated confidence and localization quality for candidate objects. Based on these cues, we define a reliable semantic value. This value integrates calibrated confidence, localization quality, class importance, and cross-UAV semantic novelty. We further formulate a joint wireless-semantic optimization problem covering subchannel assignment, power allocation, semantic token selection, and semantic quantization. To evaluate resource utilization, we introduce confidence-calibrated semantic efficiency (CCSE), which measures reliable semantic gain per unit communication cost. We solve the mixed discrete-continuous stochastic problem by using a hierarchical reinforcement learning-based resource allocation scheme. In this scheme, the upper layer allocates wireless resources, and the lower layer determines semantic content and bit depth. Extended simulation results show that the proposed framework enables efficient and trustworthy multi-UAV visual semantic communications in low-altitude networks and outperforms several baselines.
IPv6 active address discovery acts as a cornerstone for large-scale Internet measurement and network security assessment. Current strategies face a fundamental dilemma: Exhaustive probing is computationally infeasible due to the colossal 2^128 address space, while heuristic and reinforcement learning approaches are frequently misled by the widespread presence of aliased prefixes (“black holes”). To transcend these limitations, this paper proposes 6Diffusion, a novel attention-guided generative framework that reformulates address discovery as a denoising diffusion process on a continuous latent manifold. Unlike traditional methods that rely on discrete pattern matching, 6Diffusion employs a Transformer-based denoiser integrated with gated linear function multi-head self-attention (GLF-MSA) to jointly capture global hierarchical allocation semantics and fine-grained interface identifier (IID) regularities. To address the inherent latency of diffusion models, we implement a deterministic denoising diffusion implicit model (DDIM) sampling strategy, accelerating inference by 40× through step skipping without compromising generative fidelity. Extensive evaluations are conducted under a unified 100K-seed setting, utilizing active targets reliably sourced from the RouteViews and RIPE RIS public measurement projects collected between January and March 2023. Results demonstrate that 6Diffusion establishes a new state-of-the-art, achieving a 42.43
Rail sleepers are critical to the safe operation of trains but are susceptible to cracking due to prolonged load pressure and weather-related corrosion. These cracks pose serious challenges to operational safety. However, existing crack detection algorithms often struggle to accurately identify sleeper cracks in complex real-world environments. To address this issue, we propose LM2DNet, which combines the local detail feature extraction capabilities of CNNs with the global semantic context captured by transformers. Using a lightweight, multi-scale deformable CNN-Transformer two-branch fusion architecture, LM2DNet integrates both feature types effectively. This fusion enables accurate identification of fine-grained crack details while preserving semantic information about the sleepers, ultimately improving detection accuracy and recall while reducing computational overhead. LM2DNet achieves a 2.2
Vehicle ad hoc networks consist of a number of nodes, each equipped with wireless communication equipment. In these networks, the destination for some data is all the vehicles present in the network which is named data dissemination. Due to the rapid changes in the topology of these networks, the dissemination and delivery of messages to all vehicles in inter-vehicle networks is considered a significant challenge. Various methods have been proposed to overcome these challenges. Among the existing methods, clustering seems to be an appropriate approach because an entity called the cluster head is responsible for delivering packets to the nodes within the cluster. Several clustering methods have been introduced for such networks, based on metaheuristic algorithms and machine learning. However, considering the dynamic nature of vehicle networks, there is practically no time available for data collection and the execution of these algorithms. It appears that utilizing inherent information, such as social behaviors and characteristics that do not require data collection, can be suitable for clustering vehicles and selecting the cluster head. In this paper, a method based on social features is proposed, known as social clustering-based data dissemination. In this method, initially, a number of nodes are selected as cluster heads based on their social characteristics, and then other nodes connect to the cluster heads based on speed and degree of the clusters. Simulation results show that using social behaviors in clustering improves packet delivery by 15
This work analyzes four novel NOMA-based optimization frameworks: (1) relay-assisted NOMA with DF and AF techniques, showing significant outage improvement over OMA (DF superior); (2) Sensing-NOMA (SNOMA) integrating spectrum sensing with user clustering and power allocation to maximize secondary throughput while protecting primary users; (3) Hybrid NOMA (HNOMA) game-theoretic scheme enabling autonomous coalition formation and dynamic switching between NOMA and OMA modes via preference relations and sequential games, further enhanced by an energy-efficient power allocation game; and (4) deep learning-based NOMA performing signal detection and fair power allocation without explicit channel state information, outperforming conventional SIC in detection accuracy and fairness. Extensive simulations validate all four schemes over existing OMA/NOMA baselines in outage probability, throughput, SE, EE, and fairness, leading to the conclusion that integrating NOMA with relaying, sensing, game theory, and provides a robust, scalable solution for future CRNs, while open challenges and research directions are also highlighted.
As global Intelligent Connected Vehicles (ICVs) deployment accelerates, onboard communication systems face critical challenges from multi-regional standards and dynamic regulations. Conventional manual or scripted test case generation exhibits insufficient efficiency and coverage, while static large language models struggle with knowledge obsolescence and cross-domain reasoning. This paper proposes an Agentic RAG (Retrieval-Augmented Generation with intelligent agents) framework to automate automotive communication protocol testing. By constructing a heterogeneous knowledge base spanning global standards and localized scenarios, and implementing multi-agent collaboration for dynamic retrieval and adaptive case generation, the framework achieves autonomous parsing of complex test requirements. Evaluations on authentic multi-regional datasets covering China, the European Union, the USA, Japan, and Southeast Asia demonstrate superior coverage and generation efficiency, while achieving compliance accuracy comparable to manual engineering compared to conventional methods and baseline RAG models. This work provides an AI-driven solution for accelerating regional market access and reducing compliance risks in global ICV deployment.
Predominant blood glucose detection techniques are based on finger stick testing with dry taps, finger trauma, pain, and dependency on biochemicals. The prevailing microwave-based approaches lack suitable dielectric contrast sensitivity due to far-field sensing, which displays unreliable characterization of blood glucose concentrations. This study addresses these gaps by developing a near-field sensor for continuous monitoring of blood glucose in interstitial fluid for type 2 diabetic patients. The proposed microwave sensor is based on a dual L-shaped symmetric parasitic element and a complementary rectangular mesh ring resonator. For variations in interstitial fluid permittivity, the former sensor displays a sensitivity of 139 – 605 MHz/RIU and the latter sensor displays a sensitivity of 701 – 748 MHz/RIU for glucose concentrations between 0 mg/dL and 500 mg/dL, respectively. With improved sensitivity, the complementary rectangular mesh ring resonator sensor has been prototyped, demonstrating reliable resonance variation for real-time glucose concentration measurements with stable performance for continuous monitoring in diabetes management.
Industrial environments present significant challenges for computer terminals, including peripheral device compatibility, inefficient power management, limited connectivity options as well as inadequate cooling systems. This research study introduces a robust industrial computer terminal system (ICTS) framework designed to address these critical issues, optimizing performance in industrial settings. The proposed ICTS integrates a simulated black annealing (SBA) algorithm to enhance power efficiency, connectivity reliability, and cooling performance. A Simulink model of the ICTS, comprising a display unit, peripheral connectivity module, power management system, and cooling mechanism, was developed and simulated using MATLAB/Simulink. Key performance metrics, such as optimal fitness value, simulation completion time, and system efficiency, were evaluated. Compared to existing industrial terminals like those from ADVANTECH, the optimized ICTS demonstrates clear improvements in power management, peripheral compatibility, connectivity robustness, and cooling efficiency. By leveraging the SBA algorithm for advanced optimization, the ICTS framework creates a more stable, energy-efficient as well as reliable terminal solution tailored for challenging industrial environments.
The article deals with single-carrier multiple access channel (MAC), where the relative delay of sources is a fraction of the symbol period. We review the channel likelihood function and test three approaches for inter-symbol interference (ISI). We remove ISI by (I) ISI tracking, (II) by a one-step marginalization of the channel likelihood, or (III) the discrete Fourier transform (DFT) precoding. For ISI tracking, we model ISI as a virtual convolutional code observed in a multipath channel. We use all three approaches to compare joint decoding (JD) of sources and hierarchical decoding (H-decoding)—specifically, decoding of the XOR frame. For a proof of concept, we apply a pair of serially concatenated convolutional codes as a channel code and BCJR (Bahl, Cocke, Jelinek, Raviv) decoding algorithm. DFT precoding “erases” arbitrary delay and makes the equivalent system model synchronous. The proposed reduced-state hierarchical finite state machine for H-decoding brings about 2 dB gain w.r.t. simple marginalization. The proposed methods are suitable for equally strong sources particularly with BPSK alphabet.
