Search Results (340)

As a distributed approach to Artificial Intelligence (AI) model construction over wireless networks, federated learning (FL) based on multi-device collaborative training can protect data privacy, as well as increase the computing load of local model updates. In contrast, split learning (SL) with proper model splitting can adapt to the computation and transmission capabilities among devices. In this paper, while taking advantage of FL and SL, we concentrate on a semi-decentralized hybrid federated split learning (SD-HFSL) framework, in which we surpass the limitations of a single central server and allow the shared split models to be aggregated among multiple edge servers. To verify the importance of latency optimization for training efficiency, we analyze the convergence performance of SD-HFSL while jointly considering the limited computation and communication resources. Then, aiming at maximizing the long-term training efficiency, we propose an online optimization problem that includes local model splitting and device association. Considering that the training latency is unknown to the system a priori, a context-aware online training algorithm with sublinear regret is proposed based on the framework of contextual multi-armed bandit (CMAB), where the edge servers can observe the context information of device sites for latency estimation, followed by the iterative optimization based on the evaluated information in different contexts. Experiments on several neural network models show that the proposed algorithm reduces training latency and improves test accuracy compared with the selected benchmarks. Full article

Background: Insomnia and stress-related sleep disorders are increasingly recognized as systemic conditions linked to the microbiota–gut–brain axis (MGBA). With growing clinical interest in natural products that modulate the gut environment, this systematic review evaluates the efficacy and mechanisms of non-pharmacological interventions, specifically probiotics, prebiotics, dietary indices, and botanicals, in alleviating insomnia, restoring circadian rhythms, and modulating neurochemical markers. Methods: In strict accordance with PRISMA 2020 guidelines, we searched PubMed, ScienceDirect, Scopus, and The Cochrane Library for English language studies published from inception to March 31, 2026. Eligibility was restricted to studies with rigorously controlled designs, specifically randomized controlled trials (RCTs) and controlled in vivo animal studies. Interventions had to target the gut microbiota, with primary outcomes measuring sleep quality (subjective or objective) or sleep-related neurochemical markers. We excluded uncontrolled, single-arm, or observational designs; in vitro studies; non-original research; and studies involving subjects with severe medical or psychiatric comorbidities (e.g., cancer, ADHD, severe psychiatric disorders) to prevent confounding variables, though mild-to-moderate anxiety and depression were permitted. Risk of bias was assessed using the Cochrane RoB 2.0 and SYRCLE tools. Due to significant methodological heterogeneity, a narrative synthesis stratified by intervention and population was conducted. This review was not registered in PROSPERO. Results: A total of 56 studies (33 humans, 23 animals) met the inclusion criteria. Taxonomic nomenclature was updated to reflect 2020 reclassifications (e.g., Lactiplantibacillus plantarum). In human trials, interventions significantly improved subjective sleep metrics (PSQI, ISI). Recent additions demonstrated the efficacy of the Dietary Index for Gut Microbiota (DI-GM) and the improvement in N3 sleep latency by yeast mannan. Furthermore, whole-food patterns (e.g., the MIND diet) and Traditional Chinese Medicine (TCM) decoctions successfully enriched beneficial taxa, such as Bacteroides coprophilus, and increased short-chain fatty acid (SCFA) production. Animal models demonstrated that “psychobiotic” strains (Bifidobacterium breve, Lacticaseibacillus paracasei), prebiotics (GOS/PDX), and TCM formulas effectively restored GABA/5-HT profiles, lowered morning cortisol, and facilitated REM rebound in PCPA-induced models, while also consolidating non-rapid eye movement (NREM) sleep and downregulating clock genes (Per1/Per2). Conclusions: Psychobiotics, prebiotics, and botanicals represent a highly viable non-pharmacological strategy for treating insomnia. However, current evidence is constrained by a heavy reliance on subjective human questionnaires, short follow-up durations limiting insight into long-term stability, and a substantial translational gap between mechanistic rodent models and human clinical outcomes. Full article

High-Frequency Trading (HFT) dashboards require rapid reception, aggregation, and visualization of order book and trade update streams that may arrive at multi-million message rates. Conventional CPU-based and CPU-GPU hybrid visualization pipelines can suffer from significant delays during periods of burst due to CPU-mediated rendering, synchronization, kernel launch overhead, and copies on the host. This paper presents a visualization pipeline that is entirely resident on the graphics processor with zero-copy access to NIC accessible pinned buffers, persistent CUDA processing, fused stage execution of the parse-aggregate pipeline, and persistent CUDA OpenGL buffer interoperation. The goal is not to reach production status but rather to see whether host-to-host data movement can be decreased and whether the stages of GPU processing can be consolidated to improve latency, throughput and frame cadence in controlled HFT-style workloads. The evaluated workstation achieved a mean ingest-to-pixel latency of 6.3 ms using the proposed design compared to 29.4 ms for the current design, with sustained throughput of 10.2 million messages per second, which is 20 times greater than the current design, and a steady-state range of 185 to 192 frames per second with a burst floor of 178 frames per second for the proposed design. The improvement observed can be attributed to both the zero-copy ingestion and fused persistent kernel execution. Based on the obtained results, the proposed method of use of this technique in the implementation of real-time financial visualization under the proposed conditions is possible. More general testing is still required on other NICs, other generations of GPUs and PCIe configurations, workload traces, and actual exchange feeds. Full article

The application of Internet of Things (IoT) technology to Digital Twin (DT) in smart manufacturing has opened significant opportunities for real-time monitoring, predictive maintenance, and closed-loop control; however, the inherent latency that exists in these architectures (the temporal gap between a physical event and its reflection in a digital model) remains one of the most significant and least systematically understood barriers to fulfill its full potential. This paper aims to propose a formal four-layer taxonomy of latency sources in IoT-based Digital Twin systems for smart manufacturing and to review the current approaches and tools that are available for their measurement. The PRISMA protocol has been used to perform a systematic literature review, where 58 primary survey studies published between 2020 and 2026 were extracted from IEEE Xplore, Elsevier Scopus, Google Scholar and arXiv, with all the studies being coded along six dimensions (architectural layer, application domain, latency metrics reported, evaluation methodology, quantitative impact, and enabling technologies). The proposed taxonomy presents 28 different types of latencies under four layers: (L1) network, (L2) compute, (L3) data, and (L4) end-to-end (E2E), whose magnitudes vary from 0.1 ms for local network propagation to tail latencies above 500 ms in production (P99). Three categories and three cross-layer interaction patterns are formalized here and are absent from prior partial taxonomies. Among the most promising results is the finding that several high-impact interventions require no infrastructure investment: a protocol migration from Modbus to WebSocket reduces telemetry latency by 32%, while Age of Information-aware synchronization and clock drift correction deliver substantial data layer gains through software updates alone, yet remain underutilized. The review identifies a systematic under-reporting of tail-latency percentiles across the corpus, the lack of a cross-protocol jitter benchmark, and a predominance of simulation-based evaluation over real-hardware measurement. The systematic review contributions of this paper (the formal four-layer taxonomy, the proportional metric audit across the 58 papers, and the formalization of three cross-layer interaction patterns) are derived from cross-corpus analysis. The investigation also identifies three open research directions (a standardized manufacturing IoT-DT benchmark, cross-layer joint optimization frameworks, and wireless TSN validation on real manufacturing testing grounds) that together form a well-organized and practical basis to advance both the science and the application of ultra-low-latency Digital Twin technology in the industrial field. Full article

Wireless sensor networks supported by mobile edge computing are increasingly required to process heterogeneous sensing data under stringent latency, reliability, and energy constraints. However, most existing task-offloading studies are still formulated for generic user equipment and primarily focus on uplink transmission, which is insufficient for practical sensing systems where sensor nodes continuously upload measurements while simultaneously receiving control commands, model updates, and feedback from the edge. To address this gap, this paper reformulates joint computation offloading and power control as a sensor-aware optimization problem in an edge-assisted wireless sensor network. We propose a three-layer architecture consisting of sensor nodes, access points with lightweight edge servers, and a cloud coordination layer. Each sensing task is characterized by data size, computation density, latency deadline, and sensing priority, while the optimization objective jointly minimizes long-term task delay, communication and computation energy, and packet-loss penalty under transmission power, edge resource, and residual-energy constraints. To solve the resulting mixed discrete–continuous problem, we develop a multi-agent reinforcement learning framework in which each sensor node acts as an autonomous agent and learns offloading and transmission policies with clipped proximal policy optimization, while the cloud layer performs coordinated edge-resource allocation through the alternating direction method of multipliers. In addition to delay and energy, network lifetime and sensing delivery performance are incorporated into the evaluation. Simulation results in a sensor-network monitoring scenario demonstrate that the proposed framework consistently reduces latency, lowers energy consumption, and prolongs network lifetime compared with representative baselines, highlighting its effectiveness and practical potential for intelligent sensing applications that require integrated sensing, communication, and edge computing. Full article

The rapid proliferation of cloud data centers has intensified concerns over carbon emissions, energy efficiency, and sustainability. Virtual machine (VM) placement is a pivotal control lever, yet existing methods rarely couple carbon intensity signals with computationally tractable multi-objective optimization. In this paper, we propose CASO (Carbon-Aware Surrogate-Guided Optimization), a novel framework that integrates an online adaptive Radial Basis Function (RBF) surrogate model with a self-adaptive hybrid PSO-DE swarm optimizer for real-time VM placement in geo-distributed edge cloud environments. CASO simultaneously minimizes carbon emissions, energy consumption, SLA violation rate, and network latency under strict host capacity and Quality-of-Service (QoS) constraints. Three key innovations differentiate CASO: (i) an online surrogate update mechanism that refines fitness approximations incrementally as workload patterns evolve; (ii) a carbon intensity weighting scheme anchored to real-time Grid Emission Factor (GEF) signals; and (iii) an adaptive parameter controller that autonomously tunes swarm exploration–exploitation trade-offs without hand-crafting. Experiments on the publicly available Alibaba Cluster Trace (cluster-trace-v2026-GenAI) dataset within a CloudSim-Plus environment show that CASO reduces carbon emissions by up to 31.4%, energy consumption by 27.9%, and SLA violations by 18.8% compared to the strongest baseline while converging 3.8× faster than the strongest baseline (ADEDL). Full article

Soft robotics enables inherently safe, compliant interaction, yet integrating brain–computer interfaces (BCIs) remains hindered by a fundamental mismatch: BCIs typically output low-bandwidth, discrete commands, whereas soft robots possess high-dimensional, nonlinear dynamics. In this position paper, we argue that BCI–soft robot integration must move beyond direct decoder-to-actuator mapping. We propose a unified, application-oriented compatibility framework that structurally decouples hierarchical control and formally allocates authority between human neural input and local soft robotic autonomy. Crucially, we introduce verifiable, quantitative design principles that define integration as a matching problem across neural bandwidth, update frequency, latency tolerance, and control dimensionality. Through these testable hypotheses, we demonstrate that active, reactive, and passive BCIs serve distinct, complementary roles. We conclude that shared-control strategies—where the BCI provides high-level intent, target selection, or user-state feedback, while the soft robot manages low-level physical execution and interaction—offer the most practical pathway forward. We argue that future progress depends on the co-design of paradigm, decoding, control, and embodiment for neuro-adaptive and human-centred soft robotic systems. Full article

Due to the indispensable use of the internet, malicious actors have exploited URLs as a threat source of network information security and integrity. URL detection based on traditional methods has become inefficient against the uncontrolled increase of URLs, especially when facing dynamic and large-scale threats. To address the limitations of traditional methods and to provide intelligent and scalable detection of malicious URLs, this study proposes the hybrid framework (NetGuard) by integrating probabilistic data structures (PDSs) with machine learning (ML) capabilities. The proposed NetGuard utilizes PDSs to develop a Hybrid Scalable Detection Filter (HSDF), which combines the strengths of counting Bloom filters (CBFs) (deletion capability) and Scalable Bloom filters (SBFs). The proposed HSDF provides efficient membership queries under bounded false-positive rates (approximately 0.01) and ensures efficient data management and low-latency lookups on a scale of 10⁻⁵ s. On the other hand, NetGuard leverages the ML classifier capabilities to train and package a learned classifier for detecting malicious URLs. The proposed framework utilizes Decision Trees (DTs) and Random Forest (RF) classifiers. The proposed classifiers are trained by a novel SupURLsIdDs dataset which includes fifteen distinctive lexical and structural URL features extracted from four URL classes: benign, defacement, malware, and phishing URLs. The experimental results indicated the effectiveness of the HSDF in insertion and deletion operations, with minimal memory consumption (approximately 2.7 MB for 222,000 URLs) while maintaining a controlled false-positive rate (approximately 0.01 on Real-only subset up to 0.12 with synthetic data). The HSDF memory footprint represents a 99.88% enhancement compared to the RF model (which demands 2253.17 MB); thus, the HSDF complements RF as an ultra-lightweight first line of defense. The ML classifiers showed the superiority of RF, which achieved an overall classification accuracy of approximately 96% on large-scale URL data. These experiments are conducted using benchmark datasets constructed from aggregated real and synthetic data to demonstrate the scalability, adaptability, and resource efficiency of the first phase of NetGuard as a practical foundation for real-time web threat detection. The real-time integration and dynamic updates are presented as a deployment architecture and constitute future work. Full article

Access to operational information in industrial plants forces operators to interrupt their tasks, walk to the human–machine interface (HMI) terminals, and navigate heterogeneous platforms—namely programmable logic controllers (PLC), supervisory control and data acquisition (SCADA) systems, manufacturing execution systems (MES), and enterprise resource planning (ERP) systems—consuming 15–30 min per query. Previous work integrated local large language models (LLMs) into a five-layer cognitive architecture deployed in a precast concrete plant, reducing that time to 14–23 s through voice-based conversational queries; however, model inference accounted for 55.3% of total latency and the system remained reactive. This work incorporates the event-driven paradigm as a non-intrusive augmentation layer that keeps the operational context permanently updated, continuously monitoring the process and refreshing knowledge only when significant changes occur. The architecture is fully local, cloud-independent, graphics processing unit (GPU)-free, and containerized via Docker Compose. Experimental results demonstrate a 26–31% reduction in response times (means of 9.84 s, 11.23 s, and 16.47 s for simple, moderate, and complex queries), an 8.4 °C reduction in peak hardware temperature (from 79.6 °C to 71.2 °C), a 41.6% decrease in thermal variability, and an expansion of the safety margin before central processing unit (CPU) throttling from 5.4 °C to 13.8 °C. The system achieved 100% success rate and availability over 30 min of autonomous operation, validated in a real industrial environment. Full article

Blockchain oracles bridge on-chain smart contracts and off-chain data sources, but encrypted off-chain data still raises two practical challenges: how to verify retrieval integrity without exposing sensitive values, and how to keep verification information fresh when the off-chain data set changes. Existing oracle and outsourced-database retrieval mechanisms often rely on plaintext verification, heavy cryptographic proofs, or static authentication structures, which limits their applicability to latency-sensitive IoT and decentralized finance scenarios. To address these issues, this paper proposes a retrieval integrity verification mechanism based on CKKS approximate homomorphic encryption and an authenticated index named CKKS-Auth Tree. The proposed mechanism verifies encrypted query results through homomorphically aggregated metadata, while smart contracts record versioned verification commitments to detect stale or replayed results after updates. The scope of the mechanism is the integrity, completeness, privacy, and freshness of data after commitment and upload; verifying the physical authenticity of the original data source is outside the core threat model. Experimental results show that the proposed scheme reduces authentication and verification overhead compared with existing retrieval verification methods while supporting encrypted metadata updates and on-chain synchronization. Full article

Plant disease recognition models trained under laboratory conditions often degrade markedly after cross-domain transfer because of the pronounced distribution gap between source and target domains and the scarcity of labeled target-domain samples. To address the transfer task from PlantVillage (PV_100) to PlantDoc, this study develops and evaluates a lightweight cross-domain few-shot plant disease recognition method under a strict PlantVillage-to-PlantDoc protocol. The method integrates EfficientNet-B0 feature extraction, cosine-similarity-based prototypical classification, and training-time target-domain BN adaptation (TBA). During training, unlabeled target-domain images are used only for BN statistical calibration, whereas inference is limited to feature extraction and prototype matching, without gradient updates or iterative optimization. Under a unified experimental protocol, the proposed method achieved cross-split mean accuracies of 42.69 ± 0.62% for one-shot and 54.24 ± 0.72% for five-shot, where ± denotes the standard deviation across three strict data splits; it outperformed ProtoNet by 7.44 and 9.43 percentage points, respectively. Ablation results indicate that TBA is the main source of performance improvement, whereas more complex adaptation strategies do not yield stable additional gains. The core encoder can be executed entirely on the NPU, with an estimated single-sample inference latency as low as 0.658 ms, indicating strong potential for encoder-level mobile deployment. Full article

Federated learning (FL) enables distributed model training in IoT environments while keeping raw data on local devices. However, protecting model-update exchange is difficult on microcontroller-class devices due to strict latency, memory, and energy constraints. Existing studies often evaluate lightweight cryptography outside complete FL pipelines or on more powerful hardware, leaving its practical overhead on MCU-class devices insufficiently explored. This paper presents an end-to-end, hardware-validated secure framework for exchanging model updates in federated learning on resource-constrained IoT microcontrollers. Implemented on ESP32-based edge devices, the framework combines lightweight block ciphers (SPECK, SIMON, and PRESENT), HMAC-SHA256 for integrity verification, and ECDH-HKDF for session-key establishment. The evaluation assessed latency, throughput, RAM/ROM footprint, and energy consumption. Results show that SPECK provides the lowest overhead (0.13 µs/byte, 8.68 MB/s, 138.3 mJ), SIMON offers intermediate performance (0.41 µs/byte, 1.96 MB/s, 184.9 mJ), and PRESENT incurs the highest computational cost (89.37 µs/byte, 0.011 MB/s, 446.2 mJ). In the CICIoT2023 federated intrusion-detection evaluation, the secure model maintained stable convergence and achieved 85.43% accuracy after 20 rounds, remaining close to the centralized baseline. These findings demonstrate the practical feasibility of secure model-update exchange in FL on real IoT microcontrollers and provide hardware-grounded guidance for cipher selection under tight resource budgets. Full article

GNSS signals are extremely weak at the Earth’s surface and are highly vulnerable to in-band interference, particularly high-dynamic linear frequency-modulated (LFM) jamming, which may lead to receiver loss of lock. Existing anti-jamming techniques struggle to balance real-time constraints with high-precision frequency estimation. This paper proposes a Newton-iteration-enhanced three-spectral-line RIFE algorithm implemented on a heterogeneous FPGA platform (Zynq-7000 SoC). The method performs coarse frequency estimation using the three-spectral-line RIFE to mitigate FFT fence effects, followed by Newton-based quadratic refinement, enabling high estimation accuracy with reduced FFT size. A fast–slow loop architecture is adopted, where the FPGA (PL) performs real-time interference suppression and the ARM (PS) handles system control and parameter updates. Experimental results show that, under static interference, the proposed method achieves a 10.9 dB improvement over direct estimation algorithms. Under chirp interference, it significantly outperforms both direct estimation and conventional iterative methods. In GNSS closed-loop tests, the proposed approach extends the anti-jamming margin to 82 dB J/S. Overall, the proposed method effectively balances estimation accuracy and processing latency, providing a practical solution for GNSS anti-jamming in high-dynamic environments. Full article

In distributed renewable energy systems, load fluctuations caused by energy resources and energy storage increase the overload risk of distribution transformers, which may accelerate insulation aging and cause overheating, and undermine operational reliability. For transformer condition monitoring, this risk is reflected not by a single variable but by heterogeneous sensing observations acquired from electrical, thermal, and equipment status monitoring channels. Because full-scale inspection of latent defects is impractical under limited staffing and equipment resources, accurate overload risk prediction is important for sensor-driven maintenance allocation. With such motivations, this paper proposes a Transformer Overload Risk Assessment (TORA) approach for robust overload risk prediction under nonstationary load conditions. First, a feature matrix is constructed by jointly incorporating static features that capture long-term drift and dynamic features extracted from multisource sensing and supervisory signals that reflect short-term fluctuations. Then, static and dynamic features are assessed with Edge-based Static Feature Risk Assessment (E-SFRA) model and Cloud-based Dynamic Feature Risk Assessment (C-DFRA) model, respectively, according to their temporal and statistical characteristics. Next, a periodic calibration model (CE-PAA) is established through a cloud–edge loop, which uses low-latency edge updates and high-capacity cloud computation as feedback. Finally, risk score fusion (RSF) fuses generated static and dynamic risk scores to integrate cloud and edge strengths. The case study results indicate that TORA can transform heterogeneous monitoring signals into calibrated risk information in the studied single power plant scenario, providing useful support for multisource sensor data fusion, transformer condition monitoring, and maintenance decision making. Further validation using multi source field datasets is still needed to assess its cross scenario generalization ability. Full article

Vehicular Ad Hoc Networks (VANETs) face significant challenges in high-mobility environments, where dynamic channel conditions, particularly Doppler Shift (DS), degrade communication reliability and increase latency, thereby undermining safety-critical applications. To address these limitations, this paper proposes a neural network (NN)-based link adaptation strategy trained using the Scaled Conjugate Gradient (SCG) algorithm. SCG is selected as a second-order approximation optimizer that leverages curvature information to produce well-conditioned weight updates particularly suited to the small, physics-constrained training dataset. The SCG-optimized model dynamically adjusts transmission parameters to mitigate DS effects, improving real-time adaptability by explicitly incorporating Doppler Shift as a key input feature. Simulation results demonstrate that the proposed approach outperforms both the conventional Auto Rate Fallback (ARF) method and the SampleRate baseline. Specifically, the SCG-based strategy achieves an overall throughput improvement of +34.6% relative to ARF (1.77 Mbps vs. 1.32 Mbps) across all tested conditions, with condition-specific gains of +16.1% at 5 Hz Doppler (0.9 km/h), +21.7% at 750 Hz (137.3 km/h), and +35.2% at 1500 Hz (274.6 km/h), while consistently reducing transmission duration. A formal ablation study confirms that the Doppler Shift feature alone contributes +67% to +78% throughput gain at high mobility (DS > 900 Hz) compared to an SNR-only model. The main contributions of this work are threefold: (i) the explicit integration of Doppler Shift as a first-class input feature for link adaptation; (ii) the application of SCG optimization for fast, stable training of a lightweight feedforward neural network on a compact, physics-constrained dataset; and (iii) the formal ablation study that isolates and quantifies the Doppler feature’s contribution, establishing that the performance gain is attributable to feature engineering rather than the neural network architecture alone. This approach offers a scalable, real-time solution for Doppler-resilient VANET link adaptation. Full article