The rapid proliferation of emotion-aware web services has necessitated the analysis of multimodal user interactions. However, this introduces new vulnerabilities where adversaries exploit emotional signals to circumvent fraud detection systems. Despite its improved utility, the robustness of multimodal fraud detection against emotion-driven adversarial manipulation remains significantly underexplored. Existing paradigms often treat emotional cues as static features, overlooking the adversary's capability to strategically modulate multimodal signals (e.g., facial micro-expressions, vocal intonation, and textual styles) to mimic genuine behavior. Furthermore, prevalent evaluations are typically confined to unimodal perturbations and fail to account for context-consistent, cross-modal attacks, thereby compromising system reliability in real-world deployments. To bridge this gap, we propose Context-Emotion Adversarial Training (CEAT), a robust framework designed to fortify multimodal fraud detection against emotion-based attacks. CEAT leverages a Transformer-based architecture to synergistically model emotional features (e.g., visual dynamics and acoustic prosody) alongside semantic context derived from text, yielding a unified representation. Crucially, CEAT introduces a context-aware perturbation mechanism that injects noise into the emotional latent space during training. This process preserves semantic consistency while encouraging the learning of emotion-invariant and discriminative representations. Additionally, a contrastive learning objective is integrated to maximize the distributional divergence between genuine and adversarial samples within the latent manifold. Extensive experiments on multimodal benchmarks demonstrate that CEAT significantly outperforms state-of-the-art baselines, exhibiting superior robustness under simulated emotion-driven attack scenarios.

Modern cloud applications, from interative web services to mobile and WoT workloads, generate highly dynamic multi-tenant network demands. Guaranteeing priority-aware bandwidth remains challenging: legacy shapers like Linux Traffic Control Hierarchical Token Bucket are static and unscalable, while cloud-native solutions such as Cilium offer only coarse-grained rate limiting. We present BeeQoS, a cloud-native QoS system that delivers low-latency, adaptive, and scalable multi-priority bandwidth guarantees. BeeQoS consists of an eBPF-powered data plane for high-performance, fine-grained per-packet shaping, a demand-aware control plane that senses real-time flow requirements and adaptively reallocates bandwidth, and seamless Kubernetes integration for expressive policy specification and cluster-wide scalable deployment. Evaluation shows that BeeQoS scales to 1K+ flows with stable performance, boosts high-/medium-priority throughput by 14.6%/36.4%, cuts median latency by 72.4%, reduces deployment overhead, and improves video QoE by 27.3% over state-of-practice baselines.

With the continuous advancement of Industry 4.0 reshaping industrial ecosystems, Industrial Wireless Sensor Networks (IWSNs) have evolved into an indispensable cornerstone of the Industrial Internet of Things (IIoT). Nevertheless, optimizing energy efficiency and network performance in IWSNs persists as a formidable challenge—primarily stemming from the NP-hard characteristic inherent to clustering problems. This paper presents a groundbreaking multi-objective optimization approach to tackle this issue, aiming to reduce energy consumption while boosting service reliability. We introduce an advanced clustering protocol, NCPLO-HPCP, which evaluates four essential performance factors: residual energy across the network, end-to-end latency, packet loss rate, and the spatial arrangement of cluster heads with respect to the base station. At its core, NCPLO-HPCP employs a novel optimization technique, Niching Clone Polar Lights Optimization (NCPLO), which combines cloning and niching strategies to enhance global search efficiency and maintain population diversity. This innovative approach ensures faster convergence and greater network stability in dynamic industrial environments. Comprehensive simulations demonstrate that NCPLO-HPCP significantly outperforms leading protocols (MDCSFLA, EBPT-CRA, FOAEAUC-SARP, and VSSLS-SIACR), achieving a 19.57% increase in network lifetime, a 10.87% reduction in latency, and a 24.76% decrease in packet loss. Additional comparative experiments with four state-of-the-art protocols published in 2026 (DQL-iLEACH, QPSODRL, ACRM, ECAEP) further confirm its superior performance across all key metrics. These findings offer a highly effective and scalable solution for sustainable clustering in large-scale IWSNs, advancing the overall intelligence and efficiency of IIoT systems.

This article presents a multistrategy hybrid dung beetle algorithm (MHDBO) algorithm to tackle the complexities of Autonomous tractor path planning, a problem characterized by intricate optimization objectives and constraints. The proposed MHDBO effectively reduces operational costs, minimizes path lengths, and enhances coverage efficiency in agricultural operations. To improve optimization performance, a novel quantum-assisted offset estimation strategy is introduced, which significantly expands the solution space, improves population diversity, and effectively guides the evolutionary direction of the algorithm to avoid local optima. Additionally, an enhanced double-helix search strategy is designed to strengthen global search capabilities, improving both accuracy and convergence. A reverse learning strategy is integrated, in which a quasi-oppositional learning mechanism is used to avoid large positional deviations typical of traditional reverse learning, thereby maintaining solution stability while enhancing exploration. This helps prevent premature convergence and improves robustness in dynamic environments. To further optimize multitractor coordination, a multistack multitractor scheduling and field path planning model (MSPPM) is developed, addressing the challenges of large-scale agricultural operations. Experimental evaluations show that the proposed MHDBO algorithm achieves a total dispatch path length of 29 093.4007 km, an overall operational cost of 2.622543 million RMB, an execution time of only 4.128759 s, and a memory consumption of$3.588~{\times }~10{^{{9}}}$bytes in the MATLAB environment. Furthermore, MHDBO demonstrates strong performance in coverage, achieving over 95% effective farmland utilization. Compared with state-of-the-art algorithms, including adaptive elite differential evolution algorithm (AEDE), improved 2-opt ant colony optimization (IACO), and adaptive elite chaotic genetic algorithm (AECGA), MHDBO achieves a 10.28% reduction in operational costs, a 5.859% decrease in path length, and executes 15.46% faster, while consuming 16.06% less memory. These results confirm the algorithm’s superiority, efficiency, and scalability in real-time agricultural automation applications.