Haibin Yu

Industrial Internet of Things (IIoT) edge computing supports latency-sensitive services through task offloading to distributed edge resources. However, large-scale emergencies such as node failures and traffic surges may trigger cascading failures, leading to severe performance degradation and poor post-crisis recovery. Existing offloading methods mainly optimize operational efficiency under normal conditions while overlooking resilience against cascading disruptions. To address this issue, we propose Cascade-Emergency Resilient Offloading (CERO), an adversarial deep reinforcement learning framework for resilient task offloading in IIoT edge computing. Distinct from existing works, CERO introduces a structure-aware shared node encoder to capture heterogeneous topological roles of edge nodes, providing critical structural information for cascade-aware decision making, and incorporates cascade-oriented adversarial training to enhance robustness against compound disturbances. CERO integrates structure-aware state representation, minimax adversarial training, and potential-based reward shaping to learn resource-allocation policies balancing task efficiency and system resilience. By interacting with dynamically generated crisis scenarios, the agent learns resilient offloading policies and achieves high post-crisis recovery performance after cascading disruptions. All performance evaluations are conducted via discrete-event simulation experiments. Simulation results for normal, single-crisis, and compound-crisis scenarios show that CERO achieves comparable task efficiency under normal conditions and significantly superior post-crisis recovery performance compared to conventional rule-based strategies. In the hardest compound-crisis case involving simultaneous node failures and load surges, CERO achieves a post-recovery task-completion rate of 97.8%, surpassing the best rule-based baseline by more than 63 percentage points. Statistical significance is confirmed by the Wilcoxon signed-rank test with Bonferroni correction over 10 independent runs. These results demonstrate that CERO effectively improves the robustness and recoverability of IIoT edge-computing systems under cascading emergency scenarios.

Dear Editor, With the rapid development of new-generation information and communication technology, the fusion of sensing, communication, computing, and control (S3C) is becoming increasingly significant for cyber-physical systems (CPS). However, due to the non-convexity, the curse of dimensionality, and the partial observability faced by CPS, traditional convex optimization algorithms are challenging to deal with S3C co-optimization. Thus, this letter establishes the S3C problem as a partially observable Markov decision process (POMDP) and proposes a large language model (LLM)-enhanced policy transfer (PT) framework for multi-agent deep reinforcement learning (MADRL), denoted as LLMPT-MADRL. Herein, MADRL learns a co-optimization policy while LLM-enhanced policy transfer improves the exploration efficiency, as LLM utilizes large-scale dynamic neural neurons like human brain to form powerful intelligent reasoning capabilities. However, MADRL and LLM exhibit heterogeneity in neural network scale. Thus, we further design an adaptive policy transfer loss function to achieve dynamic adaptation between MADRL and LLM, avoiding the mismatch of policy transfer information. Furthermore, we conduct a case study on the co-optimization of sampling coefficient, communication bandwidth, computing frequency, and control input to minimize the system delay and control error of CPS. Experimental results show that the reward of LLMPT-MADRL is increased by 16.8%, the system delay is reduced by 24.5%, and the control error is reduced by 10.4% compared with the benchmark MADRL-based algorithms.

In industrial scenarios, various terminals continuously generate massive heterogeneous data tasks, which have complex dependencies, significant differences in real-time requirements, and distinct deadline attributes. Most existing edge computing offloading methods are not adapted to industrial scenarios, making it difficult to ensure the timely completion of high-priority tasks, particularly those with stringent real-time requirements and complex dependencies. To address this issue, this paper proposes an Adaptive Priority Attention-based Task Offloading (APATO) method for dependent industrial tasks in edge computing. Firstly, an adaptive priority-aware directed acyclic graph (DAG) model is constructed to characterize heterogeneous task attributes and dependency relationships, and a dynamic priority calculation mechanism is designed to reflect task urgency and importance. Secondly, an intelligent offloading framework based on Transformer and Proximal Policy Optimization (PPO) is proposed, where the self-attention mechanism captures long-range task dependencies and task-resource correlations, and the PPO algorithm realizes stable and efficient offloading decision optimization. Finally, a multi-objective optimization strategy is constructed to maximize the deadline satisfaction rate of high-priority tasks and minimize the total system delay. Simulation results show that compared with HEFT, DRL-RNN, and pure offloading schemes, the proposed APATO algorithm improves the deadline satisfaction rate of high-priority tasks by 15.6%–21.3% and reduces the total system delay by 22.8%–29.4%, which verifies the effectiveness, robustness, and industrial adaptability of the proposed method.

The wireless cloud robotic system (WCRS), which fully integrates sensing, communication, computing, and control capabilities as an intelligent agent, is a promising way to achieve intelligent manufacturing due to easy deployment and flexible expansion. However, the high-precision control of WCRS requires deterministic wireless communication, which is always challenging in the complex and dynamic radio space. This paper employs the reconfigurable intelligent surface (RIS) to establish a novel RIS-assisted WCRS architecture, where the radio channel is controlled to achieve ultra-reliable, low-delay, and low-jitter communication for high-precision closed-loop motion control. However, control and communication are strongly coupled and should be co-optimized. Fully considering the constraints of control input threshold, control delay deadline, beam phase, antenna power, and information distortion, we establish a stability maximization problem to jointly optimize control input compensation, RIS phase shift, and beamforming. Herein, a new jitter-oriented system stability objective with respect to control error and communication jitter is defined and the closed-form expression of control delay deadline is derived based on the Jensen Inequality and Lyapunov-Krasovskii functional. Due to the time-varying and partial observability of the channel and robot states, we model the problem as a partially observable Markov decision process (pOMDP). To solve this complex problem, we propose a multi-agent transfer reinforcement learning algorithm named LSTM-PPO-MATRL, where the LSTM-enhanced proximal policy optimization (PPO) is designed to approximate an optimal solution and the option-guided policy transfer learning is proposed to facilitate the learning process. By centralized training and decentralized execution, LSTM-PPO-MATRL is validated by extensive experiments on MuJoCo tasks for both low-mobility and high-mobility robotic control scenarios. The results demonstrate that LSTM-PPO-MATRL not only realizes high learning efficiency, but also supports low-delay, low-jitter communication for low error control, where 71.9% control accuracy improvement and 68.7% delay jitter reduction are achieved compared to the PPO-MADRL baseline.

The rapid proliferation of connected and autonomous vehicles in the 6G era demands ultra-reliable and low-latency computation with intelligent resource coordination. Unmanned Aerial Vehicle (UAV)-assisted Mobile Edge Computing (MEC) provides a flexible and scalable solution to extend coverage and enhance offloading efficiency for dynamic Internet of Vehicles (IoV) environments. However, jointly optimizing task latency, user fairness, and service priority under time-varying channel conditions remains a fundamental challenge.To address this issue, this paper proposes a novel Multi-Agent Priority-based Fairness Adaptive Delayed Deep Deterministic Policy Gradient (MA-PF-AD3PG) algorithm for UAV-assisted MEC systems. An occlusion-aware dynamic deadline model is first established to capture real-time link blockage and channel fading. Based on this model, a priority–fairness coupled optimization framework is formulated to jointly minimize overall latency and balance service fairness across heterogeneous vehicular tasks. To efficiently solve this NP-hard problem, the proposed MA-PF-AD3PG integrates fairness-aware service preprocessing and an adaptive delayed update mechanism within a multi-agent deep reinforcement learning structure, enabling decentralized yet coordinated UAV decision-making. Extensive simulations demonstrate that MA-PF-AD3PG achieves superior convergence stability, 13–57% higher total rewards, up to 46% lower delay, and nearly perfect fairness compared with state-of-the-art Deep Reinforcement Learning (DRL) and heuristic methods.

The coexistence of ultra-reliable low-latency communication (URLLC) and enhanced mobile broadband (eMBB) services in 5G-based industrial wireless networks

Sixth-generation (6G)-driven Internet of Things (IoT) networks demonstrate enormous development potential. However, complex and diverse IoT devices also pose severe cybersecurity challenges. To address the problem of low reliability and trustworthiness of existing intrusion detection systems (IDSs) in large-scale, low-latency scenarios of 6G-IoT networks, this article proposes a local–global hierarchical framework for intrusion detection. First, a reliable feature processing algorithm is designed based on fusion features and a multiobjective collaborative optimization mechanism oriented toward normal behavior modeling, enabling the binary classification model to effectively detect unknown types of attack traffic even under data anomaly conditions. Second, an adaptive classification trustworthiness assessment (ACTA) method based on classification information entropy and collaborative verification is proposed, which can accurately locate and upload suspicious traffic during the classification process. Finally, we propose a gradient-oriented feature decoupling module that effectively improves the feature learning capacity of network attack detection models in global servers for minority class samples. Experimental results demonstrate that our proposed method achieves optimal performance on five processed IDS benchmark datasets, with local IoT devices achieving an ${F}1$ score of 98.5% and global servers achieving an arithmetic mean ${F}1$ score of 96.6%, providing strong support for the reliable operation of IDS in 6G-IoT networks.

Rehabilitation exoskeletons have been demonstrated to benefit mobility-limited patients; however, accessibility is hindered by several challenges, and interaction evaluation of coupled human–exoskeleton systems remains critically understudied. In this pioneering study“, assistive normality (AN)” is introduced to characterize the cross-stage spatiotemporal features of exoskeleton-assisted gait in hemiplegic patients, and a multidimensional, low-data-cost metric framework is developed to quantify AN. Three subdivided metrics, including gait restoration (GR), phase-deviation weighting (PDW) and multijoint coordination (MJC), are proposed to assess coupled system interaction behavior across different intervention stages during rehabilitation. A homologous difference evaluation paradigm (HDEP) is introduced to capture pathological differences between healthy and hemiplegic subjects on the basis of the reusability of experimental data, providing an approach for the nominal assistance calibration and assessment of specific exoskeleton devices. In a pilot study with eight healthy individuals and nine hemiplegic patients, between-group metric differences were analyzed to determine the calibrated AN of an exoskeleton. The results provide calibration references for quantitative metrics and demonstrate the ability of framework to characterize the temporal dynamics of human–exoskeleton interactions. The proposed AN framework offers a generalizable approach to assistive robotics, potentially enhancing evaluation of human-robot interaction and advancing clinical rehabilitation applications.

Blockchain-empowered end-edge collaborative computing is a promising technology for enhancing the timeliness and trustworthiness of Industrial Internet of Things (IIoT). However, integrating task offloading with blockchain consensus inevitably escalates resource consumption across communication, computation, and energy domains. Thus, the joint optimization of task offloading, resource allocation and blockchain consensus is very important for IIoT. This paper studies a general end-edge collaborative computing scenario with multiple end devices and multiple edge servers. We first propose a novel dynamic blockchain (DBC) scheme by developing a dynamic leader election mechanism and designing a dynamic consensus waiting time window. Then, by fully considering the constraints of multi-task size and deadline, communication bandwidth, computing frequency, battery capacity, Byzantine fault tolerant and trustworthiness, we formulate the trustworthy processing efficiency (TPE) maximization problem with respect to end-edge task division, communication and computation resource allocation, leader election and consensus waiting window. To address this problem, we transform it into a Markov decision process and design a compound reward by fully considering the penalty for computing timeout and consensus failure. After that, we propose a rotating multi-agent deep reinforcement learning (R-MADRL) algorithm tailored to the proposed DBC scheme, where an entropy-based dual-critic DRL algorithm is proposed for rotating multi-agent training and decentralized execution. Extensive experiments validate the effectiveness and superiority of the proposed DBC with R-MADRL, where three benchmark DRL algorithms and three blockchain consensus schemes are compared. The results demonstrate that R-MADRL achieves stable convergence with more than 60.32% TPE reward than other algorithms while the task timeout ratio of DBC is reduced by more than 66.49% compared with other schemes.