Mufeng Wang

With the transition to Industry 5.0, there is a growing demand to deploy highly autonomous and resilient artificial intelligence (AI) systems in critical infrastructures such as power grids. In the field of load frequency control (LFC) in power grids, a hybrid control architecture in which deep reinforcement learning (DRL) controllers coexist with traditional proportionalintegral-derivative (PID) controllers can become a typical deployment model during this technological transition. However, the inherent vulnerability of DRL controllers to adversarial attacks introduces new security challenges in such complex environments: attacks not only affect the DRL-controlled areas but may also propagate to PID-controlled areas through inter-area power exchanges, potentially causing broader system instability. To accurately assess the cascading risks under this hybrid architecture, we propose a physics-constrained adversarial attack framework to simulate realistic threats targeting DRL controllers that can propagate across areas. First, we design and implement three typical hybrid control scenarios, i.e., single-agent DRL, partial-agent DRL, and full-agent DRL. Second, we propose a key-feature selection method based on gradient saliency, and we design an attack strategy that adheres to physical constraints while maintaining stealth and efficiency. Third, to enhance the system’s resiliency, we propose a two-stage active defense strategy highly compatible with the hybrid architecture. Finally, we conduct extensive simulation experiments under three typical hybrid control scenarios to evaluate the impact of the adversarial attack and the performance of the defense strategy.

随着电力系统智能化转型，引入了大量的物联网设备，带来便利的同时也引入了网络攻击问题，且大量新能源的并网导致电力系统的状态变化更加多样、频繁，对状态估计的鲁棒性和快速性提出了更高的要求。然而，目前状态估计(State Estimation, SE)方法更多关注于估计精度提升，忽略了状态估计的计算速度和鲁棒性。为解决上述问题，本文提出了一种快速、鲁棒、精确、安全的SE方案，其由攻击检测器、数据重构器和状态估计器串行组成。首先，本文提出了一种鲁棒的攻击检测器，即使电网拓扑变化也能够精确检测攻击数据。其次，本文提出了一种鲁棒的数据重构器，其能够精度、鲁棒地将攻击数据清洗为正常数据，从而保证数据的可用性。此外，本文提出了一种基于图形处理单元（graphics processing unit, GPU）的快速、鲁棒的状态估计器，结合考虑稀疏存储和重排序等方法，进一步加快SE计算。最后，我们进行了大量实验证明了上述方案的可行性和鲁棒性。

The Industrial Internet of Things (IIoT) has become a crucial technology driving the transformation of manufacturing towards intelligence. Traditional industrial networks primarily focus on data collection, whereas emerging IIoT emphasizes intelligent management and optimized control utilizing advanced technologies such as big data and artificial intelligence. With the more complex application scenarios, sole reliance on cloud computing fails to meet requirements for low latency and high security, thus leading to a trend towards edge-cloud collaborative computing. Knowledge-Defined Networking (KDN) introduces a knowledge plane above the control and data planes, which utilizes machine learning and deep learning methods to extract and analyze data and can guide the control plane in updating and optimizing network management and control decisions. This architecture enables the system to understand domain knowledge, supports knowledge-based reasoning decisions, and facilitates higher-level intelligent applications. This paper proposes a Knowledge-Defined Edge-Cloud Collaborative IoT architecture tailored for industrial monitoring scenarios. Therein, the edge is responsible for initial real-time data processing, while the cloud handles large-scale data storage and complex analytics. Knowledge definition are introduced into the entire framework and enables a comprehensive understanding of equipment status through historical data analysis and real-time feedback, which facilitates predictive issue identification and optimization of production plans. The enabling technologies that can implement the functions of the knowledge-based cloud, knowledge-based edge, knowledge-based end, and interfaces are summarized, which are expected to provide guidance for future IIoT.

With the rapid development of the Industrial Internet of Things (IIoT), programmable logic controllers (PLCs) are becoming increasingly intelligent, leading to improved productivity. However, this also brings about a growing number of security vulnerabilities. As a result, efficiently identifying potential security vulnerabilities in PLCs has become a crucial research topic for security researchers. This article proposes a method for fuzzing industrial proprietary protocols to effectively identify security vulnerabilities in PLCs’ proprietary protocols. The aim of this study is to develop a protocol fuzzing approach that can uncover security vulnerabilities in PLCs’ proprietary protocols. To achieve this, the article presents a protocol structure parsing algorithm specifically designed for PLC proprietary protocols, utilizing information theory. Additionally, a fuzzing case generation algorithm based on genetic algorithms is introduced to select test cases that adhere to the format specifications of the proprietary protocol while exhibiting a high degree of mutation. The research methodology consists of several steps. Firstly, the proposed protocol structure parsing algorithm is used to analyze two known industrial protocols, namely Modbus TCP and S7Comm. The parsing results obtained from the algorithm are then compared with the correct results to validate its effectiveness. Next, the protocol structure parsing algorithm is applied to analyze the proprietary protocol formats of two PLC models. Finally, based on the analysis results, the PLCs are subjected to fuzzing. Overall, the proposed protocol fuzzing approach, incorporating the protocol structure parsing algorithm and the fuzzing case generation algorithm, successfully identifies two denial-of-service vulnerabilities in the PLCs’ proprietary protocols. Notably, one of these vulnerabilities is a zero-day vulnerability, indicating that it was previously unknown and undisclosed.

This paper is concerned with periodic event-triggering laws in robust model predictive control (MPC) for continuous-time constrained nonlinear systems. The online optimal control problem solved at triggering times is introduced. A periodic static event-triggering condition, in which a fixed sampling time interval plays an important role in avoiding Zeno behavior, is presented to alleviate continuous checking of event detections. Then a periodic dynamic event-triggering condition is investigated to further enlarge the minimal inter-execution time. Single-mode MPC with a prediction horizon larger than the control one is considered. Sufficient conditions of recursive feasibility for the online optimal control problem are derived. In order to relax the sufficient condition of stability, Ultimately Boundedness properties are utilized in stability analysis. Finally, numerical simulation is provided to demonstrate the effectiveness of the proposed methods.

With the development of the Industrial Internet of Things (IIoT) technology, edge computing is a promising area to release the sensing and computing burdens from the overloaded center. However, in edge network scenarios, we cannot trust every node’s output since some nodes can behave maliciously by making use of the properties, such as multiple identities, heterogeneous capabilities, and mobility. In that case, trust management is widely used to solve the problem of network trustworthiness. In this article, we propose a trust evaluation framework by comprehensively considering the nodes’ malicious behaviors and heterogeneous characteristics of edge networks. Under the Bayesian framework, we use the semi-ring theory to dynamically establish mobile-edge nodes’ trust models. First, we calculate the trust value for each node with a different identity (service provider or requester). Then, we propose a security-regarded task allocation mechanism to improve the reliability of selected trusted nodes according to the matched relationship between the service requesters’ expected capability and the providers’ actual capability. Further, we conduct extensive analysis and simulations to evaluate the proposed methods in typical IIoT scenarios. The results show that the proposed method has better immunity to abnormal behaviors, including the noncooperation, malicious feedback, on–off attacks, Sybil attacks, whitewashing attack, malicious access, etc., and has higher scheduling accuracy and controllable time complexity compared to existing methods.

Trusted computing is a typical active defense technology. Trust management is a core support technology of trusted computing. However, when trust management is applied in the industrial control systems, how to identify malicious behavior effectively, model trust relationships, and make a decision based on behavior trustworthiness, meanwhile how to ensure deployed trust mechanism does not affect the control network’s availability, is a significant issue that has not been solved in the previous literature. This paper proposes a trust management method against abnormal behavior of industrial control networks under active defense architecture. Firstly, we review the difficulties of trust management when applied to industrial control networks and analyze abnormal behaviors of the control operations under unknown threats. Then we extract trust information, model the trust relationship of abnormal behaviors, and establish a trust update and decision-making mechanism under the availability constraints of industrial control networks. Furthermore, we provide a deployment method of the proposed trust management in a distributed control network. Finally, we take five typical abnormal operations on control instruction in an industrial control network as an example and perform a detailed analysis and experimental verification of the proposed method. The results prove that the proposed trust management method has good immunity to abnormal behaviors of the control flow and can be deployed in an industrial control system with availability constraints.

With the advent of Industry 4.0, water treatment systems (WTSs) are recognized as typical industrial cyber-physical systems (iCPSs) that are connected to t

Proprietary (or semi-proprietary) protocols are widely adopted in industrial control systems (ICSs). Inferring protocol format by reverse engineering is im