Large Language Models (LLMs) are increasingly deployed in time-critical systems, such as robotics, autonomous driving, embodied intelligence, and industrial automation, where generating accurate responses within a given time budget is crucial for decision-making, control, or safety-critical tasks. However, the auto-regressive generation process of LLMs makes it challenging to model and estimate the end-to-end execution time. Furthermore, existing efficient inference methods based on a fixed key-value (KV) cache eviction ratio struggle to adapt to varying tasks with diverse time budgets, where an improper eviction ratio may lead to incomplete inference or a drop in response performance. In this paper, we propose TimeBill, a novel time-budgeted inference framework for LLMs that balances the inference efficiency and response performance. To be more specific, we propose a fine-grained response length predictor (RLP) and an execution time estimator (ETE) to accurately predict the end-to-end execution time of LLMs. Following this, we develop a time-budgeted efficient inference approach that adaptively adjusts the KV cache eviction ratio based on execution time prediction and the given time budget. Finally, through extensive experiments, we demonstrate the advantages of TimeBill in improving task completion rate and maintaining response performance under various overrun strategies.

In recent years, the incorporation of early exit layers into deep neural networks has allowed inference to terminate earlier while maintaining accuracy. However, the passive decision-making involved in the these static exit placement creates a dilemma: fine-grained placement may cause high performance and energy overhead due to frequent exit layer execution, while coarse-grained placement may miss early exit opportunities. Moreover, common energy-saving techniques like adjusting processor configurations are not applicable once inference begins. To overcome these challenges and improve computation and energy efficiency, we propose LEAP, a software-hardware co-design approach. On the software side, LEAP proactively predicts exit points at runtime, reducing computation by enabling early exits without requiring every pre-placed exit layer to be executed. On the hardware side, LEAP adjusts processor settings—such as frequency and voltage—based on single or multiple predicted exits to optimize energy consumption while adhering to latency requirements. Extensive experimental results show that LEAP significantly improves efficiency. Compared to standard inference, LEAP reduces computation by up to 76.4% and saves up to 83.2% in energy. Compared to state-of-the-art early exit methods, LEAP achieves up to 27.9% less computation and 57.1% more energy savings, while maintaining similar accuracy and latency.

Edge computing empowers the new generation of industrial cyber-physical systems to perform efficient distributed sensing, even under high data loads and frequent transmission demands. In the sensing process, observability is essential for complete state estimation and subsequent precise control. However, observability guarantee has become increasingly challenging due to the growing scale of sensing networks and the deployment constraints of sensing devices in industrial environments. For this problem, an observability guaranteed hybrid wired/wireless distributed edge sensing method is proposed, which optimizes accuracy and efficiency while guaranteeing observability. The dynamics-aware structural observability is proposed to bridge dynamics and observability under sensor scheduling. The capability of the system to achieve observability is quantitatively analyzed, and a novel necessary and sufficient condition for observability guarantee is derived. Furthermore, based on observability analysis and topology of networks, an energy-efficient heuristic algorithm is developed, which assigns wired transmissions between selected sensor–edge computing unit pairs for observability guarantee. Besides, deep reinforcement learning methods are adopt to improve sensing performance in the sense of expectation for wireless sensor scheduling, overcoming the difficulty of analytically expressing the objective function. Finally, our proposed method is applied to slab temperature estimation in the industrial hot rolling process, and its effectiveness is fully verified by simulation results.

Over the past years, intelligent control tasks, such as deep neural networks (DNNs), have demonstrated significant potential in control systems. However, deploying intelligent control policies on heterogeneous computing platforms presents open challenges. These challenges extend beyond the apparent conflict between intensive computation and timing constraints and further encompass the interactions between task executions and complicated control performance. To address these challenges, this paper introduces RICH, a general and end-to-end approach to facilitate intelligent control tasks on heterogeneous computing architectures. RICH incorporates both offline Control-Oriented Computation and Resource Mapping (CCRM) and runtime Most Remaining Accelerator Segment Number First Scheduling (MRAF). Given the control tasks, the CCRM starts with balancing the computation workloads and processor resources with the goal of optimizing overall control performance. Subse-quently, the MRAF employs segment-level real-time scheduling to ensure the timely execution of tasks. Extensive experiments on the robotic arms applications (by hardware-in-the-loop simulator) demonstrate that the RICH can work as a general and end-to-end approach. These experiments reveal significant improvements in control performance, with enhancements of 50.7% observed for intelligent control applications deployed on heterogeneous computing platforms.

Wireless control systems (WCSs) are gaining rapid development in industrial automation. Compared to the star topology, mesh networks offer greater compatibility for large-scale applications that require high reliability, scalability, and extended coverage. In WCSs, multiple control loops share the multi-hop mesh network, leading to non-negligible and long- span communication latency in critical flows, which can severely degrade the overall control performance. Additionally, the criticality of each control flow largely depends on the features of the physical plant dynamics and the mesh network configuration, which is essential to properly and exactly represent. Moreover, the online scheduling and reconfiguration for large-scale mesh network for WCSs also pose unique challenges. In this paper, we propose a mesh network scheduling mechanism based on cyber-physical sensitivity. Firstly, we model each control loop as a switched system to represent the impact of arbitrary and fluctuating communication latency. Second, we propose a novel online criticality indicator, cyber-physical sensitivity (CP-Sensi), which accurately reflects the criticality of each control flow by synthesizing the switched model, runtime physical states, and network conditions. Finally, we design a CP-Sensi-based scheduling mechanism and an efficient piggyback-based network reconfiguration protocol tailored for mesh networks. Extensive studies with 12 control loops demonstrate that the proposed CP-Sensi and online mesh network scheduling achieve superior control performance compared to state-of-the-art approaches.

As industrial control applications scale up, mesh networked control systems (MNCSs) are gaining popularity, where multiple control loops share a multi-hop mesh network. However, these loops often suffer from long-span, time-varying delays caused by the multi-hop transmissions of multiple flows, which significantly degrade control performance, particularly stability. Existing studies on delay-aware stability conditions are usually independent with network scheduling, leading to a pessimistic stability analysis. Meanwhile, existing stability-aware scheduling approaches rely on coarse-grained designs, further worsening stability guarantees and limiting network capacity. In this work, we propose a stability-guaranteed scheduling mechanism for MNCSs with fine-grained timing. We first establish a stability condition that accounts for time-varying delays over an extended horizon spanning multiple superframes, which reduces the pessimism in stability analysis and enables more refined scheduling strategies. Based on this condition, We design a Long time-horizon and Fine-grained network scheduling mechanism with Stability Guarantee (LFSG), which deterministically maps the stability condition into the fine-grained network scheduling, considering fluctuating delays over the extended horizon. Furthermore, we provide a stability-capacity-aware LFSG (SCA-LFSG), which aims to maximize the number of stabilizable control loops and demonstrates its effectiveness through various application paradigms. Extensive studies demonstrate the advantages of stability analyses, LFSG, and SCA-LFSG over state-of-the-art approaches in terms of both control and timing performance.

The growing complexity and stringent real-time demands of autonomous systems, such as self-driving cars and drones, have driven the adoption of intelligent control methods based on deep neural networks (DNNs). While these methods offer improved control performance over traditional modelbased approaches, they also pose significant computational challenges, particularly for resource-constrained platforms. FieldProgrammable Gate Arrays (FPGAs) offer an attractive solution due to their energy efficiency and customizable architecture. In this work, we propose FALCON, an innovative approach for designing real-time intelligent controllers for autonomous systems using FPGA accelerators. Our approach begins with designing DNN-based intelligent controllers with varying levels of complexity and accuracy on the FPGA platform. Then, a performance function is proposed to capture the interplay among controller complexity, computational behavior, physical system characteristics, and overall control performance. Based on this performance function, we develop an algorithm-hardware codesign framework to determine the optimal control complexity, hardware configuration, and resource allocation. Finally, a case study on the co-design of intelligent controllers and FPGAbased overlay processors, together with a hardware-in-the-loop simulator, is conducted to demonstrate the advantages of the proposed methods. Compared to benchmarking controllers on other platforms, FALCON's optimized intelligent controller using FPGA accelerators shows competitive control performance with superior real-time capability and power efficiency. FALCON's optimization reduces the worst-case response time (WCRT) by up to 46.52%, improves the control performance by$1.93 \times$compared to the default setup. For performance per power efficiency, FALCON achieves a$3.67 \times$improvement compared to the DNN intelligent controller on TX2 and a remarkable$30.78 \times$improvement compared to traditional MPC on CPU.

With the improvement of edge-based autonomous systems such as mobile Industrial IoT (IIoT) networks, edge devices can capture and upload videos with increasing bitrates. Massive edge-computing end nodes are eager for adequate multimedia data to satisfy the requirements of real-time video services. However, existing encoding standards for video services in Web 2.0 are specifically designed for something other than IoT video streaming. We have improved our Adaptive Compression-Reconstruction (ACORN) framework to obtain ACORN+, based on compressed sensing and recent advances in deep learning. At end nodes, we compress multiple sequential video frames into a single frame to reduce video volume. Given that multiple kinds of intelligent tasks are expected to be finished on the device side, we also designed a device-cloud collaboration scheme where deep learning-based algorithms can be executed on both the device and server sides. Experiments reveal that video analytics can be conducted on compressed frames. Taking action recognition as a device-cloud collaboration use case, we find ACORN \(+\) obtains more than 3 \(\times\) speedup on compressed frames. The reconstruction algorithm in ACORN \(+\) is with 1– 4 dB improvements. Moreover, the encoding time cost and the encoded video volume are reduced by more than 4 \(\times\) under the ACORN \(+\) framework. 1

Heterogeneous multi-core architectures are gaining popularity in recent years as they combine the benefits of different processors, resulting in improved execution capacity and energy efficiency. However, analyzing response times and allocating resources for the typed directed acyclic graph (DAG) task, which has complex execution logic, on heterogeneous multi-core systems poses significant challenges. Major approaches may yield overly pessimistic worst-case response time (WCRT) estimates in certain scenarios while failing to adequately address critical structural characteristics inherent to typed DAG tasks. To address these limitations, this article explores the WCRT analysis and core allocations for the typed DAG task under partitioned scheduling. In this work, we first delve into the characteristics of the topology structure of the typed DAG task and propose a novel WCRT upper bound to enhance the accuracy of WCRT analysis. Then, a subtask allocation strategy is presented, which enables an effectively utilization of the resources of multi-cores. Finally, the performance of the proposed analysis algorithm and allocation strategy are tested by implementing a verification system on a real heterogeneous multi-core platform. Experimental results demonstrate that our proposed WCRT analysis algorithm exhibits substantial improvements of 38.7% and 37.43% in the theoretical analysis performance and actual analysis accuracy, respectively. Similarly, our proposed core allocation strategy improves the theoretical and the actual execution efficiency of the system by 10.6% and 7.41%, respectively. These results substantiate the practical value of our enhanced WCRT derivation methodology and allocation scheme in improving system resource utilization efficiency.

Massive field data is wirelessly transmitted to the edge side to facilitate sensing and control in the emerging Industrial Internet of Things (IIoT) systems. Under the expanding transmission scheduling space and dynamic network conditions, balancing control performance and limited transmission resources is a fundamental challenge. For this problem, we propose a novel deep reinforcement learning (DRL)-based transmission scheduling method (DTSM), where sensing performance guarantee is introduced for its criticality in ensuring complete system observation and effective control. Specifically, taking system observability as the key metric, the time slots for multi-sensor data transmission under different control demands are properly reserved with theoretically guaranteed performance. Then, the primal-dual DRL framework is adopted to further improve the overall performance of system control and resource utilization by dynamically scheduling the transmission number of each sensor. The scheduling is based on the real-time states of sensing and wireless network, and the action space is determined according to our reserved time slots. Besides, after primal-dual updates, the scheduling results can satisfy the estimation error-evaluated constraint imposed for the ultimate control effect. Finally, the proposed method is applied to the industrial laminar cooling process and its effectiveness is fully demonstrated. Note to Practitioners—This paper is motivated by the requirement of balancing control performance and scarce transmission resources in industrial automation fields such as steel manufacturing, where massive sensor data is transmitted to the edge side through wireless networks. The expanding transmission scheduling space and dynamic network conditions have led to increased interest in advanced deep reinforcement learning (DRL) methods. However, few previous works have explored the impact of control demands on intelligent transmission scheduling design. For these issues, we propose a novel DRL-based transmission scheduling method (DTSM), where the time slots for multi-sensor data transmission are delicately reserved according to different control demands and dynamic scheduling is realized based on real-time states of sensing and wireless network. The overall performance of system control and resource utilization is improved, and practitioners can easily adjust method parameters to achieve the desired balance between the two aspects according to practical demands. Case studies in the industrial hot rolling process demonstrate the superiority of DTSM. Our future work will consider the joint scheduling of uplink-downlink transmissions and design the collaboration among multiple edge computing nodes (ECNs) to address the limitations of centralized learning methods. Besides, the proposed method can be extended to other industrial applications such as flight control system testing.

Flight control system testing (FCST) is one of the most important process to check whether flight control surfaces can operate properly according to commands during aircraft assembly. Traditional testing method relies heavily on manual labor, leading to low efficiency and inconsistent quality. In this paper, we apply digital twin (DT) technology to the FCST process for the first time. We firstly design an architecture of DT-enabled FCST including four layers to support further development. Then, we present a triangular mesh alignment-based angle measurement (TMA-AM) algorithm to efficiently collect deflection angle data for DT-enabled FCST. Extensive experiments conducted on a aircraft wing subassembly platform show that the TMA-AM algorithm achieves an average angular measurement error of less than 0.1°, outperforming existing methods. Moreover, we develop a virtual experimental platform named DT-FCST aligned with a real aircraft wing subassembly platform. In addition, TMA-AM algorithm is integrated with the DT-FCST platform. By integrating real-time data from the cockpit, real-time physical-virtual interaction of aircraft control sticks and flight control surfaces are achieved, ensuring consistency between physical and virtual movements. The integration of DT technology with the TMA-AM algorithm enables real-time synchronization, monitoring, and unified data management, significantly enhancing the efficiency and accuracy of the FCST. Note to Practitioners—To address the inefficiencies and low monitoring quality associated with traditional manual testing methods in flight control system testing (FCST), we firstly introduce digital twin (DT) technology to this process. To support effective and accurate measurement during the FCST, we propose a vision-based method tailored to accurately measure deflection angles of flight control surfaces. This method replaces manual measurements with a non-contact approach, significantly improving measurement accuracy and efficiency. We provide a detailed description of the construction process of the DT-FCST platform including requirement analysis, DT model construction, and on-site experiments. This DT-based approach achieves real-time synchronization between virtual and physical testing processes, enhancing monitoring quality and overall testing effectiveness. Specifically, it can achieve a 90% reduction in the number of operators and shorten the single testing time to 16.7% of the traditional testing method.

As modern control systems, which integrate cutting-edge technologies, such as edge computing and data analytics, are transforming industries, industrial control systems have evolved into multitier cyber-physical systems. Most current work targets single control loop or multiple independent control loops. However, in practice, multiple control loops may have close relations and dependencies among each other. It is nontrivial to optimize control performances of multiple control loops with interleaving constraints under tight timing restrictions of real-time control, while guaranteeing stability. In this work, we propose a cloud-edge cooperation framework tailored for intelligent multi-loop edge control. We design an intelligent multi-loop edge controller (IMEC) to efficiently generate the control policy for multi-loop control systems, leveraging both analytical and data-driven approaches. And we apply the transfer learning (TL) method to improve the robustness and efficiency under varying control scenarios. Furthermore, we present an edge-end cooperative Simplex architecture (EESA) to guarantee the stability of the system. Extensive evaluation over two case studies, linear and nonlinear multi-loop control systems, on semi-physical simulation platforms illustrate that IMEC significantly reduces the computation cost and improves control performance compared with analytical optimal control. Transfer learning effectively enhances the robustness under different control scenarios. And EESA can guarantee the safety during the running time.

It is one of the most essential processes in the industrial cyber-physical system (ICPS) that multiple edge computing nodes (ECNs) collect field sensor information and cooperate for sensing and control. The exchange of sensing information on the edge side is critical for these ECNs that serve as multiple edge estimators and one edge controller. However, the limited transmission resources and the diverse performance demands of ECNs for sensing and control make it challenging to design the edge network topology delicately. For this issue, a novel dynamic observability (DO) condition is proposed to balance sensing-control performance and transmission cost under various demand settings. Based on the quantitative analysis of the relationship between overall transmission cost and each ECN's effective observability, DO gives the criterion for desirable network topologies with spatio-temporal dynamics. Then for the given performance demands, a dynamic observability guaranteed method (DOGM) is proposed to determine the network topology by triggering proper sensing links. In this way, the set of triggered sensing links may vary in a dynamic manner to satisfy the DO condition, and the overall performance of sensing and control is theoretically guaranteed. Finally, the comprehensive advantages of DOGM are demonstrated by the simulation study in the hot rolling laminar cooling process.

Joint estimation is crucial in the industrial Internet of Things (IIoT) by integrating data from diverse devices to improve monitoring accuracy. Network slicing can meet the heterogeneous needs of devices through logical isolation. However, existing methods often overlook the interaction of multiple slices on estimation performance, leading to potential estimation bias and ineffective resource costs. To address this, we propose an Age of Task (AoT)-driven associated network slicing method tailored for joint estimation scenarios. Specifically, we design an association-oriented slicing architecture for joint estimation that considers both the heterogeneous requirements of individual slices and the interactive effects of multiple slices. We define slice association based on the AoT to quantify the coupling relationship between slicing strategies and estimated performances. Moreover, we develop a dynamic-fitness multivariable particle swarm optimization algorithm to achieve associated slicing. Simulation results show that the associated slicing scheme achieves a flexible balance between timeliness and accuracy.

Flight control system testing (FCST) is one of the most important testing during aircraft final assembly, however, traditional testing method highly rely on manual labor, resulting in low testing quality and efficiency. Targeting at improving the testing quality and efficiency, in this paper, we apply digtial twin (DT) technology in the FCST process for the first time. A virtual experimental platform named DT-FCST is constructed which is identical to a real wing experimental platform, supporting testing elements management and physical-virtual mapping experiments. The development process of the DT-FCST platform is detailed. Several experiments are conducted to demonstrate the effectiveness of DT technology. Based on motion scripts integrating real-time data from the cockpit, the cockpit control sticks and flight control surfaces are driven to achieve consistency between physical and virtual motions. With the help of DT technology, the entire movement process of control sticks and flight control surfaces are mapped with high fidelity in the DT-FCST platform, greatly enhancing the testing efficiency and monitoring quality.

During the flight control system testing (FCST), it is crucial to accurately measure the deflection angles of flight control surfaces to determine whether they respond precisely to commands. However, the traditional measurement method, which relies on the manual use of angle measuring rulers for inspections, is inefficient, prone to wear, and lacks precision. To address this issue, we introduce a vision-based angle measurement method for FCST that replaces manual measurements, thereby significantly enhancing testing efficiency and accuracy. Our proposed triangular mesh alignment based angle measurement algorithm (TMA-AM) is a non-contact measurement method that involves two key procedures, capturing 3D coordinates from images and calculating deflection angles. The TMA-AM algorithm converts deflected angles into angular differences between two coordinate systems, while accounting for the curved characteristics of control surfaces. We evaluate TMA-AM algorithm on a 3D-printed wing test platform and an aircraft wing test platform. Experimental results demonstrate that our method achieves high accuracy, with an average angular measurement error below 0.05○.

Modern control systems, including robotics, drones, and autonomous vehicles, are increasingly incorporating intelligent controllers such as deep neural networks (DNNs) supported by heterogeneous processors. However, unlike conventional control algorithms on homogeneous platforms, the design and runtime execution of intelligent control tasks on heterogeneous computing platforms pose more rigorous demands and substantial challenges. These challenges encompass not only inherent conflicts between algorithm complexity and accuracy but also the couplings and trade-offs among run-time execution latency, end-to-end system performance, and reliability with timing constraints. To address these challenges, this paper introduces an end-to-end capability and scheduling co-design approach to efficiently design intelligent control tasks on heterogeneous computing architectures. We first introduce a novel and general control capability function, which bridges the control performance with the complexity of the intelligent controller, computation latency, and the properties of the physical plants. Subsequently, we formulate a comprehensive optimization problem to properly design algorithm capability and assign limited heterogeneous computational resources from offline heterogeneous resource allocation to run-time execution. Finally, we present a case study on the intelligent control of autonomous quadcopters (with the hardware-in-the-loop simulator built on Microsoft AirSim), and the extensive experiments demonstrate the superiority of the capability and scheduling co-design in terms of overall system performance compared with state-of-the-art design approaches.

Modern control systems are embracing multi-tier architectures integrating end devices and edge servers. However, due to the distinct control performance demands associated with each control task, it is a formidable challenge to optimize the control performance of multiple control tasks subject to stringent computation resource constraints while guaranteeing stability. Moreover, inherent contradictions exist in the timing aspect between the stability guarantee, which relies on offline analysis, and the run-time control performance, which should be enhanced online. It is essential to bridge the gap between the real-time scheduling of control tasks and their actual control performance. In this paper, we propose a novel real-time scheduling approach for multi-tier control systems, which leverages end devices for executing real-time control tasks and edge devices for runtime coordination. Specifically, we first introduce a new datadriven value function, called time/state/utility functions (TSUF), for modeling control system performance. TSUF captures not only timing but also the dynamic states of the physical plants. Subsequently, we propose value-based control scheduling (VCS), which is a multi-granularity scheduling mechanism based on our TSUF value function. VCS distinguishes the scheduling of stability jobs for ensuring system stability and performance jobs for optimizing real-time control performance based on run-time physical states. Finally, through realistic case studies involving multiple control loops, we demonstrate the advantages of VCS over existing scheduling approaches in terms of both control and real-time performance.

Industrial Internet of Things (IIoT)-based control is growing rapidly, such as smart factories and industrial automation. Sensing and transmitting physical state measurements is the first step and the prerequisite for IIoT-based control. However, sensor interference (e.g., electromagnetic interference on sensing, temperature, and humidity variations in the field) and network interference (e.g., metal obstacles and background noises) may destroy the control performance by interfering with sensing and communication processes. Most of the present upstream “fixed sensors-networking-state estimation” approaches cannot effectively deal with sensor and network interferences due to the fixed measurements/estimation and network resource limitations. To optimize the performance of IIoT-based control, we propose a smart sensing and communication co-design (SSCC) framework to select more potential sensors and establish the corresponding network scheduling. SSCC consists of a smart estimator (SE) and a sensing communication mode switching (SCMS) agent. The SE detects sensor interference and obtains resilient state estimation based on collaborative sensing. SCMS agent dynamically switches sensor selections and network configurations (routing and transmission number) in an integrated manner based on the network and plant states by solving a performance optimization problem. We propose a lightweight SCMS approach by searching a predefined mode table. We perform simulations integrating TOSSIM and MATLAB/Simulink, and semi-physical experiments on a real wireless sensor-actuator network composed of TelosB nodes. The results show that the SSCC framework can effectively improve the control performance and enhance network energy efficiency under various types of interference by dynamically selecting sensors and allocating network resources.

During the flight control system testing (FCST) process, multiple testing tasks should be completed. Battery-powered wireless sensors are used to measure the motion state of each flight control surface. In this paper, we investigate a multi-task scheduling problem to enhance overall monitoring accuracy during the FCST process. However, the decline in sensor battery levels, along with limited time slot resources, impacts the transmission quality of measurement data, leading to reduction in monitoring accuracy. Thus, we analyze the relationship among battery levels, transmission power, and monitoring accuracy to transform the original problem into an expectation probability maximization problem. Three important factors of monitoring accuracy are identified, based on which, we present the accuracy-oriented testing task scheduling (AOTS) algorithm. To validate the effectiveness of AOTS algorithm, we compare its performance among three different scheduling orders. Simulation results demonstrate that AOTS algorithm can not only improve the testing accuracy, but also reduce the fluctuation in accuracy among all testing tasks. Additionally, there are various elements in the FCST process that need to be uniformly managed to enhance the level of digitization. To address this issue, we design a digital twin enabled FCST (DT-FCST) system to manage data, models and algorithms in the FCST process. Finally, we implement the AOTS algorithm into developed DT-FCST system.

The booming development of 5G and Internet of Things (IoT) technologies significantly promotes the revolution of industrial IoT systems. Age of Information (AoI) is expected to play a critical role in industrial IoT systems, especially for time-sensitive monitoring and control applications. In addition, edge computing (EC) will be leveraged to effectively support industrial tasks in the limited communication and computing resources environment, bringing threefold benefits of shorter end-to-end delay, improving information timeliness, and reduced communication burden. Thus, we propose an edge-assisted co-design architecture of sampling-scheduling-control to improve the overall system performance. Under this architecture, a new definition, Age of Task (AoT), is proposed first to measure the timeliness of multielement and compute-intensive monitoring tasks in the industry. By analyzing the coupling relationship between AoT and control performance, an analytical expression of estimation error based on AoT is derived. Furthermore, we prove that the optimal control law could be expressed in a certain equivalent form, making it possible to decompose the design of control and network resource allocation (sampling, scheduling). According to the relation between AoT and estimation error, a co-design method, event-triggered sampling and max-age-reduce-first scheduling (ETMA), is proposed to minimize the system cost, including control cost and network energy consumption. The simulation results show that our co-design scheme has the optimal system cost among the four state-of-the-art schemes.

Along with the fourth industrial revolution, industrial automation systems are evolving into a multi-tier end-edge computing architecture. Edge controllers, which are equipped with a larger computing capacity compared to local controllers, can communicate with local plants over mainstream wireless networks such as WirelessHART, Wi-Fi, and cellular networks. Well-known challenges induced by networks, such as uncertain time delays and packet drops, have been intensively investigated from various perspectives: control synthesis, network design, or control and network co-design. The status quo is that the industry remains hesitant to close the loop between the edge controller and the actuation side due to safety concerns. This work offers an alternative perspective to address the safety concern, by exploiting the design freedom of an end-edge computing architecture. Specifically, we present a smart actuation framework, which deploys (1) an edge controller, which communicates with physical plant via wireless network, accounting for optimality, adaptation, and constraints by conducting computationally expensive operations; (2) a smart actuator, which is co-located with the physical plant on the end tier and executes a local control policy, accounting for system safety in the view of network imperfections, (3) the end-edge control co-design strategies and cooperation logic for both performance and stability. For certain classes of plants, semi-globally asymptotic stability of the resulting end-edge control systems is established when the edge controller is the model predictive control (MPC), or policy iteration-based learning control. We also provide an adaptation strategy for the end-edge control systems facing model parameter mismatches when the edge controller employs reinforcement learning. Extensive simulations demonstrate the advantages of the proposed end-edge co-design and cooperation procedures. Note to Practitioners—Edge computing is gaining momentum in areas that require low latency and high efficiency, i.e., mobile computing, video analytics, and autonomous driving. Industrial automation systems are also evolving into a multi-tier end-edge computing architecture. It pays obvious dividends to leverage the cooperation between end and edge, benefiting from fast and reliable communication on the end side, and powerful computation capacity on the edge side. The current end-edge cooperation focuses on how to partition tasks and offload computation resources in order to minimize delay and energy consumption, as well as how to balance the tradeoff between them. However, the impacts of end-edge cooperation on the safety, optimality, and cost of industrial automation have not been systematically studied. This paper aims to tailor end-edge cooperation in a smart actuation framework, for industrial automation to reconcile the above aspects by leveraging co-design of end and edge controllers and their switching logic. Extensive pure and semi-physical simulations demonstrate the advantages in performance and system stability of the proposed end-edge co-design and cooperation procedures.

Although wireless control is one of the key technologies for future industries, most wireless networks are only used for monitoring. When wireless networks are applied to transmit control commands, the uncertain link qualities and limited network resources may destroy the performance of multi-loop control systems. Hence, it is critical to allocate these resources to optimize the control performance as the network condition changes and plants evolve. This article presents comprehensive optimal scheduling strategies for wireless control systems based on adaptive dynamic programming. First, we propose an effective adaptive dynamic programming scheduling (ADPS) strategy to solve the optimal scheduling problem based on the single-step control performance at runtime while significantly reducing computational complexity. Moreover, to overcome the “short-sightedness” of single-step performance prediction, we extend ADPS to ADPS-m ( m ulti-step prediction), which optimizes multi-step performance by incorporating a longer-horizon evolution of the plants. Furthermore, we propose ADPS-H ( H eterogeneous flow scheduling) to support heterogeneous flows with different data rates and sizes and ADPS-H-m ( m ulti-step prediction for H eterogeneous flow scheduling), which schedules heterogeneous flows in a longer prediction horizon. We prove that all these scheduling strategies can achieve optimality and stability under mild assumptions. Extensive experiments integrating TOSSIM and MATLAB/Simulink are performed to evaluate all of the proposed methods in case studies of four- and ten-loop control systems. The simulation results demonstrate that these strategies can effectively improve the control performance at lower computing costs under both cyber and physical disturbances. Under the noise level of \(-\) 76 dBm, for the four-loop case, ADPS achieves the same control performance as the linear programming while saving 99.5% of the execution time. ADPS-m further improves the control performance by up to 27.0% compared with ADPS at the prediction horizon of 3, and ADPS-H-m improves the performance by up to 32.3% and 8.4% compared with round-robin and ADPS-H, respectively. The ten-loop case indicates the effectiveness and scalability of the proposed approaches.

In the industrial Internet of Things, state estimation of large-scale physical systems is performed by multiple sensors in a distributed manner. However, frequent communications during an estimation interval can increase the energy consumption of sensors, causing thermal stress and reliability issues. Although system reliability can be improved by data compression, the compression-induced distortion may lead to the divergence of the estimation error. To address these challenges, a distributed estimation and communication co-design scheme is proposed in this article, which balances estimation performance and energy efficiency under the system lifetime reliability constraint. First, a consensus-based distributed estimation algorithm is proposed to adapt the data compression configuration. Then, the impact of system dynamics, network connectivity, and data compression configuration on estimation performance is investigated. Based on the relationship, the distributed estimation algorithm and the channel allocation with power control are jointly optimized to minimize the estimation error and energy cost under the system lifetime reliability constraint. This constrained minimization problem is formulated as a mixed-integer nonlinear programming problem and solved with the designed decomposition method. Finally, simulation results demonstrate that the proposed co-design scheme shows superiority in improving both the estimation accuracy and energy efficiency under the system lifetime reliability constraint.

The harsh industrial environment and the high exposure of wireless communication networks (WCNs) seriously degrade the control performance of edge-enabled Industrial Internet of Things systems. Recently, the codesign of control and scheduling has been studied as a promising method to improve system performance. However, due to the dynamic feature of multiple unreliable factors, the impact of communication randomness on data timeliness, and the difficulty to gather sensing data, it is challenging to jointly design the schedule and control policy to mitigate the adverse effects of WCNs. To address these issues, this article presents a trust-age of information (AoI)-aware codesign scheme (TACS). We first propose a learning-based trust model with the aid of a conditional generative adversarial network to handle the sparse industrial data and a deep-neural-network-based trust online prediction to comprehensively measure the WCNs' reliability. Then, we study the impact of AoI on control performance and design the optimal controller based on the separation principle. Moreover, we derive a trust-AoI-aware scheduling policy at the edge side to dynamically select the optimal data to participate in plant control, which maximizes the control system performance and the trust of WCNs. Simulation results reveal the effectiveness of the TACS in terms of improving the system performance significantly.

Industrial Internet of Things (IIoT)-based control is growing rapidly, such as smart factories and industrial automation. In practice, imperfect wireless networks and time delay caused by delayed completion of computing tasks in IIoT may deteriorate the control performance. To enhance the performance of the control system, a multidomain resource allocation problem is formulated by co-designing control, communication, and computation resources, which is a mixed-integer nonlinear programming (MINLP) problem. In this article, a bilevel optimization framework is proposed to solve the MINLP, in which the sharing decision is derived in the upper level, and then, the optimal allocation of multidomain resources is derived in the lower level. A control-aware distributed bilevel (CADB) algorithm is developed, where these two levels interact with each other. In each round, the upper level optimization problem is updated based on the last resource allocation and solved by a primal-decomposition algorithm with provable finite-time feasibility. Then, according to the newly derived sharing decision, the lower level optimization problem is solved by the proposed mixed proximal-gradient-tracking algorithm. It is shown that CADB algorithm enables control systems to achieve enhanced control performance and energy consumption. Finally, simulations are conducted to verify the effectiveness of the proposed algorithm.

Joint monitoring, by integrating observations from multiple types of equipment, is essential for a thorough understanding of physical processes in the Industrial Internet of Things (IIoT). However, it does demand sufficient resources to ensure reliable and timely delivery of such observations. Although network slicing is widely used to meet such heterogeneous requirements, it falls short in this system, because it causes interconnected impacts on system performance across multiple slices. In this paper, we introduce an innovative associated network slicing framework for joint monitoring, which focuses on system cost minimization while accounting for slice associations. Particularly, to better understand the characteristics, we introduce a new concept, Age of Inexact Task (AoIT), to capture inter-slice associations. We then decompose the optimization variables to facilitate efficient Associated Network Slicing (ANS) algorithmic design, leading to a closed-form solution for intra-slice small-timescale resource allocation and an iterative block coordinate gradient descent algorithm for inter-slice large-timescale resource allocation. Simulation results demonstrate that our proposed ANS balances heterogeneous requirements and associations, showing significant reductions in system costs compared to existing solutions.

By adding exiting layers to the deep learning networks, early exit can terminate the inference earlier with accurate results. However, the passive decision-making of whether to exit or continue the next layer has to go through every pre-placed exiting layer until it exits. In addition, it is hard to adjust the configurations of the computing platforms alongside the inference proceeds. By incorporating a low-cost prediction engine, we propose a Predictive Exit framework for computation- and energy-efficient deep learning applications. Predictive Exit can forecast where the network will exit (i.e., establish the number of remaining layers to finish the inference), which effectively reduces the network computation cost by exiting on time without running every pre-placed exiting layer. Moreover, according to the number of remaining layers, proper computing configurations (i.e., frequency and voltage) are selected to execute the network to further save energy. Extensive experimental results demonstrate that Predictive Exit achieves up to 96.2% computation reduction and 72.9% energy-saving compared with classic deep learning networks; and 12.8% computation reduction and 37.6% energy-saving compared with the early exit under state-of-the-art exiting strategies, given the same inference accuracy and latency.

Deep learning approaches, such as convolution neural networks (CNNs), have achieved tremendous success in versatile applications. However, one of the challenges to deploy the deep learning models on resource-constrained systems is its huge energy cost. As a dynamic inference approach, early exit adds exiting layers to the networks, which can terminate the inference earlier with accurate results to save energy. The current passive decision-making for energy regulation of early exit cannot adapt to ongoing inference status, varying inference workloads, and timing constraints, let alone guide the reasonable configuration of the computing platforms alongside the inference proceeds for potential energy saving. In this paper, we propose an Energy Efficient Neural Networks (EENet), which introduces a plug-in module to the state-of-the-art networks by incorporating run-time power management. Within each inference, we establish prediction of where the network will exit and adjust computing configurations (i.e., frequency and voltage) accordingly over a small timescale. Considering multiple inferences over a large timescale, we provide frequency and voltage calibration advice, given inference workloads and timing constraints. Finally, the dynamic voltage and frequency scaling (DVFS) governor configures voltage and frequency to execute the network according to the prediction and calibration. Extensive experimental results demonstrate that EENet achieves up to 63.8% energy-saving compared with classic deep learning networks and 21.5% energy-saving compared with the early exit under state-of-the-art exiting strategies, together with improved timing performance.

Edge sensing supported by wireless transmission is one of the core enabling technologies for flexibly implementing the Industrial Internet of Things (IIoT). Balancing network resource consumption and sensing accuracy under dynamic network conditions is a critical challenge. In this work, we bridge the gap between edge sensing performance and transmission design through observability analysis and learning-based methods. Particularly, utilizing observability probability as the key metric, we design the network resource reservation for specific sensing performance demands including stability based on our derived upper and lower probability bounds. Then, to further reduce the overall cost of edge sensing and transmission, an intelligent transmission scheduling method (ITSM) based on deep reinforcement learning is provided, which dynamically schedules the number of transmissions for each sensor. In ITSM, the action space is determined according to the amount of our reserved resources, and both the states of sensing error and fading channel are taken into account. Finally, the superiority of our proposed methods is fully demonstrated through numerical simulations in a typical IIoT system of industrial hot rolling.

It is witnessed that heterogeneous architectures, such as GPUs, TPUs, and FPGAs, have made complex algorithms practical in the last decade. Despite multiple efforts to study the scheduling of these parallel and complex tasks on heterogeneous architectures, the power and energy consumption of the platforms have yet to be well managed under real-time task deadlines. To establish high schedulability in heterogeneous architectures, many scheduling strategies and models, such as multi-segment selfsuspension (MSSS), have been proposed by pioneer researchers. However, directly applying this model to heterogeneous architectures with multiple CPUs and many processing elements (PEs) suffers aggravated power consumption due to the pessimism in the scheduling algorithm and the tolerance margin in the worst-case execution time (WCET) model. Therefore, this paper presents an energy-efficient real-time scheduling approach called EESchedule, which works on heterogeneous architectures with guaranteed schedulability and improved power efficiency. In EESchedule, we build a general task execution model for the general heterogeneous architectures integrating multiple CPUs and many PEs. Then, an energy-efficient real-time scheduling strategy is introduced. Next, the response time and corresponding schedulability analysis are presented for EESchedule. Finally, extensive experiments on heterogeneous NVIDIA Jetson TX2 embedded systems and GPU servers with the Intel i9-10900x CPU and RTX 3080 GPU demonstrate that the EESchedule could achieve the same schedulability with 16.8%-40.7% and 39.0%-48.2% reduced power and energy consumption in comparison with state-of-the-art scheduling algorithms.

IoT devices are enabled to capture and upload videos with increasing bitrates. Massive IIoT is eager for effective video processing techniques to satisfy the requirements of real-time video services. With the emergence of 5G-unlicensed (5G-U), ultra-low latency video applications become possible. However, existing encoding standards for video services in Web 2.0, such as H.265, are not naturally designed for IIoT video streaming, leading to bandwidth pressure where 5G-U coexists with various other wireless signals. To tackle this problem and to support low-latency video utilization by IIoT video sources, we propose an Adaptive Compression-Reconstruction framework named ACORN, which is based on compressed sensing and recent advances in deep learning. At end nodes, we compress multiple sequential video frames into a single frame to reduce video volume. We design a QoE-aware parameter selection mechanism to deal with volatile network environments during compression. With learnable gated convolution layers and channel-wise soft-thresholding operators, ACORN also builds a real-time reconstruction module. Experimental results reveal that video analytics can be conducted on compressed frames. The reconstruction algorithm in ACORN is with $1-4 \mathrm{~dB}$ improvements. Moreover, both the encoding time cost and the encoded video volume are reduced by more than $4 \times$ under the ACORN framework.

As industrial plants embrace modern technologies, such as edge computing and wireless networks, industrial control systems have evolved into multitier cyber–physical systems. While traditional local controllers enjoy reliable connectivity to sensors/actuators, they suffer from the limited computation capacity of embedded devices. In contrast, edge servers introduce more computation resources connected to sensors/actuators through wireless networks. Offloading control functions to edge servers presents new opportunities to enhance control performance but also poses critical challenges. As wireless networks have limited bandwidth and varying reliability, it is important to optimize control performance by dynamically offloading a subset of the control functions to edge servers under the bandwidth constraint. Furthermore, the selection of offloaded control functions depends on both the cyber (wireless) and physical states of the wireless control systems. In this article, we tackle the problem of optimizing the control performance of multiple control loops through dynamic edge offloading. We establish a data-driven model to predict the control performance of each feedback control loop based on its cyber–physical states. We then develop a dynamic edge offloading approach to optimize the overall control performance of a system with multiple feedback control loops while guaranteeing their stability under fluctuating cyber–physical conditions. Finally, we demonstrate the efficacy of the data-driven model and offloading approach in case studies comprising simulations of up to 20 industrial robots.

Aircraft final assembly line (AFAL) is a typical complex manufacturing system with multiple installation and test processes operating simultaneously at each workstation. Lots of robots and sensors are connected and operated for heterogeneous processes by sharing limited communication and computing resources. How to manage devices and resources in a coordinated and efficient way is thus very challenging. Digital twin (DT) is a powerful technology for multiple objects management in the complex assembly system. It enables us to coordinate various devices and allocate communication and computing resources at workstations. In this article, two main processes, i.e., vision-assisted installation and flight control system test, are considered in the AFAL. We introduce a DT-enabled AFAL system and propose a DT-assisted heterogeneous processes coordinated (DT-HPC) framework to coordinate various devices and resources at each workstation. The wirelessly connected robots and sensors are applied for perception and information fusion. In order to minimize the total energy consumption and computing resources of all the wireless devices, joint design of the wireless channel allocation, transmission power, and computing resource allocation are proposed to satisfy the diverse Quality-of-Service (QoS) requirements. First, we propose a priority-aware channel assignment (PACA) algorithm to allocate channels for sensors and robots. Then, the optimal computing resource allocation strategy for two processes is derived while guaranteeing the processing latency requirements. Next, we derive the minimum transmission power of wireless sensors to guarantee the monitoring accuracy and calculate the transmission power of robots to obtain the satisfied transmission rate. Finally, we apply the DT-HPC framework in the DT-enabled AFAL system. The simulation results prove that our proposed algorithms can save energy while guaranteeing different QoS requirements.

Over the last two decades, as microprocessors have evolved to achieve higher computational performance, their power density has also increased at an accelerated rate. Improving energy efficiency and reducing power consumption are therefore critically important to modern computing systems. One effective technique for improving energy efficiency is dynamic voltage and frequency scaling (DVFS). With the emergence of integrated voltage regulators (IVRs), the speed of DVFS can reach microsecond ($\mu \text{s}$) timescales. However, a practical and effective strategy to guide fast DVFS remains a challenge. In this article, we propose F-LEMMA: a fast, learning-based, hierarchical DVFS framework consisting of a global power allocator in the kernel space, a reinforcement learning-based power management scheme at the architecture level, and a swift controller at the digital circuit level. This hierarchical approach leverages computation at the system and architecture levels with the short response time of the swift controller to achieve effective and rapid$\mu \text{s}$-level power management supported by the IVR. Our experimental results demonstrate that F-LEMMA can achieve significant energy savings (35.2%) across a broad range of workloads. Conservatively compared with existing state-of-the-art DVFS-based power management schemes that can only operate at millisecond timescales, F-LEMMA can provide notable (up to 11%) energy-delay product (EDP) improvements across benchmarks. Compared with state-of-the-art nonlearning-based power management, our method has a universally positive effect on evaluated benchmarks, proving its adaptability.

Modern real-time systems contain complex workloads, which are usually modeled as directed acyclic graph (DAG) tasks and deployed on multiprocessor platforms. The complex execution logic of DAG tasks results in excessive schedulability analysis overhead, and the current DAG task allocation strategy cannot efficiently utilize processor resources (inner parallelization of DAG tasks). In this article, an invalid-edge deletion (IED) method is proposed to reduce the execution complexity of the DAG tasks while guaranteeing the correctness of the execution logic. Besides, we bound the number of complete paths for DAG tasks, which re-limits the searching space of the schedulability analysis. Then, a topology-based DAG tasks allocation (TDTA) strategy is developed, which reduces the interference caused by higher-priority DAG tasks to enable the full utilization of the processor resources. The experimental results show that the IED method effectively reduces the overhead of DAG task analysis, and the performance of the TDTA strategy is better than the performance of other state-of-the-art strategies.

Despite being employed in burgeoning efforts to accelerate artificial intelligence, heterogeneous architectures have yet to be well managed with strict timing constraints. As a classic task model, multi-segment self-suspension (MSSS) has been proposed for general I/O-intensive systems and computation offloading. However, directly applying this model to heterogeneous architectures with multiple CPUs and many processing units (PEs) suffers tremendous pessimism. In this paper, we present a real-time scheduling approach, SHAPE, for general heterogeneous architectures with significant schedulability and improved utilization rate. We start with building the general task execution pattern on a heterogeneous architecture integrating multiple CPU cores and many PEs such as GPU streaming multiprocessors and FPGA IP cores. A real-time scheduling strategy and corresponding schedulability analysis are presented following the task execution pattern. Compared with state-of-the-art scheduling algorithms through comprehensive experiments on unified and versatile tasks, SHAPE improves the schedulability by 11.1% - 100%. Moreover, experiments performed on the NVIDIA GPU systems further indicate up to 70.9% of pessimism reduction can be achieved by the proposed scheduling. Since we target general heterogeneous architectures, SHAPE can be directly applied to off-the-shelf heterogeneous computing systems with guaranteed deadlines and improved schedulability.

In cloud manufacturing systems, fault diagnosis is essential for ensuring stable manufacturing processes. The most crucial performance indicators of fault diagnosis models are generalization and accuracy. An urgent problem is the lack and imbalance of fault data. To address this issue, in this article, most of existing approaches demand the label of faults asa prioriknowledge and require extensive target fault data. These approaches may also ignore the heterogeneity of various equipment. We propose a cloud-edge collaborative method for adaptive fault diagnosis with label sampling space enlarging, named label-split multiple-inputs convolutional neural network, in cloud manufacturing. First, a multiattribute cooperative representation-based fault label sampling space enlarging approach is proposed to extend the variety of diagnosable faults. Besides, a multi-input multi-output data augmentation method with label-coupling weighted sampling is developed. In addition, a cloud-edge collaborative adaptation approach for fault diagnosis for scene-specific equipment in cloud manufacturing system is proposed. Experiments demonstrate the effectiveness and accuracy of our method.

Industrial automation traditionally relies on local controllers implemented on microcontrollers or programmable logic controllers. With the emergence of edge computing, however, industrial automation evolves into a distributed two-tier computing architecture comprising local controllers and edge servers that communicate over wireless networks. Compared to local controllers, edge servers provide larger computing capacity at the cost of data loss over wireless networks. This article presents switching multitier control (SMC) to exploit edge computing for industrial control. SMC dynamically optimizes control performance by switching between local and edge controllers in response to changing network conditions. SMC employs a data-driven approach to derive switching policies based on classification models trained based on simulations while guaranteeing system stability based on an extended Simplex approach tailored for two-tier platforms. To evaluate the performance of industrial control over edge computing platforms, we have developedWCPS-EC, a real-time hybrid simulator that integrates simulated plants, real computing platforms, and real or simulated wireless networks. In a case study of an industrial robotic control system, SMC significantly outperformed both a local controller and an edge controller in face of varying data loss in a wireless network.

Industrial automation is embracing wireless sensor-actuator networks (WSANs). Despite the success of WSANs for monitoring applications, feedback control poses significant challenges due to data loss and stringent energy constraints in WSANs. Holistic control adopts a cyber-physical system approach to overcome the challenges by orchestrating network reconfiguration and process control at run time. Fundamentally, it leverages self-awareness across control and wireless boundaries to enhance the resiliency of wireless control systems. In this article, we explore efficient holistic control designs to maintain control performance while reducing the communication cost. The contributions of this work are five-fold: (1) We introduce a holistic control architecture that integrates Low-power Wireless Bus (LWB) and two control strategies, rate adaptation and self-triggered control ; (2) We present heuristics-based and optimal rate selection algorithms for rate adaptation; (3) We design novel network adaptation mechanisms to support rate adaptation and self-triggered control in a multi-hop WSAN; (4) We build WCPS-RT, a real-time network-in-the-loop simulator that integrates MATLAB/Simulink and a physical WSAN testbed to evaluate wireless control systems; (5) We empirically explore the tradeoff between communication cost and control performance in holistic control approaches. Our studies show that rate adaptation and self-triggered control offer advantages in control performance and energy efficiency, respectively, in normal operating conditions. The advantage in energy efficiency of self-triggered control, however, may diminish under harsh physical and wireless conditions due to the cost of recovering from data loss and physical disturbances.

Wireless sensor-actuator networks (WSANs) are gaining momentum in industrial process automation as a communication infrastructure for lowering deployment and maintenance costs. In traditional wireless control systems, the plant controller and the network manager operate in isolation, which ignores the significant influence of network reliability on plant control performance. To enhance the dependability of industrial wireless control, we propose a holistic cyber-physical management framework that employs runtime coordination between the plant control and network management. Our design includes a holistic controller that generates actuation signals to physical plants and reconfigures the WSAN to maintain the desired control performance while saving wireless resources. As a concrete example of holistic control, we design a holistic manager that dynamically reconfigures the number of transmissions in the WSAN based on online observations of physical and cyber variables. We have implemented the holistic management framework in the wireless cyber-physicalsimulator (WCPS). A systematic case study is presented based on two five-state plants and a load positioning system using a 16-node WSAN. Simulation results show that the holistic management design has significantly enhanced the dependability of the system against both wireless interferences and physical disturbances, while effectively reducing the number of wireless transmissions.