This paper presents an innovative framework that boosts computing power by utilizing ubiquitous computing power distribution and enabling higher computing node accessibility via adaptive UAV positioning, establishing a UAV-enabled Computing Power Network (UAV-CPN). In a UAV-CPN, a UAV functions as a dynamic relay, outsourcing computing tasks from the request zone to an expanded service zone with diverse computing nodes, including vehicle onboard units, edge servers, and dedicated powerful nodes. This approach has the potential to alleviate communication bottlenecks and overcome the "island effect" observed in multi-access edge computing. A significant challenge is to quantify computing power performance under complex dynamics of communication and computing. To address this challenge, we introduce task completion probability to capture the capability of UAV-CPNs for task computing. We further enhance UAV-CPN performance under a hybrid energy architecture by jointly optimizing UAV altitude and transmit power, where fuel cells and batteries collectively power both UAV propulsion and communication systems. Extensive evaluations show significant performance gains, highlighting the importance of balancing communication and computing capabilities, especially under dual-energy constraints. These findings underscore the potential of UAV-CPNs to significantly boost computing power.

Collaborative Perception (CP) has been shown to be a promising technique for multi-agent autonomous driving and multi-agent robotic systems, where multiple agents share their perception information to enhance the overall perception performance and expand the perception range. However, in CP, an ego agent needs to receive messages from its collaborators, which makes it vulnerable to attacks from malicious agents. To address this critical issue, we propose a unified, probability-agnostic, and adaptive framework, namely, CP-uniGuard, which is a tailored defense mechanism for CP deployed by each agent to accurately detect and eliminate malicious agents in its collaboration network. Our key idea is to enable CP to reach a consensus rather than a conflict against an ego agent's perception results. Based on this idea, we first develop a probability-agnostic sample consensus (PASAC) method to effectively sample a subset of the collaborators and verify the consensus without prior probabilities of malicious agents. Furthermore, we define collaborative consistency loss (CCLoss) for object detection task and bird's eye view (BEV) segmentation task to capture the discrepancy between an ego agent and its collaborators, which is used as a verification criterion for consensus. In addition, we propose online adaptive threshold via dual sliding windows to dynamically adjust the threshold for consensus verification and ensure the reliability of the systems in dynamic environments. Finally, we conduct extensive experiments and demonstrate the effectiveness of our framework.

Given the limited computing capabilities on autonomous vehicles, onboard processing of large volumes of latency-sensitive tasks presents significant challenges. While vehicular edge computing (VEC) has emerged as a solution, offloading data-intensive tasks to roadside servers or other vehicles faces communication-computing bottleneck, such as signal blockage from other large vehicles and limited computing resources of roadside servers. To address these challenges, Reconfigurable Intelligent Surface (RIS) can be leveraged to create line-of-sight channels, mitigate interference on the ground, and extend connectivity to more edge servers by elevating RIS adaptively. To this end, we propose RAISE, an optimization framework for RIS placement in multi-server VEC systems. Specifically, RAISE optimizes RIS altitude and tilt angle together with the optimal task assignment to maximize task throughput under deadline constraints. To find a solution, a two-layer optimization approach is proposed, where the inner layer exploits the unimodularity of the task assignment problem to derive the efficient optimal strategy while the outer layer develops a near-optimal hill climbing (HC) algorithm for RIS placement with low complexity. Extensive experiments demonstrate that the proposed RAISE framework consistently outperforms existing benchmarks.

Collaborative Perception (CP) has shown a promising technique for autonomous driving, where multiple connected and autonomous vehicles (CAVs) share their perception information to enhance the overall perception performance and expand the perception range. However, in CP, ego CAV needs to receive messages from the collaborators, which makes it easy to be attacked by malicious agents. For example, a malicious agent can send harmful information to the ego CAV to mislead it. To address this critical issue, we propose a novel method, **CP-Guard**, a tailored defense mechanism for CP that can be deployed by each agent to accurately detect and eliminate malicious agents in its collaboration network. Our key idea is that CP will lead to a consensus rather than a conflict against the ego CAV's perception results. Based on this idea, we first develop a probability-agnostic sample consensus (PASAC) method that can effectively sample a subset of the collaborators and verify the consensus without prior probabilities of malicious agents. Furthermore, we design a collaborative consistency loss (CCLoss) to calculate the discrepancy between the ego CAV and the collaborators, which is used as a verification criterion for consensus. Finally, we conduct extensive experiments in collaborative bird's eye view (BEV) tasks and the results demonstrate the effectiveness of our CP-Guard.

This paper presents an innovative framework that synergistically enhances computing performance through ubiquitous computing power distribution and dynamic computing node accessibility control via adaptive unmanned aerial vehicle (UAV) positioning, establishing UAV-enabled Computing Power Networks (UAV-CPNs). In UAV-CPNs, UAVs function as dynamic aerial relays, outsourcing tasks generated in the request zone to an expanded service zone, consisting of a diverse range of computing devices, from vehicles with onboard computational capabilities and edge servers to dedicated computing nodes. This approach has the potential to alleviate communication bottlenecks in traditional computing power networks and overcome the "island effect" observed in multi-access edge computing. However, how to quantify the network performance under the complex spatio-temporal dynamics of both communication and computing power is a significant challenge, which introduces intricacies beyond those found in conventional networks. To address this, in this paper, we introduce task completion probability as the primary performance metric for evaluating the ability of UAV-CPNs to complete ground users’ tasks within specified end-to-end latency requirements. Utilizing theories from stochastic processes and stochastic geometry, we derive analytical expressions that facilitate the assessment of this metric. Our numerical results emphasize that striking a delicate balance between communication and computational capabilities is essential for enhancing the performance of UAV-CPNs. Moreover, our findings show significant performance gains from the widespread distribution of computing nodes.

To support the development of the Low Altitude Economy (LAE), it is essential to achieve precise localization of unmanned aerial vehicles (UAVs) in urban areas where global positioning system (GPS) signals are unavailable. Vision-based methods offer a viable alternative but face severe bandwidth, memory and processing constraints on lightweight UAVs. Inspired by mammalian spatial cognition, we propose a task-oriented communication framework, where UAVs equipped with multi-camera systems extract compact multi-view features and offload localization tasks to edge servers. We introduce the Orthogonally-constrained Variational Information Bottleneck encoder (O-VIB), which incorporates automatic relevance determination (ARD) to prune non-informative features while enforcing orthogonality to minimize redundancy. This enables efficient and accurate localization with minimal transmission cost. Extensive evaluation on a dedicated LAE UAV dataset shows that O-VIB achieves high-precision localization under stringent bandwidth budgets. Code and dataset will be made publicly available: github.com/fangzr/TOC-Edge-Aerial.

Large language models (LLMs) have achieved remarkable success in various tasks, such as decision-making, reasoning, and question answering. They have been widely used in edge devices. However, fine-tuning LLMs to specific tasks at the edge is challenging due to the high computational cost and the limited storage and energy resources at the edge. To address this issue, we propose TaskEdge, a task-aware parameter-efficient fine-tuning framework at the edge, which allocates the most effective parameters to the target task and only updates the task-specific parameters. Specifically, we first design a parameter importance calculation criterion that incorporates both weights and input activations into the computation of weight importance. Then, we propose a model-agnostic task-specific parameter allocation algorithm to ensure that task-specific parameters are distributed evenly across the model, rather than being concentrated in specific regions. In doing so, TaskEdge can significantly reduce the computational cost and memory usage while maintaining performance on the target downstream tasks by updating less than 0.1% of the parameters. In addition, TaskEdge can be easily integrated with structured sparsity to enable acceleration by NVIDIA’s specialized sparse tensor cores, and it can be seamlessly integrated with LoRA to enable efficient sparse low-rank adaptation. Extensive experiments on various tasks demonstrate the effectiveness of TaskEdge.

Networks in the 5G and beyond era can use software-defined networks (SDN) to achieve network slicing (NS), so as to meet the extremely diverse service requirements of diverse applications in the Internet of Vehicles (IoV). However, the flow fluctuations in the highly dynamic IoV make it difficult to provide reliable, flexible, and scalable services for the IoV by the SDN control plane. Careful SDN controller placement can be a feasible solution to achieve its robustness and flexibility to deal with the changes in network status. Thus, this paper studies a dynamic controller placement problem to improve the performance of IoV services. To be specific, a hierarchical SDN control plane for the IoV is considered with the SDN controllers placed at the edge of networks. Under this architecture, we model the dynamic controller placement by Markov Decision Process (MDP). To efficiently solve the formulated NP-hard problems, we develop an algorithm based on Deep Deterministic Policy Gradient (DDPG) because of its advantages in solving the problem with multi-dimensional action and large solution space. Further, we incorporate a random process into the action selection strategy of DDPG to prevent it from getting trapped in local optimum. Simulation results show that the proposed DDPG-based controller placement approach can adapt to a highly dynamic IoV environment with outstanding performance.

Devices within the marine Internet of Things (MIoT) can connect to low Earth orbit (LEO) satellites and unmanned aerial vehicles (UAVs) to facilitate low-latency data transmission and execution, as well as enhanced-capacity data storage. However, without proper traffic handling strategy, it is still difficult to effectively meet the low-latency requirements. In this paper, we consider a cooperative satellite-aerial-MIoT network (CSAMN) for maritime edge computing and maritime data storage to prioritize delay-sensitive (DS) tasks by employing mobile edge computing, while handling delay-tolerant (DT) tasks via the store-carry-forward method. Considering the delay constraints of DS tasks, we formulate a constrained joint optimization problem of maximizing satellite-collected data volume while minimizing system energy consumption by controlling four interdependent variables, including the transmit power of UAVs for DS tasks, the start time of DT tasks, computing resource allocation, and offloading ratio. To solve this non-convex and non-linear problem, we propose a joint computation offloading and resource management (JCORM) algorithm using the Dinkelbach method and linear programming. Our results show that the volume of data collected by the proposed JCORM algorithm can be increased by up to 41.5% compared to baselines. Moreover, JCORM algorithm achieves a dramatic reduction in computational time, from a maximum of 318.21 seconds down to just 0.16 seconds per experiment, making it highly suitable for real-time maritime applications.

Collaborative perception enhances sensing in multi-robot and vehicular networks by fusing information from multiple agents, improving perception accuracy and sensing range. However, mobility and non-rigid sensor mounts introduce extrinsic calibration errors, necessitating online calibration, further complicated by limited overlap in sensing regions. Moreover, maintaining fresh information is crucial for timely and accurate sensing. To address calibration errors and ensure timely and accurate perception, we propose a robust task-oriented communication strategy to optimize online self-calibration and efficient feature sharing for Real-time Adaptive Collaborative Perception (R-ACP). Specifically, we first formulate an Age of Perceived Targets (AoPT) minimization problem to capture data timeliness of multi-view streaming. Then, in the calibration phase, we introduce a channel-aware self-calibration technique based on reidentification (Re-ID), which adaptively compresses key features according to channel capacities, effectively addressing calibration issues via spatial and temporal cross-camera correlations. In the streaming phase, we tackle the trade-off between bandwidth and inference accuracy by leveraging an Information Bottleneck (IB)-based encoding method to adjust video compression rates based on task relevance, thereby reducing communication overhead and latency. Finally, we design a priority-aware network to filter corrupted features to mitigate performance degradation from packet corruption. Extensive studies demonstrate that our framework outperforms five baselines, improving multiple object detection accuracy (MODA) by 25.49% and reducing communication costs by 51.36% under severely poor channel conditions. Code will be made publicly available: github.com/fangzr/R-ACP.

Collaborative perception has recently gained significant attention in autonomous driving, improving perception quality by enabling the exchange of additional information among vehicles. However, deploying collaborative perception systems can lead to domain shifts due to diverse environmental conditions and data heterogeneity among connected and autonomous vehicles (CAVs). To address these challenges, we propose a unified domain generalization framework to be utilized during the training and inference stages of collaborative perception. In the training phase, we introduce an Amplitude Augmentation (AmpAug) method to augment low-frequency image variations, broadening the model’s ability to learn across multiple domains. We also employ a meta-consistency training scheme to simulate domain shifts, optimizing the model with a carefully designed consistency loss to acquire domain-invariant representations. In the inference phase, we introduce an intra-system domain alignment mechanism to reduce or potentially eliminate the domain discrepancy among CAVs prior to inference. Extensive experiments substantiate the effectiveness of our method in comparison with the existing state-of-the-art works.

Ramp merging is one of the bottlenecks in traffic systems, which commonly cause traffic congestion, accidents, and severe carbon emissions. In order to address this essential issue and enhance the safety and efficiency of connected and autonomous vehicles (CAVs) at multi-lane merging zones, we propose a novel collaborative decision-making framework, namedAgentsCoMerge, to leverage large language models (LLMs). Specifically, we first design a scene observation and understanding module to allow an agent to capture the traffic environment. Then we propose a hierarchical planning module to enable the agent to make decisions and plan trajectories based on the observation and the agent's own state. In addition, in order to facilitate collaboration among multiple agents, we introduce a communication module to enable the surrounding agents to exchange necessary information and coordinate their actions. Finally, we develop a reinforcement reflection guided training paradigm to further enhance the decision-making capability of the framework. Extensive experiments are conducted to evaluate the performance of our proposed method, demonstrating its superior efficiency and effectiveness for multi-agent collaborative decision-making under various ramp merging scenarios.

Collaborative perception systems leverage multiple edge devices, such as surveillance cameras or autonomous cars, to enhance sensing quality and eliminate blind spots. Despite their advantages, challenges such as limited channel capacity and data redundancy impede their effectiveness. To address these issues, we introduce the Prioritized Information Bottleneck (PIB) framework for edge video analytics. This framework prioritizes the shared data based on the signal-to-noise ratio (SNR) and camera coverage of the region of interest (RoI), reducing spatial-temporal data redundancy to transmit only essential information. This strategy avoids the need for video reconstruction at edge servers and maintains low latency. It leverages a deterministic information bottleneck method to extract compact, relevant features, balancing informativeness and communication costs. For high-dimensional data, we apply variational approximations for practical optimization. To reduce communication costs in fluctuating connections, we propose a gate mechanism based on distributed online learning (DOL) to filter out less informative messages and efficiently select edge servers. Moreover, we establish the asymptotic optimality of DOL by proving the sublinearity of its regrets. To validate the effectiveness of the PIB framework, we conduct real-world experiments on three types of edge devices with varied computing capabilities. Compared to five coding methods for image and video compression, PIB improves mean object detection accuracy (MODA) by 17.8% while reducing communication costs by 82.65% under poor channel conditions.

Collaborative edge sensing systems, particularly in collaborative perception systems in autonomous driving, can significantly enhance tracking accuracy and reduce blind spots with multi-view sensing capabilities. However, their limited channel capacity and the redundancy in sensory data pose significant challenges, affecting the performance of collaborative inference tasks. To tackle these issues, we introduce a Prioritized Information Bottleneck (PIB) framework for collaborative edge video analytics. We first propose a priority-based inference mechanism that jointly considers the signal-to-noise ratio (SNR) and the camera’s coverage area of the region of interest (RoI). To enable efficient inference, PIB reduces video redundancy in both spatial and temporal domains and transmits only the essential information for the downstream inference tasks. This eliminates the need to reconstruct videos on the edge server while maintaining low latency. Specifically, it derives compact, task-relevant features by employing the deterministic information bottleneck (IB) method, which strikes a balance between feature informativeness and communication costs. Given the computational challenges caused by IB-based objectives with high-dimensional data, we resort to variational approximations for feasible optimization. Compared to TOCOM-TEM, JPEG, and HEVC, PIB achieves an improvement of up to 15.1% in mean object detection accuracy (MODA) and reduces communication costs by 66.7% when edge cameras experience poor channel conditions.

Federated learning (FL) allows multiple parties (distributed devices) to train a machine learning model without sharing raw data. How to effectively and efficiently utilize the resources on devices and the central server is a highly interesting yet challenging problem. In this paper, we propose an efficient split federated learning algorithm (ESFL) to take full advantage of the powerful computing capabilities at a central server under a split federated learning framework with heterogeneous end devices (EDs). By splitting the model into different submodels between the server and EDs, our approach jointly optimizes user-side workload and server-side computing resource allocation by considering users’ heterogeneity. We formulate the whole optimization problem as a mixed-integer non-linear program, which is an NP-hard problem, and develop an iterative approach to obtain an approximate solution efficiently. Extensive simulations have been conducted to validate the significantly increased efficiency of our ESFL approach compared with standard federated learning, split learning, and splitfed learning.

The increasingly deeper neural networks hinder the democratization of privacy-enhancing distributed learning, such as federated learning (FL), to resource-constrained devices. To overcome this challenge, in this paper, we advocate the integration of edge computing paradigm and parallel split learning (PSL), allowing multiple edge devices to offload substantial training workloads to an edge server via layer-wise model split. By observing that existing PSL schemes incur excessive training latency and a large volume of data transmissions, we propose an innovative PSL framework, namely, efficient parallel split learning (EPSL), to accelerate model training. To be specific, EPSL parallelizes client-side model training andreduces the dimension of activations' gradientsfor backpropagation (BP) vialast-layer gradient aggregation, leading to a significant reduction in server-side training and communication latency. Moreover, by considering the heterogeneous channel conditions and computing capabilities at edge devices, we jointly optimize subchannel allocation, power control, and cut layer selection to minimize the per-round latency. Simulation results show that the proposed EPSL framework significantly decreases the training latency needed to achieve a target accuracy compared with the state-of-the-art benchmarks, and the tailored resource management and layer split strategy can considerably reduce latency than the counterpart without optimization.

In unmanned aerial vehicle (UAV)-assisted multi-access edge computing (MEC) systems, where UAVs act as aerial relays to forward tasks from ground users (GUs) to remote edge servers (ESs) for processing, a crucial observation is that the computing power in the system depends on the computing capabilities at a single ES and the number of ESs covered by the UAV. The latter is essentially influenced by the UAV altitude, ES density, transmit power of the UAV, channel condition, etc. In this paper, we model a UAV-assisted MEC system featuring adjustable UAV altitude, random GU distribution, and random ES distribution. We adopt the signal-to-noise ratio-based coverage probability and derive a computing model to characterize communication-aware altitude-dependent computing power. Upon this, we model the sequential task-processing process, including task uploading, forwarding, and computing, as a three-stage tandem queue (M/D/1 →D/1 →D/1). Employing queueing theory, we derive analytical results for the end-to-end (e2e) service latency. Besides, we address the optimization problem of maximizing the number of completed tasks within the e2e latency constraint, referred to as task service throughput. Simulation and analytical results show that optimal UAV altitudes, yielding the maximum task computing throughput, can be obtained under given network parameters.

Semantic communication has been identified as a core technology for the sixth generation (6G) of wireless networks. Recently, task-oriented semantic communications have been proposed for low-latency inference with limited bandwidth. Although transmitting only task-related information does protect a certain level of user privacy, adversaries could apply model inversion techniques to reconstruct the raw data or extract useful information, thereby infringing on users’ privacy. To mitigate privacy infringement, this paper proposes an information bottleneck and adversarial learning (IBAL) approach to protect users’ privacy against model inversion attacks. Specifically, we extract task-relevant features from the input based on the information bottleneck (IB) theory. To overcome the difficulty in calculating the mutual information in high-dimensional space, we derive a variational upper bound to estimate the true mutual information. To prevent data reconstruction from task-related features by adversaries, we leverage adversarial learning to train encoder to fool adversaries by maximizing reconstruction distortion. Furthermore, considering the impact of channel variations on privacy-utility trade-off and the difficulty in manually tuning the weights of each loss, we propose an adaptive weight adjustment method. Numerical results demonstrate that the proposed approaches can effectively protect privacy without significantly affecting task performance and achieve better privacy-utility trade-offs than baseline methods.

Compressed sensing (CS) has emerged as a promising technique for reducing transmission data volume. Despite its significance, achieving a balance between the delay, energy consumption and data distortion caused by CS and transmission remains an understudied area in resource-constrained IoT systems. The emergence of multi-access edge computing provides a potential solution to the aforementioned issue by enabling the strategic implementation of CS either at IoT devices or an edge server (ES), depending on both bandwidth resources and computing resources at ES. In this paper, we investigate where to perform CS computation and how to determine the compression ratio and bandwidth allocation to minimize the weighted energy and distortion cost (WEDC) of all devices under latency requirements. We formulate a WEDC minimizing problem by jointly optimizing the task scheduling, compression ratio, and bandwidth allocation. Since the formulated problem is a mixed-integer and nonlinear programming, which is typically NP-hard, we decompose the original problem into two sub-problems and then develop an iterative algorithm to find the suboptimal solution. Extensive numerical results demonstrate the superiority of the proposed algorithm in reducing WEDC of all devices under delay constraints.

Unmanned aerial vehicles (UAVs) play a pivotal role in augmenting multi-access edge computing by facilitating low-latency services for ground units (GUs), especially in areas where the ground infrastructure is inadequate or damaged. In this context, the UAV trajectory planning and caching strategies assume paramount importance to ensure low-latency service delivery. Due to the limited caching and computing resources at a single UAV, it alone cannot provide effective services for a large number of GUs. In this paper, we design a novel cooperative framework for low-latency service provisioning by coordinating multiple UAVs's trajectories and service caching strategies. We formulate a latency minimization problem to jointly optimize both service caching and trajectory planning of multiple UAVs. However, due to the high dimension and coupling of multiple UAVs' movement and service caching, the optimization problem is a mixed-integer nonlinear programming, which is typically an NP-hard problem, and we propose an effective algorithm based on deep deterministic policy gradient to solve the high dimensional, non-convex, and continuous long-term optimization problem. Numerous experiments confirm that the proposed algorithm achieves significantly better performance in reducing the total system delay than other baseline algorithms.

The multiaccess edge computing/mobile-edge computing (MEC) is becoming a key technology toward “full 5G.” However, as it gets widely used, a fundamental problem is how to support as many service requests as possible under stringent Quality-of-Service (QoS) requirements and limited communications and computing resources. In this article, we study the long-term throughput maximization problem for multicell multiuser MEC systems. Different from most of the existing works that focus on energy or latency minimization problem for a single-edge system, a novel design is proposed from the service provider’s perspective to maximize the system-wide throughput under latency bounds by jointly taking user association and resource allocation for both communications and computing into account. To capture the stochastic nature of MEC environments, a Markov decision process (MDP) is employed to model the queuing states for both mobile devices and MEC servers. By combining MDP and matching theory, a joint user association and resource allocation algorithm is given, where the resource allocation policy under given user-server association is solved. Extensive numerical results demonstrate the superiority of the proposed scheme in comparison with several existing approaches.

Communication limitation at the edge is widely recognized as a major bottleneck for federated learning (FL). Multi-hop wireless networking provides a cost-effective solution to enhance service coverage and spectrum efficiency at the edge, which could facilitate large-scale and efficient machine learning (ML) model aggregation. However, FL over multi-hop wireless networks has rarely been investigated. In this paper, we optimize FL over wireless mesh networks by taking into account the heterogeneity in communication and computing resources at mesh routers and clients. We present a framework that each intermediate router performsin-networkmodel aggregation before sending the data to the next hop, so as to reduce the outgoing data traffic and hence aggregate more models under limited communication resources. To accelerate model training, we formulate our optimization problem by jointly considering model aggregation, routing, and spectrum allocation. Although the problem is a non-convex mixed-integer nonlinear programming, we transform it into a mixed-integer linear programming (MILP), and develop a coarse-grained fixing procedure to solve it efficiently. Simulation results demonstrate the effectiveness of the solution approach, and the superiority of the in-network aggregation scheme over the counterpart without in-network aggregation.

The computation offloading in mobile-edge computing (MEC) systems emerges as a promising technology to enhance users' quality-of-experience over mobile devices (MDs). However, the design of computation offloading policy for MEC systems inevitably faces challenges with respect to the gap between dynamic task generation in MDs and the limited resources at an MEC server, especially for a multi-user MEC system. More specifically, whether or not offload a task to a nearby MEC server and how much communication and computing resources are allocated to the selected MDs should be carefully investigated to optimize the long-term system performance. In this paper, we handle this issue based on the Markov decision process, where collaborated resource allocations are determined according to both the queueing state of the task buffer at the MDs and the MEC server. By analyzing the average task delay of each user and the average throughput of the system, we formulate a throughput maximization problem with the constraints on delay, spectrum resource, and computing resource, and develop a throughput-optimal resource allocation policy. Simulation results show that the proposed joint communication and computing resource allocation policy is highly effective and efficient.

Fog computing is a distributed computing model that can utilise the storage, analysis and processing capabilities of fog nodes near edge devices. Although fog computing can support task processing for various Internet of Things (IoT) systems, Fog‐IoT architecture faces several new challenges with the rapid development of IoT systems, especially delay‐sensitive IoT systems, such as stochastic and dynamic data arrival, optimal utility and deadline of tasks. To address these challenges, workload scheduling toward worst‐case delay and optimal utility for single‐hop Fog‐IoT architecture are studied and the workload dynamic scheduling algorithm (WDSA) is proposed. The proposed WDSA algorithm can maximise the average throughput utility while guarantees the worst‐case delay of task processing. In addition, it is online and needs no prior information about future. The algorithm performance is analysed from the perspective of optimality and worst‐case delay, demonstrating that the proposed WDSA algorithm can get an approximate optima and worst‐case delay guarantees. Finally, simulation results demonstrate that the efficiency and efficacy of this kind of the algorithm can meet the requirement.