关键词： Mobile edge computing   Task offloading   Multi-UAV network   Multi-agent deep reinforcement learning   Distance to task location and capability match  

摘要： Mobile edge computing (MEC) has emerged as a key solution for addressing the demands of computation-intensive network services by providing computational resources at the network edge, thereby minimizing service delays. Leveraging their flexible deployment, wide coverage, and reliable wireless communication, unmanned aerial vehicles (UAVs) have been integrated into MEC systems to enhance performance. This paper investigates the task offloading problem in a Multi-UAV-assisted MEC environment and proposes a collaborative optimization framework that integrates the Distance to Task Location and Capability Match (DTLCM) mechanism with a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm. Unlike traditional task priority-based offloading schemes, the proposed approach ensures optimal UAV selection based on both computational capability and spatial proximity. The system gain is defined in terms of energy efficiency and task delay with the optimization formulated as a mixed-integer programming problem. To efficiently solve this complex problem, a Multi-Agent Deep Reinforcement Learning framework is employed, combining MADDPG with DTLCM to jointly optimize UAV trajectories, task offloading decisions, computational resource allocation, and communication resource management. Comprehensive simulations demonstrate that the proposed MADDPG-DTLCM framework significantly outperforms four state-of-the-art methods (MADDPG-DTLCM,MADQN, MADDPG without DTLCM, and Greedy offloading), achieving 18% higher task completion rates and 12% lower latency under varying network conditions, particularly in high-user-density scenarios with UAV collaboration.

关键词： Mobile edge computing  

摘要： The combination of Integrated Sensing and Communication (ISAC) and Mobile Edge Computing (MEC) enables devices to simultaneously sense the environment and offload data to the base stations (BS) for intelligent processing, thereby reducing local computational burdens. However, transmitting raw sensing data from ISAC devices to the BS often incurs substantial fronthaul overhead and latency. This paper investigates a three-tier collaborative inference framework enabled by Integrated Sensing, Communication, and Computing (ISCC), where cloud servers, MEC servers, and ISAC devices cooperatively execute different segments of a pre-trained deep neural network (DNN) for intelligent sensing. By offloading intermediate DNN features, the proposed framework can significantly reduce fronthaul transmission load. Furthermore, multiple-input multiple-output (MIMO) technology is employed to enhance both sensing quality and offloading efficiency. To minimize the overall sensing task inference latency across all ISAC devices, we jointly optimize the DNN partitioning strategy, ISAC beamforming, and computational resource allocation at the MEC servers and devices, subject to sensing beampattern constraints. We also propose an efficient two-layer optimization algorithm. In the inner layer, we derive closed-form solutions for computational resource allocation using the Karush-Kuhn-Tucker conditions. Moreover, we design the ISAC beamforming vectors via an iterative method based on the majorization–minimization and weighted minimum mean square error techniques. In the outer layer, we develop a cross-entropy-based probabilistic learning algorithm to determine an optimal DNN partitioning strategy. Simulation results demonstrate that the proposed framework substantially outperforms existing two-tier schemes in inference latency. © 2025, CC0.

关键词： Mobile edge computing   Decentralized task offloading   Meta reinforcement learning   Deep reinforcement learning  

摘要： In mobile edge computing (MEC), the limited resources of individual edge servers and the uneven workload can significantly impact the system performance and user quality of experience (QoE), necessitating greater consideration for edge-edge and edge-cloud collaboration. However, existing collaborative task offloading strategies still fall short in performance and lack adaptability to dynamic environments. To address these challenges, this paper proposes a meta-reinforcement learning-based two-stage task offloading algorithm, named MRL-TSO, for decentralized task offloading in a generic MEC framework that integrates edge-edge and edge-cloud cooperation. MRL-TSO is the first to leverage MRL to address collaborative task offloading for edge-edge and edge-cloud at the edge layer and innovatively divides the offloading decision into two stages: in the first stage, a trained decision model is used to make preliminary decisions based on local information;in the second stage, the initial decision is combined with the status of edge servers to filter out unavailable resources, thereby improving the reliability of the decision. Experimental results indicate that, compared to the state-of-the-art work, MRL-TSO increases the task success rate by at least 1.43% and reduces latency by 9.69%. Especially in the case of edge server failures, it improves the task success rate by at least 45.60% compared to the competitive algorithms, demonstrating better environmental adaptability and robustness.

关键词： Unmanned Aerial Vehicle   Non-orthogonal multiple access   Mobile edge computing   Clustering   Multi-agent learning  

摘要： Unmanned Aerial Vehicle (UAV)-assisted Mobile Edge Computing (MEC) is emerging as a viable technology for computation-intensive and latency-critical jobs in the future Internet of Things (IoT), driven by the proliferation of smart devices and the expansion of high-bandwidth applications. However, the surge in data volume places a heavy burden on the scarce radio spectrum. This paper explores a UAV-assisted edge computing framework that incorporates non-orthogonal multiple access (NOMA), specifically a framework where multiple UAVs are deployed as edge servers to provide computational assistance to terrestrial Smart Mobile Devices (SMDs). Three-dimensional (3D) trajectory design, power allocation, and joint offloading decision optimization problems are proposed to minimize the system latency. Due to their movable nature, UAVs and SMDs must be redeployed in accordance with the movement of SMDs in a timely manner. to address this associated dynamic issue. The best offloading approach is first determined theoretically and mathematically, and the SMDs are then clustered using a K-means based clustering algorithm. Then, for every UAV server, we develop a cooperative multi-agent Twin Delayed Deep Deterministic policy gradient algorithm (MATD3) and formulate a multi- agent extension of the Markov Decision Process (MDP) using centralized training and distributed execution to jointly determine the UAV flight trajectory and power allocation. The simulation results demonstrate that: (1) the proposed algorithm can achieve lower latency than other offloading schemes while satisfying the task fault-tolerant energy consumption;(2) the system latency of the UAV-assisted MEC-NOMA network is 14% lower than the Orthogonal Multiple Access (OMA) case when the decoding order and power allocation are implemented;and (3) the proposed MATD3 algorithm's optimization of the UAV trajectory reduces the total system latency by 6% compared to the two-dimensional (2D)trajectory.

关键词： Resource management   Satellites   Heuristic algorithms   6G mobile communication   Reconfigurable intelligent surfaces   Energy consumption   Low earth orbit satellites   Delays   Real-time systems   Servers   6G networks   deep reinforcement learning   mobile edge computing   orbital edge computing   intelligent autonomous transportation systems  

摘要： The integration of terrestrial and non-terrestrial networks with mobile edge computing (MEC) and orbital edge computing (OEC) technologies is essential for advancing 6G communication networks. This paper introduces a network architecture that combines terrestrial and non-terrestrial networks by integrating drones (also known as UAV)-carried reconfigurable intelligent surfaces (RIS) and satellite-based MEC to optimize resource allocation in intelligent autonomous transportation systems (IATS). The primary objective is to minimize total system utility costs through the optimal allocation of bandwidth, computational power at the base station and low Earth orbit (LEO) satellite, and offloading decisions, all while adhering to strict performance and delay constraints. We address the complex resource optimization challenge by formulating a nonlinear programming (NLP) problem. To solve this problem, we employ long short-term memory (LSTM)-enhanced deep deterministic policy gradient (DDPG) and LSTM-enhanced twin delayed deep deterministic policy gradient (TD3) algorithms, which enable dynamic and adaptive resource management. These LSTM-enhanced algorithms improve convergence speed by 44.44% and 73.81%, respectively, compared to their conventional counterparts, while significantly enhancing cost efficiency. Our simulation results demonstrate substantial improvements in system performance, with effective resource allocation and minimal utility costs, providing a robust solution for ensuring high-quality, low-latency communication in diverse 6G IATS environments.

关键词： mobile edge computing   computing offload   deep reinforcement learning   service cache  

摘要： The rapid development of mobile Internet technology has made users' requirements for quality of service (QoS) continuously improve. The task unloading process of mobile edge computing has the problem that it is impossible to balance delay and energy consumption for task unloading under the condition of fluctuating network bandwidth. To address this issue, this paper proposes a task offloading scheme based on the Proximal Policy Optimization (PPO) algorithm. On the basis of traditional cloud edge collaborative architecture, the collaborative computing mechanism between edge node devices is further integrated, and the concept of service caching is introduced to reduce duplicate data transmission, reduce communication latency and network load, and improve overall system performance. Firstly, this article constructs an energy efficiency function with a certain weight ratio of energy consumption and latency as the core optimization objective. Then, the task offloading process of mobile terminal devices is modeled as a Markov Decision Process (MDP). Finally, the deep reinforcement learning PPO algorithm is used for training and learning, and the model is solved. The simulation results show that the proposed scheme has significant advantages in reducing energy consumption and latency compared to the comparative scheme.

关键词： mobile edge computing   Markov decision process   reinforcement learning   deep reinforcement learning  

摘要： Mobile edge computing (MEC) has emerged as a promising paradigm to enhance computational capabilities at the network edge, enabling low-latency services for users while ensuring efficient resource utilization for operators. One of the key challenges in MEC is optimizing offloading decisions and resource allocation to balance user experience and operator profitability. In this paper, we integrate software-defined networking (SDN) and MEC to enhance system utility and propose an SDN-based MEC network framework. Within this framework, we formulate an optimization problem that jointly maximizes the utility of both users and operators by optimizing the offloading decisions, communication and computation resource allocation ratios. To address this challenge, we model the problem as a Markov decision process (MDP) and propose a reinforcement learning (RL)-based algorithm to optimize long-term system utility in a dynamic network environment. However, since RL-based algorithms struggle with large state spaces, we extend the MDP formulation to a continuous state space and develop a deep reinforcement learning (DRL)-based algorithm to improve performance. The DRL approach leverages neural networks to approximate optimal policies, enabling more effective decision-making in complex environments. Experimental results validate the effectiveness of our proposed methods. While the RL-based algorithm enhances the long-term average utility of both users and operators, the DRL-based algorithm further improves performance, increasing operator and user efficiency by approximately 22.4% and 12.2%, respectively. These results highlight the potential of intelligent learning-based approaches for optimizing MEC networks and provide valuable insights into designing adaptive and efficient MEC architectures.

关键词： Mobile edge computing   Computation offloading   Deep learning   Parallel task offloading   Mobile devices  

摘要： The growing demand for real-time computing applications on mobile devices is burdening their processing power and battery life. Mobile edge computing helps by allowing these tasks to be offloaded to nearby servers having more processing power. However, when it comes to multiple servers and tasks, choosing the optimal components for offloading becomes challenging. This is because we need to balance between reducing the amount of data transferred and keeping communication latency low. To address this problem, an energy- efficient parallel computation offloading mechanism through deep learning (EPCOD), is proposed. An algorithm using deep learning (DL) is developed and trained as a decision-making system. This system selects the best combination of application components taking into account various factors, such as energy consumption, network conditions, computational load, data transfer volume, and communication latency. A cost function that includes all these factors is developed to calculate the cost for each possible offloading policy combination. By analyzing a large dataset, we find the best policies. Additionally, we use a DL network to efficiently handle this computational task. Simulation results demonstrate that EPCOD effectively minimizes both latency and energy consumption, achieving a high accuracy of deep neural network of up to 73.5%.

关键词： mobile edge computing   heterogeneous network   task offloading   queueing theory   stochastic geometry  

摘要： Mobile edge computing architecture (MEC) can provide users with low latency services by integrating computing, storage and processing capabilities near users and data sources. As such, there has been intense interest in this topic, especially in single-server and homogeneous multi-server scenarios. The impact of network heterogeneity and load fluctuations is ignored, and the performance evaluation system relies too much on statistical mean indicators, ignoring the impact of real-time indicators. In this paper, we propose a new heterogeneous edge computing network architecture composed of multi-core servers with varying transmission power, computing capabilities and waiting queue length. Since it is necessary to evaluate and analyze the service performance of MEC to guarantee Quality of Service (QoS), we design some indicators by solving the probability distribution function of response time, such as average task offloading delay, immediate service probability and blocking probability. By analyzing the impact of bias factors and network parameters associated with MEC servers on network performance, we provide insights for MEC design, deployment and optimization.

关键词： Mobile edge computing  

摘要： With the rapid development of building information modeling (BIM) technology, its application in construction projects has become widespread, particularly in the collaborative work during the design, construction, and operation phases. However, traditional BIM models, due to their large data volume and complex structure, present significant challenges in data transmission and real-time processing during cross-regional collaboration. Meanwhile, mobile edge computing (MEC), as an emerging computational paradigm, reduces data transmission latency and enhances processing efficiency by deploying computing and storage resources at the network edge. The integration of a BIM lightweight engine and MEC offers a more efficient solution for cross-regional cost collaboration. While previous studies have explored the combination of BIM with cloud and edge computing, most have focused on single-region scenarios, with the lightweight methods for BIM data and MEC deployment for cross-regional collaboration still lacking efficiency and practicality. Furthermore, existing optimization algorithms often fail to fully address the unique demands of cross-regional collaboration, leading to suboptimal overall performance. To address these issues, a cross-regional cost collaborative optimization model based on the BIM lightweight engine and MEC was proposed in this study. The main content of this study includes a) the design of a BIM lightweight engine based on point cloud compression, which improves data transmission efficiency through efficient compression techniques;b) the investigation of the MEC server deployment strategy in cross-regional cost collaborative scenarios, optimizing the distribution and scheduling of computational resources;and c) the development of a cross-regional cost collaborative optimization method based on an improved genetic algorithm to enhance the intelligence and efficiency of resource scheduling. This research provides not only a new theoretical framework for