关键词： Mobile edge computing   Internet of vehicles   Deep reinforcement learning   Content caching   Task offloading  

摘要： Considering the dynamic variability of the vehicular edge environment and the limited edge servers resources, this paper proposes a joint task caching and computation offloading scheme based on deep reinforcement learning (DRL). Considering that the motion trajectories of different vehicles overlap and their task requests may be the same, this paper designs a vehicle-edge-cloud computing framework to fully use the cache resources of vehicles, edge servers, and clouds to reduce task processing delays and energy consumption. Secondly, this paper adopts partial offloading and collaboration between edge servers to make full use of the computational resources of vehicles, edge servers and clouds, avoiding the waste of resources and reducing the burden of vehicles and edge servers. In addition, this paper proposes a DRL-based task offloading scheme to obtain better task caching and offloading strategies. The simulation results show that the scheme proposed in this article performs better compared to other schemes and effectively reduces the latency and energy consumption of task processing.

关键词： Uncertainty   Servers   Resource management   Optimization   Costs   Delays   Signal processing algorithms   Machine learning algorithms   Vehicle-to-everything   Vehicle dynamics   Mobile edge computing   two-time scale task offloading/delivery   RSU activation/deactivation   demand and channel uncertainties   task partitioning   task migration  

摘要： In this paper, a novel two-time scale task offloading/delivery framework is presented which dynamically manages radio, storage, computing, and cost aspects through multiple Mobile Edge Computing (MEC) servers distributed across Road Side Units (RSUs), Base Stations (BSs), vehicles, and pedestrians. To enhance flexibility, we introduce a smart dynamic task partitioning and cooperative offloading/delivery mechanism, enabling users to offload task partitions to various MEC servers. Uncertainties stemming from users' high mobility and traffic variations may lead to unpredictable task offloading demands and Imperfect Channel State Information (I-CSI), potentially causing unsuccessful task offloading/delivery. In response, we propose a robust and cooperative two-time-scale task offloading/delivery approach, adept at handling demand uncertainty and I-CSI. The problem formulation involves a two-time scale Markov decision process, aimed at minimizing task offloading/delivery costs while maximizing task completion rates. Large time-scale slots incorporate intelligent RSU activation/deactivation, while short time-scale slots focus on task offloading/delivery. To efficiently implement this framework, we propose Federated Reinforcement Learning (FRL) algorithm based on Deep Deterministic Policy Gradient (DDPG);namely Low overhead Multi-Agent DDPG (LoMADDPG) where at each slot, all agents tune their Actor and Critic neural network weights with the help of Global Software-defined Networking (SDN) Controller (GSC) which acts as a global weight tuner. Simulation results illustrate that our proposed task offloading/delivery scheme can result in significant improvement in task completion rate and cost reduction.

关键词： Mobile edge computing   resource allocation   task offloading   unmanned aerial vehicle   federated reinforcement learning   digital twin   Mobile edge computing   resource allocation   task offloading   unmanned aerial vehicle   federated reinforcement learning   digital twin  

摘要： The computationally intensive tasks generated by AIoTD face challenges due to its limitations in battery power and computing capabilities. These devices typically operate in resource-constrained environments where energy efficiency (EE), computational efficiency, and real-time responsiveness are critical factors. To address these challenges, MEC is a promising solution as it allows tasks to be transferred from IoT devices to MEC for processing. By utilizing UAVs in MEC systems, computing and storage resources are brought closer to end devices, particularly in remote areas. This study investigates the TO issue within a AIoT-enabled UAV-assisted MEC system empowered by DT technologies. The DT concept is employed to provide environmental information and facilitate data exchange for agents linked to AIoT devices. The TO problem is structured to optimize EE and distribute workloads effectively among edge servers. Subsequently, the problem is reformulated as a MDP and an energy-efficient TO strategy, termed FedRL-EETO (FedRL based), is proposed. Moreover, a DT-enabled FedRL-EETO approach is implemented to enhance privacy protection and train decentralized RA strategies for communication networks, further improving network efficiency. Experimental evaluations are conducted to represents the benefits of the FedRL-EETO algorithm in enhancing EE and workload distribution, demonstrating its superior performance in comparison to existing method.

关键词： Industrial internet of things   Mobile edge computing   Computation offloading   Security   Deep reinforcement learning  

摘要： In the industrial internet of things, the rapid development of smart mobile devices and 5G network technology has driven the application of mobile edge computing, reducing the delay in task computation offloading to some extent. However, the increasing complexity of the IIoT environment presents challenges for communication management and offloading performance. To achieve efficient computation offloading communication, we designed a cloud-edge-device IIoT system model, utilizing Voronoi diagrams to partition the service areas of edge servers, thereby adapting to the complex IIoT environment and improving communication efficiency. Additionally, considering that different offloading strategies may result in varying levels of offloading security risks, we developed a principal component analysis-based offloading security evaluation model (PCA-OSEM) to analyze potential security risks during the offloading process and identify key factors. Finally, to optimize offloading strategies to reduce offloading delay and security risks, we proposed a dual-Q with Kolmogorov- Arnold networks in deep reinforcement learning computation offloading (D2KCO). This method enhances the neural network's approximation capability and training stability. Experimental results show that the proposed PCA-OSEM is effective, and the D2KCO method can reduce offloading delay by 13% and 23.52% compared to the D3PG and DDPG algorithms, respectively, while also reducing security risks.

关键词： QUIC protocol   Mobile edge computing   PID control   queue management   queue management   consumer electronic devices   consumer electronic devices   queue management   consumer electronic devices  

摘要： The Quick UDP Internet Connection (QUIC) protocol has been utilized in traditional cloud computing environments. However, in consumer electronic devices under edge cloud scenarios, QUIC may struggle to take advantage of its benefits due to its unique network characteristics and resource constraints. This problem arises when the network quality fluctuates or resources are insufficient to handle incoming data, causing rapid buffer expansion, significant delays, and potential packet loss. This study proposes a queue management algorithm inspired by the classical control theory of the Proportional-Integral-Differential (PID) control, which aims to support QUIC data flows to further enhance delay control and improve goodput. The algorithm adds differential operations to the traditional PI control to predict the error trend to respond to queue changes in advance. Combining expert experience in integral separation and queue error management, it makes the control strategy more relevant to the specific needs of real application scenarios. Simulation results demonstrate that the PID-Delay algorithm achieves an average goodput improvement of 1.98 times and reduces the standard deviation by 26.1% than the classical algorithm. In comparison to other delay algorithms, it exhibits an average 1.81 times increase in goodput and an 18.5% reduction in standard deviation.

关键词： Mobile edge computing   Task scheduling   Containers   Optimization   Task offloading  

摘要： Mobile application offloading is a critical issue in mobile edge computing environments. This study explores the concept of scheduling tasks based on containers in a Mobile Edge Computing, using container technology. In this paper, we introduce the Optimized Container Selection (OCS) algorithm, which integrates the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The OCS algorithm is based on a heuristic algorithm composed of two stages and leveraging linear programming techniques to tackle the computational complexity. The primary objective of the research is to develop strategies for selecting containers that are optimised for performance. By following a standardised platform for container management across cloud and edge servers, all mobile applications are deployed solely within containers. The task prioritisation process for tasks with imminent deadlines is determined by considering multiple factors, such as response time and task execution cost. This procedure entails assessing and choosing the best container to carry out the assigned activities. Adjusting the number of edge servers and tasks is necessary while conducting experimental assessments in a mobile edge cloud system. The outcomes show improvements in four baseline algorithms Quality of Service (QoS), penalty cost, and violation time..

关键词： Microservice architectures   Optimization   Quality of service   Dynamic scheduling   Servers   Reinforcement learning   Resource management   Cloud computing   Time factors   Load modeling   Mobile edge computing   microservices   minimal deployment unit   two-phase deployment   reinforcement learning  

摘要： In mobile edge computing (MEC) environment, effective microservices deployment significantly reduces vendor costs and minimizes application latency. However, existing literatures overlook the impact of dynamic characteristics such as the frequency of user requests and geographical location, and lack in-depth consideration of the types of microservices and their interaction frequencies. To address these issues, we propose TP-MDU, a novel two-stage deployment framework for microservices. This framework is designed to learn users' dynamic behaviors and introduces, for the first time, a minimal deployment unit. Initially, TP-MDU generates minimal deployment units online, tailored to the types of microservices and their interaction frequencies. In the initial deployment phase, aiming for load balancing, it employs a simulated annealing algorithm to achieve a superior deployment plan. During the optimization scheduling phase, it utilizes reinforcement learning algorithms and introduces dynamic information and new optimization objectives. Previous deployment plans serve as the initial state for policy learning, thus facilitating more optimal deployment decisions. This paper evaluates the performance of TP-MDU using a real dataset from Australia's EUA and some related synthetic data. The experimental results indicate that TP-MDU outperforms other representative algorithms in performance.

关键词： Mobile edge computing   Unmanned aerial vehicle communications   High altitude platform   Service caching   Resource allocation  

摘要： The exponential growth of Internet of Things devices has triggered an unprecedented surge in mobile data traffic, posing significant challenges for latency-sensitive services. Mobile Edge Computing (MEC) has emerged as a promising solution by decentralizing computation and caching resources to the network edge. However, traditional terrestrial MEC systems struggle with limited coverage and flexibility. To overcome these issues, this paper proposes a novel dual-layer aerial MEC architecture, where multiple Unmanned Aerial Vehicles (UAVs) provide computation and caching support for resource-constrained terminal devices, and a high- altitude platform serves as a central hub for long-term service storage and retrieval. The system aims to minimize total latency by jointly optimizing service caching, task offloading, resource allocation, and 3D UAV deployment, formulated as a mixed-integer nonlinear programming problem and efficiently solved using an iterative algorithm based on linear relaxation and successive convex approximation. Simulation results demonstrate that the proposed scheme converges quickly across different scales and outperforms all baselines with minimal runtime increase, reducing total latency by 42.86% compared to the random UAV deployment.

关键词： Downlink   Millimeter wave communication   Wireless communication   Uplink   Optimization   Energy harvesting   Servers   Mobile edge computing   simultaneous wireless information and power transfer   mmWave   computation offloading   energy harvesting   beam management  

摘要： In this paper, we investigate the SWIPT (simultaneous wireless information and power transfer) based MEC (mobile edge computing) system in the mmWave band, where wireless devices offload partial tasks to the MEC server in the uplink, receive the computing result and harvest energy in the downlink. We focus on exploiting multiple beams in the mmWave band to enhance the multi-device downlink transmissions and maintain fairness among multiple devices. Specifically, we concentrate on the joint optimization of the uplink offloading, downlink beam management for multi-device transmissions, downlink power splitting within each beam for computing result receiving and energy harvesting, and time allocation between uplink and downlink. We formulate the joint optimization problem to maximize the minimum residual device energy. As the problem is a mixed-integer non-convex optimization program, we design a heuristic algorithm with low computing complexity, by relaxing the integer constraint and exploiting the successive convex approximation method. We carry out extensive simulations to verify our proposed algorithm performance. The simulations demonstrate that the joint optimization of beam management can increase the minimum residual energy and ensure the max-min fairness among multiple devices within the mmWave SWIPT-based MEC system.

关键词： Mobile edge computing   Mobile augmented reality   Task offloading   Multi-cache placement  

摘要： As mobile augmented reality (MAR) applications continue to develop and spread, more and more data-intensive and computationally intensive tasks are sensitive to delay and energy consumption. The emergence of mobile edge computing provides an effective solution for the demand of low latency and low energy consumption. This paper studies the task offloading and multi-cache placement of MAR applications in a three-tier transmission system composed of edge servers, edge data centers and remote cloud. Under the constraint of computing resources and cache space, the task offloading and cache placement problems are formulated to maximize the energy efficiency function including task completion rate and energy consumption. To solve this problem, we design a task offloading and multi-cache placement algorithm based on block coordinate descent. Firstly, we generate the priority queue for the tasks. Then, caches are placed according to the popularity of each cache and the cache space ratio of each edge node. Tasks are offloaded according to the priority of each edge node. After the initialization, the task offloading strategy and cache placement strategy are optimized using block coordinate descent. Finally, we update the optimized cache placement strategy and task offloading strategy until the objective function converges. Simulation results show that our algorithm can significantly shorten service delay and reduce energy consumption compared with other algorithms.