关键词： Task analysis   Data models   Cloud computing   Training data   Training   Computational modeling   Costs   Mobile cloud computing   Mobile edge computing   computation offloading   deep learning   incremental learning  

摘要： MobileCloud Computing (MCC) provides computing, storage, and other fruitful services to end users. Offloading such tasks to cloud servers can help to fulfill the demands of extensive computing resources, but may also lead to network congestion and high latency. Mobile Edge Computing (MEC) places the computing nodes near the end users to enable low-latency services, whereas it cannot execute too many computing tasks due to limited computing resources. Therefore, MCC and MEC are highly complementary. For computing offloading problems in a collaborative cloud-edge environment, traditional optimization algorithms require multiple iterations to obtain results, which leads to excessive time spent to obtain offloading strategies. Deep Neural Network (DNN) based offloading algorithms can provide low latency offloading strategies, but training data is difficult to be obtained and the cost of retraining is too high. Therefore, in this article, we adopt an incremental training method to overcome the problem of insufficient training data and high retraining costs in DNN-based offloading algorithms. An incremental DNN-based computation offloading (MEDIA) algorithm is proposed to derive near-optimal offloading strategies for collaborative cloud-edge computing. The task information on the real scenarios is sent to the central cloud to generate training data, and the powerful computing resources of the central cloud improve the efficiency of training model. The continuous incremental training can maintain a high accuracy of the DNN model and reduce the time for training the model. The evaluation results demonstrate that the proposed algorithm substantially reduces the cost for updating the model without loss of performance.

关键词： Autonomous aerial vehicles   Optimization   Task analysis   Energy consumption   Bandwidth   Quality of experience   Trajectory planning   Lyapunov optimization   mobile edge computing   unmanned aerial vehicle   successive convex approximation  

摘要： This letter aims to minimize the summation energy consumption of user equipments under controllable task backlog queues with unmanned aerial vehicles (UAV)-assisted mobile edge computing system. Based on Lyapunov optimization, the original stochastic optimization problem can be transformed into a deterministic optimization problem, and system stability is also ensured. To address this nonconvex problem, we employ alternating optimization to achieve the near-optimal transmit power, bandwidth ratio, and trajectory planning, respectively. Furthermore, considering the accuracy of the solution and algorithm complexity, we propose a two-stage operating mode for the UAV based on a recursive successive convex approximation method to optimize trajectory. Finally, the convergence and effectiveness of the proposed algorithm are validated through simulation results.

关键词： Edge computing   Edge computing enabled IoT   Authentication   Key agreement   Provably  

摘要： Edge computing server deployed at the edge of the network has promoted the development of the Internet of Things (IoT) but also brings some security problems. Traditional IoT security solutions have not met the requirements of the current rapid growth of equipment, real-time and lightweight. Therefore, the development of new authentication and key agreement is becoming an important means to solve the security problems of the IoT. To address this challenge, we propose an authentication framework for the edge computing-enabled IoT environment. This framework aims to facilitate the establishment of secure communication among devices and between devices and edge servers. This authentication framework achieves secure communication and identity verification among edge computing-enabled IoT devices and between devices and edge servers by incorporating a portion of the local private key during registration and leveraging a distributed blockchain network. Formal and informal security analyses of the proposed protocol show that the protocol effectively responds to numerous security threats while satisfying various security attributes. In addition, the study provides a comprehensive assessment of the computational and transmission overheads associated with the protocol. The results show that the protocol's commendable performance in terms of both computational efficiency and transmission cost-effectiveness.

关键词： Federated deep reinforcement learning   mobile edge computing   multiple base stations collaboration   task offloading   user mobility  

摘要： With the proliferation of the Internet of Things (IoT), mobile edge computing (MEC) has great potential to achieve low latency, high reliability, and low energy consumption. However, in collaborative MEC environments, user movement and task migration may cause task transmission and processing delays, resulting in elevated task response times. Therefore, system performance and user experience need to be ensured by rational task offloading and resource management. At the same time, the protection of user data privacy is becoming increasingly important as a challenge to be overcome. To address the problems of intense resource competition and privacy leakage in MEC, the federated learning for the TD3-based task offloading (FedTO) algorithm is proposed. The algorithm has a dual objective of energy consumption and task response time while protecting user privacy. It employs a cryptographic local model update and aggregation mechanism and uses deep reinforcement learning (DRL) to obtain an efficient task offloading decision. Based on the mobile trajectories of real devices, and the pre-deployment of base station locations, experimental results show that the FedTO algorithm ensures task data security. It also effectively reduces the total energy consumption and average task response time of the system, which further improves the system utility.

关键词： Computational Offloading   Internet of vehicles   Task management   Deep reinforcement learning   Decentralized Framework   Mobile edge computing  

摘要： This paper focuses on the problem of computation offloading in a high-mobility Internet of Vehicles (IoVs) environment. The goal is to address the challenges related to latency, energy consumption, and payment cost requirements. The approach considers both moving and parked vehicles as fog nodes, which can assist in offloading computational tasks. However, as the number of vehicles increases, the action space for each agent grows exponentially, posing a challenge for decentralised decision-making. The dynamic nature of vehicular mobility further complicates the network dynamics, requiring joint cooperative behaviour from the learning agents to achieve convergence. The traditional deep reinforcement learning (DRL) approach for offloading in IoVs treats each agent as an independent learner. It ignores the actions of other agents during the training process. This paper utilises a cooperative three-layer decentralised architecture called Vehicle-Assisted Multi-Access Edge Computing (VMEC) to overcome this limitation. The VMEC network consists of three layers: the fog, cloudlet, and cloud layers. In the fog layer, vehicles within associated Roadside Units (RSUs) and neighbouring RSUs participate as fog nodes. The middle layer comprises Mobile Edge Computing (MEC) servers, while the top layer represents the cloud infrastructure. To address the dynamic task offloading problem in VMEC, the paper proposes using a Decentralized Framework of Task and Computational Offloading (DFTCO), which utilises the strength of MADRL and NOMA techniques. This approach considers multiple agents making offloading decisions simultaneously and aims to find the optimal matching between tasks and available resources.

关键词： mobile edge computing   UAV path planning   task offloading   communication   collision avoidance  

摘要： Due to CPU and memory limitations, mobile IoT devices face challenges in handling delay-sensitive and computationally intensive tasks. Mobile edge computing addresses this issue by offloading tasks to the wireless network edge, reducing latency and energy consumption. UAVs serve as auxiliary edge clouds, providing flexible deployment and reliable wireless communication. To minimize latency and energy consumption, considering the limited resources and computing capabilities of UAVs, a multi-UAV and multi-edge cloud system was deployed for task offloading and UAV trajectory optimization. A joint optimization model for computing task offloading and UAV trajectory was proposed. During model training, a UAV communication mechanism was introduced to address potential coverage issues for mobile user devices through multiple UAVs or complete coverage. Considering the fact that decisions made by UAVs during trajectory planning may lead to collisions, a MADDPG algorithm with an integrated safety layer was adopted to obtain the safest actions closest to the joint UAV actions under safety constraints, thereby avoiding collisions between UAVs. Numerical simulation results demonstrate that the optimization method based on safety reinforcement learning considering communication cost outperforms other optimization methods. Communication between UAVs effectively addresses the issue of redundant or incomplete coverage for mobile user devices, reducing computation latency and energy consumption for task offloading. Additionally, the introduction of safety reinforcement learning effectively avoids collisions between UAVs.

关键词： Internet of Things   Mobile edge computing   Lyapunov optimization theory   Deep reinforcement learning   Smart buildings  

摘要： The Internet of Things (IoT) has become an infrastructure that makes smart cities possible. is both accurate and efficient. The intelligent production industry 4.0 period has made mobile edge computing (MEC) essential. Computationally demanding tasks can be delegated from the MEC server to the central cloud servers for processing in a smart city. This paper develops the integrated optimization framework for offloading tasks and dynamic resource allocation to reduce the power usage of all Internet of Things (IoT) gadgets subjected to delay limits and resource limitations. A Federated Learning FL-DDPG algorithm based on the Deep Deterministic Policy Gradient (DDPG) architecture is suggested for dynamic resource management in MEC networks. This research addresses the optimization issues for the CPU frequencies, transmit power, and IoT device offloading decisions for a multi-mobile edge computing (MEC) server and multi-IoT cellular networks. A weighted average of the processing load on the central MEC server (PMS), the system's overall energy use, and the task-dropping expense is calculated as an optimization issue. The Lyapunov optimization theory formulates a random optimization strategy to reduce the energy use of IoT devices in MEC networks and reduce bandwidth assignment and transmitting power distribution. Additionally, the modeling studies demonstrate that, compared to other benchmark approaches, the suggested algorithm efficiently enhances system performance while consuming less energy.

关键词： Energy minimization   federated learning   machine learning based algorithms   mobile edge computing   UE selection and aggregator placement  

摘要： To unveil the hidden value in the datasets of user equipments (UEs) while preserving user privacy, federated learning (FL) is emerging as a promising technique to train a machine learning model using the datasets of UEs locally without uploading the datasets to a central location. Customers require to train machine learning models based on different datasets of UEs, through issuing FL requests that are implemented by FL services in a mobile edge computing (MEC) network. A key challenge of enabling FL in MEC networks is how to minimize the energy consumption of implementing FL requests while guaranteeing the accuracy of machine learning models, given that the availabilities of UEs usually are uncertain. In this paper, we investigate the problem of energy minimization for FL in an MEC network with uncertain availabilities of UEs. We first consider the energy minimization problem for a single FL request in an MEC network. We then propose a novel optimization framework for the problem with a single FL request, which consists of (1) an online learning algorithm with a bounded regret for the UE selection, by considering various contexts (side information) that influence energy consumption;and (2) an approximation algorithm with an approximation ratio for the aggregator placement for a single FL request. We third deal with the problem with multiple FL requests, for which we devise an online learning algorithm with a bounded regret. We finally evaluate the performance of the proposed algorithms by extensive experiments. Experimental results show that the proposed algorithms outperform their counterparts by reducing at least 13% of the total energy consumption while achieving the same accuracy.

关键词： Deep Reinforcement Learning   Mobile Edge Computing   Multi-user   Collaborative inference   Hybrid action space  

摘要： This study tackles the problem of increasing efficiency and scalability in deep neural network (DNN) systems by employing collaborative inference, an approach that is gaining popularity because to its ability to maximize computational resources. It involves splitting a pre-trained DNN model into two parts and running them separately on user equipment (UE) and edge servers. This approach is advantageous because it results in faster and more energy-efficient inference, as computation can be offloaded to edge servers rather than relying solely on UEs. However, a significant challenge of collaborative belief is the dynamic coupling of DNN layers, which makes it difficult to separate and run the layers independently. To address this challenge, we proposed a novel approach to optimize collaborative inference in a multi-agent scenario where a single-edge server coordinates the assumption of multiple UEs. Our proposed method suggests using an autoencoder-based technique to reduce the size of intermediary features and constructing tasks using the deep policy inference Q-inference network's overhead (DPIQN). To optimize the collaborative inference, employ the Deep Recurrent Policy Inference Q-Network (DRPIQN) technique, which allows for a hybrid action space. The results of the tests demonstrate that this approach can significantly reduce inference latency by up to 56% and energy usage by up to 72% on various networks. Overall, this proposed approach provides an efficient and effective method for implementing collaborative inference in multi-agent scenarios, which could have significant implications for developing DNN systems.

关键词： Internet of vehicle   Mobile edge computing   Offloading strategy   Latency   Energy consumption   Smart IoV  

摘要： The current task offloading technique needs to be performed more effectively. Onboard terminals cannot execute efficient computation due to the explosive expansion of data flow, the quick increase in vehicle population, and the growing scarcity of spectrum resources. As a result, this study suggests a task-offloading technique based on reinforcement learning computing for the Internet of Vehicles edge computing architecture. The system framework for the Internet of Vehicles has been initially developed. Although the control centre gathers all vehicle information, the roadside unit collects vehicle data from the neighborhood and sends it to a mobile edge computing server for processing. Then, to guarantee that job dispatching in the Internet of Vehicles is logical, the computation model, communications approach, interfering approach, and concerns about confidentiality are established. This research examines the best way to analyze and design a computation offloading approach for a multiuser smart Internet of Vehicles (IoV) based on mobile edge computing (MEC). We present an analytical offloading strategy for various MEC networks, covering one-to-one, one-to-two, and two-to-one situations, as it is challenging to determine an analytical offloading proportion for a generic MEC-based IoV network. The suggested analytic offload strategy may match the brute force (BF) approach with the best performance of the Deep Deterministic Policy Gradient (DDPG). For the analytical offloading design for a general MEC-based IoV, the analytical results in this study can be a valuable source of information.