关键词： Task analysis   Privacy   Topology   Delays   Costs   Vehicle dynamics   Processor scheduling   Mobile edge computing   directed acyclic graph   dependency-aware   computation offloading   deep reinforcement learning   vehicular edge computing networks  

摘要： The proliferation of computation-intensive and delay-sensitive applications in the Internet of Vehicles (IoV) poses great challenges to resource-constrained vehicles. To tackle this issue, Mobile Edge Computing (MEC) enabling offloading on-vehicle tasks to edge servers has emerged as a promising approach. MEC jointly augments network computing capabilities and alleviates resource utilization for IoV. Nevertheless, the efficacy of MEC depends heavily on the adopted offloading scheme, especially in the presence of complex subtask dependencies. Existing research has largely overlooked the crucial dependencies among subtasks, which significantly influences the decision making for offloading. This work attempts to schedule subtasks with guaranteed dependencies while minimizing system latency and energy costs in multi-vehicle scenarios. First, we introduce a subtask priority scheduling method based on the Directed Acyclic Graph (DAG) topological structure to ensure the priority order of subtasks, especially in the scenarios with complex interdependencies. Second, in light of privacy concerns and limited information sharing, we propose an Optimized Distributed Computation Offloading (ODCO) scheme based on deep reinforcement learning (DRL), alleviating the conventional requirement for extensive vehicle-specific information sharing to achieve optimal offloading performance. The adaptive k-step learning approach is further presented to enhance the robustness of the training process. Numerical experiments are presented to illustrate the advantages of the proposed scheme compared to the existing state-of-the-art offloading schemes. For instance, the ODCO achieved a system utility of approximately 0.80 within 300 episodes, obtaining utility gains of about 0.05 compared to the distributed earliest-finish time offloading (DEFO) algorithm with around 500 episodes.

关键词： deep reinforcement learning   local differential privacy   machine learning   mobile edge computing   task offloading  

摘要： As machine learning (ML) technologies continue to evolve, there is an increasing demand for data. Mobile crowd sensing (MCS) can motivate more users in the data collection process through reasonable compensation, which can enrich the data scale and coverage. However, nowadays, users are increasingly concerned about their privacy and are unwilling to easily share their personal data. Therefore, protecting privacy has become a crucial issue. In ML, federated learning (FL) is a widely known privacy-preserving technique where the model training process is performed locally by the data owner, which can protect privacy to a large extent. However, as the model size grows, the weak computing power and battery life of user devices are not sufficient to support training a large number of models locally. With mobile edge computing (MEC), user can offload some of the model training tasks to the edge server for collaborative computation, allowing the edge server to participate in the model training process to improve training efficiency. However, edge servers are not fully trusted, and there is still a risk of privacy leakage if data is directly uploaded to the edge server. To address this issue, we design a local differential privacy (LDP) based data privacy-preserving algorithm and a deep reinforcement learning (DRL) based task offloading algorithm. We also propose a privacy-preserving distributed ML framework for MEC and model the cloud-edge-mobile collaborative training process. These algorithms not only enable effective utilization of edge computing to accelerate machine learning model training but also significantly enhance user privacy and save device battery power. We have conducted experiments to verify the effectiveness of the framework and algorithms.

关键词： Mobile edge computing   Task graph   Computation offloading   Reinforcement learning  

摘要： Various mobile applications that comprise dependent tasks are gaining widespread popularity and are increasingly complex. These applications often have low-latency requirements, resulting in a significant surge in demand for computing resources. With the emergence of mobile edge computing (MEC), it becomes the most significant issue to offload the application tasks onto small-scale devices deployed at the edge of the mobile network for obtaining a high-quality user experience. However, since the environment of MEC is dynamic, most existing works focusing on task graph offloading, which rely heavily on expert knowledge or accurate analytical models, fail to fully adapt to such environmental changes, resulting in the reduction of user experience. This paper investigates the task graph offloading in MEC, considering the time-varying computation capabilities of edge computing devices. To adapt to environmental changes, we model the task graph scheduling for computation offloading as a Markov Decision Process (MDP). Then, we design a deep reinforcement learning algorithm (SATA-DRL) to learn the task scheduling strategy from the interaction with the environment, to improve user experience. Extensive simulations validate that SATA-DRL is superior to existing strategies in terms of reducing average makespan and deadline violation.

关键词： Internet of Vehicles   mobile edge computing   iterative double auction   evolutionary game   cloud mining   Internet of Vehicles   mobile edge computing   iterative double auction   evolutionary game   cloud mining  

摘要： With the growing prominence of the Internet of Vehicles (IoV), there is an unprecedented generation of vehicular data, which harbors immense potential for diverse applications. Considering the convergence of blockchain technology and IoV, blockchain can be used as a distributed ledger to establish a decentralised and autonomous trading platform for IoV networks. However, the limitations in the power and computational capacities of IoV devices hinder their direct involvement in the Proof-of-Work (PoW) process. Addressing this challenge, offloading the computational demands of PoW to Mobile Edge Computing (MEC) servers emerges as a feasible solution. This scenario has prompted us to conceptualize the computational resource management in blockchain consensus as a two-layered resource game, aimed at optimizing both the profits of computing power providers and the utility for vehicles. In the Blockchain Market Layer subgame, we employ a game-theoretic framework to analyze the interactions between MEC servers and vehicles within a PoW-based blockchain network. To optimize the system utility, we introduce a centralized evolutionary game-based pool selection algorithm. Simultaneously, in the Edge Cloud Network Layer subgame, a reverse auction algorithm is implemented to align the computing power supply with the demand within the blockchain market. Here, the cloud data center functions as the seller of computing power, while the edge servers aggregate the computing demand from IoV devices. A broker is then tasked with the allocation and pricing of computing resources, in accordance with the bids, ensuring adherence to economic principles of authenticity and individual rationality. Simulation results demonstrate that our mechanism enhances system efficiency while ensuring revenue, thereby enabling an effective distribution of computing resources between vehicles and cloud data centers.

关键词： Streaming media   Quality of experience   Bit rate   Channel allocation   Bandwidth   Wireless communication   Resource management   Mobile edge computing   video streaming   resource allocation   deep reinforcement learning  

摘要： This letter investigates a video streaming system enabled by mobile edge computing (MEC), where we aim to maximize the user quality of experience (QoE) under dynamic mobile environments, through user association, bitrate adaptation and bandwidth allocation. To tackle this hybrid optimization problem, we first decompose it into discrete inter-frame and continuous intra-frame problems. In further, an in-depth analysis of intra-frame bandwidth allocation is conducted, guiding the design of a modified equal bandwidth allocation (MEBA) algorithm. Furthermore, we adopt a deep reinforcement learning (DRL) approach to cope with inter-frame user association and bitrate adaptation in dynamic mobile environments, while embedding MEBA in the DRL algorithm to jointly tackle the hybrid problem. Simulations are finally conducted to verify the superiority of the proposed method under the dynamic environment.

关键词： Mobile edge computing   two-timescale   smart grid   resource allocation   Lyapunov optimization   Mobile edge computing   two-timescale   smart grid   resource allocation   Lyapunov optimization  

摘要： Powering mobile edge computing (MEC) with a hybrid supply of smart grid (SG) and renewable energy source (RES) offers an opportunity to utilize clean energy and cut down energy expenses under two-way energy transactions. We propose two-timescale online resource allocation and energy management (TSRE) for MEC with a hybrid power supply, adapting to dynamic task and RES arrivals, wireless channels, and energy prices. The TSRE minimizes the time-averaged cost of predictive energy planning and real-time energy trading of base stations (BSs), and the energy usage of mobile users. By generalizing the Lyapunov optimization and stochastic subgradient method, the energy planning (sub)problem is solved upon RES arrivals using only historical data. The real-time offloading schedules, energy trading decisions and CPU configurations are decoupled over time and (asymptotically) optimally made in a distributed manner. The feasibility and the asymptotic optimality of the TSRE are proved. Numerical results demonstrate that the TSRE saves system cost significantly by 33.7%, compared to its baselines.

关键词： Mobile edge computing   D2D underlay communication   Mean field game   Latency-critical applications   Dynamic system  

摘要： With the development of 5 G mobile users are increasing massively. Some mobile applications like healthcare are latency-critical and requires real-time data processing. A preference-based task offloading framework in mobile edge computing with a device-to-device offloading (MECD2D) system has been proposed to fulfill the latency demands of such applications for minimum energy consumption ensuring resiliency. The problem is formulated as a constraint-based non-linear optimization problem which is complex. The resources are allocated in two steps. In the first step, resources are allocated based on latency demand to ensure resiliency. In the second step, allocated resources are optimized using a non-cooperative mean field game for dynamic system. To ensure the performance of the system for dynamic network, the results are executed on a real-time Shanghai dataset. The computational results indicate that the proposed algorithm performs better in terms of energy consumption. Other parameters such as throughput, network utilization and task computation are also analysed. The results are verified by performing the proposed algorithm with existing Q learning and mean-field game algorithms. The results performed on the dataset indicate an improvement in energy consumption by 5-10 %, and 10-50 % as compared to Q learning and mean-field game respectively.

关键词： blockchain   consensus algorithm   mobile edge computing   reputation mechanism  

摘要： Mobile edge computing (MEC) promotes the development and popularity of Internet of Things (IoT) devices with higher connectivity and ultra-low latency through network topology optimization and real-time data analysis. With the exponential growth of data traffic generated by IoT devices, it is essential to prevent edge servers (ESs) from overloading through efficient task scheduling to ensure the quality of service (QoS). However, the lack of trust and incentive mechanisms between ESs deployed and managed by different infrastructure providers makes it challenging to perform peer-offloaded tasks. Moreover, to meet the low-latency requirement of many IoT applications, the execution efficiency of the task scheduling scheme in MEC needs to be improved. To address the above challenges systematically, a QoS-driven task scheduling scheme with blockchain in MEC named SharpEdge is proposed. In SharpEdge, ESs publish tasks with rewards and then select the most reliable executors through a historical performance-based reputation mechanism to perform peer offloading. The performance of the executor will be recorded into the blockchain after the task is completed. In addition, a concurrent consensus mechanism using sharding technology is designed, improving task scheduling efficiency. We implement SharpEdge based on Hyperledger Fabric and verify its performance in a simulated MEC environment.

关键词： computing offloading   deep reinforcement learning   internet of things   mobile edge computing  

摘要： In the last few years, the rapid advancement of the Internet of Things (IoT) and the widespread adoption of smart cities have posed new challenges to computing services. Traditional cloud computing models fail to fulfil the rapid response requirement of latency-sensitive applications, while mobile edge computing (MEC) improves service efficiency and customer experience by transferring computing tasks to servers located at the network edge. However, designing an effective computing offloading strategy in complex scenarios involving multiple computing tasks, nodes, and services remains a pressing issue. In this paper, a computing offloading approach based on Deep Reinforcement Learning (DRL) is proposed for large-scale heterogeneous computing tasks. First, Markov Decision Processes (MDPs) is used to formulate computing offloading decision and resource allocation problems in large-scale heterogeneous MEC systems. Subsequently, a comprehensive framework comprising the "end-edge-cloud" along with the corresponding time-overhead and resource allocation models is constructed. Finally, through extensive experiments on real datasets, the proposed approach is demonstrated to outperform existing methods in enhancing service response speed, reducing latency, balancing server loads, and saving energy.

关键词： Mobile edge computing   Deep reinforcement learning   Parallel exploration   Asynchronous training   Offloading decision  

摘要： In mobile edge computing (MEC), randomly offloading tasks to edge servers (ES) can cause wireless devices (WD) to compete for limited bandwidth resources, leading to overall performance degradation. Reinforcement learning can provide suitable strategies for task offloading and resource allocation through exploration and trial-and-error, helping to avoid blind offloading. However, traditional reinforcement learning algorithms suffer from slow convergence and a tendency to get stuck in suboptimal local minima, significantly impacting the energy consumption and data timeliness of edge computing task unloading. To address these issues, we propose Parallel Exploration with Asynchronous Training-based Deep Reinforcement Learning (PEATDRL) algorithm for MEC network offloading decisions. Its objective is to maximize system performance while limiting energy consumption in an MEC environment characterized by time-varying wireless channels and random user task arrivals. Firstly, our model employs two independent DNNs for parallel exploration, each generating different offloading strategies. This parallel exploration enhances environmental adaptability, avoids the limitations of a single DNN, and addresses the issue of agents getting stuck in suboptimal local minima due to the explosion of decision combinations, thereby improving decision performance. Secondly, we set different learning rates for the two DNNs during the training phase and trained them at various intervals. This asynchronous training strategy increases the randomness of decision exploration, prevents the two DNNs from converging to the same suboptimal local solution, and improves convergence efficiency by enhancing sample utilization. Finally, we examine the impact of different parallel levels and training step differences on system performance metrics and explain the parameter choices. Experimental results show that the proposed method provides a viable solution to the performance issues caused by slow converge