关键词： Collaborative inference   computation partition   mobile edge computing   multi-agent reinforcement learning  

摘要： Collaborative inference in mobile edge computing (MEC) enables mobile devices to offload the computation tasks for the computation-intensive perception services, and the inference policy determines the inference latency and energy consumption. The optimal inference policy depends on the inference performance model of deep learning, the data generation model and the network model that are rarely known by mobile devices in time. In this paper, we propose a multi-agent reinforcement learning (RL) based energy-efficient MEC collaborative inference scheme, which enables each mobile device to choose both the partition point of deep learning and the collaborative edge of each mobile device based on the image quantity, the channel conditions and the previous inference performance. A learning experience exchange mechanism exploits the Q-values of the neighboring mobile devices to accelerate the inference policy optimization with less energy consumption. We also provide a deep multi-agent RL based inference scheme to accelerate learning for large-scale MEC networks, in which an actor network yields the collaborative inference policy probability distribution and a critic network guides the weight update of the actor network to enhance sample efficiency. We provide the inference performance bound and analyze the computational complexity. Both simulation and experimental results show that our proposed schemes reduce the inference latency and save the MEC energy consumption.

关键词： Task analysis   Satellites   Delays   Servers   Computer architecture   Cloud computing   Computational modeling   Satellite-terrestrial networks   mobile edge computing   computation offloading   delay optimization  

摘要： The integrated satellite-terrestrial network is promising to provide global broadband communication service. However, the long propagation delay of satellite-terrestrial links will lead to high communication delay when users access the Internet via satellites. In this paper, we investigate the cooperative multi-tier computing in the integrated satellite-terrestrial network, in which the computation tasks of users are processed by leveraging the cooperation of devices, edge nodes, and cloud servers. Based on the proposed three-tier computing framework, we formulate the cooperative edge-cloud offloading problem to minimize the total delay of the network. Considering the computation task is dividable, we propose the optimal task splitting strategy based on the partial offloading model, in which the closed-form solution is derived for each computation task. With the optimal task splitting strategy obtained, the original optimization problem is reformulated as the problem of the time slot allocation strategy and the computation capacity allocation strategy. Then, we further propose the cooperative edge-cloud computing strategy to optimize the delay performance of the network. Finally, numerical results are presented to demonstrate the performance of the proposed three-tier computing architecture and the offloading strategies.

关键词： Array signal processing   Task analysis   Internet of Things   Optimization   Servers   Resource management   Cloud computing   Intelligent reflecting surface (IRS)   mobile edge computing   dynamic beamforming   resource allocation   binary offloading   computation rate  

摘要： In this paper, we develop a unified dynamic intelligent reflecting surface (IRS) beamforming framework to boost the sum computation rate of an IRS-aided mobile edge computing (MEC) system, where each device follows a binary offloading policy. Specifically, the task of each device has to be either executed locally or offloaded to MEC servers as a whole with the aid of given number of IRS beamforming vectors available. By flexibly controlling the number of times for IRS reconfiguring phase-shifts, the system can achieve a balance between the performance and associated signalling overhead. We aim to maximize the sum computation rate by jointly optimizing the computational mode selection for each device, offloading time allocation, and IRS beamforming vectors across time. Since the resulting optimization problem is non-convex and NP-hard, there are generally no standard methods to solve it optimally. To tackle this problem, we first propose a penalty-based successive convex approximation algorithm, where all the associated variables in the inner-layer iterations are optimized simultaneously and the obtained solution is guaranteed to be locally optimal. Then, we further derive the offloading activation condition for each device by deeply exploiting the intrinsic structure of the original optimization problem. According to the offloading activation condition, a low-complexity algorithm based on the successive refinement method is proposed to obtain high-quality suboptimal solutions, which are more appealing for practical systems with a large number of devices and IRS elements. Moreover, the optimal condition for the proposed low-complexity algorithm is revealed. The effectiveness of the proposed algorithms is demonstrated through numerical examples. In addition, the results illustrate the practical significance of the IRS in MEC systems for achieving coverage extension and supporting multiple energy-limited devices for task offloading, and also unveil the fundamental perf

关键词： Mobile edge computing   multimodal optimization   particle swarm algorithm   server placement problem  

摘要： Mobile edge computing (MEC) deploys computing and storage resources close to mobile devices, enabling resource demanding applications to run on mobile devices with short network latency. In the past few years, large numbers of research works focused on the research hotspots in MEC, such as computation offloading and energy efficiency. However, few researchers have investigated the deployment of edge servers. On the one hand, blindly deploying numerous edge servers will result in a large amount of capital expenditure. On the other hand, the deployment of edge servers is a multimodal problem that should provide decision makers with multiple deployment options to deal with the impact of unmeasured real-world factors. Considering these factors, we study the multimodal optimization problem of edge server placement (MESP) with the goal of minimizing the average system response time in this work. Regarding the difficulty of the MESP problem, we propose a heuristic algorithm that combines particle swarm optimization and niching technology to obtain a set of competitive placement solutions. Extensive experiments over a real-world dataset show that the proposed algorithm can significantly reduce the system response time.

关键词： edge AI   mobile edge computing   deep learning   citrus orchards  

摘要： Deep Learning models have presented promising results when applied to Agriculture 4.0. Among other applications, these models can be used in disease detection and fruit counting. Deep Learning models usually have many layers in the architecture and millions of parameters. This aspect hinders the use of Deep Learning on mobile devices as they require a large amount of processing power for inference. In addition, the lack of high-quality Internet connectivity in the field impedes the usage of cloud computing, pushing the processing towards edge devices. This work describes the proposal of an edge AI application to detect and map diseases in citrus orchards. The proposed system has low computational demand, enabling the use of low-footprint models for both detection and classification tasks. We initially compared AI algorithms to detect fruits on trees. Specifically, we analyzed and compared YOLO and Faster R-CNN. Then, we studied lean AI models to perform the classification task. In this context, we tested and compared the performance of MobileNetV2, EfficientNetV2-B0, and NASNet-Mobile. In the detection task, YOLO and Faster R-CNN had similar AI performance metrics, but YOLO was significantly faster. In the image classification task, MobileNetMobileV2 and EfficientNetV2-B0 obtained an accuracy of 100%, while NASNet-Mobile had a 98% performance. As for the timing performance, MobileNetV2 and EfficientNetV2-B0 were the best candidates, while NASNet-Mobile was significantly worse. Furthermore, MobileNetV2 had a 10% better performance than EfficientNetV2-B0. Finally, we provide a method to evaluate the results from these algorithms towards describing the disease spread using statistical parametric models and a genetic algorithm to perform the parameters' regression. With these results, we validated the proposed pipeline, enabling the usage of adequate AI models to develop a mobile edge AI solution.

关键词： Augmented reality   Offloading   Mobile edge computing   Energy efficiency   Latency   5G  

摘要： Augmented reality (AR) is a recent communication paradigm that is considered one of the primary functions of the fifth-generation cellular system (5G). AR is one of the ultra-reliable low latency communications since it requires a total delay of about 5 ms for AR applications. Since mobile devices are battery-operated and have many limitations in computing resources, i.e., processing, storage, and energy, computing tasks associated with mobile AR applications have to be offloaded to nearby devices with sufficient computing resources. To this end, this paper provides a system structure for mobile AR and web-based AR applications;based on the modified multi-level mobile edge computing system (MM-MEC). Furthermore, the required offloading algorithms for managing data offloading from mobile devices to edge servers are introduced. The algorithms are latency and energy-aware, as the selection of the executing device for the AR computing tasks achieves the highest energy and latency efficiency. The proposed structure and offloading techniques have been validated through simulation processes. The developed framework and the proposed offloading schemes have been simulated over a reliable environment for various simulation The suggested system and its offloading methods were simulated in a dependable environment for a number of simulation situations.

关键词： Mobile edge computing   Task offloading   Location prediction   Machine learning  

摘要： Several applications have emerged with the proliferation of mobile devices to provide communication, learning, social networking, entertainment, and community computing services. Such applications include augmented reality, online gaming, and other real-time applications that need higher computational resources. These applications, executing on mobile devices, often need to access external computing resources and offload the application tasks to the cloud or mobile edge computing (MEC) servers. However, delivering task offloading results to the users in the MEC environment is a challenge, certainly when user mobility is high. Sub-optimal server selection at the offloading stage increases latency, energy consumption and deteriorates both quality of experience and quality of service. Existing techniques proposed in the literature handle computation offloading and mobility management separately. Without considering the real-time mobility factors, the solutions produced are sub-optimal. Some solutions exist to manage mobility, but they involve higher time complexity. We consider the user mobility in offloading decisions and present a lightweight mobility prediction and offloading (LiMPO) framework that offloads the compute-intensive tasks to the predicted user location using artificial neural networks with less complexity. In addition, we propose a multi-objective genetic algorithm based server selection technique that jointly optimizes latency and energy consumption while improving the resource utilization of MEC servers. The performance of the proposed framework is compared with two other techniques task-assignment with optimized mobility and dynamic mobility-aware offloading algorithm for edge computing. The simulation results show that LiMPO outperforms the others by latency reduction, energy efficiency, and enhanced resource utilization.

关键词： deep reinforcement learning   energy harvesting   mobile edge computing   service caching  

摘要： Mobile user terminals in 5G networks can generate massive computational workloads, which require sufficient computation and caching resources, and the processors of user terminals cannot tackle these workloads. Emerging mobile edge computing (MEC) has become the key to solving the computation problem by offloading computation-intensive workloads to the MEC server. To make full use of the limited resources on the MEC side, service caching can pre-store specific executable programs, databases, or libraries for executing offloaded workloads from user terminals. In this study, a service caching-assisted MEC model is designed. According to the proposed MEC model, a decentralized model-free deep reinforcement learning algorithm-based server caching optimization policy (DDSCOP) is proposed to minimize the long-term weighted average cost. Considering the time-varying request workloads from user terminals, the stochastic channel state, and the arrival of renewable resources of the MEC server, DDSCOP can find the near-optimal service caching decision-making policy by training the neural networks, that is, whether the service caching can be hosted by the MEC server, and which MEC server hosts the service caching. The numerical experimental results verify the convergence and effectiveness of DDSCOP through extensive parameters configuration, and DDSCOP outperforms the three baseline algorithms.

关键词： computation offloading   mobile edge computing   privacy   security   smart campus   virtual reality  

摘要： Nowadays, as the smart campus concept becomes a reality, virtual reality (VR) applications are being applied to connect students with the virtual teaching world via VR devices (VDs), enhancing learning efficiency. Nevertheless, VR applications are latency-sensitive while VDs are subjected to shortcomings to handle many VR applications simultaneously. Fortunately, mobile edge computing (MEC) has been recognized as a promising solution that can bring abundant resources to VDs to relieve hardware limits. However, the computing resources of edge nodes in MEC are limited, and thus how to allocate resources effectively is critical. Meanwhile, some sensitive information of students collected by VD needs to be protected. In view of this, we investigate computation offloading for educational VR applications in MEC-enabled smart campus. The aims of this issue are to optimize the motion-to-photon latency, energy consumption, and resource utilization while satisfying the privacy and security constraints. To this end, we propose a new multi-objective optimization method using MoCell. Eventually, the experimental evaluations are designed to illustrate the effectiveness and superiority our proposed method.

关键词： Internet of Vehicles   mobile edge computing   blockchain   resource allocation   asynchronous advantage actor-critic  

摘要： Recently, the rise of the Internet of Vehicles (IoV) has driven the broad development of intelligent transportation and smart cities. In order to promote the computing power of mobile vehicles and decrease the content delivery latency of suppliers, mobile edge computing (MEC) is recognized as a promising computational paradigm and used in vehicular networks. However, there are some essential issues to be considered: 1) privacy and authenticity of data transmission in IoV, and 2) reasonable resource allocation for collaborative computing and caching. In this paper, to solve above issues, blockchain technology is introduced and adopted to ensure the accuracy and reliability of data transmission and interaction. Meanwhile, we develop an intelligent framework of resource allocation about computing and caching for blockchain-enabled MEC systems in IoV. Through jointly considering and optimizing offloading decision of computation task carried by vehicle, caching decision, the number of offloaded consensus nodes, block interval and block size, the weighted consumption costs of energy consumption and computation overheads can be decreased, and the transactional throughput of the blockchain can be increased. Moreover, due to the continuity and dynamic of the available resources of mobile vehicles and computing servers, the optimization problem is modeled as a Markov decision process (MDP). Facing the large-scale and dynamic characteristics of the system, the asynchronous advantage actor-critic (A3C) approach is introduced to deal with the optimization problem. Simulation results show that our proposed scheme achieves significant advantages over other comparison schemes, such as the total reward of the proposed scheme is about 14% higher than that of the deep Q-network based scheme.