关键词： Mobile edge computing   inference acceleration   DNN partition   task offloading   distributed SAC  

摘要： In mobile edge computing environment, intelligent inference services driven by DNN are highly sensitive to latency. Recently, collaborative inference between User Devices and Edge Servers (ESs) based on Deep Neural Networks (DNN) partition has achieved success in service acceleration. However, most of the existing collaborative acceleration schemes are partitioned for a single DNN inference task, which cannot quickly make partition decisions for a set of concurrent inference tasks, and often sacrifice inference accuracy. In addition, due to the limited resources of ESs, there is resource competition among concurrent requests, which makes the partitioned tasks cannot be offloaded to ESs in time for processing. Therefore, designing an efficient offloading scheme becomes essential. The task offloading schemes based on deep reinforcement learning can solve complex decision-making problems in high-dimensional state space, but they have problems such as insufficient sample diversity and easily falling into local optimum. In this paper, a Collaborative Inference Acceleration Scheme integrating DNN Partitioning and Task Offloading (CIAS-PnO) is proposed. First, while ensuring inference accuracy, the Collaborative DNN Layer Partitioning (CDLP) algorithm is designed with the goal of optimal latency. CDLP can reduce the problem scale of concurrent inference tasks partition by pruning operation and determine the partition decisions in time. Then, the Distributed Soft Actor-Critic (SAC)-based Partition Task Offloading algorithm (DSACO) is designed. DSACO supports SAC Agents to explore samples in parallel and share learning experiences, and uses the automatic entropy adjustment mechanism to improve the exploration efficiency of Agents, so as to avoid falling into local optimum and achieve efficient offloading of partition tasks. Experimental results on DNN benchmarks show that compared with the baseline acceleration schemes, CIAS-PnO achieves more than 19.8% acceleration performa

关键词： mobile edge computing   wireless energy transfer   reconfigurable intelligent surfaces   offloading   unmanned aerial vehicle   rate-splitting multiple access  

摘要： In this study, a reconfigurable intelligent surface (RIS)-assisted wireless-powered mobile edge computing (WP-MEC) system is proposed, where a single-antenna unmanned aerial vehicle (UAV)-mounted cloudlet provides offloading opportunities to K user equipments (UEs) with a single antenna, and the K UEs can harvest the energy from the broadcast radio-frequency signals of the UAV. In addition, rate-splitting multiple access is used to provide offloading opportunities to multiple UEs for effective power control and high spectral efficiency. The aim of this paper is to minimize the total energy consumption by jointly optimizing the resource allocation in terms of time, power, computing frequency, and task load, along with the UAV trajectory and RIS phase-shift matrix. Since coupling issues between optimization variable designs are caused, however, an alternating optimization-based algorithm is developed. The performance of the proposed algorithm is verified via simulations and compared with the benchmark schemes of partial optimizations of resource allocation, path planning, and RIS phase design. The proposed algorithm exhibits high performance in WP-MEC systems with insufficient resources, e.g., achieving up to 40% energy reduction for a UAV with eight elements of RIS.

关键词： Servers   Task analysis   Resource management   Optimization   Stability analysis   Channel allocation   Delays   Lyapunov optimization   mobile edge computing   deep reinforcement learning (DRL)   queueing networks  

摘要： Mobile Edge Computing (MEC) is a key technology towards delay-sensitive and computation-intensive applications in future cellular networks. In this paper, we consider a multi-user, multi-server system where the cellular base station is assisted by a UAV, both of which provide additional MEC services to the terrestrial users. Via dual connectivity (DC), each user can simultaneously offload tasks to the macro base station and the UAV-mounted MEC server for parallel computing, while also processing some tasks locally. We aim to propose an online resource management framework that minimizes the average power consumption of the whole system, considering long-term constraints on queue stability and computational delay of the queueing system. Due to the coexistence of two servers, the problem is highly complex and formulated as a multi-stage mixed integer non-linear programming (MINLP) problem. To solve the MINLP with reduced computational complexity, we first adopt Lyapunov optimization to transform the original multi-stage problem into deterministic problems that are manageable in each time slot. Afterward, the transformed problem is solved using an integrated learning-optimization approach, where model-free Deep Reinforcement Learning (DRL) is combined with model-based optimization. Via extensive simulation and theoretical analyses, we show that the proposed framework is guaranteed to converge and can produce nearly the same performance as the optimal solution obtained via an exhaustive search.

关键词： Mobile edge computing  

摘要： Artificial intelligence (AI) technologies have emerged as pivotal enablers across a multitude of industries, including consumer electronics, healthcare, and manufacturing, largely due to their significant resurgence over the past decade. The transformative power of AI is primarily derived from the utilization of deep neural networks (DNNs), which require extensive data for training and substantial computational resources for processing. Consequently, DNN models are typically trained and deployed on resource-rich cloud servers. However, due to potential latency issues associated with cloud communications, deep learning (DL) workflows (e.g., DNN training and inference) are increasingly being transitioned to wireless edge networks in proximity to end-user devices (EUDs). This shift is designed to support latency-sensitive applications and has given rise to a new paradigm of edge AI, which will play a critical role in upcoming sixth-generation (6G) networks to support ubiquitous AI applications. Despite its considerable potential, edge AI faces substantial challenges, mostly due to the dichotomy between the resource limitations of wireless edge networks and the resource-intensive nature of DL. Specifically, the acquisition of large-scale data, as well as the training and inference processes of DNNs, can rapidly deplete the battery energy of EUDs. This necessitates an energy-conscious approach to edge AI to ensure both optimal and sustainable performance. In this paper, we present a contemporary survey on green edge AI. We commence by analyzing the principal energy consumption components of edge AI systems to identify the fundamental design principles of green edge AI. Guided by these principles, we then explore energy-efficient design methodologies for the three critical tasks in edge AI systems, including training data acquisition, edge training, and edge inference. Finally, we underscore potential future research directions to further enhance the energy efficiency of

关键词： Resource allocation   sustainable   smart city   5G communication resources   privacy data   mobile edge computing  

摘要： In the field of Internet of Things, mobile communication will be deeply integrated with industrial construction, biomedicine, agricultural production, transportation and other industries to fully realize the "Internet of Everything" and realize smart city and sustainable development. This paper intends to solve problems in increasing latency and consumption of energy and security after deploying Mobile Edge Computing (MEC) servers in single cell multi-user network model, and has advocated a strategy of allocating the resources of 5G communication in regards to the protection mechanism for privacy in the environment of edge computing. First of all, solution model for the problem is characterized as a model with time delay and energy consumption. Second, privacy is used to measure the uncertainty of data in order to quantify privacy data while computing the allocation of tasks and facilitate the design of objective function. Finally, the task migration problem of minimizing energy consumption under the limitation of completion time is constructed as a nonlinear 0-1 programming problem. Besides, we design the optimal resource allocation strategy of discrete algorithm of binary particle swarm optimization (BPSO) for problem solution. Results indicate that while the quantity of users is 100, the total user expenditure of proposed method is 18.3J, which is lower than 20.0J and 30.7J of the compared methods. Moreover, the proposed method has lower latency and energy consumption, which can well balance the load of MEC servers.

关键词： Mobile edge computing  

摘要： To support the running of human-centric metaverse applications on mobile devices, Unmanned Aerial Vehicle (UAV)- assisted Wireless Powered Mobile Edge Computing (WPMEC) is promising to compensate for limited computational capabilities and energy supplies of mobile devices. The high-speed computational processing demands and significant energy consumption of metaverse applications require joint resource scheduling of multiple devices and UAVs, but existing WPMEC solutions address either device or UAV scheduling due to the complexity of combinatorial optimization. To solve the above challenge, we propose a two-stage alternating optimization algorithm based on multi-task Deep Reinforcement Learning (DRL) to jointly allocate charging time, schedule computation tasks, and optimize trajectory of UAVs and mobile devices in a wireless powered metaverse scenario. First, considering energy constraints of both UAVs and mobile devices, we formulate an optimization problem to maximize the computation efficiency of the system. Second, we propose a heuristic algorithm to efficiently perform time allocation and charging scheduling for mobile devices. Following this, we design a multi-task DRL scheme to make charging scheduling and trajectory design decisions for UAVs. Finally, theoretical analysis and performance results demonstrate that our algorithm exhibits significant advantages over representative methods in terms of convergence speed and average computation efficiency. Copyright © 2023, The Authors. All rights reserved.

关键词： Mobile edge computing  

摘要： Federated Learning (FL) enables many resource-limited devices to train a model collaboratively without data sharing. However, many existing works focus on model-homogeneous FL, where the global and local models are the same size, ignoring the inherently heterogeneous computational capabilities of different devices and restricting resource-constrained devices from contributing to FL. In this paper, we consider model-heterogeneous FL and propose Federated Partial Model Training (FedPMT), where devices with smaller computational capabilities work on partial models (subsets of the global model) and contribute to the global model. Different from Dropout-based partial model generation, which removes neurons in hidden layers at random, model training in FedPMT is achieved from the back-propagation perspective. As such, all devices in FedPMT prioritize the most crucial parts of the global model. Theoretical analysis shows that the proposed partial model training design has a similar convergence rate to the widely adopted Federated Averaging (FedAvg) algorithm, O(1/T), with the sub-optimality gap enlarged by a constant factor related to the model splitting design in FedPMT. Empirical results show that FedPMT significantly outperforms the existing benchmark FedDrop. Meanwhile, compared to the popular model-homogeneous benchmark, FedAvg, FedPMT reaches the learning target in a shorter completion time, thus achieving a better trade-off between learning accuracy and completion time. © 2023, CC BY-NC-SA.

关键词： Internet of vehicles   Load balancing   Reusable task caching   Mobile edge computing  

摘要： In order to solve the problem of limited resources in the vehicle, mobile edge computing (MEC) is integrated into the cellular-vehicle-to-everything (C-V2X) system to provide users with low-latency services. Considering the resource limitation of computing, storage and communication of MEC servers, it is necessary to improve the system performance through the collaborative optimization of multi-domain resources. Therefore, for applications with double dependence on time and data in C-V2X scenarios, we design a MEC hierarchical resource management framework to jointly optimize system offloading decision, scheduling decision, and caching decision to minimize the system delay. The optimization problem is decomposed into a resource allocation problem within a single MEC server and a load balancing problem among multiple MEC servers. For the former, we present a task scheduling algorithm based on the latest start time and a caching decision algorithm based on dynamic programming to minimize the average task completion delay;for the latter, we present a load balancing algorithm based on the coalition game to minimize the global system delay. Simulation results show that the proposed scheme can significantly improve the performance in terms of application completion delay, application failure rate and resource utilization compared with the benchmark schemes.

关键词： Mobile edge computing   distributed edge caching   mean-field game theory   dynamic requirements   optimal control  

摘要： Mobile edge computing (MEC) can use wireless access network (RAN) to provide users with nearby information technology (IT) services and cloud computing functions, which creates a high-performance and low latency service environment. By caching the popular content at small base station (SBS) can reduce the heavy backhaul load and the content retransmission. However, the time-varying and dynamic of the content requests may lead to the base station to cache the useless contents. In this paper, we study a distributed caching optimization problem in edge networks (ENs) with the spatio-temporal requirements. In the considered ENs, the cache control is described as a stochastic differential game (SDG) in which each SBS defines a caching strategy to reduce the total cost in terms of the service delay and backhaul link load. To reduce the computational complexity, the original optimization problem is transformed into a mean field game (MFG). We propose a distributed caching iterative control algorithm that decouples the information interaction between the general SBS and others through the mean field distribution. In addition, we obtain the optimal edge caching control strategy, while the existence and uniqueness of the mean field equilibrium (MFE) can also be guaranteed. Simulation results demonstrate that our proposed caching control algorithm can average reduce 27.12% storage cost and achieve better performance than other existing schemes.

关键词： Discrete power allocation   online user allocation   mobile edge computing   non-orthogonal multiple access (NOMA)  

摘要： This study tackles the online user allocation problem in mobile edge computing (MEC) systems powered by non-orthogonal multiple access. App vendors need to determine a proper wireless channel in a base station/edge server and sufficient transmit power for every user. We consider a stochastic MEC system where users arrive and depart over time. When an edge server runs out of computing resources, some users will have to wait until the resources become available again, which incurs an allocation delay cost. This cost is often not investigated in many studies, which also do not consider a multi-cell, multi-channel system as we do in this work, due to its complexity. We aim to minimize the allocation delay and transmit power costs, increasing the system's energy efficiency. To achieve this objective while guaranteeing users' data rate requirements over time, we adopt the Lyapunov framework to convert this long-term optimization problem into a series of subproblems to be solved in every time slot. To solve the aforementioned subproblems efficiently, we present a distributed game theory-based approach. The proposed algorithm is theoretically evaluated and experimentally demonstrated to outperform several baseline and state-of-the-art methods, highlighting the significance of systematic consideration for both computation and communication aspects of this problem.