关键词： Online learning   Internet of Things   Task offloading   Energy efficiency   Mobile edge computing  

摘要： User equipment produces a series of tasks that are processed locally or remotely, falling into three categories: (i) local computing only, (ii) a fraction of the task is computed locally and the remaining task unprocessed is offloaded for remote computation, and (iii) the entire task is offloaded. Each case has attracted substantial attention in recent studies, where a delay-constrained non-linear optimization problem is often formulated. The solutions employed are either based on Lagrange duality, heuristic search, or dynamic programming. To our knowledge, there is no unifying task-processing orchestrator that is an online tailored solver for learning the model-free problems, encapsulating the three cases above. We fill this gap and present the first attempt on an innovative actor-critic reinforcement learning approach in consideration of the energy-efficiency, to compute the asymptotically optimal solutions via decomposing the comprehensive optimization into sub-problems. Rigorous theoretical analyses and experience-driven simulations demonstrate significant advantages over the benchmark approaches, in terms of task-processing delay, power efficiency, and convergence time.

关键词： Mobile edge computing  

摘要： Limited battery life and poor computational resources of mobile terminals are challenging problems for the present and future computation-intensive mobile applications. Wireless powered mobile edge computing is one of the solutions, in which wireless energy transfer technology and cloud server's capabilities are brought to the edge of cellular networks. In wireless powered mobile edge computing systems, the mobile terminals charge their batteries through radio frequency signals and offload their applications to the nearby hybrid access point in the same time slot to minimize their energy consumption and ensure uninterrupted connectivity with hybrid access point. However, the smart division of application into k subtasks as well as intelligent partitioning of time slot for harvesting energy and offloading data is a complex problem. In this paper, we propose a novel deep-learning-based offloading and time allocation policy (DOTP) for training a deep neural network that divides the computation application into optimal number of subtasks, decides for the subtasks to be offloaded or executed locally (offloading policy), and divides the time slot for data offloading and energy harvesting (time allocation policy). DOTP takes into account the current battery level, energy consumption, and time delay of mobile terminal. A comprehensive cost function is formulated, which uses all the aforementioned metrics to calculate the cost for all k number of subtasks. We propose an algorithm that selects the optimal number of subtasks, partial offloading policy, and time allocation policy to generate a huge dataset for training a deep neural network and hence avoid huge computational overhead in partial offloading. Simulation results are compared with the benchmark schemes of total offloading, local execution, and partial offloading. It is evident from the results that the proposed algorithm outperforms the other schemes in terms of battery life, time delay, and energy consumption, with

关键词： Task analysis   Reinforcement learning   Quality of service   Energy consumption   Optimization   Markov processes   Computational modeling   Energy   mobile edge computing   mobility   multi-agent reinforcement learning   task migration  

摘要： Closer to mobile users geographically, mobile edge computing (MEC) can provide some cloud-like capabilities to users more efficiently. This enables it possible for resource-limited mobile users to offload their computation-intensive and latency-sensitive tasks to MEC nodes. For its great benefits, MEC has drawn wide attention and extensive works have been done. However, few of them address task migration problem caused by distributed user mobility, which can't be ignored with quality of service (QoS) consideration. In this article, we study task migration problem and try to minimize the average completion time of tasks under migration energy budget. There are multiple independent users and the movement of each mobile user is memoryless with a sequential decision-making process, thus reinforcement learning algorithm based on Markov chain model is applied with low computation complexity. To further facilitate cooperation among users, we devise a distributed task migration algorithm based on counterfactual multi-agent (COMA) reinforcement learning approach to solve this problem. Extensive experiments are carried out to assess the performance of this distributed task migration algorithm. Compared with no migrating (NM) and single-agent actor-critic (AC) algorithms, the proposed distributed task migration algorithm can achieve up 30-50 percent reduction about average completion time.

关键词： Internet of things (IoT)   Reinforcement learning   Markov decision process   Computation offloading   Deep learning   Mobile edge computing  

摘要： Recently, in order to distribute computing, networking resources, services, near terminals, mobile fog is gradually becoming the mobile edge computing (MEC) paradigm. In a mobile fog environment, the quality of service affected by offloading speeds and the fog processing, however the traditional fog method to solve the problem of computation resources allocation is difficult because of the complex network states distribution environment (that is, F-AP states, AP states, mobile device states and code block states). In this paper, to improve the fog resource provisioning performance of mobile devices, the learning-based mobile fog scheme with deep deterministic policy gradient (DDPG) algorithm is proposed. An offloading block pulsating discrete event system is modeled as a Markov Decision Processes (MDPs), which can realize the offloading computing without knowing the transition probabilities among different network states. Furthermore, the DDPG algorithm is used to solve the issue of state spaces explosion and learn an optimal offloading policy on distributed mobile fog computing. The simulation results show that our proposed scheme achieves 20%, 37%, 46% improvement on related performance compared with the policy gradient (PG), deterministic policy gradient (DPG) and actor- critic (AC) methods. Besides, compared with the traditional fog provisioning scheme, our scheme shows better cost performance of fog resource provisioning under different locations number and different task arrival rates.

关键词： Protocols   Data models   Training   Cloud computing   Performance evaluation   Distributed databases   Computer architecture   Federated learning   mobile edge computing   distributed computing   machine learning  

摘要： Mobile Edge Computing (MEC), which incorporates the Cloud, edge nodes, and end devices, has shown great potential in bringing data processing closer to the data sources. Meanwhile, Federated learning (FL) has emerged as a promising privacy-preserving approach to facilitating AI applications. However, it remains a big challenge to optimize the efficiency and effectiveness of FL when it is integrated with the MEC architecture. Moreover, the unreliable nature (e.g., stragglers and intermittent drop-out) of end devices significantly slows down the FL process and affects the global model's quality in such circumstances. In this article, a multi-layer federated learning protocol called HybridFL is designed for the MEC architecture. HybridFL adopts two levels (the edge level and the cloud level) of model aggregation enacting different aggregation strategies. Moreover, in order to mitigate stragglers and end device drop-out, we introduce regional slack factors into the stage of client selection performed at the edge nodes using a probabilistic approach without identifying or probing the state of end devices (whose reliability is agnostic). We demonstrate the effectiveness of our method in modulating the proportion of clients selected and present the convergence analysis for our protocol. We have conducted extensive experiments with machine learning tasks in different scales of MEC system. The results show that HybridFL improves the FL training process significantly in terms of shortening the federated round length, speeding up the global model's convergence (by up to 12x) and reducing end device energy consumption (by up to 58 percent).

关键词： C-RAN   Mobile cloud computing   Mobile Edge Computing   Fine-grain offloading  

摘要： Cloud Radio Access Network (C-RAN) and Mobile Edge Computing (MEC) have recently emerged as promising leading technologies for next generation mobile networks. Due to its low access latency, MEC is not only a convenient candidate for deployment of C-RAN, but it can also be served by Mobile Users (MUs) to offload their computation-intensive applications. This convergence can facilitate the utilization of knowledge acquired through inter-BBU information sharing to improve the quality of offloading decision. In this paper, we propose an end-to-end communication and computation offloading architecture which takes the full advantage of C-RAN to solve the MEC offloading problem with regard to both partitioning as well as sending and return RRH assignment problems. Based on the proposed architecture, we model the end-to-end offloading problem as an ILP with the objective of minimizing the cost of offloading considering the intra and inter cluster handover costs besides other factors. Due to the complexity of the end-to-end offloading problem, we propose a combination of utility functions and modified min-cut algorithms to solve the aforementioned problems in a timely manner. Simulation results demonstrate that the proposed approach outperforms significantly other alternatives in terms of execution time, energy consumption and aggregated cost under scenarios with different amounts of normalized throughput, invocation data and workload.

关键词： mobile edge computing   resource allocation   computing offloading   data caching   optimization  

摘要： Mobile terminal applications with high computing complexity and high time delay sensitivity are developing quite fast today, which aggravates the load of mobile cloud computing and storage and further leads to network congestion and service quality decline. Mobile edge computing (MEC) is a way of breaking through the limits of computing and storage resources of mobile cloud and alleviating the load of mobile cloud. Computing time costs and transmission time costs are considered to be the main issues for the mobile cloud when carrying out computing offloading and data caching. Therefore, an efficient resource management strategy, which could minimize the system delay, is proposed in this paper. The new scheme offloads reasonably computing tasks and caches the tasks' data from the mobile cloud to mobile edge computing-enabled base stations. An intelligence algorithm, genetic algorithm, is being used to solve the global optimization problem which would cause transmission delay and computing resources occupation, and to determine the computing offloading and data caching probability. The simulation of the system using MATLAB is conducted in 8 different scenarios with different parameters. The results show that our new scheme improves the system computing speed and optimizes the user experience in all scenarios, compared with the scheme without data caching and the scheme without computing offloading and data caching.

关键词： Federated Learning   Distributed machine learning   Mobile edge computing   Non-IID data  

摘要： This paper presents new asynchronous methods to the Federated Learning (FL), one of the next generation paradigms for Artificial Intelligence (AI) systems. We consider the two-fold challenges lay ahead. First, non-IID (non-Independent and Identically Distributed) data across devices cause unstable performance. Second, unreliable and slow environments not only slow the convergence but also cause staleness issues. To address these challenges, this study uses a bottom-up approach for analysis and algorithm design. We first reformulate FL by unifying both synchronous and asynchronous updating schemes with an asynchrony-related parameter. We theoretically analyze this new form and find practical strategies for optimization. The key findings include: 1) a two-stage training strategy to accelerate training and reduce communication overhead;2) strategies of choosing key hyperparameters optimally for these stages to maintain efficiency and robustness. With these theoretical guarantees, we propose FedSA (Federated Staleness -Aware), a novel asynchronous federated learning algorithm. We validate FedSA on different tasks with non-IID/IID and staleness settings. Our results indicate that, given a large proportion of stale devices, the proposed algorithm presents state-of-the-art performance by outperforming existing methods on both non-IID and IID cases. (C) 2021 Elsevier B.V. All rights reserved.

关键词： Trust assessment   Community   Mobile edge computing   Vehicular networks  

摘要： In vehicular networks, trustworthy information sharing between vehicles is an important security issue. We find that existing trust systems in vehicular networks have the disadvantages of high assessment latency and high maintenance cost. In this paper, by introducing mobile edge computing (MEC), we propose a tree-searching based trust assessment method through communities, named TTAC method, for vehicular networks. The proposed TTAC method includes two parts. First, based on information interactions, TTAC gives a direct trust assessment method by utilizing Dempster-Shafer (D-S) evidence theory. Second, with the assistance of MEC base stations, TTAC designs a tree-searching based indirect trust calculation method by utilizing two neural networks through vehicles' communities. In experiments, we use a dataset of Shenzhen taxicab traffic and simulate information interactions among vehicles. The experimental results show that TTAC method can ensure fast calculation time with high assessment accuracy in a distributed manner. Especially, in terms of the accuracy of the indirect trust assessment, the mean square error (MSE) of TTAC method is lower than that of two popular trust assessment methods, compared with 3VSL by 41.4% and with MoleTrust by 71.4%.

关键词： Computational modeling   Wireless communication   Servers   Performance evaluation   Learning systems   Machine learning   Data models   Distributed machine learning   mobile edge computing   timely computing   computation-communication tradeoff   distributed optimization  

摘要： We consider distributed machine learning at the wireless edge, where a parameter server builds a global model with the help of multiple wireless edge devices that perform computations on local dataset partitions. Edge devices transmit the result of their computations (updates of current global model) to the server using a fixed rate and orthogonal multiple access over an error prone wireless channel. In case of a transmission error, the undelivered packet is retransmitted until successfully decoded at the receiver. Leveraging on the fundamental tradeoff between computation and communication in distributed systems, our aim is to derive how many edge devices are needed to minimize the average completion time while guaranteeing convergence. We provide upper and lower bounds for the average completion and we find a necessary condition for adding edge devices in two asymptotic regimes, namely the large dataset and the high accuracy regime. Conducted experiments on real datasets and numerical results confirm our analysis and substantiate our claim that the number of edge devices should be carefully selected for timely distributed edge learning.