Search

Edge computing has received wide attention due to the benefits it brings compared to the existing cloud computing model. With the emergence of this new paradigm, content service providers (CSPs) can extend their existing cloud-based CDN (content delivery network) to edge environments to achieve cost-effectiveness. Specifically, for those regions with many access requests, edge services can be rented for saving bandwidth costs. However, the cost of renting edge services is not negligible. If edge services are rented in places with few access requests, high renting costs will cause economic losses instead. Therefore, CSPs need to make rental decisions dynamically without any knowledge of the future. For solving this problem, we summarize it as an online edge service renting problem, and propose a randomized online edge-renting algorithm for CSPs to rent edge services cost-effectively. Through theoretical analysis, we prove that the cost of our algorithm would not exceed e / (e − 1 + α) times compared to the corresponding optimal algorithm, where α is the bandwidth discount of edge service compared to the cloud. Lastly, by verifying through experiments, the results show that our online algorithm can help CSPs save 11.9% of the total cost in edge service renting. [ABSTRACT FROM AUTHOR]

The cloud–edge–client hierarchical computing architecture takes into account the dual advantages of cloud and edge computing. However, in realistic scenarios, the explosive increase of distributed data promotes the widespread application of the cloud–edge–client architecture, but also brings a variety of security threats. Relying on the advantages of federated learning technology, this has led to a growing interest in federated learning as the intersection of machine learning and cloud–edge–client fields. However, recent studies found that federated learning is potentially vulnerable to data poisoning attacks. Existing explorations of federated learning poisoning attacks and defense techniques in the cloud–edge–client environment are largely attack capabilities-specific and privacy protection-based : they are designed but fail to consider the characteristics of federated learning vulnerability and the difference in attack intensity, and lack flexibility. Yet few different attack capabilities and general defense strategies have been developed. To address these concerns, we introduce federated cloud–edge–client learning collaborative framework. Based on this architecture, a client-launched multi-label flipping attack method is discussed, which exhibits different attack capabilities. Further, we propose SPAD, a Server-based Parameter Adaptive Defense method in response to different attack intensities, to identify malicious attackers and benign participants, thereby improving the security of the federated learning image classification task. The results on image classification tasks demonstrate that in cloud–edge–client scenario, the multi-label flipping attack method presents different degrees of poisoning effect, and SPAD method has strong feasibility and effectiveness against poisoning attacks, which verify the significant advantages of the proposed method. • We explore the different attack intensities caused by MLFA methods under different settings, which are superior to non-MLFA. • Based on the FCECL collaborative framework, we explore the threats brought by the vulnerability of distributed learning. • We further propose a defense method, SPAD, which can adapt to poisoning attacks of different intensities. • Extensive experimental results demonstrate that MLFA and SPAD methods show outstanding advantages. [ABSTRACT FROM AUTHOR]

In the intelligent transportation system (ITS), the vehicular fog computing network (VFCN) can effectively alleviate the bottleneck existing in the cloud computing framework, such as high latency-sensitive applications, through edge computing offloading. It uses vehicles as the infrastructure, and fog nodes can communicate, perceive and share resources, so resource orchestration has become an essential issue of VFCN. To reduce the communication transportation cost and improve the resource utilization of VFCN, we propose a spectral graph theory-based resource orchestration algorithm by combining Virtual Network Embedding (VNE) and Deep Reinforcement Learning (DRL). Specifically, we propose a four-layer strategy network based on Graph Convolutional Networks (GCNs) for computing node embedding probability, where fog nodes fully mine spatial structure information by fusing themselves with neighborhood information to compensate for the lack of traditional heuristic VNE. Moreover, fog link embedding is performed by breadth-first search (BFS). Finally, the effectiveness of the proposed strategy is scientifically and rigorously proved in terms of long-term average revenue, long-term average revenue-cost ratio, and VNR acceptance rate through simulation cases, which can reasonably arrange the resources of VFCN. [ABSTRACT FROM AUTHOR]