Numerous studies have demonstrated the susce, of deep neural networks (DNNs) to subtle adversar turbations, prompting the development of many at adversarial defense methods aimed at mitigating ac ial attacks. Current defense strategies usually train for a specific adversarial attack method and can good robustness in defense against this type of ac ial attack. Nevertheless, when subjected to evan involving unfamiliar attack modalities, empirical e reveals a pronounced deterioration in the robust DNNs. Meanwhile, there is a trade-off between the classification accuracy of clean examples and adversarial examples. Most defense methods often sacrifice the accuracy of clean examples in order to improve the adversarial robustness of DNNs. To alleviate these problems and enhance the overall robust generalization of DNNs, we propose the Test-Time Pixel-Level Adversarial Purification (TPAP) method. This approach is based on the robust overfitting character-istic of DNNs to the fast gradient sign method (FGSM) on training and test datasets. It utilizes FGSM for adversarial purification, to process images for purifying unknown adversarial perturbations from pixels at testing time in a “counter changes with changelessness” manner, thereby enhancing the defense capability of DNNs against various unknown adversarial attacks. Extensive experimental results show that our method can effectively improve both overall robust generalization of DNNs, notably over pre-vious methods. Code is available https://github.com/tly18/TPAP.

Abstract Fog computing has become an emerging environment that provides data storage, computing and some other services on the edge of network. It not only can acquire data from terminal devices, but also can provide computing services to users by opening computing resources. Compared with cloud computing, fog devices can collaborate to provide users with powerful computing services through resource allocation. However, as many of fog devices are not monitored, there are some security problems. For example, since fog server processes and maintains user information, device information, task parameters and so on, fog server is easy to perform illegal resource allocation for extra benefits. In this paper, we propose a secure computing resource allocation framework for open fog computing. In our scheme, the fog server is responsible for processing computing requests and resource allocations, and the cloud audit center is responsible for auditing the behaviors of the fog servers and fog nodes. Based on the proposed security framework, our proposed scheme can resist the attack of single malicious node and the collusion attack of fog server and computing devices. Furthermore, the experiments show our proposed scheme is efficient. For example, when the number of initial idle service devices is 40, the rejection rate of allocated tasks is 10% and the total number of sub-tasks is changed from 150 to 200, the total allocation time of our scheme is only changed from 15 ms to 25 ms; additionally, when the task of 5000 order matrix multiplication is tested on 10 service devices, the total computing time of our scheme is $\sim$250 s, which is better than that of single computer (where single computer needs more than 1500 s). Therefore, our proposed scheme has obvious advantages when it faces some tasks that require more computational cost, such as complex scientific computing, distributed massive data query, distributed image processing and so on.

Fog computing needs to seamlessly integrate heterogeneous computing resources widely distributed in edge networks, and then, it can provide unified resources and services for users. However, due to a large number of fog nodes, there are still many security problems in the fog computing process. For example, when fog servers make resource allocation between users' tasks and fog nodes, some fog nodes can falsely claim that they have more resources than their actual ability, so as to get more task processing qualifications. Thus, it will greatly damage the interests of users and affect the quality of service. In this article, we propose a resource allocation scheme for community-based fog computing based on a reputation mechanism. When fog network provides computing services for users, we use a reputation mechanism to enable users to obtain reliable resources in fog computing. In our proposed scheme, a user first submits his/her task request to the community-based fog network, and then, the fog server makes a reliable resource allocation process based on multiple-layer communities and reputation calculation. Based on our experiments, our scheme enables users to obtain reliable resources and improves the service quality of computing resources in fog computing.