A federated learning attack method based on edge collaboration via cloud

Abstract

Federated learning (FL) is widely used in edge‐cloud collaborative training due to its distributed architecture and privacy‐preserving properties without sharing local data. FLTrust, the most state‐of‐the‐art FL defense method, is a federated learning defense system with trust guidance. However, we found that FLTrust is not very robust. Therefore, in the edge collaboration scenario, we mainly study the poisoning attack on the FLTrust defense system. Due to the aggregation rule, FLTrust, with trust guidance, the model updates of participants with a significant deviation from the root gradient direction will be eliminated, which makes the poisoning effect on the global model not obvious. To solve this problem, under the premise of not being deleted by the FLTrust aggregation rules, we construct malicious model updates that deviate from the trust gradient to the greatest extent to achieve model poisoning attacks. First, we utilize the rotation of high‐dimensional vectors around axes to construct malicious vectors with fixed orientations. Second, the malicious vector is constructed by the gradient inversion method to achieve an efficient and fast attack. Finally, a method of optimizing random noise is used to construct a malicious vector with a fixed direction. Experimental results show that our attack method reduces the model accuracy by 20%, severely undermining the usability of the model. Attacks are also successful hundreds of times faster than the FLTrust adaptive attack method.

Edge AI: A survey

“…AI at the edge is the use of AI in real-world devices. Edge AI refers to the practice of doing AI computations near the users at the network's edge instead of centralised location like a cloud service provider's data centre or a company's own private data warehouse [24]. The Internet's worldwide reach means that any region might be considered its periphery [25].…”

“…Management and Monitoring: Standardizing the management and monitoring tools can help simplify deployment and reduce the risk of human error. This includes standardizing remote management protocols, monitoring and reporting tools, and alerting mechanisms [24]. [83].…”

“…In the broadest view, there is the requirement to develop experimental test beds for a wide variety of purposes, including deployment scenarios and economic modelling [57,205]. Similarly, generic processing functionalities would aid developers to mix and match when devising edge solutions [24]. New offerings such as SDN and NFV could be used to achieve dynamic configurations for edge processing [26].…”

Special issue on collaborative edge computing for secure and scalable Internet of Things

“…The third paper titled “A federated learning attack method based on edge collaboration via cloud” by Yang et al 3 focuses on poisoning attacks, where attackers try to manipulate the training process to produce a faulty model. FLTrust's defense mechanisms can eliminate updates that deviate significantly from the expected direction.…”

Special issue on efficient management of microservice‐based systems and applications

“…In the sixth paper, Zhang et al 6 proposed a chunk reuse mechanism aimed at efficiently identifying node‐local duplicate data during container updates. This mechanism contributes to a reduction in the volume of data transmission needed for image building.…”