Biblio

In federated learning, machine learning and deep learning models are trained globally on distributed devices. The state-of-the-art privacy-preserving technique in the context of federated learning is user-level differential privacy. However, such a mechanism is vulnerable to some specific model poisoning attacks such as Sybil attacks. A malicious adversary could create multiple fake clients or collude compromised devices in Sybil attacks to mount direct model updates manipulation. Recent works on novel defense against model poisoning attacks are difficult to detect Sybil attacks when differential privacy is utilized, as it masks clients' model updates with perturbation. In this work, we implement the first Sybil attacks on differential privacy based federated learning architectures and show their impacts on model convergence. We randomly compromise some clients by manipulating different noise levels reflected by the local privacy budget ε of differential privacy with Laplace mechanism on the local model updates of these Sybil clients. As a result, the global model convergence rates decrease or even leads to divergence. We apply our attacks to two recent aggregation defense mechanisms, called Krum and Trimmed Mean. Our evaluation results on the MNIST and CIFAR-10 datasets show that our attacks effectively slow down the convergence of the global models. We then propose a method to keep monitoring the average loss of all participants in each round for convergence anomaly detection and defend our Sybil attacks based on the training loss reported from randomly selected sets of clients as the judging panels. Our empirical study demonstrates that our defense effectively mitigates the impact of our Sybil attacks.

Cyber-Physical Systems (CPS) such as intelligent connected vehicles, smart farming and smart logistics are constantly generating tons of data and requiring real-time data processing capabilities. Therefore, Edge Computing which provisions computing resources close to the End Devices from the network edge is becoming the ideal platform for CPS. However, it also brings many issues and one of the most prominent challenges is how to ensure the development of trustworthy smart services given the dynamic and distributed nature of Edge Computing. To tackle this challenge, this paper proposes a novel Federated Learning based Edge Computing platform for CPS, named “FengHuoLun”. Specifically, based on FengHuoLun, we can: 1) implement smart services where machine learning models are trained in a trusted Federated Learning framework; 2) assure the trustworthiness of smart services where CPS behaviours are tested and monitored using the Federated Learning framework. As a work in progress, we have presented an overview of the FengHuoLun platform and also some preliminary studies on its key components, and finally discussed some important future research directions.