21 November 2023 in IEEE Transactions on Industrial Informatics
Federated learning (FL) is a promising paradigm for collaboratively training networks on distributed clients while retaining data locally. Recent work has shown that personal data can be recovered even though clients only send gradients to the server. To against the gradient leakage issue, differential privacy (DP)-based solutions are proposed to protect data privacy by adding noise to the gradient before sending it to the server. However, the introduced noise affects the training efficiency of local clients, resulting in low model accuracy. Moreover, the identity privacy of clients has not been seriously considered in FL. In this article, we propose FL2DP, a privacy-preserving scheme focusing on protecting the data privacy as well as the identity privacy of clients. Different from the current schemes that add noise sampled from the Gaussian or Laplace distribution, in our scheme the noise is added to the gradient based on the exponential mechanism to achieve high training efficiency. Then, clients upload the perturbed gradients to a shuffler, which reassigns these gradients with different identities. We give a formal privacy definition called gradient indistinguishability to provide strict unlinkability for gradients shuffle. We propose a new gradient shuffling mechanism by adapting the DP-based exponential mechanism to satisfy gradient indistinguishability using the designed utility function. In this case, an attacker cannot infer the real identity of the client via the shuffled gradient. We conduct extensive experiments on two real-world datasets, and the results demonstrate the effectiveness of the proposed scheme.
26 August 2024 in IEEE Transactions on Intelligent Transportation Systems
The emergence of vehicular applications such as collision warning enhances traffic efficiency and the driving experience of users. Due to the features of decentralization, complexity, and heterogeneity in intelligent transportation systems, the key issue is ensuring the legality and trustworthiness of all participating parties. Traditional methodologies such as public key infrastructure provide services for verifying the legitimacy of entities with certificates. However, even if a vehicle is legally registered, trust in its behavior is not always assured. Recently decentralized trust management systems (DTMS) are proposed to solve the trust issue. However, achieving both dependability and efficiency remains a challenge. In this paper, we propose a novel DTMS based on the consortium blockchain. We in particular focus on the dependability of trust evaluation, where trust computation is performed on both vehicles and RSUs. To achieve a sustainable trust environment, we propose an incentive model whereby the raters can earn rewards for providing honest ratings. Instead of utilizing the consensus algorithm based on the Proof of Work (PoW) or Proof of Stake (PoS), we implement a verifiable delay function (VDF)-based consensus model within the trusted execution environment (TEE) to ensure efficiency and security for the consortium blockchain. In addition, we design a smart contract on top of the blockchain to assist the system in detecting a specific attack against the trust system, known as the on-off attack. Through extensive experiments, the results of trust assessment and blockchain performance demonstrate the dependability and efficiency of our proposed DTMS.
2 February 2023 in IEEE Transactions on Vehicular Technology
Traffic monitoring services show great potential for improving traffic efficiency in vehicular networks. With the assistance of 5G innovation, vehicles can report information such as driving misbehavior and traffic status to the traffic monitoring server which then returns feedback to the network to enhance the efficiency of traffic management. However, some challenging issues in information reporting need to be addressed first. Trivial reported information significantly increases the workload of the server. Moreover, a server is difficult to distinguish the truth of reported messages. As a result, the server is incapable of returning immediate feedback to the network, thereby reducing the quality of traffic services. Meanwhile, uploading location-based information to the server exposes vehicles' privacy. Although some existing techniques such as trust management benefit information reporting, a complete framework is still missing. In this paper, we propose an information reporting framework for solving both efficiency and privacy issues. Specifically, messages are probabilistically reported to the server depending on their importance. Furthermore, the DAG-based blockchain ensures that messages reported for a specific event are all gathered in the same ledger. We also present a scheme based on the geo-indistinguishability to protect the location privacy of vehicles. Extensive experiments demonstrate that the proposed framework preserves the location privacy of vehicles and achieves efficiency with acceptable overheads.
1 October 2018 in Future Generation Computer Systems