Opportunistic Federated Learning (OppCL) is a widely adopted distributed learning approach for Intelligent Transportation Systems (ITS). However, the exchange of information among distributed learning parties raises privacy concerns, and OppCL training faces drawbacks such as a high number of iterations, a long training time, and low efficiency. This paper proposes an innovative framework for privacy-preserving reputation-based Opportunistic Federated Learning (PR-OppCL) to address these challenges. Firstly, we introduce a reputation value considering factors like time and model loss. This value guides our incentive mechanism, which encourages high-quality data-seeking clients to join the training process. Secondly, we implement an auction mechanism that allows clients to auction local training tasks to fog nodes, improving efficiency, and addressing performance imbalances. Thirdly, we integrate lightweight cryptographic primitive Oblivious Transfer (OT) into gradient sharing to protect vehicle privacy. Finally, the experiments demonstrate a significant reduction in training rounds and total time, outperforming existing methods.

In this paper, we present a novel blockchain-enabled approach to opportunistic federated learning (OppCL) for intelligent transportation systems (ITS). Our approach integrates blockchain with OppCL to streamline the learning of autonomous vehicle models while addressing data privacy and trust challenges. We deploy resilient countermeasures, incentivized mechanisms, and a secure gradient distribution to combat single-point failure verification attacks. Additionally, we integrate the Byzantine fault-tolerant algorithm (BFT) into the node verification component of the delegated proof of stake (DPoS) to minimize verification delays. We validate our approach through experiments on the MNIST, SVHN, and CIFAR-10 datasets, showing convergence rates and prediction accuracy comparable to traditional OppCL approaches.

As Artificial Intelligence (AI) progresses, the application of federated learning in areas such as consumer app security and intelligent transportation systems is increasing rapidly. Federated learning allows model training without necessitating the sharing of local data, yet security issues present obstacles to its advancement. This paper presents a federated learning method that merges gradient compression with homomorphic encryption. Firstly, a unique gradient compression technique is proposed to reduce data transfer by compressing the model parameters exchanged among clients. Then, homomorphic encryption is utilized to prevent breaches of gradient privacy. Experimental results demonstrate that our proposed approach has a minimal impact on the accuracy of the global model, while it reduces data transmission and improves the privacy and security of federated learning.

Public-key Encryption with Keyword Search (PEKS) enables secure keyword searches within encrypted data. At the same time, Public-key Authenticated Encryption with Keyword Search (PAEKS) enhances security by permitting authorized users to search specific keyword sets, protecting against Internal Keyword Guessing Attacks (IKGA). However, to the best of our knowledge, existing PEKS and PAEKS schemes typically require to generate a distinct set of keyword ciphertext for each data user, leading to storage, computation, and communication costs and the lack of support for multiple-keyword search. In this paper, we introduce a novel, efficient public-key searchable encryption scheme from the private set intersection (PSI) with scalable proxy servers, using a PSI protocol with multiple proxy server settings, which achieves sub-linear complexity. Our scheme is secure against IKGA and supports multiple keyword searches and sharing one encrypted keyword set by multiple users. We introduce an efficient system model with scalable proxy servers, significantly reducing computational overhead through a divide-and-conquer approach. Our proposed scheme supports multiple data users, and multiple keyword searches, utilizing a single set of keyword ciphertext for multiple data users. We formally define a security model and present a comprehensive security proof to demonstrate that our scheme maintains ciphertext-indistinguishability and trapdoor-indistinguishability.