Active learning is a technique for maximizing performance of machine learning with minimal labeling effort and letting the machine automatically and adaptively select the most informative data for labeling. Since the labels on records may contain sensitive information, privacy-preserving mechanisms should be integrated into active learning. We propose a privacy-preservation scheme for active learning using homomorphic encryption-based federated learning. Federated learning provides distributed computation from multiple clients, and homomorphic encryption enhances the privacy preservation of user data with a strong security level. The experimental result shows that the proposed homomorphic encryption-based federated learning scheme can preserve privacy in active learning while maintaining model accuracy. Furthermore, we also provide a Deep Leakage Gradient comparison. The proposed scheme has no gradient leakage compared to the related schemes that have more than 74% gradient leakage.

The Internet of Things faces significant security challenges, particularly in device authentication. Traditional methods of PUF-based authentication protocols do not fully address IoT’s unique security needs and resource constraints. Existing solutions like Identity-Based Encryption with Physically Unclonable Functions enhance security but still struggle with protecting data during transmission. We show a new protocol that leverages PUFs for device authentication by utilizing Paillier homomorphic encryption or the plaintext equality test to enhance security. Our approach involves encrypting both the challenge–response pairs (CRPs) using Paillier homomorphic encryption scheme or ElGamal encryption for plaintext equality testing scheme. The verifier does not need access to the plaintext CRPs to ensure that sensitive data remain encrypted at all times and our approach reduces the computational load on IoT devices. The encryption ensures that neither the challenge nor the response can be deciphered by potential adversaries who obtain them during the transmission. The homomorphic property of the Paillier scheme or plaintext equality testing scheme allows a verifier to verify device authenticity without decrypting the CRPs, preserving privacy and reducing the computational load on IoT devices. Such an approach to encrypting both elements of the CRP provides resistance against CRP disclosure, machine learning attacks, and impersonation attacks. We validate the scheme through security analysis against various attacks and evaluate its performance by analyzing the computational overhead and the communication overhead. Comparison of average computational and communication time demonstrates Paillier scheme achieves approximately 99% reduction while the plaintext equality test achieves approximately 94% reduction between them.

Passing through a rural area with a limited network infrastructure may disrupt fog computing support for vehicles. As a result, some applications on vehicles may turn off and bother the performance of the vehicular systems. In order to escape from this kind of situation, vehicular fog computing is discussed in recent times as an alternative of fog computing support while passing through a blank spot of network infrastructure. However, it is not feasible to establish a trusted vehicular fog computing service among vehicles without mutual trust. To deal with this situation, this paper proposes a method called Bidding-Price-based Transaction (BPT) for vehicular fog computing service in rural areas. This method is composed of bidding-price-based mutual trust establishment between client vehicle and server vehicle and also payoff assignment based on transaction evaluation. By applying this method, trusted fog computing service transactions between two vehicles can be achieved without the direct assistance of any trusted third party as a validating entity. The simulation results and feasibility analysis then validate the performance of the BPT scheme in rural areas. Based on feasibility analysis, we claim that the BPT scheme can be realized by adjusting vehicle speed and transmission range with respect to the size of offloaded data.

Named Data Networking (NDN), where addressable content name is used, is considered as a candidate of next-generation Internet architectures. NDN routers use In-Network cache to replicate and store passing packets to make faster content delivery. Because NDN uses a human-readable name, it is easy for an adversary to guess what kind of content is requested. To solve this issue, we develop a PEKS-based strategy for forwarding packets, where PEKS stands for public key encryption with keyword search. We implement the PEKS-based strategy based on the best route strategy and multicast strategy of NDN and show the performance of the PEKS-based NDN strategy. We also discuss the issues of the PEKS-based NDN strategy.

NDN is one of the new emerging future internet architectures which brings up new solutions over today’s internet architecture, facilitating content distribution, in-network caching, mobility support, and multicast forwarding. NDNs ubiquitous in-network caching allows consumers to access data directly from the intermediate router’s cache. However, it opens content privacy problems since data packets replicated in the router are always accessible by every consumer. Sensitive contents in the routers should be protected and accessed only by authorized consumers. Although the content protection problem can be solved by applying an encryption-based access control policy, it still needs an efficient content distribution scheme with lower computational overhead and content retrieval time. We propose an efficient and secure content distribution (ES_CD), by combining symmetric encryption and identity-based proxy re-encryption. The analysis shows that our proposed scheme achieves content retrieval time reduction up to 20% for the cached contents in our network simulation environment and a slight computational overhead of less than 19 ms at the content producer and 9 ms at the consumer for 2 KB content. ES_CD provides content confidentiality and ensures only legitimate consumers can access the contents during a predefined time without requiring a trusted third party and keeping the content producer always online.