Vehicular ad hoc networks (VANETs) are the communication foundation for future intelligent transportation systems and guarantee safe driving of intelligent networked vehicles. Moreover, VANETs face a series of security challenges related to the protection of vehicle privacy, authenticity of transmitted information, and bandwidth limitations. To resolve these contradictions, many certificateless aggregate signature (CLAS) schemes have been proposed. However, the majority of them can neither resist malicious-but-passive key generation center attacks, replay attacks, and link attacks, nor track the actual identities of malicious vehicles. Meanwhile, the security of previous CLAS schemes in VANETs is only formally provided in the random oracle model (ROM), which might be insecure in actual implementation. In addition, most CLAS schemes still have problems of large verification delays and high communication overhead. To address the above-mentioned problems, a new conditional privacy-preserving CLAS scheme in VANETs is proposed, which adopts full aggregation technology to reduce computation and bandwidth resources. According to the formal security proofs under the computational Diffie-Hellman problem (CDHP) given in the standard model (SM), the security level of this scheme is higher than that of other CLAS schemes under ROM in practical applications. Additionally, the use of pseudonym mechanism realizes conditional privacy protection in VANETs. The performance analysis shows that this scheme has a higher efficiency in terms of computation and communication overhead compared with several previous CLAS schemes.

Vehicular ad-hoc networks (VANETs) can help facilitate traffic flow, reduce accidents, and enhance the driving experience. However, VANETs have some problems in terms of the authenticity and integrity of transmitted information and the preservation of vehicles’ privacy. Many certificateless aggregate signature (CLAS) schemes have been proposed to address these concerns. Nevertheless, most of these schemes suffer from security and efficiency challenges, such as the inability, to resist forgery attacks and high-computation costs. Recently, an efficient CLAS scheme with conditional privacy protection has been put forward by Chen et al. However, there is a security flaw in this scheme. In this article, we give a specific attack algorithm to indicate that Chen et al.’s proposal cannot resist a public key replacement attack initiated by external adversaries and then put forward a security-enhanced scheme. Furthermore, an efficient invalid signature identification algorithm is designed to identify invalid signatures after an aggregate verification has failed. Through rigorous security analysis, it has been verified that the scheme put forward can satisfy the fundamental security requirements of VANETs. Compared with other related schemes, our proposal improves efficiency while providing privacy and security guarantees for VANETs.

The issue of how to improve the usability of data publishing under differential privacy has become one of the top questions in the field of machine learning privacy protection, and the key to solving this problem is to allocate a reasonable privacy protection budget. To solve this problem, we design a privacy budget allocation algorithm based on out-of-bag estimation in random forest. The algorithm firstly calculates the decision tree weights and feature weights by the out-of-bag data under differential privacy protection. Secondly, statistical methods are introduced to classify features into best feature set, pruned feature set, and removable feature set. Then, pruning is performed using the pruned feature set to avoid decision trees over-fitting when constructing an $ϵ$-differential privacy random forest. Finally, the privacy budget is allocated proportionally based on the decision tree weights and feature weights in the random forest. We conducted experimental comparisons with real data sets from Adult and Mushroom to demonstrate that this algorithm not only protects data security and privacy, but also improves model classification accuracy and data availability.