Vehicular social network (VSN) has been emerged in recent years with the prosperity of Internet of Vehicles (IoV). VSN service providers can exploit the social relationships of vehicles to improve the performance of vehicular networks, such as improving the order-taking efficiency of taxis. However, when social relationships are captured by attackers, other contact information is passively disclosed without their knowledge, seriously threatening user privacy. This article proposes a$(\theta ,\epsilon)$privacy technique for protecting the social relationships of vehicles that can be used to shorten the average order-taking distance of taxis during order allocation. Two variants are included in the proposed technique, which combines graph theory and graph differential privacy (GDP) mechanisms. 1)$(\theta ,\epsilon)$-GDP–a projection-based GDP algorithm is proposed to protect social relationships among the vehicles. 2) Furthermore, an improved order allocation scheme named$(\theta ,\epsilon)$-PrivOT is proposed to reduce the average order-taking distance. Real data evaluations are provided to verify the outperformance of the proposed technique. The privacy level of the proposed algorithm is improved by up to 8.77% while its data utility is improved by up to 22.94% compared with the edge removal scheme. The proposed dispatching scheme can shorten the average order-taking distance by up to 6.89% compared with the scheme of original data with 100 vehicles.

In heterogeneous Internet of Things (IoT) networks, various communication technologies lead to different transmission ranges of nodes, and they can cooperatively provide seamless communication. Unfortunately, the flexible communication mode offers more chances for malware to invade the IoT devices. To maintain its cyber security, two key issues, i.e. the critical threshold of the onset of malware propagation and the lowest-cost defense strategy after it occurs, need to be addressed. To solve the first challenge, we construct a dynamics model by using the degree-based mean-field theory and point process theory, to study the malware propagation-defense process among heterogeneous IoT devices. We analytically derive the closed-form expression of the critical malware transmission rate which can be used to predict whether the malware can propagate or not. For the latter problem, we investigate the equivalent conditions of degree-related patching strategies in defense effectiveness, which can be used to determine the lowest-cost one more directly. In addition, We compare different transmission ways of malware, and find that the malware infection is hard to avoid under the way of centralized transmission, but the decentralized transmission has a greater risk of large-scale infection. We also explore how the immunization measures and the different communication frequencies affect malware propagation. The results show that it is significant to identify important devices for immunization and infected cluster may form in some cases. Our results offer a theoretical foundation to predict and defense malware propagation in heterogeneous IoT networks.

Consensus is one of the most important issues of a blockchain system because it is a necessary process to reach an agreement between a group of separated nodes that do not trust each other in a decentralized framework. Most existing blockchain consensus works assume that the time of block propagation among separated nodes during the consensus process is ignorable, i.e., a block always successfully reaches every participating node during a period of time that is far shorter than the mining time. However, when blockchain is used in vehicularad hocnetworks (VANETs), the block propagation time is no longer negligible since the dynamic connectivity of the moving nodes in a wireless environment brings opportunistic communication to blockchain consensus. In this article, we study the impact of mobility on block propagation under the single-chain structure in VANET. Specifically, we investigate the dynamics of block propagation from the macroscopic view and derive the closed-form expression of the single-block propagation time. Then, we characterize the blockchain forking as the multiblock competitive propagation. In this way, an approximate result on multiblock propagation time is discussed. An interesting finding is that higher mobility and more moving vehicles can speed up the block propagation. In addition, we also discover that distinct propagation capabilities of moving nodes contribute to the forking reduction in the blockchain consensus.