With the popularity of 5G and 6G ^[1] and the application of programmable V2X environments and blockchain-based V2X (vehicle to everything) technologies ^[2], the IoV has embraced rapid development. In IoV, vehicles can share their perceived information with other nodes, including traffic safety information, weather information, road information, etc., and obtain services from other nodes ^[3], thus, improving traffic safety and efficiency.

However, vehicles can be unreliable, and we need to solve the problems of how to evaluate the reliability of the message sent by the vehicle and quantify an evaluation measure ^[4] (i.e., trust value) based on the historical behavior of the vehicle before utilizing IoV. For example, in IoV, vehicles may be controlled by attackers to spread false information for selfish reasons, thus, leading to false environmental perception and driving decision-making and thus, endangering the safety of drivers and causing serious traffic accidents ^[5].

In IoV, the basic principle behind the trust model is to ensure the reliable transmission of data by identifying and canceling malicious vehicles and the false news generated by them ^[6]. The trust management mechanism can help vehicles calculate the credibility of received messages ^[7] to improve the accuracy of vehicles in decision-making. In summary, the existing trust management mechanisms can be generally divided into centralized trust management and distributed trust management ^[8]. Centralized trust management has problems such as single points of failure, while distributed trust management has problems such as the delayed update of trust value.

Blockchain, as bitcoin’s core technology, is a distributed ledger ^[9]. Due to its decentralized and immutable characteristics, blockchain can record and update vehicles’ trust values. With blockchain, even if a small number of RSUs have storage errors or are controlled by attackers, the consensus results of the entire network can still be protected. Therefore, some researchers combine blockchain with trust management mechanisms to solve the above problems of centralized and distributed trust management.

However, there are still some problems in the research of trust management mechanisms based on the blockchain or single-layer blockchain. First of all, the vehicles need to store a complete blockchain ledger or send a request to the adjacent full node for verification every time the transaction is verified, which will undoubtedly increase the burden of the vehicle and waste the vehicle’s resources. Secondly, because the number of blockchain nodes is very large and the coverage is active and wide, it is also difficult to conduct hierarchical management according to objective factors such as geographical location, communication traffic, and node density. Finally, because the importance of vehicle data is not the same, the data storage and data sharing between vehicles and RSUs is inefficient if the system does not distinguish the importance of messages. Therefore, how to enable the system to store and share data of different levels of importance is a problem.

2. Trust Management Mechanism of Double-Layer Blockchaodel in IoV

2.1 Centralized Trust Management

Mahmoud et al. ^[10] adopted an incentive and punishment strategy (TRIPO) to prevent intentional packet loss attacks in rational cases and unintentional packet loss attacks in irrational cases. TRIPO uses small payments to reward rational nodes that correctly forward packets from other nodes. For irrational nodes, TRIPO uses a reputation system to measure, i.e., a new monitoring technique to monitor the nodes. However, all of these operations are centralized in the offline trusted party. Based on the malicious behavior detection system running on vehicles and RSUs, Bißmeyer et al. ^[11] proposed a centralized trust management model, which uses the malicious behavior report to establish trust relationships and reach the goal of identifying and removing attackers in IoV. Li et al. ^[12] proposed a reputable ad hoc network announcement scheme that consists of a centralized reputation server, access point (physical wireless communication equipment), and vehicle. The centralized reputation server’s role is to collect and aggregate feedback to generate reputation and spread reputation. The access point acts as the communication interface between the vehicle and the reputation server, and the vehicle broadcasts and receives information from neighboring vehicles. The credibility of the received information is evaluated and then reported to the reputation server.

2.2 Distributed Trust Management 

Consider a scenario where vehicle A needs to know about the traffic and business situation on street A. Vehicle A sends a request to the nearby RSU (we assume that each vehicle is equipped with an Onboard Unit (OBU), which uses Dedicated Short Range Communication (DSRC) or Cellular-V2X (C-V2X) communication technology for micro-wave communication with the RSU). Upon receiving the request, the RSU queries the trust value of vehicle A and allows vehicle A to use the service if its trust value is above a certain threshold. RSU queries relevant data in the RSU blockchain and returns it to vehicle A through a secure transmission channel. Vehicle A can then fully understand the situation of the front area according to the data returned by RSU and the data in its own vehicle blockchain.

Huang et al. ^[13] proposed a distributed reputation management system (DREAMS), in which basic reputation management tasks are performed by local authorities

(LA) in different locations. LA acts as the trusted authority and arranges the vehicle edge computing server for local reputation display and updates. Oluoch et al. ^[14] also proposed a reputation model to help vehicles in the network evaluate the reliability of other vehicles, that is, each receiving vehicle requests other vehicles within its communication range to give reliability to the sending vehicle, or the receiving vehicle obtains the corresponding results from the RSU. Raya et al. ^[15] proposed a data-centric trust management model, which calculates the trust of each data, aggregates multiple related but possibly contradictory data, and finally obtains the final trust value.

However, vehicles and RSUs can both become malicious nodes and act maliciously, affecting other nodes in the system. For example, malicious RSUs may tamper with vehicle trust values, and malicious vehicles may send false messages. To solve the problem of malicious nodes, we propose a trust management mechanism based on the double-layer blockchain. This mechanism is divided into three parts. Of which, the first part is the double-layer blockchain, the second part is the system architecture, and the last part is the consensus mechanism. We introduce the proposed double-layer blockchain trust management mechanism through these three parts.

2.3  Combination of Blockchain and Trust Management

In this paper, we propose a double-layer blockchain-based trust management mechanism DLBTM to solve the malicious attacks targeted on the communication among vehicles and RSUs in IoV. Using the double-layer blockchain, we can reduce the burden of vehicles in IoV, realize hierarchical management of nodes in IoV, protect vehicular privacy, implement hierarchical data storage and sharing, and realize effective trust evaluation and management of vehicle nodes. Remarkably, our message classification on type algorithm and message evaluator selection algorithm has lower time complexity compared with similar algorithms, and simulation experiments show that our trust management mechanism can effectively identify malicious nodes. Therefore, our DLBTM is effective and feasible in complex IoV environments. For future research, we will introduce an incentive mechanism into our model to promote cooperative behavior.

Yang et al. ^[16] proposed a decentralized trust management model for IoV based on blockchain technology. The receiving vehicle uses Bayesian inference to verify the results for the messages received from adjacent vehicles. Then according to this result, the receiving vehicle generates scores for each vehicle sending messages and uploads them to the nearby RSU, which is responsible for calculating the variation of trust value of each vehicle according to the scores and packaging these data into a “block”. RSUs compete to become miners using the POW Consensus algorithm. Zhang et al. ^[17] proposed a trust management system for the IoV based on blockchain, which solves the problem of calculating message credibility. Moreover, this system can detect vehicles sending malicious messages and reduce their credit value according to the rating mechanism. In addition, a combination of the consensus mechanisms of PoW and PoS is used to ensure that vehicles with significant changes in reputation can be updated to the blockchain more quickly. Kang et al. ^[18] proposed a credit-based data sharing scheme, which considers the three weights of interaction frequency, event timeliness, and trajectory similarity, adopting the three-weight subjective logic (TWSL) model to select more reliable data sources and improve data credibility. In addition, the alliance blockchain is utilized to establish a secure and distributed vehicle blockchain and smart contracts are deployed on the vehicle blockchain to realize safe and efficient data storage of RSUs and data sharing among vehicles.

Highlights of the paper:

2.4. Combination of Double-Layer Blockchain and Trust Management 

1. Proposes a double-layer blockchain structure that allows selective data storage based on message importance, reducing storage pressure.
2. Uses logistic regression algorithms to compute vehicle node trust values, accurately judging good and bad nodes.
3. Employs advanced consensus mechanisms like Ouroboros to make the system secure and reliable.
4. Tested through simulations, can effectively identify over 90% of malicious nodes under various conditions.
5. Compared to existing algorithms, this method has lower time and space complexity.

Lee et al ^[19] proposed a two-layer blockchain trust management model for the Internet of Vehicles, which is composed of the local one-day message blockchain and the global vehicle reputation blockchain. The data in the global vehicle reputation blockchain are generated by RSUs located in different regions, which consist of the vehicle’s reputation score based on the vehicle’s historical behavior. Therefore, each vehicle’s reputation is updated and permanently stored in the global vehicle reputation blockchain for further query. In the local one-day message blockchain, vehicles and RSUs store and share local traffic information in a short period of time. RSUs and vehicles in the same region act as blockchain nodes. This blockchain creates a new block at a set time every day and deletes the previously recorded blockchain data. Kandah et al ^[20] also proposed a two-layer blockchain trust management model composed of platoon blockchain and global blockchain. The participating nodes of a platoon blockchain are a group of vehicles with a small gap in proximity and speed. They store the localized trust consensus (trust value of vehicles), while the global blockchain stores the trust factors of all vehicles in the system, that is, the data in the platoon blockchain is added to the global blockchain through mining. In the mining stage, RSU mines the block using the trust bidding system.