A Blockchain-Based Digital Twin Trust Evaluation Method for Internet of Vehicles

Abstract

The present invention discloses a blockchain-based digital twin trust evaluation method for the vehicle networking. This method first calculates the node combination trust value and the authoritative combination trust value of each node in the digital twin network of the vehicle networking through a base station. Secondly, the base station combines the node combination trust value and the authoritative combination trust value according to weights to generate the comprehensive trust value of each node. Finally, the base station packages the comprehensive trust value together with the timestamp or verification information into a new block and adds it to the blockchain, completing the digital twin trust evaluation of the vehicle networking. The present invention ensures the accuracy and reliability of the trust evaluation, effectively eliminates the interference of malicious nodes, and guarantees the fairness and accuracy of the recommended trust value.

Description

Technical Field

[0001] The present invention relates to the field of vehicle networking security, and in particular, to a digital twin trust evaluation method for vehicle networking based on blockchain. Background Art

[0002] The core of digital twin technology lies in constructing a virtual space equivalent to the physical space, thereby enhancing the stability of the virtual environment. However, in the process of realizing the stable mapping between the virtual and physical spaces, many technical challenges still exist. In particular, in the digital twin model, vehicles may have malicious behaviors, and such attacks may have a significant impact on the overall reliability of the mobile network.

[0003] In vehicle ad hoc networks, the mutual trust between vehicles is the basis for the stable operation of the network. However, due to the high-speed movement of vehicles and the extremely short communication time, it is difficult for the system to evaluate the credibility of interaction information in real time and accurately. Trust evaluation, as a key technology, is widely regarded as an effective means to solve the trust and privacy problems in vehicle ad hoc digital twin networks. An efficient trust evaluation system can evaluate the historical behaviors of entities by assigning trust values, and then accurately judge their credibility.

[0004] Existing trust evaluation models are mainly divided into two categories: centralized trust models and decentralized trust models. Centralized trust models rely on a central server to manage the reputation of vehicles. However, with the continuous expansion of the coverage of intelligent transportation systems, such models have exposed the following problems: high latency, performance bottlenecks in multi-request scenarios, and single-point failure risks, etc. In contrast, decentralized trust evaluation models have gradually become a hot research direction due to their distributed architecture.

[0005] In view of this, there is an urgent need to design a digital twin trust evaluation method for vehicle networking based on blockchain technology. The method aims to enable the authoritative institutions in the vehicle ad hoc digital twin network to identify and isolate potential malicious behaviors in real time through a distributed trust evaluation mechanism, thereby maximizing the security and stability of the network.

[0006] Currently, in the prior art such as the patent with the application publication number CN113380024A, a reputation update method and a trust degree calculation method based on vehicle networking are proposed. It realizes the trust evaluation between vehicles by classifying vehicle sets and calculating the reputation value using a three-valued subjective logic algorithm. However, this method mainly focuses on the interaction and trust evaluation between vehicles and lacks consideration for the evaluation between base stations and vehicles. In addition, the prior art does not fully combine the application scenarios of digital twin technology in vehicle networking, resulting in an incomplete trust evaluation system and being difficult to cover the multi-dimensional dynamic trust requirements in complex vehicle networking environments. Summary of the Invention

[0007] Aiming at the technical bottleneck of single solution and difficulty in identifying malicious nodes in the existing technology, the present invention proposes a brand-new technical framework, aiming to efficiently solve the problem of trust evaluation in the vehicle-mounted ad hoc digital twin network. By introducing blockchain technology, this method provides a digital twin trust evaluation method for vehicle-to-everything network based on distributed trust evaluation. The present invention makes up for the above deficiencies. Based on the trust evaluation between vehicles, this method introduces a trust evaluation mechanism between the base station and the vehicle, comprehensively improving the accuracy and dynamic adaptability of trust evaluation. At the same time, blockchain technology is used to ensure the security and immutability of trust data, and technologies such as time decay and trust recommendation filtering are combined to optimize the trust evaluation process, which is applicable to the complex and dynamic environment of the vehicle-to-everything digital twin system.

[0008] The technical solution adopted by the present invention is as follows:

[0009] S1. Calculate the combined node trust value of each node in the vehicle-to-everything digital twin network through the base station.

[0010] Each node in the vehicle-to-everything digital twin network reports the events received from other nodes to the authoritative base station in the current domain. The base station calculates the direct trust value of the node through the Bayesian inference method to quantify its trust level. And a time decay function is introduced. The target node calculates the recommended trust value based on the historical evaluations of the recommended nodes interacting with it. The recommended trust filtering algorithm based on k-means is used to filter out malicious nodes, and the similarity between each trusted node and the source node is calculated among the filtered trusted nodes. The node recommended trust value is calculated according to the similarity and the latest comprehensive trust value of the recommended nodes on the blockchain. The node direct trust and the node recommended trust are combined by weights to obtain the combined node trust value.

[0011] S2. Calculate the authoritative combined trust value of each node in the vehicle-to-everything digital twin network through the base station.

[0012] S21. Determine the time t when the node joins the vehicle-to-everything digital twin network in , calculate the time interval t - t from joining the network to the current time t in . Through the exponential decay function , assign a lower initial trust value to the newly joined node and gradually increase the trust value over time. Count the event reporting frequency R of node i i , and the average reporting frequency of all nodes in the network By calculating evaluate the deviation degree of the node behavior from the overall network expectation. Finally, based on the node joining time and the reporting frequency deviation, the authoritative direct trust value of the node is comprehensively obtained

[0013] S22. Maintain the historical path base station information of each vehicle on the blockchain. The base station where the current node is located obtains the latest H base stations it has passed through from the blockchain, and requests these base stations to obtain the trust value of the target node i when it passed through this base station in the past. where k represents the base stations that the node has passed through, H represents the first H base stations that the node has passed through, and according to the time elapsed since the node left each base station Use the time decay factor γ2 to calculate the decay weight Calculate the authoritative recommended trust at time t

[0014] S23. Through the preset weight ω for the authoritative direct trust value and the authoritative recommended trust value Perform weighted summation to calculate the authoritative combined trust value of the target node

[0015] S3. The base station combines the node combined trust value and the authoritative combined trust value according to the weight to generate the comprehensive trust value of each node.

[0016] S4. The base station packs the comprehensive trust value together with the timestamp or verification information into a new block and adds it to the blockchain to complete the trust evaluation of the digital twin of the vehicle network.

[0017] Advantages of the present invention:

[0018] (1) Trust value calculation based on interaction behavior: A direct trust value calculation method based on interaction behaviors such as message authenticity, data tampering, and packet loss is proposed to ensure the accuracy and reliability of trust evaluation.

[0019] (2) Introduction of the recommended trust filtering algorithm: A recommended trust filtering mechanism based on the k-means clustering algorithm is adopted to effectively exclude the interference of malicious nodes, thereby ensuring the fairness and accuracy of the recommended trust value.

[0020] (3) Introduction of authoritative direct trust evaluation: The base station dynamically calculates the direct trust value according to the node behavior frequency and joining time, encourages legitimate nodes to actively report real events, and prevents the stealth attack of malicious nodes.

[0021] (4) Global evaluation of authoritative recommended trust: Based on the collaborative trust mechanism between multiple base stations, the global historical behavior records of the target node are evaluated to improve the ability to identify malicious nodes.

[0022] (5) Comprehensiveness of the combined trust model: Through the dual combination of direct trust and recommended trust, the node and authoritative combined trust values are respectively generated, and a comprehensive trust evaluation system is constructed from the micro and macro levels to make up for the limitations of a single trust model and enhance the robustness and adaptability of the overall system.

[0023] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. Description of the Drawings

[0024] Figure 1 is a flowchart of an embodiment of the present invention;

[0025] Figure 2 is a schematic diagram of an embodiment of the present invention. Detailed Embodiment

[0026] To facilitate the understanding of the present invention, the present invention will be described more comprehensively below with reference to the relevant accompanying drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in different forms and is not limited to the embodiments described in the text. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0027] Embodiment: A blockchain-based digital twin trust evaluation method for vehicle-to-everything (V2X), as Figure 1 and Figure 2 shown, includes the following steps:

[0028] S1. Calculate the node combination trust value of each node in the digital twin network of vehicle-to-everything through a base station;

[0029] Specifically, the process of calculating the node combination trust value of each node in the digital twin network of vehicle-to-everything through a base station in step S1 includes:

[0030] S11. Each node in the digital twin network of vehicle-to-everything reports the events received from other nodes to the authoritative base station in the current domain. The base station calculates the direct trust value of the node through the Bayesian inference method to quantify its trust level. Use α and β to represent the results of direct interaction between the source node i and the target node j at time t. Considering the existence of the cold start problem that two nodes have never interacted, the initial trust value is set to 0.5. The direct trust value NDT of node i for node j based on the Beta distribution ij is calculated as:

[0031]

[0032] where α ij and β ijDenote the cumulative positive and negative interactions of node i with respect to node j. Use message authenticity verification to determine whether the interaction is positive or negative: When node j observes the occurrence of event x somewhere, it broadcasts the message to nearby nodes including node i. When node i receives the message about event x from node j, it reports the message passed by j to the base station. The base station aggregates the messages about event x from all nodes at that moment based on Bayesian inference, judges the authenticity of each event, and thus determines whether the interaction between node i and node j is positive or negative.

[0033] Introduce a time decay function and maintain an interaction history record table to store the results of the past M interactions, and calculate α at time t ij and β ij values, so as to reduce the influence of past interaction results and enhance the influence of existing interaction results:

[0034]

[0035] where and represent the cumulative positive and negative interactions of node i with respect to node j at time t, and λ1 and λ2 are decay factors used to control the decay rate of e.

[0036] S12. The target node calculates the recommended trust value based on the historical evaluations of the recommended nodes with which it interacts. Use the k-means-based recommended trust filtering algorithm to filter out malicious nodes, and calculate the similarity between each trusted node and the source node among the filtered trusted nodes. The value of the similarity indicates how similar the evaluations of node i and the recommended node are for the common target node:

[0037]

[0038] where Sim(i, r′ F ) represents the similarity between i and r′ F , Set(i, r′ F ) represents the set of common target nodes of i and r′ F , and represent the direct trust values of node i and node r′ F for node x respectively, and Num(Set(i, r′ F )) represents the number of nodes in Set(i, r′ F ).

[0039] Then, calculate the node recommended trust value based on the calculated similarity and the latest comprehensive trust value of the recommended nodes on the blockchain:

[0040]

[0041] Among them represents the node recommendation trust value of node i for node j at time t. r′ F is the recommended node in the trust set R tc . is the trust value of this recommended node recorded on the blockchain. Sim(i, r′ F ) is the similarity between node i and node r′ F . is the direct node trust value of r′ F and node j.

[0042] S13. Combine the direct node trust and the recommended node trust according to the weight ω n to form the combined node trust of the target node, improve the accuracy of trust value evaluation, effectively eliminate the limitations of a single trust model, and enhance the comprehensiveness and reliability of the evaluation. The value of ω n is dynamically adjusted according to the actual road environment:

[0043]

[0044] Among them is the combined node trust degree of node i for node j at time t.

[0045] S2. Calculate the authoritative combined trust value of each node in the digital twin network of the vehicle network through the base station;

[0046] Specifically, in step S2, the process of calculating the authoritative combined trust value of each node in the digital twin network of the vehicle network through the base station includes:

[0047] S21. Determine the time t in when the node joins the digital twin network of the vehicle network, and calculate the time interval t - t in from joining the network to the current time t. Assign a lower initial trust value to the newly joined node through the exponential decay function , and gradually increase the trust value over time. Count the event reporting frequency R i of node i, and the average reporting frequency of all nodes in the network Calculate to evaluate the deviation degree of the node behavior from the overall network expectation. Finally, based on the node joining time and reporting frequency deviation, comprehensively obtain the authoritative direct trust value

[0048]

[0049] Among them is the authoritative direct trust, R iis the reporting frequency of node i, is the average node reporting frequency, ω ad is the weight value dynamically adjusted according to the road environment, and γ1 is the attenuation factor.

[0050] S22. Maintain the historical passing base station information of each vehicle on the blockchain. The base station where the current node is located obtains the latest H base stations it has passed through from the blockchain, and requests the trust value of the target node i when it passed through this base station from these base stations where k represents the base station that the node has passed through, and H represents the first H base stations that the node has passed through. According to the time elapsed since the node left each base station Use the time decay factor γ2 to calculate the decay weight To reduce the impact of earlier trust records and thus emphasize more recent trust evaluations, the authoritative recommended trust at time t is calculated as:

[0051]

[0052] where is the authoritative recommended trust, is the trust value of the base station that the node has passed through in history for this node, is the time elapsed since node i left this base station.

[0053] S23. Through the preset weight ω, the authoritative direct trust value of the target node at the current time t calculated in step S21 and the authoritative recommended trust value of the target node at the current time t calculated in step S22 are weighted and summed to calculate the authoritative combined trust value of the target node ω a The value of is dynamically adjusted according to the actual road environment:

[0054]

[0055] S3. Integrate the node combined trust value and the authoritative combined trust value, comprehensively fuse the behavior performance of the node itself and the evaluation data of the base station, and generate the final comprehensive trust value of the node;

[0056] Through the preset weight ω, the node combined trust of the target node at the current time t calculated in step S1 and the authoritative combined trust value of the target node at the current time t calculated in step S2 are weighted and summed to calculate the comprehensive trust value of the target node

[0057]

[0058] where ω c is the weight.

[0059] S4. The base station packages the comprehensive trust value together with the timestamp into a new block and adds it to the blockchain;

[0060] Specifically, the process in which the base station packages the comprehensive trust value together with the timestamp into a new block and adds it to the blockchain in step S4 includes:

[0061] S41. The base station aggregates the comprehensive trust values of each node and packages them with the corresponding timestamps to form a new block, and adds the new block to the blockchain through a distributed consensus algorithm to ensure the immutability and security of the data:

[0062]

[0063] where is the latest trust value of node j, is the trust value stored last time for this node on the blockchain, is the comprehensive trust value of node j, and p is the number of nodes that make a trust evaluation of j at time l + 1.

[0064] S42. The base station checks the latest trust value TV of each node in the new block l+1 If the trust value of a node is lower than the preset trust value threshold, the node is marked as a malicious node to enhance the network security and the effectiveness of trust evaluation.

[0065] The present invention has been described exemplarily in combination with the accompanying drawings. Obviously, the specific implementation of the present invention is not limited by the above methods. As long as such non-substantive improvements are made by adopting the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

[0066] The present invention constructs a digital twin network simulation environment of an Internet of Vehicles containing 200 vehicle nodes and 20 base stations by means of the urban traffic simulator (SUMO) and the network simulator (NS2). In this environment, the moving speed of the vehicle nodes is set in the range of 10 m / s - 20 m / s, the coverage range of the base stations is 800 m × 800 m, and the communication range of the vehicle nodes is 150 m. At the same time, a blockchain simulation platform is built by using Hyperledger Fabric to store and manage trust evaluation data.

[0067] The method of the present invention is compared with three baseline methods, namely BTM, RBTM, and DTM. Among them, BTM calculates the trust value of surrounding vehicle nodes based on the rationality of the query space, the rationality of the query frequency, and the location; RBTM calculates the node trust value using a recommendation-based method, which is achieved by filtering malicious recommendations and combining the confidence value, deviation value, and closeness centrality value; DTM has a classical weighted voting method and is commonly used in trust management.

[0068] 40 rounds of experiments are conducted for each scenario and the average value is taken. When the number of vehicle nodes is 200, in the attack mode where malicious nodes share false trust ratings with the source node, the present invention is higher than the baseline methods in terms of both recall rate and accuracy. Even when the malicious nodes reach 50%, the performance of the present invention can still achieve a precision rate and recall rate of more than 80%, and in the case of a large number of malicious nodes, it can still provide accurate trust value judgments for normal nodes. In the intermittent malicious attack mode, since it is more difficult to identify malicious nodes, as the number of malicious nodes increases, the present invention can still reach an accuracy rate and recall rate of more than 80%, and this performance is also higher than the recall rate and accuracy rate of the baseline methods. In summary, it is fully proved that the method of the present invention is effective.

Claims

1. A blockchain-based digital twin trust evaluation method for the Internet of Vehicles, characterized in that, The steps are as follows: S1. Calculate the node combined trust value of each node in the digital twin network of the vehicle networking through the base station; S2. Calculate the authoritative combined trust value of each node in the digital twin network of the vehicle networking through the base station. The specific implementation process is as follows: S21. Determine the time t when the node joins the digital twin network of the vehicle network, and calculate the time interval t - t from joining the network to the current time t in , and calculate the time interval t - t from joining the network to the current time t in ; Through the exponential decay function assign an initial trust value to the newly joined node, and gradually increase the trust value over time. Statistically record the event reporting frequency R of node i i , as well as the average reporting frequency of all nodes in the network By calculating evaluate the deviation degree of the node behavior from the overall network expectation. Finally, based on the node joining time and the reporting frequency deviation, comprehensively obtain the authoritative direct trust value of the node Among them is the direct trust of authority, R i is the reporting frequency of node i, is the average reporting frequency of nodes, ω ad is the weight value dynamically adjusted according to the road environment, and γ1 is the attenuation factor; S22. Maintain the historical path base station information of each vehicle on the blockchain. The base station where the current node is located obtains the latest H base stations that the vehicle has passed through from the blockchain, and requests these base stations to obtain the trust value of the target node i when the target node passed through the base station in the past. where k represents the base station that the node has passed through, according to the time that has passed since the node left each base station Use the time decay factor γ2 to calculate the decay weight The authoritative recommended trust at time t is calculated as: wherein is the trust value of the base station that the node has passed through in its history for this node, is the time elapsed since node i left this base station; S23. Perform a weighted sum on the authoritative direct trust value of the target node at the current moment t and the authoritative recommendation trust value of the target node at the current moment t to calculate the authoritative combined trust value of the target node ω a The value of is dynamically adjusted according to the actual road environment: S3. The base station combines the node combined trust value and the authoritative combined trust value according to the weight to generate the comprehensive trust value of each node; S4. The base station packages the comprehensive trust value together with the timestamp or verification information into a new block and adds it to the blockchain to complete the trust evaluation of the digital twin of the vehicle networking.

2. The digital twin trust evaluation method for the vehicle networking based on blockchain according to claim 1, wherein The specific implementation process of step S1 is as follows: S11. Each node in the digital twin network of the vehicle Internet of Things reports the events received from other nodes to the base station within the current domain. The base station calculates the direct trust value of the node through the Bayesian inference method. The direct trust value NDT of node i for node j based on the Beta distribution is ij calculated as: where α ij and β ij represent the cumulative positive and negative interactions of node i with respect to node j; Introduce a time decay function, maintain an interaction history record table to store the results of the past M interactions, and calculate α at time t ij and β ij values: Among them, and represent the cumulative positive and negative interactions experienced by node i towards node j at time t, and λ1 and λ2 are decay factors that control the decay rate of e; S12. Use the recommendation trust filtering algorithm based on k-means to filter out malicious nodes, and calculate the similarity between each trusted node and the source node among the filtered trusted nodes: Among them, Sim(i, r′ F ) represents the similarity between i and r′ F , Set(i, r′ F ) represents the set of common target nodes of i and r′ F , and respectively represent the direct trust values of node i and node r′ F for node x, Num(Set(i, r′ F )) represents the number of nodes in Set(i, r′ F ); Then calculate the node recommendation trust value according to the calculated similarity and the latest comprehensive trust value of the recommended nodes on the blockchain: Among them represents the node recommendation trust value of node i for node j at time t; r′ F is the recommended node in the trust set R tc , is the trust value of this recommended node recorded on the blockchain; Sim(i, r′ F ) is the similarity between node i and node r′ F , is the direct trust value of r′ F to node j; S13. According to the weight ω n Combine the direct trust of the node and the recommended trust of the node to form a combined node trust value for the target node: Among them, is the combined node trust degree of node i for node j at time t.

3. The method for evaluating the digital twin trust of the vehicle networking based on blockchain according to claim 2, wherein, The specific implementation process of step S3 is as follows: The combined trust of the target node at the current time t and the authoritative combined trust value of the target node at the current time t are weighted and summed to calculate the comprehensive trust value of the target node Among them, ω c is the weight.

4. The method for evaluating the digital twin trust of the vehicle networking based on the blockchain according to claim 3, characterized in that The specific implementation process of step S4 is as follows: S41. The base station aggregates the comprehensive trust values of each node and packages them with the corresponding timestamps to form a new block, and adds the new block to the blockchain through a distributed consensus algorithm: Among them is the latest trust value of node j, is the previous trust value of node j on the blockchain, is the comprehensive trust value of node j, and p is the number of nodes that make a trust assessment of j at time l + 1; S42. The base station checks the latest trust value TV of each node in the new block. If the trust value of a node is lower than the preset trust value threshold, the node is marked as a malicious node. l+1 ​

Images