Internet of vehicles trust management method of pre-reward and punishment mechanism based on block chain

By constructing a blockchain-based vehicle-to-everything (V2X) trust management system, combined with a pre-reward and punishment mechanism and CNNs models, the problem of insufficient trust management in V2X systems has been solved, achieving accurate prediction of vehicle behavior and secure and reliable information exchange, and improving the stability and fault tolerance of the system.

CN118890625AInactive Publication Date: 2024-11-01NANTONG UNIV
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202410907397.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-08
Publication Date
2024-11-01
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing vehicle-to-everything (V2X) trust management methods are inadequate in terms of accurate prediction of vehicle behavior, design of reward and punishment mechanisms, and efficiency and security of consensus algorithms, leading to system instability and security issues.

Method used

A blockchain-based vehicle network trust management system is constructed, which combines a pre-reward and punishment mechanism, and achieves secure and reliable information exchange between vehicles by building a vehicle trust evaluation system, deploying CNNs models to predict message authenticity, and designing a trust consensus mechanism.

Benefits of technology

It improves the accuracy of predicting vehicle behavior, identifies and punishes malicious nodes, enhances system security and stability, ensures the reliability and sharing of information transmission, and improves system fault tolerance and performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118890625A_ABST
    Figure CN118890625A_ABST
Patent Text Reader

Abstract

The invention provides an Internet of Vehicles trust management method based on a block chain pre-reward and punishment mechanism, and the method comprises the steps: firstly constructing a comprehensive vehicle trust evaluation system, and considering a plurality of factors, such as initialization trust, local trust and global trust; and secondly, introducing a convolutional neural network based message accuracy prediction method, and performing preventive reward and punishment on the message of the vehicle node by comprehensively considering the number of historical generation blocks of the vehicle, the number of successful verification transactions, the trust value and other factors and the prediction result of the message authenticity. And finally, providing a consensus mechanism based on pre-reward and punishment and trust to ensure that the verified reliable traffic information and vehicle information are packaged into a new block and are safely stored in a block chain network. Through the optimization of the consensus algorithm, the throughput and safety of the system are improved, and a solid guarantee is provided for the reliability and safety of the Internet of Vehicles system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of blockchain and Internet of Vehicles, and specifically is a trust management method of a pre-reward and punishment mechanism in an Internet of Vehicles system based on blockchain. Background Art

[0002] With the rapid development of Internet of Vehicles (IoV) technology, information exchange between vehicles has become an important part of intelligent transportation systems. However, the IoV environment faces many challenges, including data security, information credibility, and trust management. Traditional IoV systems often rely on centralized trust agencies to manage communication and data exchange between vehicles, but this approach is vulnerable to single point failures and malicious attacks, leading to system instability and security issues. In order to solve these problems, researchers have begun to explore IoV trust management methods based on blockchain technology. Blockchain, as a decentralized distributed ledger technology, can ensure data security and credibility, making communication between vehicles more secure and reliable. At the same time, the pre-reward mechanism, as an incentive measure, can effectively improve the behavioral compliance and integrity of vehicle nodes and promote the benign operation of the system.

[0003] However, there are still some problems in the existing IoV trust management methods, such as the accurate prediction of vehicle behavior, the reasonable design of reward and punishment mechanisms, and the efficiency and security of consensus algorithms, which still need to be further improved. Therefore, conducting research on IoV trust management based on blockchain has important theoretical and practical significance, and can provide more reliable and secure technical support for the development and application of intelligent transportation systems. Summary of the invention

[0004] Purpose of the invention: The present invention aims to build an innovative trust management model for Internet of Vehicles, which combines blockchain technology with a pre-reward mechanism to solve the problems of data security, information credibility, and trust management in traditional Internet of Vehicles systems. By establishing a safe, reliable, intelligent, and efficient trust management system, it promotes information exchange and data sharing between vehicles, promotes the development and application of smart transportation systems, and realizes safer and more reliable smart transportation.

[0005] Technical solution: A trust management method for a pre-reward and punishment mechanism in a blockchain-based Internet of Vehicles system: including the following steps:

[0006] 1) Build a blockchain-based Internet of Vehicles trust management model;

[0007] 2) A vehicle trust evaluation system that comprehensively considers initialization trust, local trust and global trust is constructed;

[0008] 3) Construct a pre-reward and punishment mechanism based on the CNNs model to predict the authenticity of messages;

[0009] 4) Design a consensus mechanism based on pre-reward and punishment and trust.

[0010] Furthermore, in step 1), a blockchain-based Internet of Vehicles trust management model is constructed. The above model mainly includes a vehicle, a roadside unit node (RSU) and a trusted certification authority (TA).

[0011] (1) Vehicle: In the trust management system, vehicles not only provide data such as road event information, traffic conditions, driving safety, and terrain services to other vehicle nodes, but also establish trust relationships based on the initial trust and local trust of other vehicle nodes, and upload these data to the roadside unit. The vehicle as the leader node is responsible for generating new blocks and uploading them to the roadside unit, while the consensus participating vehicle nodes work together with other vehicles to verify and confirm the validity of traffic information. In this process, vehicle nodes jointly maintain the safe and stable operation of the entire Internet of Vehicles system through mutual cooperation and trust establishment.

[0012] (2) Roadside Unit Node (RSU): RSU plays a vital role in the Internet of Vehicles trust management system. First, it calculates the global trust based on the initial trust and local trust data uploaded by the vehicle node, and broadcasts the global trust to the vehicles in the area, thereby establishing a trust relationship network between vehicles. Secondly, RSU deploys the CNNs model-based prediction vehicle information accuracy and pre-punishment mechanism to improve the credibility and accuracy of vehicle information transmission, and further ensure the security and stability of the system. Finally, as a data transfer and storage node, RSU is responsible for receiving block data from vehicle nodes, transferring and storing it, and providing support for subsequent data processing and analysis.

[0013] (3) Trusted Certification Authority (TA): The Trusted Certification Authority (TA) has the important responsibility of authenticating and establishing trust for vehicle nodes. By verifying the identity information of vehicle nodes and issuing digital certificates and other credentials, TA ensures that only legitimate vehicle nodes can participate in the Internet of Vehicles system and establishes a trust relationship between vehicle nodes. During the vehicle node registration phase, TA generates initial trust and reviews the vehicle registration information after a certain period or after a violation occurs. After the review is completed, TA updates the initial trust of the vehicle and sends the updated information to the roadside unit through blockchain broadcast to ensure the legitimacy and trustworthiness of the vehicle node. The role of TA is not only to confirm the authenticity of the vehicle identity, but also to maintain the trust network between vehicle nodes, providing reliable protection for the security and stability of the system.

[0014] The Internet of Vehicles Trust Management System uses a variety of technologies and algorithms, including trusted authentication authorities (TA), roadside units (RSU), prediction models, consensus algorithms, and blockchain technology. TA is responsible for authenticating and establishing trust with vehicle nodes, while RSU is responsible for receiving and verifying traffic information uploaded by vehicle nodes, and executing prediction models and pre-reward and punishment mechanisms. The system uses blockchain technology to achieve secure storage and non-tamperability of traffic information. By combining these technologies and algorithms, the Internet of Vehicles Trust Management System provides important support for the safety, reliability, and trustworthiness of intelligent transportation systems.

[0015] Furthermore, in step 2), a vehicle trust evaluation system is constructed that comprehensively considers initialization trust, local trust and global trust.

[0016] Definition of vehicle trust: In this invention, a region is used to illustrate the definition of trust, assuming that there are multiple vehicle nodes in this specified region. In such a region, roadside units can coexist with multiple vehicles to form an integrated system for trust management and calculation.

[0017] Step 2-1: Initialize Trust: The vehicle type, brand, model, insurance, driving age, and violation record that the vehicle needs to submit when registering. This information constitutes the Trust Attribute Matrix (TAM).

[0018] TAM=ξta1,ta2,…,ta l ,…,ta m} (1)

[0019] use Represents the trust attribute ta g The relevant metric function of the trust measurement. m represents the number of trust attributes.

[0020] Therefore, the initial trust value of the vehicle can be expressed as:

[0021]

[0022] It is worth noting that the vehicle management department will review the vehicle registration information in a certain period or after a violation occurs, and assign a value to the vehicle's initial trust after the review is completed. Therefore, the vehicle's initial trust is not fixed. ω represents the weight of each attribute.

[0023] Step 2-2: Local Trust: At regular intervals, the vehicle sends the trust feedback generated in the area to the roadside unit, which forms a trust feedback list for the local trust of all other devices in an area. Local trust is defined by path transmission loss and the accuracy of the provided message.

[0024] Assume that the local trust of vehicle node i to vehicle node i in the region is defined as LT i j.

[0025] Path loss: LP ij =L0+10nlog 10 (d ij ) (3)

[0026] Where L0 represents the path loss when the reference distance is 1 meter, n represents the path loss exponent, d ij Represents the distance between vehicle node i and vehicle node,.

[0027] Message accuracy:

[0028] Where α and β represent the number of correct and incorrect messages received by vehicle node i from vehicle node j, respectively.

[0029]

[0030] Where ω1 and (1-ω1) represent the weights of path loss and message transmission accuracy, and γ and ε are additional parameters used to adjust the impact of path loss and message transmission accuracy.

[0031] Since local trust is time-sensitive, the set of local trust of vehicle node i to vehicle node j with respect to time t is defined as If no information interaction occurs within time t, the initial trust is used for assignment.

[0032] Updates to local trust:

[0033] in Represents the average value of the local trust in the window time. ΔT represents the update degree used to adjust the trust value, λ ΔT The selected value needs to consider the characteristics of the specific application scenario and the sensitivity to changes in trust values. ΔT represents the difference between the current local trust and the average value of the historical local trust, that is:

[0034]

[0035] The update method can be expressed as follows:

[0036]

[0037] represents the local trust value of the τth history within the window time, and σ is a decay parameter used to control the weight of the historical trust value.

[0038] Step 2-3: When vehicle node i enters a roadside unit area, it will upload its local trust list to the roadside unit in the area. The roadside unit will then calculate the global trust. In order to prevent collusion between malicious nodes, each node has a certain node trust similarity.

[0039] In the present invention, Euclidean distance is used to calculate the similarity of vehicle node trust:

[0040]

[0041] Global trust can be expressed as:

[0042]

[0043] It represents the set of nodes in a certain area that conduct local trust evaluation with vehicle nodes i and j within time t.

[0044] Furthermore, in step 3), a pre-reward and punishment mechanism based on the CNNs model predicts the authenticity of the message, thereby improving the accuracy and credibility of the vehicle node messages in the Internet of Vehicles system.

[0045] Step 3-1: Use the number of blocks generated by the vehicle node and the number of transactions successfully verified to measure the node's contribution.

[0046] The roadside unit records the number of blocks generated by the vehicle node and the number of successfully verified transactions. After reaching a consensus in the area, it is uploaded to the roadside unit for storage and broadcast by the roadside unit in the blockchain.

[0047] The contribution of the node C i =ρ·M i +(1-ρ)·N i (11)

[0048] Among them, M i Represents the number of blocks generated by vehicle node i, N i It represents the number of transactions successfully verified by vehicle node i, and ρ is a parameter to be determined, which is used to adjust the impact of the number of blocks and transactions on the node contribution.

[0049] Step 3-2: Use the CNNs model to predict the authenticity of the message provided by the vehicle.

[0050] In the vehicle trust evaluation system and the contribution of the node, it can be seen that the above evaluation feature parameters are all based on time series. If you want to predict the authenticity of the message provided by the vehicle, and then give pre-rewards and punishments, thereby reducing the vehicle's intention to participate in malicious behavior and improving the security of the Internet of Vehicles. Therefore, in this invention, we use the authenticity of message events based on CNNs for prediction.

[0051] Assume there is a set of message event data sets {(x (1) ,y (1) ), (x (2) ,y (2) ),…,(x (m) ,y (m) )}, where x (i) is the characteristic of the i-th message event, including the initial trust, local trust and global trust of the message sender, and whether the message signature is successfully verified, y (i) is the true label of the i-th message event (0 for false and 1 for true).

[0052] First, this method defines a message authenticity prediction model based on CNNs, which can transform the message feature x (i) Mapping to ground truth predictions The parameters of the model are represented by θ. The input layer of the CNNSS model is defined as the feature x that receives the message event. (i) , CNNs are used to learn sequence features, and the output layer is used to predict the probability of the authenticity of the message event.

[0053] For a given message event feature x (i) , perform forward propagation calculations to obtain the model's predicted output

[0054] c κ =F(W κ *x (i) +b κ ) (12)

[0055] p κ =mean(c κ ) (13)

[0056]

[0057] Where W κ and b κ represents the weight and bias of the kth convolution kernel, F represents the RELU activation function, and W y and b y are the weights and biases of the fully connected layer, softmax is the activation function, and k is the number of categories. is the probability distribution of the prediction model.

[0058] Secondly, the cross entropy loss function is defined in the method to measure the model prediction output and the true label y (i) The difference between.

[0059]

[0060] Finally, using the stochastic gradient descent optimization algorithm, a sample (x (i) ,y (i) ), calculate the objective function At the current parameter θ (t) The gradient of the selected sample Using the gradient descent update rule, update the parameter vector:

[0061]

[0062] Where μ is the learning rate, which controls the compensation of parameter update in each iteration. Repeat the iteration until the gradient size is less than 10 -4 , the model is considered to have converged and training is stopped.

[0063] Step 3-3: Pre-reward and punishment mechanism

[0064] In this method, the designed reward and punishment function aims to take into account the contribution and trust of the vehicle node and is adjusted according to the authenticity of the predicted message event.

[0065]

[0066] in, and δ are weight coefficients used to adjust the importance of different factors.

[0067] C i is the contribution of the vehicle; T i is the global trust of the vehicle; Confidence(x) is the confidence of the model in predicting the authenticity of the message; is the authenticity of the message event predicted by the model; theshold is the threshold of the authenticity of the message event; Sig(x) is the sigmoid function, which maps x to the range of (0,1).

[0068]

[0069] Confidence is the negative value of the entropy of the predicted probability distribution of the model's prediction of the authenticity of the message. When the entropy value is lower, it means that the model has a higher confidence in the prediction result; when the entropy value is higher, it means that the model has a lower confidence in the prediction result.

[0070] The design of this method allows for more fine-tuned adjustments of rewards and penalties to more effectively incentivize good performance in vehicle message transmission behavior and punish bad behavior.

[0071] Furthermore, in step 4), a pre-reward and punishment and trust consensus mechanism is designed to improve the fault tolerance and performance of the Internet of Vehicles system to ensure that the system can still reach a consensus in the presence of a maximum of f malicious nodes.

[0072] Step 4-1: Trust identification and role division:

[0073] Based on the global trust and pre-reward and punishment of vehicle nodes, consensus participants and candidate participants are identified.

[0074] The number of consensus participants is more than 1 / 3 of the total number of nodes and greater than 2f+1, where f is the total number of faulty nodes and malicious nodes, and the remaining nodes are candidate participants. Malicious nodes do not participate in consensus. Specifically: global trust + reward / - penalty, sort from large to small, select more than 1 / 3 of the sorting and greater than 2f+1 vehicle nodes as consensus participants.

[0075] Step 4-2: Leader Selection:

[0076] The vehicle node with the highest trust and pre-reward is selected as the leader from the consensus participants, and the node with the second highest trust is selected as the candidate leader. If the leader is attacked, the vehicle node of the candidate leader becomes the leader.

[0077] The leader is responsible for broadcasting traffic information within the consensus time, and authenticates the information with its own signature and sends it to other consensus participant nodes in all regions. If the message verification is successful, the consensus participant vehicle node enters the preparation phase; otherwise, the information will be discarded.

[0078] Step 4-3: Message Verification and Response

[0079] After receiving the leader's broadcast information, the consensus participant node performs public key verification to ensure the integrity of the information and the credibility of the source of the message. If the verification fails, the node will discard the information to prevent the information from being tampered with or forged. The verified node continues to verify the information and broadcasts the verification results to other participants. If a node receives verification from at least one-third of the participants, it broadcasts a commit message. After receiving enough commit messages, a reply message will be sent to the leader to confirm the transaction. After receiving the commit messages from at least one-third of the participants, the leader will package the transaction into a block. If not enough result messages are received, the leadership will be transferred to the second most trusted node. This process is repeated until a new block is created or the leader is transferred three times. If consensus is not reached after three leader changes, the current round of consensus is abandoned and the next round or rollback is started.

[0080] Step 4-4: Trust and Reward and Punishment Update

[0081] After consensus is completed and blocks are generated, the leader node updates its trust evaluation. For consensus participating nodes, their trust values ​​are dynamically adjusted based on their performance and the results of processing traffic information. At the same time, vehicle nodes are rewarded or punished based on the pre-reward and punishment mechanism. Global trust + reward / - punishment = new global trust.

[0082] Beneficial effects: The trust management method of the pre-reward and punishment mechanism in the blockchain-based Internet of Vehicles system of the present invention, firstly, by constructing a vehicle trust evaluation system that comprehensively considers initialization trust, local trust and global trust, can more accurately evaluate the trust level of vehicle nodes, thereby improving the system's understanding and prediction capabilities of vehicle behavior. Secondly, the pre-reward and punishment mechanism based on the CNNs model to predict the authenticity of messages can effectively identify and punish the behavior of malicious vehicle nodes, reduce the risk of malicious attacks, and improve the security and stability of the system. Finally, the design of a consensus mechanism based on pre-reward and punishment and trust can improve the fault tolerance and performance of the system, ensuring that consensus can still be reached in the presence of malicious nodes, thereby ensuring the reliable transmission and sharing of traffic information. In summary, this invention provides important technical guarantees for the safe, reliable and efficient operation of the Internet of Vehicles system.

[0083] The present invention aims to solve the problems of insufficient trust management and imperfect consensus mechanism in the Internet of Vehicles system based on blockchain. In view of the problem that nodes with low latency requirements and default security in Internet of Vehicles applications are vulnerable to malicious attacks, this paper proposes a trust management method based on a pre-reward and punishment mechanism. Through this method, the problems of insufficient vehicle trust management and information reliability can be solved, and the problem of imperfect consensus can be solved by designing a consensus mechanism based on pre-reward and punishment and trust. Therefore, the present invention aims to design a trust management method suitable for the Internet of Vehicles architecture based on blockchain, so as to improve the security, stability and reliability of the Internet of Vehicles system. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 Trust management model of Internet of Vehicles based on blockchain.

[0085] Figure 2 Flowchart of the pre-reward and punishment mechanism based on the CNNs model to predict the authenticity of messages.

[0086] Figure 3 Flowchart of the consensus mechanism based on pre-reward and trust. DETAILED DESCRIPTION

[0087] The technical method of the present invention will be further described in detail below in conjunction with the accompanying drawings.

[0088] As shown in Figure (1), a blockchain-based Internet of Vehicles trust management model is established. The model mainly includes vehicles, roadside unit nodes (RSU) and trusted certification authorities (TA).

[0089] (1) Vehicle: In the trust management system, vehicles not only provide data such as road event information, traffic conditions, driving safety, and terrain services to other vehicle nodes, but also establish trust relationships based on the initial trust and local trust of other vehicle nodes, and upload these data to the roadside unit. The vehicle as the leader node is responsible for generating new blocks and uploading them to the roadside unit, while the consensus participating vehicle nodes work together with other vehicles to verify and confirm the validity of traffic information. In this process, vehicle nodes jointly maintain the safe and stable operation of the entire Internet of Vehicles system through mutual cooperation and trust establishment.

[0090] (2) Roadside Unit Node (RSU): RSU plays a vital role in the Internet of Vehicles trust management system. First, it calculates the global trust based on the initial trust and local trust data uploaded by the vehicle node, and broadcasts the global trust to the vehicles in the area, thereby establishing a trust relationship network between vehicles. Secondly, RSU deploys the CNNs model-based prediction vehicle information accuracy and pre-punishment mechanism to improve the credibility and accuracy of vehicle information transmission, and further ensure the security and stability of the system. Finally, as a data transfer and storage node, RSU is responsible for receiving block data from vehicle nodes, transferring and storing it, and providing support for subsequent data processing and analysis.

[0091] (3) Trusted Certification Authority (TA): The Trusted Certification Authority (TA) has the important responsibility of authenticating and establishing trust for vehicle nodes. By verifying the identity information of vehicle nodes and issuing digital certificates and other credentials, TA ensures that only legitimate vehicle nodes can participate in the Internet of Vehicles system and establishes a trust relationship between vehicle nodes. During the vehicle node registration phase, TA generates initial trust and reviews the vehicle registration information after a certain period or after a violation occurs. After the review is completed, TA updates the initial trust of the vehicle and sends the updated information to the roadside unit through blockchain broadcast to ensure the legitimacy and trustworthiness of the vehicle node. The role of TA is not only to confirm the authenticity of the vehicle identity, but also to maintain the trust network between vehicle nodes, providing reliable protection for the security and stability of the system.

[0092] The Internet of Vehicles Trust Management System uses a variety of technologies and algorithms, including trusted authentication authorities (TA), roadside units (RSU), prediction models, consensus algorithms, and blockchain technology. TA is responsible for authenticating and establishing trust with vehicle nodes, while RSU is responsible for receiving and verifying traffic information uploaded by vehicle nodes, and executing prediction models and pre-reward and punishment mechanisms. The system uses blockchain technology to achieve secure storage and non-tamperability of traffic information. By combining these technologies and algorithms, the Internet of Vehicles Trust Management System provides important support for the safety, reliability, and trustworthiness of intelligent transportation systems.

[0093] Step 2, a vehicle trust evaluation system is constructed that comprehensively considers initialization trust, local trust and global trust.

[0094] Definition of vehicle trust: In this invention, a region is used to illustrate the definition of trust, assuming that there are multiple vehicle nodes in this specified region. In such a region, roadside units can coexist with multiple vehicles to form an integrated system for trust management and calculation.

[0095] Step 2-1: Initialize Trust: The vehicle type, brand, model, insurance, driving age, violation record and other information that the vehicle needs to submit when registering. This information constitutes the Trust Attribute Matrix (TAM).

[0096] TAM={ta1,ta2,…,ta l ,…,ta m} (1)

[0097] use Represents the trust attribute ta g The relevant metric function of the credible measurement.

[0098] Therefore, the initial trust value of the vehicle can be expressed as:

[0099]

[0100] It is worth noting that the vehicle management department will review the vehicle registration information at a certain period or after a violation occurs, and assign a value to the vehicle's initial trust after the review is completed. Therefore, the vehicle's initial trust is not fixed.

[0101] Step 2-2: Local Trust: At regular intervals, the vehicle sends the trust feedback generated in the area to the roadside unit, which forms a trust feedback list for the local trust of all other devices in an area. Local trust is defined by path transmission loss and the accuracy of the provided message.

[0102] Assume that the local trust of vehicle node i to vehicle node j in the region is defined as LT ij .

[0103] Path loss: LP ij =L0+10nlog 10 (d ij ) (3)

[0104] Where L0 represents the path loss when the reference distance is 1 meter, n represents the path loss exponent, d ij Represents the distance between vehicle node i and vehicle node j.

[0105] Message accuracy:

[0106] Where α and β represent the number of correct and incorrect messages received by vehicle node i from vehicle node j, respectively.

[0107]

[0108] Where ω1 and (1-ω1) represent the weights of path loss and message transmission accuracy, and γ and ε are additional parameters used to adjust the impact of path loss and message transmission accuracy.

[0109] Since local trust is time-sensitive, the set of local trust of vehicle node i to vehicle node j with respect to time t is defined as If no information interaction occurs within time t, the initial trust is used for assignment.

[0110] Updates to local trust:

[0111] in represents the average value of the historical local trust within the window time. ΔT represents the difference between the current local trust and the average value of the historical local trust, that is:

[0112]

[0113] The update method can be expressed as follows:

[0114]

[0115] Represents the local trust value of the τth history within the window time.

[0116] Step 2-3: When vehicle node i enters a roadside unit area, it will upload its local trust list to the roadside unit in the area. The roadside unit will then calculate the global trust. In order to prevent collusion between malicious nodes, each node has a certain node trust similarity.

[0117] In the present invention, Euclidean distance is used to calculate the similarity of vehicle node trust:

[0118]

[0119] Global trust can be expressed as:

[0120]

[0121] Step 3, a pre-reward and punishment mechanism based on is designed to deal with the evaluation and management of the authenticity of messages provided by vehicles in the Internet of Vehicles, and is deployed in roadside units. The algorithm covers both reward and punishment. By comprehensively considering factors such as the number of historically generated blocks, the number of successfully verified transactions, the trust value, and the prediction results of the authenticity of the message, the vehicle's behavior of transmitting messages is rewarded and punished in a preventive manner.

[0122] Step 3-1: Use the number of blocks generated by the vehicle node and the number of transactions successfully verified to measure the node's contribution.

[0123] The roadside unit records the number of blocks generated by the vehicle node and the number of successfully verified transactions. After reaching a consensus in the area, it is uploaded to the roadside unit for storage and broadcast by the roadside unit in the blockchain.

[0124] The contribution of the node C i =ρ·M i +(1-ρ)·N i (11)

[0125] Among them, M i Represents the number of blocks generated by vehicle node i, N i It represents the number of transactions successfully verified by vehicle node i, and ρ is a parameter to be determined, which is used to adjust the impact of the number of blocks and transactions on the node contribution.

[0126] Step 3-2: Use the CNNs model to predict the authenticity of the message provided by the vehicle.

[0127] In the vehicle trust evaluation system and the contribution of nodes, we can see that the above evaluation feature parameters are all based on time series. If you want to predict the authenticity of the message provided by the vehicle, and then give pre-rewards and punishments, thereby reducing the vehicle's intention to participate in malicious behavior and improving the security of the Internet of Vehicles. Therefore, in this invention, we use the authenticity of message events based on convolutional neural networks (CNNs) for prediction.

[0128] Assume there is a set of message event data sets {(x (1) ,y (1) ), (x (2) ,y (2) ),…,(x (m) ,y (m) )}, where x (i) is the characteristic of the i-th message event, including the initial trust, local trust and global trust of the message sender, and whether the message signature is successfully verified, y (i) is the true label of the i-th message event (0 for false and 1 for true).

[0129] First, this method defines a neural network model based on CNNs, which can transform the message feature x (i) Mapping to ground truth predictions The parameters of the model are represented by θ. The input layer of the CNNs model is defined as the feature x that receives the message event. (i) , the CNNs layer is used to learn sequence features, and the output layer is used to predict the probability of the authenticity of the message event.

[0130] For a given message event feature x (i) , perform forward propagation calculations to obtain the model's predicted output

[0131] c κ =F(W κ *x (i) +b κ ) (12)

[0132] p κ =mean(c κ ) (13)

[0133]

[0134] Where Wκ and bκ represent the weight and bias of the kth convolution kernel, F represents the RELU activation function, and W y and b y are the weights and biases of the fully connected layer, softmax is the activation function, and k is the number of categories. is the probability distribution of the prediction model.

[0135] Secondly, the cross entropy loss function is defined in the method to measure the model prediction output and the true label y (i) The difference between.

[0136]

[0137] Finally, using the stochastic gradient descent optimization algorithm, a sample (x (i) ,y (i) ), calculate the objective function At the current parameter θ (t) The gradient of the selected sample Using the gradient descent update rule, update the parameter vector:

[0138]

[0139] Where μ is the learning rate, which controls the compensation of parameter update in each iteration. Repeat the iteration until the gradient size is less than 10 -4 , the model is considered to have converged and training is stopped.

[0140] Step 3-3: Pre-reward and punishment mechanism

[0141] In this method, the reward and punishment function we designed aims to take into account the contribution and trust of the vehicle node and adjust it according to the authenticity of the predicted message event.

[0142]

[0143] in, and δ are weight coefficients used to adjust the importance of different factors.

[0144] C i is the contribution of the vehicle; T i is the global trust of the vehicle; Confidence(x) is the confidence of the model in predicting the authenticity of the message; is the authenticity of the message event predicted by the model; threshold is the threshold of the authenticity of the message event; Sig(x) is the sigmoid function, which maps × to the range of (0,1).

[0145]

[0146] Confidence is the negative value of the entropy of the predicted probability distribution of the model's prediction of the authenticity of the message. When the entropy value is lower, it means that the model has a higher confidence in the prediction result; when the entropy value is higher, it means that the model has a lower confidence in the prediction result.

[0147] The design of this method allows for more fine-tuned adjustments of rewards and penalties to more effectively incentivize good performance in vehicle message transmission behavior and punish bad behavior.

[0148] Step 4, a pre-reward and trust consensus mechanism is designed to solve the Byzantine fault tolerance problem in the regional traffic information sharing system. It reaches consensus by establishing a trust relationship between the nodes of the vehicles in the region and ensures the correctness of the system in the presence of at most f malicious nodes.

[0149] Step 4-1: Trust identification and role division:

[0150] Based on the global trust of vehicle nodes, consensus participants and candidate participants are identified.

[0151] The number of consensus participants is more than 1 / 3 of the total number of nodes and greater than 2f+1, where f is the number of faulty or malicious nodes, and the remaining nodes are candidate participants. Penalized nodes do not participate in consensus.

[0152] Step 4-2: Leader Selection:

[0153] The vehicle node with the highest trust and pre-reward is selected as the leader from the consensus participants, and the node with the second highest trust is selected as the candidate leader. If the leader is attacked, the vehicle node of the candidate leader becomes the leader.

[0154] The leader is responsible for broadcasting traffic information within the consensus time, and signs the information with its own signature and sends it to other consensus participant nodes in all regions. If the message verification is successful, the consensus participant vehicle node enters the preparation phase; otherwise, the information will be discarded.

[0155] Step 4-3: Message Verification and Response

[0156] After receiving the leader's broadcast information, the consensus participant node performs public key verification to ensure the integrity of the information and the credibility of the source of the message. If the verification fails, the node will discard the information to prevent the information from being tampered with or forged. The verified node continues to verify the information and broadcasts the verification results to other participants. If a node receives verification from at least one-third of the participants, it broadcasts a commit message. After receiving enough commit messages, a reply message will be sent to the leader to confirm the transaction. After receiving the commit messages from at least one-third of the participants, the leader will package the transaction into a block. If not enough result messages are received, the leadership will be transferred to the second most trusted node. This process is repeated until a new block is created or the leader is transferred three times. If consensus is not reached after three leader changes, the current round of consensus is abandoned and the next round or rollback is started.

[0157] Step 4-4: Trust and Reward and Punishment Update

[0158] After consensus is reached and blocks are generated, the leader node updates its trust evaluation. For consensus participating nodes, the trust value is dynamically adjusted based on their performance and the results of processing traffic information. At the same time, vehicle nodes are rewarded or punished according to the pre-reward and punishment mechanism.

[0159] The above is only an implementation method of the present invention under the blockchain-based Internet of Vehicles model. The protection scope of the present invention is not limited to the above implementation method. Any equivalent modifications and other modifications made by ordinary technicians in this field based on the contents disclosed by the present invention should be included in the protection scope recorded in the claims.

Claims

1. A trust management method for Internet of Vehicles based on blockchain pre-reward and punishment mechanism, characterized in that: The steps include: Obtain the initialization trust, local trust and global trust of vehicle nodes; The contribution of the node is calculated using the number of blocks generated by the vehicle node and the number of transactions successfully verified; Construct a historical message event dataset provided by vehicle nodes in the Internet of Vehicles (x (1) ,y (1) ),(x (2) ,y (2) ),…,(x (m) ,y (m) )}, where x (i) is the characteristic of the i-th message event, including the initial trust, local trust and global trust of the message sender, and whether the message signature is successfully verified; y (i) is the true label of the i-th message event, false or true; Construct a neural network model based on CNNs, with the input layer as feature x (i) , the output layer is used to predict the probability of the authenticity of the message event Training the neural network model using a message event dataset; The pre-reward and pre-penalty values ​​for the vehicle node are calculated based on the contribution of the vehicle node, global trust, and the authenticity of the message event predicted by the trained neural network model; Based on the global trust of each vehicle node and its accumulated pre-reward and pre-penalty times, consensus participants are identified, and message transmission, verification and response are carried out between consensus participants. After reaching a consensus, the message is uploaded to the roadside unit.

2. According to claim 1, a vehicle network trust management method based on a blockchain pre-reward and punishment mechanism is characterized in that: The initial trust value of the vehicle is expressed as: Represents the trust attribute ta g The relevant metric function of the trust measurement, m represents the number of trust attributes, ω l Represents the weight of each attribute.

3. According to claim 1, a vehicle networking trust management method based on a blockchain pre-reward and punishment mechanism is characterized in that: Local trust is expressed as: Where ω1 and (1-ω1) represent the weights of path loss and message transmission accuracy, γ and ε are additional parameters used to adjust the impact of path loss and message transmission accuracy; path loss LP ij =L0+10nlog 10 (d ij ), where L0 represents the path loss when the reference distance is 1 meter, n represents the path loss exponent, and d ij Represents the distance between vehicle node i and vehicle node j; message accuracy Where α and β represent the number of correct and incorrect messages received by vehicle node i from vehicle node j, respectively; Define the set of local trusts of vehicle node i on vehicle node j at time t If no information exchange occurs within time t, the initial trust IT is used i Assign values; Local trust is updated based on the following formula: in, represents the average value of the local trust in the history within the window time, λ ΔT represents the update degree used to adjust the trust value, λ ΔT The selected value needs to consider the characteristics of the specific application scenario and the sensitivity to the change of trust value. ΔT represents the difference between the current local trust and the average value of the historical local trust, that is: The weighted update method of historical trust value is expressed as follows: represents the local trust value of the τth history within the window time, and σ is a decay parameter used to control the weight of the historical trust value.

4. The method for trust management of Internet of Vehicles based on blockchain pre-reward and punishment mechanism according to claim 3 is characterized in that: Global trust is expressed as: in, It represents the set of nodes in a certain area that conduct local trust evaluation with vehicle nodes i and j within time t.

5. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 1 is characterized in that: The contribution of the node C i =ρ·Μ i +(1-ρ)·Ν i Among them, i represents the number of blocks generated by vehicle node i, Ν i It represents the number of transactions successfully verified by vehicle node i, and ρ is a parameter to be determined, which is used to adjust the impact of the number of blocks and transactions on the node contribution.

6. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 1 is characterized in that: The specific steps of predicting the authenticity of the message provided by the vehicle based on the CNNs model include: For a given message event feature x (i) , perform forward propagation calculation: c k =F(W κ *x (i) +b κ ) Using average pooling for feature compression p κ =mean(c κ ) Get the model's predicted output: Where W κ and b k represents the weight and bias of the kth convolution kernel, F represents the RELU activation function, and W y and b y are the weights and biases of the fully connected layer, softmax is the activation function, k is the number of categories, is the probability distribution of the prediction model; Given the cross entropy loss function, it is used to measure the model prediction output and the true label y (i) The difference between: Using the stochastic gradient descent optimization algorithm, a sample (x (i) ,y (i) ), calculate the objective function At the current parameter θ (t) The gradient of the selected sample Using the gradient descent update rule, update the parameter vector: Where μ is the learning rate, which controls the compensation of parameter updates in each iteration.

7. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 1 is characterized in that: The calculation formula for the pre-reward value Reward and the pre-penalty value Penalty is: in, and δ are weight coefficients used to adjust the importance of different factors; C i is the contribution of the vehicle; T i is the global trust of the vehicle; Confidence(x) is the confidence of the model in predicting the authenticity of the message; is the authenticity of the message event predicted by the model; threshold is the threshold of the authenticity of the message event; Sig(x) is the sigmoid function, which maps x to the range of (0,1); Confidence is the negative value of the entropy of the predicted probability distribution of the model's prediction of the authenticity of the message; when the entropy value is lower, it means that the model has a higher confidence in the predicted result; when the entropy value is higher, it means that the model has a lower confidence in the predicted result.

8. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 1 is characterized in that: Identify consensus participants, transmit, verify and respond to each other, and upload messages to roadside units after reaching consensus. The specific steps include: 1) Identify consensus participants based on the global trust of vehicle nodes and their accumulated pre-reward and pre-penalty times; 2) Select the vehicle node with the highest trust and pre-reward from the consensus participants as the leader; the leader is responsible for broadcasting traffic information within the consensus time, and authenticates the information with its own signature and sends it to other consensus participant nodes in all areas; if the message verification is successful, the consensus participant vehicle node enters the preparation stage; otherwise, the information will be discarded; 3) After receiving the leader's broadcast information, the consensus participant nodes perform public key verification to ensure the integrity of the information and the credibility of the source of the message; if the verification fails, the node will discard the information to prevent the information from being tampered with or forged; the verified node continues to verify the information and broadcasts the verification results to other participants; if a node receives verification from at least one-third of the participants, it will broadcast a commit message; after receiving enough commit messages, it will send a reply message to the leader to confirm the transaction; after receiving the commit messages from at least one-third of the participants, the leader will package the transaction into a block; if not enough result messages are received, the leadership will be transferred to the node with the second highest global trust value; this process is repeated until a new block is created or the leader is transferred three times; if consensus is not reached after three leader changes, the current round of consensus is abandoned and the next round or rollback is started.

9. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 8 is characterized in that: Add the pre-reward value to the global trust value of the consensus participant and subtract the pre-penalty value; sort the data of each consensus participant from large to small, and select the vehicle nodes that are above the top 1 / 3 and greater than 2f+1 as consensus participants, where f is the total number of faulty nodes and malicious nodes.

10. The Internet of Vehicles trust management method based on blockchain pre-reward and punishment mechanism according to claim 8, characterized in that: After consensus is completed and blocks are generated, the leader node updates its own trust evaluation; for consensus participating nodes, its trust value is dynamically adjusted based on its performance and the results of its processing of traffic information; at the same time, vehicle nodes are rewarded or punished based on the pre-reward and punishment mechanism.

Citation Information

Cited By

  • Rail transit accident monitoring system and method based on wireless communication

    CN119160250A

  • Internet of Things encryption method and system combined with distributed trust mechanism

    CN120342632A