Blockchain car networking cross-region authentication system and method based on ai dynamic trust model

Abstract

The application provides a blockchain vehicle networking cross-region authentication system and method based on an AI dynamic trust model, and belongs to the technical field of vehicle networking secure communication. The system continuously collects and analyzes multi-dimensional behavior data of vehicles in a communication network, uses a deep learning algorithm to calculate and update a comprehensive trust score that can accurately reflect the behavior trustworthiness of the vehicles in real time, then deeply couples the AI evaluation result with a blockchain platform, records the trust data on a distributed ledger in a secure and tamper-proof manner, and automatically executes a differentiated authentication and resource control strategy based on the trust level through a smart contract. The application solves the communication security problem in vehicle networking from multiple dimensions, significantly improves the security, reliability and intelligent level of cross-region communication of vehicles in intelligent transportation, autonomous driving and other scenarios, and builds a solid and flexible security and trust foundation for the healthy development of the vehicle networking ecosystem.

Description

Technical Field

[0001] This invention belongs to the field of vehicle-to-everything (V2X) secure communication technology, specifically relating to a blockchain-based cross-regional authentication system and method for V2X based on an AI dynamic trust model. Background Technology

[0002] With the rapid development of vehicle-to-everything (V2X) technology, interconnectivity between vehicles and between vehicles and infrastructure has gradually become a core direction in the field of intelligent transportation. Compared with traditional transportation modes, this technology can greatly improve traffic efficiency and travel safety by enabling real-time monitoring of vehicle operating status, efficient sharing of traffic information, and intelligent optimization of driving decisions. However, the inherent openness of V2X makes its communication links face severe security and privacy challenges, especially in scenarios involving cross-regional communication and multi-entity collaboration, where threats such as identity forgery and data tampering are becoming increasingly prominent.

[0003] To address these challenges, blockchain technology, due to its decentralized and immutable characteristics, is widely regarded as a powerful tool for building a trust foundation for connected vehicles. Existing research has proposed various blockchain-based solutions, such as using smart contracts to restrict secure communication protocols for malicious vehicles, or constructing connected vehicles as consortium blockchains and setting "trust scores" for vehicle nodes in a polygonal network protocol. However, these existing solutions share a common and critical technical flaw: their trust assessment mechanisms are overly simplistic and static. Typically, these trust scores are calculated based on preset, fixed rules (such as simple message forwarding success rates), making it difficult to accurately reflect the true trustworthiness of vehicles in complex and dynamic environments. This also hinders the effective identification of new types of coordinated attacks, resulting in significant deficiencies in the "intelligence" and "accuracy" of the system's trust management.

[0004] Traditional static models are ill-suited to the dynamic changes in the Internet of Vehicles (IoV) environment. Their limitations are particularly pronounced in scenarios such as cross-domain authentication. They are not only vulnerable to new types of attacks but also prone to low authentication efficiency, severely hindering the further development of IoV technology. Artificial intelligence (AI) technology, on the other hand, can accurately identify vehicle behavior patterns and abnormal activities through in-depth analysis and learning of massive amounts of IoV data, providing data support and intelligent decision-making capabilities for dynamic trust assessment. Summary of the Invention

[0005] The purpose of this invention is to provide a blockchain-based cross-regional authentication system and method for connected vehicles based on an AI dynamic trust model. It utilizes AI technology to perform in-depth analysis of massive vehicle behavior data to achieve dynamic and accurate assessment of trust levels. At the same time, by leveraging the distributed ledger and smart contracts of blockchain, it ensures the immutability and cross-domain verifiability of the assessment results and automatically executes differentiated authentication strategies.

[0006] To achieve the above objectives, the present invention employs the following technical solution:

[0007] This invention provides a blockchain-based cross-regional authentication system for connected vehicles based on an AI dynamic trust model, comprising:

[0008] The vehicle behavior data acquisition module is used to collect multi-dimensional data reflecting vehicle behavior characteristics in real time. This multi-dimensional data includes basic message characteristics, message content credibility, network interaction behavior, and historical behavior records. The basic message characteristics include message sending timestamp, message type identifier, unique identifier ID, message byte length, and message sending cycle threshold. The message content credibility includes message checksum, continuous checksum values ​​of message sequence number, and semantic rationality score of message content. The network interaction behavior includes message sending frequency, message interval time statistical characteristics, message response latency, and network load rate. The historical behavior records include historical message frequency patterns based on time windows, frequency of abnormal behavior occurrence, and baseline characteristics of normal behavior.

[0009] The data preprocessing and feature vector construction module is used to preprocess the collected multidimensional data and construct feature vectors.

[0010] The AI ​​trust assessment engine is used to evaluate the dynamic behavior of vehicles based on a pre-trained deep learning model, using the feature vector as input, and to calculate a comprehensive trust score.

[0011] The blockchain storage and sharing module is used to store the comprehensive trust score, key behavioral evidence hash, and evaluation process parameters calculated by the AI ​​trust assessment engine on the blockchain.

[0012] The smart contract authentication decision module is used to make intelligent decisions on cross-regional vehicle authentication behavior based on on-chain trusted data, and to execute differentiated authentication and dynamic allocation strategies of network resources based on trust level.

[0013] Preferably, the data preprocessing and feature vector construction module preprocesses the collected multidimensional data and constructs feature vectors, including:

[0014] The multidimensional data collected by the vehicle in real time during the communication process is shown below:

[0015] ,

[0016] in, Representing multidimensional data, Indicates the first Each feature data, The total number of features;

[0017] For multidimensional data For missing values, imputation is performed using K-nearest neighbor-based interpolation or the median; for outliers, imputation is performed using... Identification and rejection are based on principles or box plots;

[0018] An improved min-max normalization method is used for data with different dimensions, and a dynamic scaling factor is introduced. A linear mapping to the interval [0.1, 0.9] is expressed as follows:

[0019] , ,

[0020] in, and Representing multidimensional data The minimum and maximum values ​​in express The new result value after normalization;

[0021] Based on the normalized data, a feature vector comprehensively representing the instantaneous behavior state of the vehicle is constructed, denoted as . .

[0022] Preferably, the AI ​​trust assessment engine calculates a comprehensive trust score, as follows:

[0023] The feature vector is input into a pre-trained deep learning model to obtain a direct confidence output, which is a normalized value between 0 and 1.

[0024] Based on the direct trust score output, the overall trust score is calculated as follows:

[0025] ,

[0026] in, For at any time The overall trust score, For at any time The overall trust score, For time intervals, The trust decay rate coefficient, , It is a moment The input feature vector, For deep learning models based on feature vectors The calculated direct trust level output, Historical trust weighting factor, , The penalty coefficient is... , The dynamic penalty value for a vehicle whose malicious behavior is proven by reliable evidence. The incentive coefficient for trust growth. , For a moment Positive behavioral incentive value;

[0027] The calculation is as follows: ,

[0028] in, As the incentive benchmark coefficient, , For a moment The number of compliant actions within the previous sliding window. For a moment Total number of actions within the previous sliding window It is a factor of behavioral importance.

[0029] Preferably, the deep learning model uses a hybrid dataset containing massive amounts of historical vehicle normal behavior data, simulated attack behavior data, and real malicious behavior data, and is obtained through a two-stage training method combining supervised learning and semi-supervised learning.

[0030] Preferably, the dynamic penalty value is quantitatively calculated based on the type of malicious behavior, severity level, and scope of impact, as shown below:

[0031] ,

[0032] in, The number of confirmed malicious behavior types. For the first Weighting coefficients for malicious behaviors No. The severity score of malicious behavior. For the first The impact range coefficient of malicious behavior.

[0033] Preferably, the data processing procedure of the blockchain storage and sharing module is as follows:

[0034] The relevant data that needs to be stored on the blockchain is combined into a structured transaction. The details are as follows:

[0035] ,

[0036] in, Evidence of key behavioral events cryptographic hash value, For the time the transaction was generated, It serves as a unique identifier for the vehicle;

[0037] Transaction data After its validity is verified through the consensus mechanism, it is included in the candidate block;

[0038] During block construction, the Merkle root of a block is linked to the hash value of the previous block to form an immutable chain structure, represented as:

[0039] ,

[0040] in, For the current new block, It is the cryptographic hash of all the data in the block header of the previous block. It's the version number. It is the block height. It's a timestamp. Indicates the first This transaction, This indicates that all blocks within this region will be packaged together. Each transaction is ultimately calculated into a unique root hash value using pairwise hashing. It is a random number. It is the verifier's signature.

[0041] Preferably, the smart contract authentication decision module encodes and deploys smart contracts in the distributed ledger of the blockchain network. The smart contract defines rules for mapping the dynamic trust score calculated in real time by the vehicle to authentication strategies of different trust levels. When the preset authentication triggering conditions are met, the smart contract is triggered to automatically execute the differentiated authentication and dynamic allocation strategy of network resources based on the trust level.

[0042] The specific strategy for differentiated authentication and dynamic allocation of network resources based on trust levels is as follows:

[0043] The smart contract queries the blockchain for the latest comprehensive trust score corresponding to the vehicle's unique identifier ID. And obtain the sequence of all historical comprehensive trust scores of the vehicle within the specified sliding historical time window. , The sequence length;

[0044] The smart contract performs a trust assessment based on a pre-defined weighted average strategy, calculating the dynamic trust level as follows:

[0045] ,

[0046] in, For dynamic trust levels, It is the weighted average of all historical comprehensive trust scores of the vehicle within a specified sliding historical time window. and Let be the weighting coefficient, satisfying ,and , The standard deviation of the historical overall trust score. The recent trend value of trust changes is denoted as... , The recent window length;

[0047] Smart contracts will dynamically determine trust levels. Mapped to one of three predefined discrete trust levels and triggering the corresponding authentication policy;

[0048] Based on the trust level, the corresponding authentication and network resource control policies are automatically executed.

[0049] Preferably, the smart contract will dynamically assign trust levels. Mapped to one of three predefined discrete trust levels, including:

[0050] like This indicates a high level of trust.

[0051] like If so, it is a medium trust level;

[0052] like If so, it is considered a low trust level.

[0053] Preferably, the step of automatically executing authentication and network resource control policies corresponding to the trust level includes:

[0054] For vehicles with high trust levels, fast-track authentication is implemented, including: eliminating the handshake process during authentication, prioritizing the forwarding of sent messages, and granting them priority access to network bandwidth resources;

[0055] For vehicles with a medium trust level, a standard certification process is implemented.

[0056] For vehicles with low trust levels, enhanced authentication is implemented, including: performing multi-factor authentication and certificate chain verification; allocating a sandboxed network channel to the low-trust-level vehicle, to which all communication data is redirected; restricting or prohibiting the low-trust-level vehicle from sending broadcast messages and prohibiting it from acting as a relay node to forward data for other vehicles; temporarily placing the low-trust-level vehicle on a watch list, and if malicious behavior is confirmed, immediately terminating network access, adding the low-trust-level vehicle's digital certificate to the revocation list, and permanently recording the malicious behavior event on the blockchain.

[0057] This invention also provides a blockchain-based cross-regional authentication method for connected vehicles based on an AI dynamic trust model, implemented based on the aforementioned blockchain-based cross-regional authentication system for connected vehicles using an AI dynamic trust model. The method includes:

[0058] The system collects multidimensional data reflecting vehicle behavior characteristics in real time. This multidimensional data includes basic message characteristics, message content credibility, network interaction behavior, and historical behavior records. The basic message characteristics include message sending timestamp, message type identifier, unique identifier (ID), message byte length, and message sending cycle threshold. The message content credibility includes message checksum, continuous checksum values ​​for message sequence number, and semantic rationality score of message content. The network interaction behavior includes message sending frequency, message interval time statistics, message response latency, and network load rate. The historical behavior records include historical message frequency patterns based on time windows, frequency of abnormal behavior, and baseline characteristics of normal behavior.

[0059] The collected multidimensional data is preprocessed and feature vectors are constructed;

[0060] Using the feature vector as input, a pre-trained deep learning model is used to evaluate the dynamic behavior of the vehicle and calculate a comprehensive trust score.

[0061] The comprehensive trust score, key behavioral evidence hash, and assessment process parameters calculated by the AI ​​trust assessment engine are stored on the blockchain in a chain.

[0062] Intelligent decision-making is made based on trusted on-chain data to enable cross-regional vehicle authentication, and a differentiated authentication and dynamic allocation strategy for network resources based on trust level is implemented.

[0063] The beneficial effects achieved by this invention are as follows:

[0064] This invention proposes a blockchain-based cross-regional authentication system for connected vehicles based on an AI dynamic trust model. From data collection, trust calculation, on-chain evidence storage to authentication decision-making, the entire process requires no manual intervention, forming a self-driven and intelligent authentication closed loop.

[0065] This invention uses AI models to continuously calculate and update the trust level of vehicles based on real-time and historical behavior data, enabling the vehicle identity trustworthiness assessment results to change dynamically with their behavior, effectively identifying and curbing unauthorized vehicle identity impersonation attacks using static credentials; the system can automatically adjust the parameters and weights of the trust assessment model according to changes in vehicle behavior patterns to adapt to different driving scenarios and safety requirements.

[0066] This invention leverages the distributed ledger characteristics of blockchain to ensure that the trust level updated in any region can be synchronized to other regions in real time, securely, and immutably. This achieves consistency in cross-domain trust states. When any region detects abnormal or malicious vehicle behavior and updates its trust level to low, this information is rapidly synchronized to the entire network via the blockchain consensus mechanism. Other regions can then use this information to implement enhanced authentication or access restrictions in advance, forming a cross-domain collaborative proactive defense system. This invention combines the internal behavior assessment of an AI dynamic trust model with the tamper-proof and traceable external data protection of blockchain technology, constituting a dual security mechanism that addresses both internal and external threats.

[0067] This invention automatically executes differentiated authentication strategies through smart contracts, granting different network access permissions to vehicles with different trust levels, thereby achieving refined security management, minimizing the potential attack surface, avoiding the delays caused by traditional centralized authentication authorities, and enabling rapid authorization for cross-domain vehicle access. Attached Figure Description

[0068] Figure 1 A schematic diagram of a blockchain-based vehicle network cross-regional authentication system architecture based on an AI dynamic trust model is provided for this invention.

[0069] Figure 2 This is a schematic diagram of data processing by the AI ​​trust assessment engine in the authentication system provided by the present invention;

[0070] Figure 3 This is a schematic diagram of data processing in the smart contract authentication decision module of the authentication system provided by the present invention. Detailed Implementation

[0071] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0072] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0073] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0074] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0075] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0076] It should be emphasized here that the step markers mentioned below are not a limitation on the order of the steps, but should be understood as meaning that the steps can be executed in the order mentioned in the embodiments, or in a different order than in the embodiments, or several steps can be executed simultaneously.

[0077] This invention provides a blockchain-based cross-regional authentication system for connected vehicles based on an AI dynamic trust model. Specifically, it is a closed-loop security authentication system integrating "data collection → AI intelligent assessment → blockchain notarization → smart contract decision-making." The system includes: a vehicle behavior data collection module, a data preprocessing and feature vector construction module, an AI trust assessment engine, a blockchain storage and sharing module, and a smart contract authentication decision-making module. The overall data processing architecture of the system is as follows: Figure 1 As shown, the details are as follows:

[0078] The vehicle behavior data acquisition module is used to collect multi-dimensional data reflecting vehicle behavior characteristics in real time. This multi-dimensional data includes four categories: basic message characteristics, message content credibility, network interaction behavior, and historical behavior records.

[0079] The data preprocessing and feature vector construction module is used to preprocess the collected multidimensional data and construct feature vectors.

[0080] The AI ​​trust assessment engine is used to evaluate the dynamic behavior of vehicles based on a pre-trained deep learning model, using the feature vectors constructed above as input, and to calculate a comprehensive trust score.

[0081] The blockchain storage and sharing module is used to store the comprehensive trust score, key behavioral evidence hashes, and assessment process parameters calculated by the AI ​​trust assessment engine on the blockchain.

[0082] The smart contract authentication decision module is used to make intelligent decisions on cross-regional vehicle authentication behavior based on on-chain trusted data, and to execute differentiated authentication and dynamic allocation strategies of network resources based on trust level.

[0083] In this invention, the vehicle behavior data acquisition module collects multi-dimensional data reflecting vehicle behavior characteristics in real time. The specific implementation process is as follows:

[0084] During communication between the vehicle and roadside units and other vehicles, multi-dimensional data reflecting the vehicle's behavioral characteristics are collected in real time and recorded as a set, as shown below:

[0085] ,

[0086] in, This refers to the collection of all characteristic data that can reflect vehicle behavior and are collected in real time from vehicle communication, i.e., "multidimensional data"; Indicates the first Each feature data point is a specific indicator describing vehicle behavior, which together form a complete data vector. , The total number of features.

[0087] In this invention, multidimensional data includes four categories: basic message characteristics, message content credibility, network interaction behavior, and historical behavior records, denoted as follows: , , , , represented as:

[0088] ,

[0089] Among them, the basic characteristics of a message This includes message sending timestamp, message type identifier, unique identifier ID, message byte length, and message sending period threshold; message content credibility. This includes message checksum, consecutive checksum values ​​of message sequence number, and semantic reasonableness score of message content; network interaction behavior. This includes message sending frequency, message interval statistics (mean, variance), message response latency, and network load rate; historical behavior records. This includes historical message frequency patterns, frequency of abnormal behavior, and baseline characteristics of normal behavior, all statistically analyzed based on a time window (the past thirty days). This data is continuously recorded and updated periodically through a real-time acquisition system.

[0090] In this invention, the data preprocessing and feature vector construction module preprocesses the collected multidimensional data and constructs feature vectors. The specific implementation process is as follows:

[0091] The collected raw multidimensional data Preprocessing is performed, including data cleaning, normalization, and feature engineering, as detailed below:

[0092] Data cleaning: handling missing and outlier values. For missing values, imputation is performed using K-nearest neighbor-based interpolation or median; for outliers, imputation is performed using... Identification and rejection are based on principles or box plots.

[0093] Normalization: An improved min-max normalization method is used for data with different dimensions, introducing a dynamic scaling factor. A linear mapping to the interval [0.1, 0.9] is expressed as follows:

[0094] , ,

[0095] in, Represents the first in the original multidimensional data One set of original feature data; and Representing the original multidimensional data The minimum and maximum values ​​in; Represents original feature data The new result value after normalization The size reflects the original feature data Relative position within the entire dataset.

[0096] Feature engineering: Constructing a high-dimensional feature vector that comprehensively represents the instantaneous behavior state of a vehicle based on cleaned and normalized data, denoted as This is used for subsequent model training.

[0097] In this invention, the AI ​​trust assessment engine uses the constructed feature vector as input, performs dynamic behavior assessment of the vehicle based on a pre-trained deep learning model, and calculates a comprehensive trust score. For details, please refer to [link to implementation details]. Figure 2 ,include:

[0098] The collected and preprocessed real-time and historical data are used as input feature vectors and fed into a pre-trained deep learning model. The calculations are performed within the dataset. This deep learning model uses a hybrid dataset containing massive amounts of historical vehicle normal behavior data, simulated attack behavior data, and real malicious behavior data. It is trained through a two-stage approach combining supervised and semi-supervised learning. The first stage uses supervised learning to train the basic trust assessment capability, while the second stage uses semi-supervised learning to improve the model's ability to generalize and identify unknown malicious behaviors. The model outputs a direct trust assessment value for the current vehicle behavior and, combined with historical trust states, calculates a dynamically updated, comprehensive, real-time trust score through a specific time decay and event penalty mechanism.

[0099] Based on the evaluation results, the overall trust score is calculated as follows:

[0100] ,

[0101] in, It is preprocessed multidimensional data At any moment The values ​​are constructed into a feature vector, which comprehensively reflects the instantaneous behavioral state of the vehicle in terms of communication, message content, network interaction and historical performance; For at any time The overall trust score, For the previous moment The overall trust score reflects the historical continuity of trust assessment. For time intervals, The trust decay rate coefficient, , For deep learning models based on time behavioral feature vector The calculated direct trust score output is a normalized value between 0 and 1. Historical trust weighting factor, Used to control the degree of influence of historical trust levels. The larger the value, the greater the weight of controlling historical trust, the more persistent the impact of historical behavior, and the smoother the change in system trust. The smaller the value, the more sensitive the system is to recent behavior; The penalty coefficient is... This is used to adjust the intensity of the negative impact of confirmed malicious behavior on the overall trust level. The dynamic penalty value for a vehicle whose malicious behavior is proven by reliable evidence. The incentive coefficient for trust growth. This is used to control the degree to which incentives affect the final level of trust. For a moment Positive behavioral incentive value;

[0102] The formula is as follows, based on the cumulative number of compliant behaviors and the quantitative calculation of the importance of those behaviors:

[0103] ,

[0104] in, As the incentive benchmark coefficient, , For a moment The number of compliant actions within the previous sliding window. For a moment Total number of actions within the previous sliding window It is a factor of behavioral importance.

[0105] It should be noted that the dynamic penalty value A three-dimensional quantitative calculation is performed based on the type of malicious behavior, severity level, and scope of impact, using the following formula:

[0106] ,

[0107] in, The number of confirmed malicious behavior types is taken in this embodiment. , For the first The weighting coefficient of a malicious behavior type reflects its relative importance within the overall threat. No. The severity score of malicious behavior, For the first The impact scope coefficient of malicious behavior, and the relevant values ​​are shown in Table 1 below.

[0108] Table 1. Malicious Behavior Weights and Severity Scores

[0109]

[0110] In this invention, the data processing procedure of the blockchain storage and sharing module is as follows:

[0111] To prevent vehicle node trust data from being tampered with locally and to provide globally trusted data support for cross-domain identity authentication, the blockchain storage and sharing module combines trust-related data that needs to be stored on-chain into a structured transaction data. The details are as follows:

[0112] ,

[0113] in, Evidence of key behavioral events The cryptographic hash value is obtained by using a cryptographic hash function to hash the original evidence. It is generated by performing one-way irreversible operations to ensure data integrity and source verifiability. Any modification to the original evidence will result in a drastic change in the hash value. The transaction generation time is used to ensure data timeliness; It serves as a unique identifier for the vehicle.

[0114] Transaction data After passing the validity verification through the consensus mechanism, it is included in the candidate block. During the block construction process, the system generates a Merkle hash tree for all packaged transactions and encodes the root hash value of the tree with the hash pointer of the previous block header, thereby constructing an ordered data chain with cryptographic self-verification properties in the distributed ledger.

[0115] It should be noted that evidence of key behavioral events It originates from communication interaction records between vehicles and infrastructure within the Internet of Vehicles, external physical perception and environmental data, and consensus reports in the trust network.

[0116] Direct communication evidence includes untampered original messages sent or forwarded by the vehicle (such as safety messages BSM and cooperative awareness messages) and network interaction logs (complete signaling logs generated by the vehicle and the RSU or other vehicles during interactions such as security handshakes, data forwarding, and request-response).

[0117] Environmental perception evidence is provided by sensors such as cameras and radar in the roadside unit (RSU) and is used to verify the authenticity of the communication content using physical world data (such as whether there is a vehicle-reported accident).

[0118] Consensus reporting in a trust network refers to the independent reporting records of multiple independent, geographically separated RSUs or vehicles for the same anomalous event (such as disinformation broadcasting).

[0119] To ensure the validity and processability of the evidence, the acquired raw data must be standardized and packaged into a structured data package containing the event type, timestamp, location, perpetrator, raw data, and witness information. This data package is known as "key behavioral event evidence." ".evidence Standardized packaging process:

[0120] Data scraping: scrape raw data from the above sources in real time or near real time.

[0121] Spatiotemporal alignment: Attach high-precision timestamps and geospatial coordinates to all data.

[0122] Formatted Description: The event is described in a structured manner using a standardized data description framework. Key fields include:

[0123] event_type: Event type;

[0124] actor_id: The vehicle ID of the actor;

[0125] timestamp: The time when the action occurred;

[0126] location: the location where the action occurs;

[0127] raw_data: The original data or its characteristic hash;

[0128] witness_ids: A list of witness IDs.

[0129] Then its hash value Calculated and together with other elements, constitute transaction data. Stored on-chain. Original evidence. It can be stored off-chain, based on the hash value. The blockchain acts as a "digital fingerprint" of the evidence, ensuring that the evidence itself is immutable once generated and can be verified at any time.

[0130] Transaction data After verification and packaging by the consensus mechanism, the block is linked to the hash value of the previous block through the Merkle tree root, forming an immutable chain structure. The core of this process is defined by the following formula:

[0131] ,

[0132] in, For the current new block, It is the cryptographic hash of all the data in the block header of the previous block;

[0133] It is the version number, an integer field used to identify the protocol version of the block structure. Its function is to ensure that all nodes in the network follow the same set of data parsing and verification rules, so as to achieve forward compatibility and smooth upgrades.

[0134] It is the block height, a monotonically increasing integer, uniquely identifying the block's sequential position in the blockchain. Within the block structure... The field stores the height value of the block, denoted as the first... Each block The field value is Its value is generated by incrementing the height value of the previous block, satisfying... Among them, the genesis block ;

[0135] It is a timestamp that records the precise time when this block was generated. Its core function is to provide global timing logic, prevent transaction replay, and serve as the basis for judgment in some smart contract logic based on time windows.

[0136] Indicates the first This transaction, This indicates that all blocks within this region will be packaged together. Each transaction is hashed pairwise to generate a unique root hash value. This root hash represents the digital fingerprint of all transactions within the block, and even the slightest change to any transaction will result in... A complete change;

[0137] It is a random number used to adjust the mining difficulty value so that the hash value of the block header meets a specific difficulty target;

[0138] It is the validator signature, a digital signature generated by the block-producing node in this round for the entire block header data except for this signature field. Its function is to provide proof of the block creator and guarantee its integrity.

[0139] In this invention, the smart contract authentication decision module makes intelligent decisions on vehicle authentication behavior based on blockchain, and executes differentiated authentication decisions. The specific decision-making process is as follows: Figure 3 As shown, it includes:

[0140] The smart contract authentication decision module encodes and deploys smart contracts in the distributed ledger of the blockchain network. These smart contracts explicitly define the rules for mapping the vehicle's real-time calculated dynamic trust score to different levels of authentication strategies. When the preset authentication trigger conditions are met, the roadside unit or verification node will invoke and run the smart contract, thereby enabling the enforcement of corresponding differentiated authentication control instructions based on the dynamic trust score stored on the chain without manual intervention.

[0141] The differentiated certification decision-making process is as follows:

[0142] (1) Query,

[0143] The smart contract queries the blockchain for the latest comprehensive trust score corresponding to the vehicle's unique identifier ID. And obtain the sequence of all historical comprehensive trust scores for the vehicle within a specified sliding historical time window. Plot the historical trust level change curve. This represents the total number of historical comprehensive trust scores recorded and stored by the system for this vehicle within that time window, i.e., the sequence length, which is the number of sampling points. The preceding text describes historical behavior records. This includes historical message frequency patterns based on a time window (the past thirty days). The value is not directly equal to the number of days in the time window, but rather equal to the time window length divided by the sampling interval, i.e. ,in Indicates the time window, setting , The system uses a fixed sampling interval. Take a snapshot of the vehicle's overall trust level to obtain... in The earliest recorded score within that time window. This is the score from the most recent record.

[0144] (2) Evaluation,

[0145] The smart contract performs a trust assessment based on a pre-defined weighted average strategy, calculating the dynamic trust level as follows:

[0146] ,

[0147] in, For dynamic trust levels, It is the weighted average of all historical comprehensive trust scores of the vehicle within the specified sliding historical time window, denoted as ; and Let be the weighting coefficient, satisfying ,and This ensures the security and fairness of the system. The standard deviation of the historical overall trust score. The recent trend value of trust changes is denoted as... , The recent window length, A value greater than 0 indicates a positive trend, while a value less than 0 indicates a negative trend.

[0148] (3) Decision-making and execution

[0149] Smart contracts will dynamically determine trust levels. It maps to one of three predefined discrete trust levels and triggers the corresponding authentication policy, specifically,

[0150] like If it is, then it is a high level of trust. If it is, then it is a medium trust level. If so, it is considered a low trust level.

[0151] Based on the trust level, automatically implement differentiated authentication and network resource control policies, as follows:

[0152] High-trust level vehicles: perform fast-channel authentication (such as eliminating the complex handshake process), their messages are forwarded first, and they have priority access to network bandwidth resources.

[0153] Medium trust level vehicles: Follow standard certification procedures.

[0154] For vehicles with low trust levels: implement enhanced authentication and restrict their broadcast message privileges, or temporarily place them on a watchlist. If malicious behavior is confirmed, a penalty mechanism is triggered, and the event is permanently recorded on the blockchain for reference by all network nodes, achieving collaborative defense.

[0155] It should be noted that "Fast Track Authentication" is an optimized authentication and resource priority mechanism activated for vehicles assessed as having high trust levels by the system. Its core design philosophy is: based on trust, efficiency is gained; while ensuring security, communication latency and access overhead for highly trusted vehicles are minimized to the greatest extent possible, thereby improving the efficiency and response speed of the entire vehicle-to-everything (V2X) network. This includes optimizations in three dimensions:

[0156] A. Optimized authentication process, eliminating the need for complex handshakes.

[0157] B. Message forwarding priority is increased, meaning that messages from high-trust vehicles will be given higher priority in the forwarding queues of routing nodes (RSU, other vehicles). This means their messages will have shorter queuing times and can be delivered to the target recipient more quickly, thereby improving the timeliness of critical security information.

[0158] C. Prioritized allocation of network bandwidth resources: When wireless channel resources are scarce, the network infrastructure will prioritize the communication needs of high-trust vehicles. This is reflected in allocating them with better wireless channels or time slots, giving them a higher probability of access during contention for access, and ensuring minimum bandwidth guarantees for their critical services.

[0159] The standard authentication process specifically refers to the complete two-way identity authentication and key negotiation protocol that has been widely standardized and adopted in the Internet of Vehicles.

[0160] Enhanced authentication is a set of enhanced, multi-dimensional, and deep verification security access and control processes implemented for vehicles assessed by the system as having a low trust level. Its core objective is to allow suspicious vehicles limited access to the network while verifying their identity and intent with a higher level of security, and to strictly monitor their behavior and restrict their access, thereby isolating potential threats within a controllable range.

[0161] When a smart contract assesses the trust level of a vehicle's evaluation. At that time, the enhanced authentication process is triggered:

[0162] S1, Enhanced authentication, specifically includes:

[0163] S11. Multi-factor authentication: Requires vehicles to provide additional credentials (such as one-time passwords or hardware token signatures) in addition to digital certificates.

[0164] S12, Historical Behavior Inquiry: Based on suspicious behaviors recorded on the blockchain, the smart contract launches targeted challenges to the vehicle (such as explaining abnormal messages at a specific point in time).

[0165] S13, Certificate Chain Verification: Deeply verify its certificate chain to check if any intermediate certificates have been revoked or expired.

[0166] S2, Restricted Access and Communication Monitoring

[0167] S21, Sandboxed Network Channel: Assigns a logically isolated, restricted network channel or "security sandbox" to low-trust vehicles, and all their communication data will be redirected to this sandboxed network channel.

[0168] S22. Deep Packet Inspection and Behavioral Analysis: In this sandboxed network channel, the system performs deep packet inspection and real-time behavioral analysis on its communication traffic to check whether its message content and sending patterns further exhibit malicious characteristics.

[0169] S23. Minimize privileges: Strictly restrict or prohibit it from sending broadcast messages (such as collaborative sensing messages, emergency event warnings) to prevent it from spreading false information. Prohibit it from acting as a relay node to forward data for other vehicles, cutting off any further damage it may cause to the network.

[0170] S3. Included in the observation list and subject to behavioral monitoring

[0171] S31, Mandatory Observation Period: The vehicle is automatically added to the system's observation list, and an observation period is set.

[0172] S32. Continuous behavior recording: During the observation period, all its communication behaviors are recorded in detail and uploaded to the blockchain.

[0173] S33. Dynamic reassessment of trust: At the end of the observation period, a new round of trust assessment is automatically triggered based on behavioral data.

[0174] S4, Punishment Mechanism

[0175] S41. Immediate Penalty: If malicious behavior is confirmed during enhanced authentication or observation period, the network access must be terminated immediately, the digital certificate must be added to the local or global revocation list, and the trust score must be reduced to the minimum threshold.

[0176] S42. Event Permanently Recorded on the Blockchain: Complete evidence of malicious behavior, punishment decisions, and execution results are permanently recorded on the blockchain as tamper-proof transactions for reference by all network nodes, enabling cross-domain collaborative defense.

[0177] S5, Trust Restoration

[0178] If a vehicle performs well during the observation period and does not trigger further penalties, its trust score can be gradually restored according to a preset strategy (increasing by 0.1 with each assessment), and it can be removed from the observation list to resume the standard certification process.

[0179] Based on the aforementioned technical means, this invention constructs a highly efficient, secure, and automated cross-domain authentication closed loop, enabling dynamic and adaptive assessment of vehicle identity trustworthiness and cross-domain consistency management. By combining machine learning algorithms with a distributed consensus mechanism, this invention establishes a differentiated policy execution model based on fine-grained trust levels, ultimately achieving efficient and reliable dynamic security authentication. This strategy significantly improves the security, privacy protection capabilities, and adaptability of vehicle-to-everything (V2X) communication in cross-regional communication, intelligent transportation, and autonomous driving scenarios. For future dynamic and ever-changing environments, the authentication system proposed in this invention can be combined with continuous learning technology to dynamically optimize the trust assessment model, addressing novel collaborative attacks and various sudden security threats.

[0180] Based on the same inventive concept, this invention also provides a blockchain-based cross-regional authentication method for connected vehicles using an AI dynamic trust model. Implemented based on the aforementioned authentication system, the method includes:

[0181] The system collects multidimensional data reflecting vehicle behavior characteristics in real time. This multidimensional data includes basic message characteristics, message content credibility, network interaction behavior, and historical behavior records. The basic message characteristics include message sending timestamp, message type identifier, unique identifier (ID), message byte length, and message sending cycle threshold. The message content credibility includes message checksum, continuous checksum values ​​for message sequence number, and semantic rationality score of message content. The network interaction behavior includes message sending frequency, message interval time statistics, message response latency, and network load rate. The historical behavior records include historical message frequency patterns based on time windows, frequency of abnormal behavior, and baseline characteristics of normal behavior.

[0182] The collected multidimensional data is preprocessed and feature vectors are constructed;

[0183] Using the feature vector as input, a pre-trained deep learning model is used to evaluate the dynamic behavior of the vehicle and calculate a comprehensive trust score.

[0184] The comprehensive trust score, key behavioral evidence hash, and assessment process parameters calculated by the AI ​​trust assessment engine are stored on the blockchain in a chain.

[0185] Intelligent decision-making is made based on trusted on-chain data to enable cross-regional vehicle authentication, and a differentiated authentication and dynamic allocation strategy for network resources based on trust level is implemented.

[0186] It is worth noting that the system embodiment corresponds to the above method embodiment. The implementation methods of the above method embodiments are all applicable to the system embodiment and can achieve the same or similar technical effects, so they will not be described in detail here.

[0187] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0188] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0189] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0190] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A blockchain-based cross-regional authentication system for connected vehicles based on an AI dynamic trust model, characterized in that: include: The vehicle behavior data acquisition module is used to collect multi-dimensional data reflecting vehicle behavior characteristics in real time. This multi-dimensional data includes basic message characteristics, message content credibility, network interaction behavior, and historical behavior records. The basic message characteristics include message sending timestamp, message type identifier, unique identifier ID, message byte length, and message sending cycle threshold. The message content credibility includes message checksum, continuous checksum values ​​of message sequence number, and semantic rationality score of message content. The network interaction behavior includes message sending frequency, message interval time statistical characteristics, message response latency, and network load rate. The historical behavior records include historical message frequency patterns based on time windows, frequency of abnormal behavior occurrence, and baseline characteristics of normal behavior. The data preprocessing and feature vector construction module is used to preprocess the collected multidimensional data and construct feature vectors. The AI ​​trust assessment engine is used to evaluate the dynamic behavior of vehicles based on a pre-trained deep learning model, using the feature vector as input, and to calculate a comprehensive trust score. The blockchain storage and sharing module is used to store the comprehensive trust score, key behavioral evidence hash, and evaluation process parameters calculated by the AI ​​trust assessment engine on the blockchain. The smart contract authentication decision module is used to make intelligent decisions on cross-regional vehicle authentication behavior based on on-chain trusted data, and to execute differentiated authentication and dynamic allocation strategies of network resources based on trust level.

2. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 1, characterized in that, The data preprocessing and feature vector construction module preprocesses the collected multidimensional data and constructs feature vectors, including: The multidimensional data collected by the vehicle in real time during the communication process is shown below: ， in, Representing multidimensional data, Indicates the first Each feature data, The total number of features; For multidimensional data For missing values, imputation is performed using K-nearest neighbor-based interpolation or the median; for outliers, imputation is performed using... Identification and rejection are based on principles or box plots; An improved min-max normalization method is used for data with different dimensions, and a dynamic scaling factor is introduced. A linear mapping to the interval [0.1, 0.9] is expressed as follows: ， ， in, and Representing multidimensional data The minimum and maximum values ​​in express The new result value after normalization; Based on the normalized data, a feature vector comprehensively representing the instantaneous behavior state of the vehicle is constructed, denoted as . .

3. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 2, characterized in that, The AI ​​trust assessment engine calculates a comprehensive trust score, as follows: The feature vector is input into a pre-trained deep learning model to obtain a direct confidence output, which is a normalized value between 0 and 1. Based on the direct trust score output, the overall trust score is calculated as follows: ， in, For at any time The overall trust score, For at any time The overall trust score, For time intervals, The trust decay rate coefficient, , It is a moment The input feature vector, For deep learning models based on feature vectors The calculated direct trust level output, Historical trust weighting factor, , The penalty coefficient is... , The dynamic penalty value for a vehicle whose malicious behavior is proven by reliable evidence. The incentive coefficient for trust growth. , For a moment Positive behavioral incentive value; The calculation is as follows: , in, As the incentive benchmark coefficient, , For a moment The number of compliant actions within the previous sliding window. For a moment Total number of actions within the previous sliding window It is a factor of behavioral importance.

4. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 3, characterized in that, The deep learning model uses a hybrid dataset containing massive amounts of historical vehicle normal behavior data, simulated attack behavior data, and real malicious behavior data, and is acquired through a two-stage training method combining supervised learning and semi-supervised learning.

5. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 3, characterized in that, The dynamic penalty value is quantitatively calculated based on the type of malicious behavior, severity level, and scope of impact, as shown below: ， in, The number of confirmed malicious behavior types. For the first Weighting coefficients for malicious behaviors No. The severity score of malicious behavior. For the first The impact range coefficient of malicious behavior.

6. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 3, characterized in that, The data processing procedure of the blockchain storage and sharing module is as follows: The relevant data that needs to be stored on the blockchain is combined into a structured transaction. The details are as follows: ， in, Evidence of key behavioral events cryptographic hash value, For the time the transaction was generated, It serves as a unique identifier for the vehicle; Transaction data After its validity is verified through the consensus mechanism, it is included in the candidate block; During block construction, the Merkle root of a block is linked to the hash value of the previous block to form an immutable chain structure, represented as: ， in, For the current new block, It is the cryptographic hash of all the data in the block header of the previous block. It's the version number. It is the block height. It's a timestamp. Indicates the first This transaction, This indicates that all blocks within this region will be packaged together. Each transaction is ultimately calculated into a unique root hash value using pairwise hashing. It is a random number. It is the verifier's signature.

7. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 6, characterized in that, The smart contract authentication decision module encodes and deploys smart contracts in the distributed ledger of the blockchain network. The smart contract defines rules for mapping the dynamic trust score calculated in real time by the vehicle to authentication strategies of different trust levels. When the preset authentication triggering conditions are met, the smart contract is triggered to automatically execute the differentiated authentication and dynamic allocation strategy of network resources based on the trust level. The specific strategy for differentiated authentication and dynamic allocation of network resources based on trust levels is as follows: The smart contract queries the blockchain for the latest comprehensive trust score corresponding to the vehicle's unique identifier ID. And obtain the sequence of all historical comprehensive trust scores of the vehicle within the specified sliding historical time window. , The sequence length; The smart contract performs a trust assessment based on a pre-defined weighted average strategy, calculating the dynamic trust level as follows: ， in, For dynamic trust levels, It is the weighted average of all historical comprehensive trust scores of the vehicle within a specified sliding historical time window. and Let be the weighting coefficient, satisfying ,and , The standard deviation of the historical overall trust score. The recent trend value of trust changes is denoted as... , This is the recent window length; Smart contracts will dynamically determine trust levels. Mapped to one of three predefined discrete trust levels and triggering the corresponding authentication policy; Based on the trust level, the corresponding authentication and network resource control policies are automatically executed.

8. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 7, characterized in that, The smart contract will dynamically assign trust levels. Mapped to one of three predefined discrete trust levels, including: like This indicates a high level of trust. like If so, it is a medium trust level; like If so, it is considered a low trust level.

9. The blockchain-based cross-regional authentication system for vehicle networking based on an AI dynamic trust model according to claim 8, characterized in that, The automatic execution of authentication and network resource control policies corresponding to the trust level includes: For vehicles with high trust levels, fast-track authentication is implemented, including: eliminating the handshake process during authentication, prioritizing the forwarding of sent messages, and granting them priority access to network bandwidth resources; For vehicles with a medium trust level, a standard certification process is implemented. For vehicles with low trust levels, enhanced authentication is implemented, including: performing multi-factor authentication and certificate chain verification; allocating a sandboxed network channel to the low-trust-level vehicle, to which all communication data is redirected; restricting or prohibiting the low-trust-level vehicle from sending broadcast messages and prohibiting it from acting as a relay node to forward data for other vehicles; temporarily placing the low-trust-level vehicle on a watch list, and if malicious behavior is confirmed, immediately terminating network access, adding the low-trust-level vehicle's digital certificate to the revocation list, and permanently recording the malicious behavior event on the blockchain.

10. A blockchain-based cross-regional authentication method for connected vehicles based on an AI dynamic trust model, characterized in that, The implementation of the blockchain-based vehicle network cross-regional authentication system based on the AI ​​dynamic trust model as described in claim 1, the method comprising: The system collects multidimensional data reflecting vehicle behavior characteristics in real time. This multidimensional data includes basic message characteristics, message content credibility, network interaction behavior, and historical behavior records. The basic message characteristics include message sending timestamp, message type identifier, unique identifier (ID), message byte length, and message sending cycle threshold. The message content credibility includes message checksum, continuous checksum values ​​for message sequence number, and semantic rationality score of message content. The network interaction behavior includes message sending frequency, message interval time statistics, message response latency, and network load rate. The historical behavior records include historical message frequency patterns based on time windows, frequency of abnormal behavior, and baseline characteristics of normal behavior. The collected multidimensional data is preprocessed and feature vectors are constructed; Using the feature vector as input, a pre-trained deep learning model is used to evaluate the dynamic behavior of the vehicle and calculate a comprehensive trust score. The comprehensive trust score, key behavioral evidence hash, and assessment process parameters calculated by the AI ​​trust assessment engine are stored on the blockchain in a chain. Intelligent decision-making is made based on trusted on-chain data to enable cross-regional vehicle authentication, and a differentiated authentication and dynamic allocation strategy for network resources based on trust level is implemented.

Images