Unmanned vehicle group coordination model construction method based on blockchain security reputation management

Abstract

The application belongs to the field of unmanned driving, and proposes a method for constructing an unmanned vehicle group coordination model based on blockchain security reputation management, comprising: step 1. Related definitions; step 2. Unmanned vehicle group dynamic evolution method based on blockchain security reputation management; step 3. Safety coordination model of unmanned vehicle group. The application proposes a reputation proof consensus mechanism, aiming to avoid the centralization of reputation weight; proposes a distributed vehicle group dynamic evolution algorithm based on vehicle comprehensive safety reputation evaluation; combines state conversion and event triggering conditions to construct an unmanned vehicle group safety coordination model, and analyzes the properties and safety of the model. The application improves the overall safety of the vehicle group, ensures that the unmanned vehicle group can effectively resist the invasion of malicious nodes, and provides effective safety protection for the stable and orderly driving of unmanned vehicle movement. It has important significance and application value for promoting the healthy and rapid development of unmanned driving.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving, specifically to a method for constructing a collaborative model for autonomous vehicle groups based on blockchain-based security and reputation management. Background Technology

[0002] (1) Current Status of Blockchain Research in the Security Field

[0003] Information security refers to the protection of computer systems, network systems, and the data contained therein from unauthorized access, use, disclosure, damage, or tampering through various technical and management measures. With the development of blockchain technology, it has wide applications in the field of information security, especially in identity authentication, access control, data protection, and encryption.

[0004] Identity authentication is the process of confirming a user's identity to ensure its legitimacy and authenticity. Access control, based on identity authentication, controls and restricts access to system resources to protect sensitive information from unauthorized access or use. Hkeem et al. designed a vehicle-to-everything (V2X) identity authentication scheme that combines biometrics, tamper-proof technology, and dual authentication, ensuring information security through hash chains and message signatures. Jiang et al. proposed a traceable personal information blockchain registration method that protects user privacy while ensuring the non-repudiation of transaction records. Gu et al. developed a grid signature mechanism that protects privacy by obfuscating user identities into similar information groups; this scheme improves the efficiency of signature generation. Li et al. used ring signatures to protect user privacy on the blockchain but did not optimize system performance or defend against attacks. Wang et al. proposed a blockchain incentive model that uses k-anonymity technology to protect data privacy and further protects user identity through ring signatures. Cui et al. proposed a multi-WSN identity authentication scheme that utilizes a hierarchical IoT node and hybrid blockchain model to protect identity.

[0005] Data protection technologies are dedicated to ensuring the confidentiality and integrity of data. Confidentiality guarantees that data cannot be accessed by unauthorized persons; while integrity ensures the authenticity, validity, and tamper-free nature of the data. Currently, many fields are exploring how to use blockchain technology to protect data security and privacy, proposing various blockchain-based data encryption and storage schemes, including using smart contracts to implement data encryption and storing data in shards on the blockchain network to improve data security.

[0006] Although blockchain has been the subject of much research and development in the field of information security, it still cannot be directly applied to safety methods for autonomous vehicle swarms for the following reasons:

[0007] Blockchain consensus can easily lead to the Matthew effect: In blockchain consensus mechanisms, some consensus algorithms (such as proof-of-stake) may cause nodes with more computing resources to more easily obtain block rewards, thereby further increasing their computing resources and power, forming a positive feedback loop. This may result in a few nodes or entities controlling most of the network's computing power and decision-making power, while other nodes gradually lose the opportunity to participate in decision-making, thus exacerbating the Matthew effect.

[0008] (2) Method for constructing autonomous vehicle swarms based on safety reputation assessment

[0009] The patent application filed by Cheng Jiujun et al. on October 20, 2023, entitled "A Method for Constructing an Autonomous Vehicle Swarm Based on Security Reputation Evaluation" (Applicant: Tongji University, Patent Application No.: 202410025918.5), proposes a reputation evaluation utility to measure the security of a distributed autonomous vehicle swarm. It also proposes a method for constructing a basic node set and a core node chain based on the state of autonomous vehicle nodes. Furthermore, it proposes an autonomous vehicle swarm formation algorithm based on reputation evaluation utility (REUCCF) using blockchain for identity authentication. The application provides evaluation indicators such as vehicle swarm reputation evaluation utility, reputation evaluation credibility, vehicle swarm survival time, and trusted node survival time. Simulation experiments were conducted to verify the security of the proposed autonomous vehicle swarm, improving the overall security of the autonomous vehicle swarm and enabling trusted vehicle nodes to perform identity authentication, thus accelerating the response speed of identity verification. Summary of the Invention

[0010] To address the vulnerabilities of existing vehicle swarms to single-point attacks and the Matthew effect inherent in blockchain consensus mechanisms, this invention proposes a method for constructing a collaborative model for autonomous vehicle swarms based on blockchain-based secure reputation management. This method proposes a reputation proof consensus mechanism to avoid the centralization of reputation weights; it also proposes a distributed dynamic evolution algorithm for vehicle swarms based on comprehensive vehicle safety reputation assessment; and by combining state transitions and event triggering conditions, a secure collaborative model for autonomous vehicle swarms is constructed, and its properties and security are thoroughly analyzed and proven. Through simulation experiments, using evaluation metrics such as vehicle swarm uniformity, vehicle swarm security reputation assessment utility, security reputation assessment credibility, vehicle swarm survival time, and trusted node survival time, the proposed secure autonomous vehicle swarm collaborative model is validated in low-speed, medium-speed, and high-speed highway scenarios. This demonstrates improved overall vehicle swarm security and ensures that the autonomous vehicle swarm effectively resists intrusions by malicious nodes.

[0011] The technical solution of the present invention:

[0012] A method for constructing a collaborative model for autonomous vehicle swarms based on blockchain-based security and reputation management is characterized by the following steps:

[0013] Step 1. Relevant Definitions;

[0014] Step 2. A dynamic evolution method for autonomous vehicle swarms based on reputation management; including:

[0015] Step 2.1 Reputation Management;

[0016] Step 2.2 Dynamic Evolution Events;

[0017] Step 2.3 Vehicle group topology optimization;

[0018] Step 2.4 Autonomous vehicle swarm evolution algorithm.

[0019] Step 3: Construction of a safety collaboration model for the autonomous vehicle swarm; including:

[0020] Step 3.1 Vehicle group status transition;

[0021] Step 3.2 Security Collaboration Model.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] This invention discloses a method for constructing a collaborative model for autonomous vehicle swarms based on blockchain-based security and reputation management. This method provides effective safety assurance for the intelligent movement behavior of autonomous vehicle swarms, helps to address potential traffic hazards, avoid collisions, and improve road traffic efficiency, thereby reducing the risk of traffic accidents. It has significant practical application value for promoting the healthy and rapid development of autonomous driving technology. Attached Figure Description

[0024] Figure 1 Reputation proof consensus mechanism;

[0025] Figure 2 The process of vehicle group merging algorithm;

[0026] Figure 3 The process of the vehicle group splitting algorithm;

[0027] Figure 4 The process of vehicle group topology optimization algorithm;

[0028] Figure 5 State transition diagram of the security collaboration model;

[0029] Figure 6 SUMO simulation software and simulation scene diagram;

[0030] Figure 7 Vehicle group uniformity as a function of time (maximum vehicle speed 65 km / h);

[0031] Figure 8 Vehicle group uniformity variation curve over time (maximum vehicle speed 75km / h);

[0032] Figure 9 Vehicle group uniformity as a function of time (maximum vehicle speed 85 km / h);

[0033] Figure 10 Vehicle uniformity over time (maximum vehicle speed 95 km / h);

[0034] Figure 11 Vehicle uniformity over time (maximum vehicle speed 105 km / h);

[0035] Figure 12 Vehicle uniformity over time (maximum vehicle speed 115 km / h);

[0036] Figure 13 The curve showing the change in the safety reputation assessment utility of a vehicle group over time (maximum vehicle speed 70km / h);

[0037] Figure 14 The curve showing the change in the safety reputation assessment utility of a vehicle group over time (maximum vehicle speed is 90 km / h);

[0038] Figure 15 The curve showing the change in the safety reputation assessment utility of a vehicle group over time (maximum vehicle speed is 110 km / h);

[0039] Figure 16 The credibility of security reputation assessment.

[0040] Figure 17 Lifetime of trusted nodes;

[0041] Figure 18 Trusted node stability duration;

[0042] Figure 19 Vehicle platoon survival time;

[0043] Figure 20 A flowchart of an embodiment of the present invention. Detailed Implementation

[0044] The technical solutions provided in this application will be further described below with reference to specific embodiments and accompanying drawings. The advantages and features of this application will become clearer from the following description.

[0045] The concepts and specific algorithms related to the safety reputation assessment utility of autonomous vehicle swarms and the formation of autonomous vehicle swarms can be found in the patent application filed by Cheng Jiujun et al. on October 20, 2023, entitled "Method for Constructing Autonomous Vehicle Swarms Based on Safety Reputation Assessment" (Applicant: Tongji University, Patent Application No.: 202410025918.5).

[0046] This invention specifically includes the following four aspects:

[0047] Step 1. Relevant Definitions

[0048] Step 2. Dynamic Evolution Method of Autonomous Vehicle Flock Based on Safety Reputation Management

[0049] Step 2.1 Reputation Management

[0050] Step 2.2 Dynamic Evolution Events

[0051] Step 2.3 Vehicle Group Topology Optimization

[0052] Step 2.4 Autonomous Vehicle Swarm Evolution Algorithm

[0053] Step 3: Safety Collaboration Model for Autonomous Vehicle Swarms

[0054] Step 3.1 Vehicle group status transition

[0055] Step 3.2 Security Collaboration Model

[0056] Step 3.3 Related Properties

[0057] Step 3.4 Safety and Reasonableness Proof

[0058] Step 4. Simulation Experiment and Result Analysis

[0059] Step 4.1 Simulation Experiment Data and Methods

[0060] Step 4.2 Simulation Experiment Results and Analysis

[0061] Details are as follows:

[0062] The symbols used in this invention are explained in Table 1.

[0063] Table 1

[0064]

[0065]

[0066] Step 1 Security Trust Assessment

[0067] Definition 1: Vehicle Formation Event (Formation) A vehicle corps forms when autonomous vehicle nodes establish connections with each other and conduct safety and reputation assessments through cooperative events. The conditions for a vehicle corps formation event are:

[0068]

[0069] Where G represents the vehicle group; v represents the vehicle node; N V Represents the set of neighboring nodes; σ represents the role and function of the autonomous vehicle, σ=1 indicates that the autonomous vehicle has not joined any vehicle group and is a detached node, σ=2 indicates that the autonomous vehicle is in the basic node set of a vehicle group, and σ=3 indicates that the vehicle node is a node in the core node chain of a vehicle group; REU i，j It is node v i For node v j The direct credit assessment utility; REU th It is the direct reputation assessment utility threshold used for security reputation assessment.

[0070] After the basic node set is formed, the autonomous vehicle swarm will further construct a core node chain to ensure the security of the vehicle swarm's identity authentication information. The conditions for its construction are as follows:

[0071]

[0072] Where BSN represents the basic node set; τ represents the threshold number of nodes; Represents node v i The overall credit assessment utility; CREU th It is used to determine node v j Whether it is a trusted node or not. Trust threshold.

[0073] Define 2 Node Join Event: When a free autonomous vehicle node encounters an autonomous vehicle group with a core node chain during its journey, a node join event will occur if it has the same motion similarity as its neighboring nodes in the group and meets the safety reputation assessment requirements for joining the group.

[0074]

[0075] Define a 3-node leave event: A vehicle in an autonomous vehicle swarm may leave the swarm, either voluntarily or involuntarily, for various reasons during operation. If a vehicle node's safety reputation assessment utility or weight is too low, it will be expelled from the swarm. If a vehicle node cannot maintain stable communication with the swarm due to external interference or a different path, it will voluntarily leave the swarm. The conditions for a node leave event are:

[0076]

[0077] Among them, W j Represents vehicle node v j The credit rating weight.

[0078] Definition 4: Vehicle group merging event (Merge): When two autonomous vehicle groups meet during their journey, if the distance between vehicle nodes in different groups is within the communication range and their speeds are similar, a communication connection is established, and a safety and reputation assessment is further conducted. If the conditions are met, they are merged into a new vehicle group.

[0079]

[0080] Definition 5: Split event: If the communication distance between nodes in an autonomous vehicle group is too far, the speed difference is too large, or the credit assessment utility between them is too low, a split event will occur, and other nodes in the vehicle group will join different vehicle groups according to the conditions.

[0081]

[0082] Among them, Dis th Sim represents the maximum distance between two vehicle nodes; th This represents the threshold for the speed similarity of nodes.

[0083] Definition 6: Topology Optimization Event (Adjust): During the dynamic evolution process, autonomous vehicle swarms frequently undergo changes in their network topology, which can easily lead to unreasonable topologies that cause the distributed structure of the vehicle swarm to become centralized. Therefore, topology optimization is needed to strengthen the connections between neighboring nodes.

[0084]

[0085] Among them, e ij Represents node v of the autonomous vehicle swarm i With v j There is a communication connection between them. B(v) j ) represents node v j Betweenness centrality is a measure of graph centrality in a network based on the shortest path, and it is calculated as follows:

[0086]

[0087] Where, N jk It is v i to v j The number of shortest paths, N jk (v i These paths pass through v i The number of times. The higher the betweenness centrality, the more concentrated the vehicle group is at that point.

[0088] Definition 7: Disappearance of a vehicle swarm: When the number of vehicle nodes in an autonomous vehicle swarm is small and they cannot communicate with other autonomous vehicles, the swarm structure will disintegrate.

[0089]

[0090] Step 2: Reputation Management-Based Dynamic Evolution Method for Autonomous Vehicle Swarms

[0091] Step 2.1 Reputation Management

[0092] Maintaining the safety and reputation assessment effectiveness of an autonomous vehicle corps requires continuous comprehensive safety and reputation assessments of all nodes within the corps. Malicious nodes that fail to meet the assessment criteria are removed from the corps to ensure its security. This invention proposes a blockchain consensus mechanism based on proof-of-reputation, using safety and reputation assessment weights as transaction tokens for the distributed management of the corps' safety and reputation assessment effectiveness. The proof-of-reputation consensus process is as follows: Figure 1 As shown. The calculation method for transaction token rewards is as follows:

[0093]

[0094] Where T0 is the reputation error threshold, neighboring nodes whose security reputation assessment error exceeds this threshold will have their reputation tokens deducted, that is, their security reputation assessment weight will be reduced; γ>0 is the reward decay factor, which controls the nodes with more reputation tokens to receive relatively less reward, thus avoiding reputation centralization.

[0095] The method for handling safety reputation assessment events of autonomous vehicle convoys is shown in Algorithm 1, and the specific steps are as follows:

[0096] (1) Neighboring nodes communicate with node v j The success rate of collaborative events leads to a direct assessment of security and reputation utility (lines 1-3).

[0097] (2) Based on the reputation verification mechanism, according to the security reputation assessment weight W of different nodes. i The final comprehensive security reputation assessment utility of the node is calculated. Nodes whose comprehensive security reputation assessment utility reaches the threshold and are not in the core node set will be added to the core node set (lines 4-11).

[0098] (3) The consensus mechanism will issue reputation token rewards to neighboring nodes that participate in the computation, which increases the security reputation assessment weight of the node, and record the result as a transaction in the blockchain (lines 12-15).

[0099] (4) Nodes with insufficient reputation tokens or low overall security reputation assessment utility are considered malicious nodes that affect the security of the vehicle group and will be expelled from the vehicle group. The information of such nodes will be recorded in the Certificate Revocation List (CRL) and they will be refused rejoining the vehicle group (lines 16-21).

[0100]

[0101]

[0102] Step 2.2 Dynamic Evolution Events

[0103] Step 2.2.1 Adding a node

[0104] When a lone autonomous vehicle node encounters a convoy of vehicles, it will proactively send a node joining request. As the convoy continuously receives cooperation event information from lone nodes, it updates its security and reputation assessment, leading to node joining events. New nodes joining the convoy's basic node set obtain identity information through authentication on the convoy's core node chain. This authentication information is stored in the blockchain, and the immutability of transaction blocks ensures the security of node authentication.

[0105] The method for handling the joining event of autonomous vehicle swarm nodes is shown in Algorithm 2, and the specific steps are as follows:

[0106] (1) Free node v j Towards the car group G k Node v inside i The sending node adds the request REQ_ADD and its own status information (line 1).

[0107] (2) Node v i After receiving a node join request, calculate the relative distance Dis between the two nodes. i，j and speed similarity Sim i，j It then checks whether the certificate is on the certificate revocation list. If it does not meet the requirements, it sends a request to node v. j Send a request rejection message RESP_REF (lines 2-3).

[0108] (3) If the relative motion between the two nodes meets the requirements, then further calculate node v. i For v j The direct security reputation assessment utility of REU i，j If the security credit assessment utility is below the threshold REU th If so, the node is rejected from joining the vehicle group (line 4).

[0109] (4) When all the above conditions are met, send a signal to node v. j Send the RESP_AGREE message to agree to join. Node v j This generates a public key pair (PK).user and private key SK user It is used for secure communication between nodes and transmits its own user information (userInfo) and public key (PK). user Send to vehicle group node v i It is then further forwarded to node v, which is the core node of the vehicle group. p (Lines 5-9)

[0110] (5) Node v p The received user information is encrypted to generate the distributed identity (DID) of the newly joined node. The node's distributed ID, encrypted identity information, node public key, and timestamp are recorded as a transaction ledger for storage in the blockchain (lines 10-16).

[0111] (6) Return the updated information (lines 17-21).

[0112]

[0113] Step 2.2.2 Vehicle group merging

[0114] When two groups of autonomous vehicles meet, if all neighboring nodes satisfy the conditions of motion similarity and direct safety reputation assessment utility reaching a threshold, then the safety reputation assessment utility of the merged group will not be lower than the threshold, and the group can be merged to maximize the group size and share vehicle node information while ensuring group safety. The merging of identity authentication records and certificate revocation lists in the group merging algorithm will differ depending on whether a core node chain exists in the group or not.

[0115] The handling method for vehicle group merging events is as follows: Figure 2 As shown in Algorithm 3, the specific steps are as follows:

[0116] (1) Group of vehicles G m Node v inside i With car group G n Node v inside j Meeting, node v i First, calculate the vehicle group G within the communication range. n If the relative status information of the node meets the requirements, a vehicle group merging request REQ_MERGE and its own status information (lines 1-7) are sent to it.

[0117] (2) Node v j Upon receiving a node join request, the relative state information between the two nodes is calculated. If the requirements are not met, a request is sent to node v. i Send a rejection message RESP_REF for the merge request (lines 8-10).

[0118] (3) If the two nodes are within the range of neighboring nodes, then the direct safety reputation assessment utility between the two nodes is further calculated. If the direct safety reputation assessment utility of any one of the neighboring nodes of the two vehicle groups meeting is lower than the threshold REU, then the direct safety reputation assessment utility between the two nodes is calculated. th If so, then the vehicle group merging is rejected (lines 11-14).

[0119] (4) When the direct security reputation assessment utility among all nodes meets the conditions, send a signal to node v. i Send the RESP_AGREE message to agree to join. The two nodes send each other the reputation tokens held by each vehicle node in their respective vehicle group, clear historical data records, and use the total number of reputation tokens as the new starting block (lines 15-17).

[0120] (5) If neither vehicle group has a core node chain, the algorithm ends; if one vehicle group has a core node chain and the other does not, the vehicle group with the core node chain will authenticate the nodes of the other vehicle group and store them in the blockchain; if both vehicle groups have core node chains, the authentication records of the two vehicle groups are merged as a new blockchain ledger, and the certificate revocation lists (CRLs) of the two vehicle groups are merged (lines 18-28).

[0121]

[0122]

[0123] Step 2.2.3 Vehicle Group Splitting

[0124] Some dynamic evolutionary behaviors of autonomous vehicle swarms are determined by the states of vehicle nodes within a local area. It's possible for a vehicle node to be relatively stable relative to the overall state of its multiple neighboring nodes, but the connections between two nodes may be weak. If both nodes are in the core node chain, the swarm as a whole remains unstable or unsafe. In this case, the swarm should be split into smaller swarms in a timely manner. The swarm splitting algorithm's processing is as follows: Figure 3 As shown.

[0125] The method for handling vehicle group splitting events is shown in Algorithm 4, and the specific steps are as follows:

[0126] (1) Node v i Node v was found in another core node chain. j If the relative node state is unstable, it broadcasts a vehicle group split request REQ_SPLIT and v to other nodes in the vehicle group. i With v j The node status information (lines 1-6).

[0127] (2) Nodes in the vehicle group are arranged according to their distance from node v i v jSort the network distance (i.e., hop count) and determine the groups of vehicles to be added to the two split nodes in order from nearest to farthest (lines 7-9).

[0128] (3) If the node to be assigned has a direct or indirect connection with both nodes and the network connection distance is equal, then compare the relative state of the node with the directly connected nodes in the two split vehicle groups, and add the node to the vehicle group where the relative node state is closer (lines 10-14).

[0129] (4) If the node to be assigned is reachable from only one of the nodes, it is directly added to the split vehicle group where that node is located, and its communication connection is maintained (lines 15-20).

[0130] (5) When all nodes in the vehicle group have selected the split vehicle group to join, the nodes will broadcast and update their own status information within the vehicle group. The two vehicle groups will update their certificate revocation lists respectively, and the vehicle group split is completed (lines 21-22).

[0131]

[0132] Step 2.3 Vehicle Group Topology Optimization

[0133] After dynamic evolution events such as node addition or departure, the topology of an autonomous vehicle swarm may become either too concentrated or too dispersed. When the swarm is dispersed, the edge connectivity may be 1, meaning a node has only one connection to every other node. When the swarm is concentrated, if multiple communication paths require passing through a particular node, that node's betweenness centrality is too high. In both cases, without topology optimization, the distributed structure will tend towards centralization, reducing swarm stability. Therefore, when a node in the autonomous vehicle swarm detects an unreasonable network structure, a swarm topology optimization algorithm needs to be executed. The execution process of the swarm topology optimization algorithm in both cases is as follows: Figure 4 As shown.

[0134] The method for handling vehicle group topology optimization events is shown in Algorithm 5, and the specific steps are as follows:

[0135] (1) If node v i Only with another node v in the vehicle group j To communicate, v i Check if there are other neighboring nodes that meet the trusted connection requirements (lines 1-2).

[0136] (2) If there exists another node v k If so, a connection establishment request REQ_LINK is sent to it (lines 3-4).

[0137] (3)v kAfter calculating that the direct security reputation assessment utility meets the conditions, a connection establishment agreement message RESP_AGREE is sent to increase network stability (lines 5-8).

[0138] (4) If node v i Discover neighbor node v j It is a central node, and it is necessary to reduce the centralization of the vehicle group and establish connections with other nodes as much as possible, such as steps (2)(3)(lines 9-11).

[0139]

[0140] Step 2.4 Autonomous Vehicle Swarm Evolution Algorithm

[0141] Combining reputation management and different event handling algorithms 1-5, the complete dynamic evolution algorithm for the autonomous vehicle swarm is obtained as shown in Algorithm 6. The specific steps are as follows:

[0142] (1) The vehicle group periodically executes the reputation management algorithm to obtain the safety reputation assessment utility of each node (line 1).

[0143] (2) Based on the different triggering conditions met by the vehicle group during operation, execute the corresponding node addition, node departure, vehicle group merging or vehicle group splitting algorithm (lines 2-10).

[0144] (3) After a dynamic evolution event occurs in the vehicle group, the vehicle group topology optimization algorithm should be executed to optimize the unreasonable vehicle group topology structure (lines 11-12).

[0145]

[0146]

[0147] Step 3: Autonomous Vehicle Swarm Collaboration Model Based on Blockchain Security and Reputation Management

[0148] Step 3.1 Vehicle group status transition

[0149] To maintain the safety and stability of autonomous vehicle swarms during operation, different dynamic evolution behaviors are proposed. When the swarm meets the event triggering conditions, a corresponding event occurs, and the swarm's state transitions accordingly. For example, when a new node joins, the swarm's state changes from a stable state to an expanding state. At this point, the swarm maintains a safe and stable state through a dynamic evolution algorithm. The following section introduces the definition of the overall attribute set of the swarm and the transitions between different states.

[0150] Define 8. Vehicle group attribute set P: The set of all attributes involved in the safe collaborative model of autonomous vehicle groups, which can be represented as the following six-tuple:

[0151] P =<V,E,F,A,S,BC> (11)

[0152] Where, V={v i |i=1,2,…,n} represents the set of all autonomous vehicle nodes, which can further form the basic node set BNS and the core node chain CNC; E={E i |i = 1, 2, ..., n}, represents the set of trusted connections between all vehicle nodes; F = {f i |i=1,2,…,8} represents the set of evolutionary events of the vehicle group, mainly including eight events: vehicle group formation f1, reputation management f2, node joining f3, node leaving f4, topology optimization f5, vehicle group merging f6, vehicle group splitting f7, and vehicle group extinction f8. Evolutionary events will lead to transitions between vehicle group states; A={a i |i=1,2,…,6} represents the set of dynamic evolution algorithms for the vehicle group, including safety and reputation management algorithm a1, node joining algorithm a2, node leaving algorithm a3, vehicle group merging algorithm a4, vehicle group splitting algorithm a5, and vehicle group topology optimization algorithm a6; S={S i |i=0,1,…,8} represents the set of vehicle group states, including initial state S0, safe and cooperative state S1, unsafe state S2, expansion state S3, shrinkage state S4, topology optimization state S5, merger state S6, split state S7, and extinction state S8; BC represents the vehicle group blockchain, which is used to store the identity authentication information of autonomous vehicle nodes and the utility of security reputation assessment.

[0153] (1) Initial state S0: At time t, there are at least two free nodes, such as v i v j Initially, a basic node set BNS is constructed using a vehicle group formation algorithm, and the newly formed vehicle group G is at this point. k In the initial state, the number of nodes has not yet reached the threshold τ. The set of vehicle group attributes is as follows:

[0154]

[0155] (2) Safe Cooperative State S1: At time t, the initial state of the vehicle group G k The core node chain (CNC) is constructed through a vehicle corps formation algorithm, or the vehicle corps in a safe and cooperative state is restored to a safe and cooperative state through other dynamic evolution algorithms. At this time, the attribute set of the vehicle corps is as follows:

[0156]

[0157] Among them, ACREU represents the utility of vehicle group reputation assessment;

[0158] (3) Unsafe state S2: At time t, the vehicle group G kA reputation management algorithm was used to detect a malicious node v within the vehicle group. i At this point, the attributes of the vehicle group are as follows:

[0159]

[0160] (4) Expansion state S3: At time t, node v i Apply to join car group G k The attributes of the vehicle group are as follows:

[0161]

[0162] (5) Reduced state S4: At time t, node v i Actively or passively leaving the vehicle group G k The attributes of the vehicle group are as follows:

[0163]

[0164] (6) Topology optimization state S5: At time t, the vehicle group G k Middle node v i The vehicle group topology was found to be unreasonable, and its relationship with neighboring node v was also found to be unreasonable. j Establishing more connections increases the utility of vehicle group safety and reputation assessment. The vehicle group's attribute set is as follows:

[0165]

[0166] (7) Merged state S6: At time t, the vehicle group G m G n They met and merged to form the new car group G. k The size of the vehicle group has increased. The attributes of the vehicle group are as follows:

[0167]

[0168] Among them, E mn Indicates vehicle group G m G n A newly established trusted connection between them.

[0169] (8) Split state S7: At time t, the vehicle group G k The vehicle group may split into multiple new vehicle groups, either actively or passively, such as vehicle group G. m G n The attributes of the vehicle group are as follows:

[0170]

[0171] (9) Demise state S8: At time t, node v i Actively or passively leaving the vehicle group G kAt this point, the attributes of the vehicle group are as follows:

[0172]

[0173] Step 3.2 Security Collaboration Model

[0174] Based on the dynamic evolution method of the autonomous vehicle swarm given in step 2, and the changes in the attribute set of the vehicle swarm state transitions mentioned above, a safe cooperative model for the autonomous vehicle swarm can be obtained, as follows:

[0175]

[0176] in, This represents the combination of different attributes of a vehicle group. The cooperative model consists of a series of changes caused by the vehicle group state and safety reputation assessment, with different vehicle group states S. t-1 and the utility of security reputation assessment between nodes REU t This will change the attributes of the vehicle group, leading to vehicle group evolution event f. t When the trigger condition is met, the vehicle group attributes are updated to temporary state S. t Execute the corresponding dynamic evolution algorithm a t The car group eventually maintained the new state S. t+1 Vehicle Group Blockchain BC t+1 Corresponding changes may also occur.

[0177] The state transition diagram of the autonomous vehicle swarm safety collaboration model is as follows: Figure 5 As shown:

[0178] Step 3.3 Related Properties

[0179] The safety properties of the autonomous vehicle swarm cooperative model are: conditional privacy, data integrity, message reliability, and robustness. Specifically:

[0180] (1) Conditional Privacy: The core node chain provides identity authentication for vehicle nodes, generates distributed identity (DID) for vehicles, and stores the real identities of all vehicles, thus providing conditional privacy protection for vehicles. When the core node chain exists in the vehicle group, information in the vehicle group will be transmitted in encryption to protect the privacy information of vehicle nodes. If a malicious node is detected in the vehicle group through security reputation assessment, the core node chain can track the vehicle's identity based on the recorded identity information correspondence, add it to the Certificate Revocation List (CRL), and refuse its rejoining to the vehicle group based on the identity information.

[0181] (2) Data Integrity: The security reputation assessment utility of the vehicle group and the encrypted identity information of the vehicles will be stored through the blockchain, which can effectively ensure the integrity of the data and prevent the data from being tampered with. The blockchain records the digital digest of all underlying data through Merkle trees, and any minor modification can be quickly detected. Therefore, it is difficult for attackers to attack the vehicle group by modifying historical data.

[0182] (3) Message reliability: The vehicle group formation algorithm and the dynamic evolution algorithm are combined with a reputation management system, which constrains the reputation conditions of vehicle nodes to ensure the reliability of vehicle messages. Only vehicle nodes whose comprehensive security reputation assessment utility reaches the threshold can exist stably in the vehicle group and communicate with other members. At the same time, the vehicles that receive the messages can also judge the authenticity of the messages through the security reputation assessment utility of the message sender.

[0183] (4) Robustness: Robustness refers to the ability of the vehicle corps to maintain safe operation through dynamic evolution when facing various attack methods. When facing a single point of attack, the autonomous vehicle corps can identify and eliminate malicious nodes through security reputation assessment. Even if multiple malicious nodes exist, because the initial security reputation assessment weight of each node is low, a large error in the security reputation assessment will lead to the rapid detection and removal of malicious nodes. If a malicious node has already infiltrated the core node chain, surrounding nodes can also initiate a corps split to reduce the attacks received by the corps.

[0184] Step 3.4 Safety and Reasonableness Proof

[0185] Since semi-centralized vehicle groups are vulnerable to single-point attacks, and considering that the security collaboration model can avoid single-point failures and that the dynamic evolution method can ensure the safety of the vehicle group under attack, the security collaboration model for vehicle groups is proven to be reasonable in the face of threats. Specifically:

[0186] Proposition 1: The safe collaborative model of autonomous vehicle swarms can prevent the distributed vehicle swarm from gradually becoming centralized, thereby resisting single-point attacks.

[0187] prove:

[0188] (1) When node v i Participating node v j During the comprehensive security and reputation assessment, node v i Reputation tokens are awarded to incentivize participation in security reputation assessments, as can be obtained from formula (10):

[0189]

[0190] If node v i The security reputation assessment error is small, i.e., |REU i，j (t)-CREUj (t)|<T0, at this time node v i Bonus (v) of reputation tokens earned i If γ > 0, and γ > 1, then the reputation token reward Bonus(v) will be 0. i Regarding the weighting of safety and reputation assessment W i The partial derivatives of node v are less than 0. i The reputation token rewards earned will be weighted according to their security reputation assessment weight W. i It decreases as it increases.

[0191] If node v i The security reputation assessment has a large error, i.e., |REU i，j (t)-CREU j (t)|>T0, at this time node v i Bonus (v) of reputation tokens earned i If γ < 0, and γ < 1, then the reputation token reward Bonus(v) is calculated. i Regarding the weighting of safety and reputation assessment W i The partial derivatives of node v are less than 0. i Safety and reputation assessment weight W i The larger the value, the greater the Bonus (v) reward in reputation tokens. i The smaller the value, the higher the safety and credit assessment weight W. i The more you lose.

[0192] Therefore, the weighting for security and credit assessment is set as follows:

[0193]

[0194] Safety and reputation assessment weight W i Larger nodes receive fewer token rewards for accurate assessments but are penalized more for inaccurate assessments, effectively avoiding the Matthew effect in blockchain.

[0195] (2) During the operation of the autonomous vehicle swarm, the network topology may gradually become concentrated. According to formula (7), when the edge connectivity K(G) of the vehicle swarm is 1 or the betweenness centrality B(v) of the nodes in the vehicle swarm is 1, the network topology may gradually become concentrated. j () greater than threshold B th When the vehicle group triggers a topology optimization event, the vehicle group state transitions to the topology optimization state and begins executing Algorithm 5. This increases trusted connections between nodes in the vehicle group, reduces the number of critical nodes, and the vehicle group returns to the safe collaborative state after completing the topology optimization algorithm.

[0196] In summary, the safe collaborative model for autonomous vehicle swarms ensures the network topology and distributed storage characteristics of the distributed vehicle swarm, prevents the centralization of the distributed vehicle swarm, reduces the harm of single-point attacks, and avoids single-point failure problems.

[0197] Proposition 2: The autonomous vehicle swarm collaboration model can ensure safe collaboration of the swarm when attacked.

[0198] prove:

[0199] (1) When there are malicious nodes in the basic node group, the reputation management event f2 is triggered, which can identify the malicious nodes in the vehicle group whose safety reputation assessment utility does not meet the conditions, and enter the reduction state S4. The malicious nodes will be expelled from the vehicle group according to the dynamic evolution algorithm, and the identity information of the node will be added to the certificate revocation list CRL. The vehicle group returns to the safe collaboration state S1.

[0200] (2) When there is a malicious node in the core node chain, event f2 is triggered. If another core node evaluates its security reputation too low, it enters the split state S7. The nodes in the vehicle group select the split vehicle group according to the communication connection distance and the security reputation evaluation utility. After the split is completed, the vehicle group returns to the security collaboration state S1. If the neighboring node identifies a malicious node, it enters the reduction state S4. The identity of the malicious node is traced according to the identity authentication information stored in other core nodes. After the malicious node is expelled, the vehicle group returns to the security collaboration state S1.

[0201] Therefore, in both of the above situations, the vehicle group safety collaboration model can ensure safe collaboration even when there are malicious nodes in the vehicle group.

[0202] Based on the above two propositions, the autonomous vehicle swarm safety collaboration model can resist single-point attacks and ensure the safe collaboration of the vehicle swarm when attacked, thus proving the safety and rationality of the vehicle swarm safety collaboration model.

[0203] Step 4: Simulation Experiment and Result Analysis

[0204] To verify the effectiveness of the autonomous vehicle swarm safety cooperation model, SUMO and NS3 were used to simulate the motion and communication behavior of the vehicle swarm, and the experimental data during the simulation process were statistically analyzed.

[0205] Step 4.1 Simulation Experiment Data and Evaluation Indicators

[0206] This invention uses NS3 (version 3.33) and SUMO to simulate vehicle motion and communication processes. NS3 is a network simulator used to simulate network communication between vehicle nodes. SUMO is an open-source, free discrete-time traffic system simulation software capable of simulating continuous, microscopic traffic flow, planning routes for each vehicle, and automatically determining driving behavior based on factors such as current speed, direction, maximum acceleration, and distance to the vehicle in front. This chapter simulates vehicle motion behavior using SUMO, and then inputs the SUMO vehicle motion information into NS3 to simulate network communication between vehicles.

[0207] like Figure 6 As shown, this invention uses SUMO to construct a 10km long, one-way, four-lane highway with an exit every 2km. 100 unmanned vehicles are injected into the road, including 80% trusted vehicle nodes and 20% malicious vehicle nodes. Vehicles randomly select an exit to leave the highway. A 200s interval is taken during the simulation, starting from when all vehicles enter the highway, to ensure that each vehicle passes at least one exit during its movement. To verify the effectiveness of the vehicle group formation and maintenance methods at different speeds, considering the minimum speed limit of 60km / h on the highway, the maximum speed of all vehicles is set between 65km / h and 120km / h, increasing in 5km / h increments. In the network simulation using NS3, vehicles communicate via the WAVE protocol, with a maximum communication range of 250m. During the experiment, the position and speed information of the vehicles are collected every 1s, and the specific parameter configurations are shown in Table 2.

[0208] Table 2

[0209]

[0210] To simulate a scenario where the central node is more vulnerable to network attacks and can be transformed into a malicious node, a mechanism for node network attack is added. All nodes in the vehicle group have a certain probability of being transformed into malicious nodes due to network attacks. Because key nodes in the vehicle group possess more information about other nodes, they are more susceptible to attacks. Let node v be set at time t. i The probability of a node becoming malicious due to a cyberattack is:

[0211]

[0212] As shown in equation (24), the higher the Comprehensive Security Reputation Evaluation Utility (CREU) of a vehicle, the lower the probability of it being successfully attacked; the betweenness centrality B(v) of the vehicle node i The higher the value, the closer it is to the center of the vehicle group, and the higher the probability of being successfully attacked, with a maximum probability of 100%.

[0213] To verify the performance of the RMCCM model, this invention uses the following metrics to conduct a comparative experiment with the SOCM vehicle group collaboration model:

[0214] (1) Vehicle group uniformity: Vehicle group uniformity measures the evenness of the distribution of nodes in a vehicle group. The formula is as follows:

[0215]

[0216] Where, d i Represents node v i The degree, that is, the distance between the vehicle group and node v i The number of connected edges, d max Indicates vehicle group G k The highest node degree in the vehicle group. The higher the vehicle group uniformity, the more uniform the topological distribution of the vehicle group, and the smaller the impact of a single point attack.

[0217] (2) Safety reputation assessment utility: The safety reputation assessment utility and the credibility of the safety reputation assessment utility reflect the overall safety and credibility of the vehicle group.

[0218] (3) Different survival time indicators: The survival time of the vehicle group reflects the overall stability of the vehicle group, while the survival time and stability time of trusted nodes reflect the trustworthiness and stability of the vehicle group in an unsafe environment.

[0219] Step 4.2 Simulation Experiment Results and Analysis

[0220] To cover various speed scenarios from low to high speeds, the simulation experiment uniformly selected representative speed scenarios within the speed limit range of 60km / h to 120km / h, tracked and recorded safety evaluation indicators such as vehicle group uniformity, vehicle group safety reputation assessment effectiveness, and vehicle group survival time, and conducted statistical analysis to evaluate the safety effect of the REMCCM vehicle group model.

[0221] (1) Security

[0222] Vehicle group uniformity reflects the evenness of the node topology distribution within a vehicle group. A higher vehicle group uniformity value indicates a more uniform distribution of nodes, higher consistency of node degrees, and a stronger ability of the vehicle group collaborative model to resist single-point attacks. When all nodes in a vehicle group have the same degree, the node distribution is completely uniform, there is no special central node, and the vehicle group uniformity is 1.

[0223] In the simulation experiment, the maximum speed of the vehicle group was set at 10 km / h intervals within the range of [65 km / h, 115 km / h] to comprehensively test the model performance under different vehicle motion scenarios. The uniformity of the vehicle group at each time point during the simulation was statistically analyzed to assess the uniformity of the vehicle group's distribution.

[0224] like Figure 7 As shown, the maximum speed limit for vehicles is 65 km / h. The red line represents the vehicle group uniformity of the RMCCM model, which gradually increases from the start of the simulation as the vehicle group forms and evolves, reaching a maximum value of 0.63 around 60 seconds, and then gradually decreasing. During the simulation, the vehicle group uniformity of the RMCCM model remained above 0.5 for most of the time. The green line represents the vehicle group uniformity of the SOCM model, which remained basically within the range of 0.45 to 0.55, lower than that of the RMCCM model. This indicates that the RMCCM model outperforms the SOCM method in terms of vehicle group uniformity.

[0225] like Figure 8 As shown, the maximum speed limit for vehicles is 75 km / h. The red line represents the RMCCM model, which has a higher vehicle group uniformity than the green line represents the SOCM model, and the trend is similar to that when the maximum speed is 65 km / h. In low-speed scenarios, the connections between vehicles change less, and the network topology of the vehicle group is relatively stable. The vehicle group uniformity of the two vehicle group safety cooperation models does not change drastically. The vehicle uniformity of the RMCCM model is consistently higher than that of the SOCM model, indicating that in low-speed scenarios, the RMCCM model is more conducive to maintaining the uniform distribution of the distributed vehicle group.

[0226] like Figure 9 As shown, the maximum speed limit for vehicles is 85 km / h. The red line represents the RMCCM model, whose vehicle group uniformity is higher than that of the green line (SOCM model) for the vast majority of the time. In the later stages of the simulation, as vehicles spend more time in the simulation environment, the distance between vehicle nodes increases accordingly, and the vehicle density decreases. At this point, the vehicle group uniformity of the RMCCM model decreases slightly, becoming similar to that of the SOCM model. This indicates that the performance of the RMCCM model decreases in low-density scenarios, becoming closer to that of the SOCM model.

[0227] like Figure 10 As shown, the maximum speed limit for vehicles is 95 km / h. Increased vehicle speed leads to faster changes in the vehicle group topology, and the dynamic evolution behavior has a greater impact on the uniformity of the distributed vehicle group. The vehicle group uniformity of the RMCCM model (represented by the red line) changes drastically in the first half of the simulation experiment, with oscillations exceeding 0.1. In the second half of the simulation experiment, the vehicle group structure tends to stabilize, gradually approaching the vehicle group uniformity of the SOCM model (represented by the green line). Figure 9 Experimental data shows that in medium-speed scenarios, the vehicle uniformity of the RMCCM model changes more drastically compared to low-speed scenarios. Furthermore, in the later stages of the simulation, as vehicle density decreases, the vehicle uniformity of the RMCCM model gradually decreases, narrowing the gap with the SOCM model. Overall, the vehicle uniformity level of the driving vehicle group is improved by 23.1%, maintaining a high degree of distributed characteristics in the vehicle group to resist single-point attacks.

[0228] The safety reputation assessment utility of a vehicle group is a quantitative indicator for measuring the overall safety of the vehicle group. The safety collaboration model simulation experiment selected three speed scenarios representing low speed, medium speed and high speed in the range of 60km / h to 120km / h, and tracked and recorded the changes in the safety reputation assessment utility of the vehicle group in real time.

[0229] like Figure 13 As shown, the maximum speed limit for vehicles is 70 km / h. The safety reputation assessment utility of the RMCCM model is above 0.9 most of the time, while the safety reputation assessment utility of the SOCM model is basically stable at around 0.7. The vehicle group using the RMCCM model achieves a high safety reputation assessment utility at all times. This indicates that in low-speed scenarios, the RMCCM model can enable the vehicle group to have a high level of safety.

[0230] like Figure 14 As shown, the maximum speed limit for the vehicle group is 90 km / h. The vehicle group safety reputation assessment utility of the RMCCM model is higher than 0.85 most of the time, while that of the SOCM model is generally in the range of 0.6 to 0.8. The vehicle group safety reputation assessment utility of the RMCCM model is higher than that of the SOCM model throughout the simulation. These results indicate that the RMCCM model can provide a higher level of safety for the vehicle group in medium-speed scenarios.

[0231] like Figure 15 As shown, the maximum speed limit of the vehicle group is 110 km / h. During the initial stage of vehicle group formation and evolution, the safety reputation assessment utility of the vehicle group using the RMCCM model gradually increases and remains at around 0.9, significantly outperforming the SOCM model. This indicates that the RMCCM model can ensure the safety of the vehicle group in high-speed scenarios.

[0232] Based on the analysis of the simulation results above, it can be seen that under the various speed scenarios mentioned above, the RMCCM model improves the safety reputation assessment effectiveness of the vehicle group by 27.7%, and the safety of the vehicle group is stronger.

[0233] The reliability of safety reputation assessment utility reflects the degree of deviation of neighboring nodes in assessing the safety reputation utility of the same vehicle. The higher the reliability of safety reputation assessment utility, the more reliable the safety reputation assessment utility of the vehicle group. For example... Figure 16 As shown, the reliability of the safety reputation assessment utility of RMCCM remained around 0.04, lower than that of the SOCM model. The reliability increased by 51.9%. Experimental results indicate that the RMCCM model has stronger vehicle group safety reliability under various speed scenarios.

[0234] (2) Stability

[0235] The lifetime of a trusted node reflects how long it remains in the vehicle group during the simulation. A longer lifetime indicates a more stable safety collaboration model. For example... Figure 17 As shown, the red line represents the RMCCM model, where the trusted node survival time remains high within the speed range of [65, 105], but decreases in high-speed scenarios. The rapid changes in vehicle group structure at high speeds reduce node survival time. The green line represents the SOCM model, where the trusted node survival time remains above 140s within the speed range of [65, 80], and is approximately around 120s within the speed range of [85, 120]. Overall, the RMCCM model's trusted node survival time is 23% longer than the SOCM model, indicating that the RMCCM model provides stronger stability for vehicle groups.

[0236] The duration of trusted node stability reflects how long a trusted node remains stably active within the same vehicle group. A longer node stability duration indicates a more stable vehicle group. For example... Figure 18 As shown, the stable duration of trusted nodes in RMCCM is basically maintained between 30-40 seconds, which is significantly better than the SOCM model by 42.9%, indicating that the RMCCM model can maintain higher node stability in the vehicle group.

[0237] Vehicle pod survival time reflects the average survival time from vehicle pod formation to disintegration. A longer survival time indicates a stronger survivability of the vehicle pod in unsafe environments and a more stable collaborative model. For example... Figure 19 As shown, the red line represents the RMCCM model, whose vehicle corps survival time remains around 65 seconds across different speed scenarios. The green line represents the SOCM model, which performs better at low speeds (65-70 km / h), with a vehicle corps survival time exceeding 60 seconds, but only maintains a survival time of around 45-55 seconds at other speeds. Analysis of the experimental results shows that the RMCCM model's vehicle corps survival time is 19.6% higher than the SOCM model, indicating that the RMCCM model can better maintain the vehicle corps structure and stability.

[0238] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. A method for constructing a collaborative model for autonomous vehicle swarms based on blockchain-based security and reputation management, characterized in that, Includes the following steps: Step 1. Define vehicle group formation events, node joining events, node leaving events, vehicle group merging events, vehicle group splitting events, topology optimization events, and vehicle group resolution events; Step 2. Dynamic evolution of autonomous vehicle swarm based on blockchain security and reputation management, which includes, in sequence: security and reputation management, dynamic evolution events, vehicle swarm topology optimization, and autonomous vehicle swarm evolution algorithm; Step 3: Construction of a safety collaboration model for the autonomous vehicle swarm, including: vehicle swarm state transitions and a safety collaboration model; Specifically, Step 1: Definition 1: Vehicle Formation Event (Formation): Autonomous vehicle nodes establish connections with each other and conduct safety and reputation assessments through cooperative events, thereby forming a vehicle corps; the conditions for a vehicle corps formation event are: (1) in, Indicates a group of vehicles; Indicates the vehicle node; Represents the set of neighboring nodes; This indicates the role and function of driverless vehicles. This indicates that the autonomous vehicle has not joined any vehicle group and is a free node. This indicates that the autonomous vehicle belongs to the basic node set of a certain vehicle group. This indicates that a vehicle node is a node in the core node chain of a certain vehicle group; It is a node For nodes The direct credit assessment utility; REU th It is the direct credit assessment utility threshold used for security credit assessment; After the basic node set is formed, the autonomous vehicle swarm will further construct a core node chain to ensure the security of the vehicle swarm's identity authentication information. The conditions for its construction are as follows: (2) in, Represents the basic node set; Indicates the threshold number of nodes; Represents a node The overall credit assessment effectiveness; It is used to determine nodes Is it a trusted node based on the trusted threshold? Define 2 Node Join Event: When a free autonomous vehicle node encounters an autonomous vehicle group with a core node chain during its journey, a node join event will occur if it has the same motion similarity as its neighboring nodes in the group and meets the safety reputation assessment requirements for joining the group. (3) Define three node leave events: During the operation of an autonomous vehicle swarm, if a vehicle node's safety reputation assessment utility or weight is too low, it will be expelled from the swarm; if a vehicle node cannot maintain stable communication with the swarm due to external interference or a different path, it will voluntarily leave the swarm. The conditions for a node leave event are: (4) in, Represents vehicle node The credit assessment weight; Definition 4: Vehicle group merging event (Merge): When two autonomous vehicle groups meet during driving, if the distance between vehicle nodes in different vehicle groups is within the communication range and their speeds are similar, a communication connection is established, and a safety reputation assessment is further carried out. If the conditions are met, they are merged into a new vehicle group. (5) Definition 5: Split event: If the communication distance between nodes in an autonomous vehicle group is too far, the speed difference is too large, or the safety reputation assessment utility between them is too low, a split event will occur, and other nodes in the vehicle group will join different vehicle groups according to the conditions. (6) in, This represents the maximum distance between two vehicle nodes; The threshold representing the speed similarity of nodes; Definition 6: Topology Optimization Event (Adjust): During the dynamic evolution process, the autonomous vehicle swarm performs topology optimization to strengthen the connections between neighboring nodes; (7) in, Represents nodes in an autonomous vehicle swarm and There are communication connections between them; Represents a node Betweenness centrality is a measure of graph centrality in a network based on the shortest path, and it is calculated as follows: (8) in, yes arrive The number of shortest paths, These paths pass through The number of times; the higher the betweenness centrality, the more concentrated the vehicle group is at that point; Definition 7: Disappearance of vehicle swarm: When the number of vehicle nodes in an autonomous vehicle swarm is small and they cannot communicate with other autonomous vehicles, the vehicle swarm structure will disintegrate. (9) Step 2 includes: Step 2.1 Security and Reputation Management The autonomous vehicle swarm maintenance mechanism assesses the safety and reputation of the swarm by comprehensively evaluating the safety and reputation of each node. Malicious nodes that fail to meet the assessment criteria are removed from the swarm, thus ensuring its safety. Based on a blockchain consensus mechanism using security and reputation proofs, the security and reputation assessment weights are used as transaction tokens for the distributed management of the swarm's safety and reputation assessment effectiveness. The calculation method for transaction token rewards is as follows: (10) in, This is the security reputation error threshold. Neighboring nodes whose security reputation assessment error exceeds this threshold will have their reputation tokens deducted, which means reducing the security reputation assessment weight of that node. It is a reward decay factor that controls the rewards received by nodes with more reputation tokens, thereby avoiding the centralization of security reputation. The security reputation management algorithm is Algorithm 1, and the specific steps are as follows: (1) Neighboring nodes communicate with nodes The success rate of collaborative events directly contributes to the effectiveness of security and reputation assessment. (2) The security reputation verification mechanism is used to assess the security reputation of different nodes based on their respective weights. The final comprehensive security reputation assessment utility of the node is calculated; nodes whose comprehensive security reputation assessment utility reaches the threshold and are not in the core node set will be added to the core node set. (3) The consensus mechanism will issue reputation token rewards to neighboring nodes that participate in the computation, which will increase the security reputation assessment weight of the node and record the result as a transaction in the blockchain; (4) Nodes with insufficient security reputation tokens or low overall security reputation evaluation utility are considered malicious nodes that affect the security of the vehicle group and will be expelled from the vehicle group. The information of such nodes will be recorded in the Certificate Revocation List (CRL) and they will be refused rejoining the vehicle group. Step 2.2 Dynamic Evolution Events Step 2.2.1 Adding the node When a free autonomous vehicle node encounters a swarm of vehicles, it will actively send a node joining request. After the swarm continuously receives cooperation event information from free nodes, it updates its security reputation assessment effect, which in turn triggers a node joining event. New nodes joining the basic node set of the swarm will obtain identity information through the core node chain of the swarm, and the authentication information will be stored in the blockchain. The security of node authentication is guaranteed by the immutable property of transaction blocks. Step 2.2.2 Vehicle group merging When two groups of autonomous vehicles meet, if all neighboring nodes meet the conditions of similar motion and direct safety reputation assessment utility reaching a threshold, the safety reputation assessment utility of the merged group will not be lower than the threshold, and the group will be merged to maximize the group size and share vehicle node information while ensuring group safety. The merging of identity authentication records and certificate revocation lists in the group merging algorithm will differ depending on whether a core node chain exists in the group or not. Step 2.2.3 Vehicle Group Splitting Some dynamic evolution behaviors of autonomous vehicle swarms are determined by the state of vehicle nodes within a local range. There may be situations where a vehicle node is relatively stable in the overall state of multiple neighboring nodes, but the connection between two nodes is not tight. If both nodes are in the core node chain, the swarm as a whole is still in an unstable or unsafe state. In this case, the swarm should be split into smaller swarms in time. Step 2.3 Vehicle group topology optimization: After dynamic evolution events such as node addition or departure, the topology of an autonomous vehicle swarm may become too concentrated or too dispersed. When the swarm structure is dispersed, the edge connectivity of the swarm may be 1, meaning that there is a node with only one connection to other nodes. When the swarm structure is concentrated, if multiple communication paths need to pass through a certain node, that node's betweenness centrality is too high. In both cases, without topology optimization, the distributed structure will tend to become centralized, reducing the stability of the swarm. Therefore, when a node in an autonomous vehicle swarm detects an unreasonable network structure, it is necessary to execute a swarm topology optimization algorithm. The vehicle group topology optimization algorithm is Algorithm 5, and the specific steps are as follows: (1) If node Only with another node in the car group To communicate, Check if there are other neighboring nodes that meet the trusted connection requirements; (2) If another node exists If so, a connection establishment request REQ_LINK is sent to it; (3) After calculating that the direct security reputation assessment utility meets the conditions, a connection establishment agreement message RESP_AGREE is sent, thereby increasing network stability. (4) If node Discover neighboring nodes It is a central node, and it is necessary to reduce the centralization of the vehicle group and establish connections with other nodes as much as possible, such as in steps (2) and (3). Step 2.4 Autonomous Vehicle Swarm Evolution Algorithm: Combining the safety reputation management algorithm in step 2.1 with the processing algorithms for different events in steps 2.2.1 (node ​​addition), 2.2.2 (vehicle group merging), and 2.2.3 (vehicle group splitting), the complete vehicle group dynamic evolution algorithm for autonomous vehicles is obtained as Algorithm 6, with the specific steps as follows: (1) The vehicle group periodically executes the safety reputation management algorithm to obtain the safety reputation evaluation utility of each node; (2) Execute the corresponding node addition, node departure, vehicle group merging or vehicle group splitting algorithms according to the different triggering conditions met by the vehicle group during operation; (3) After a dynamic evolution event occurs in the vehicle group, a vehicle group topology optimization algorithm should be executed to optimize the unreasonable vehicle group topology structure; Step 3 includes: Step 3.1 Vehicle group status transition To maintain the safety and stability of the autonomous vehicle swarm during operation, different dynamic evolution behaviors of the autonomous vehicle swarm were analyzed. When the swarm reaches the event triggering condition, the swarm experiences a corresponding event and performs a state transition accordingly. At this time, the swarm maintains a safe and stable state through a dynamic evolution algorithm. The overall attribute set of the vehicle group and the transitions between different states are defined as follows: Define the set of vehicle group attributes The set of all attributes involved in the safe collaborative model of autonomous vehicle swarms is represented by the following six-tuple: (11) in, , representing the set of all autonomous vehicle nodes, further forming the basic node set BNS and the core node chain CNC; , represents the set of trusted connections between all vehicle nodes; This represents the set of evolutionary events of a vehicle corps, including corps formation. Credit Management Node addition Node leaving Topology optimization Vehicle group merging vehicle group split The disappearance of the car group Eight types of events, including evolutionary events, will lead to transitions between vehicle states; This represents a set of algorithms for the dynamic evolution of vehicle groups, including safety and reputation management algorithms. Node addition algorithm Node leaving algorithm Vehicle group merging algorithm Vehicle group splitting algorithm Vehicle group topology optimization algorithm ; Represents the set of vehicle group states, including the initial state. , secure collaborative status Unsafe state Expansion state Reduced state Topology optimization status Merged status Split state Extinction state ; This refers to a vehicle-to-everything blockchain, used to store the identity authentication information and reputation assessment utility of autonomous vehicle nodes. (1) Initial state :exist At time 1, at least two free nodes initially construct a basic node set (BNS) using a vehicle group formation algorithm, and the newly formed vehicle group is... In the initial state, the number of nodes has not yet reached the threshold. The vehicle group attributes are as follows: (12) (2) Secure Coordination Status :exist At that moment, the initial state of the vehicle group The core node chain (CNC) is constructed through a vehicle corps formation algorithm, or the vehicle corps in a safe and cooperative state is restored to a safe and cooperative state through other dynamic evolution algorithms. At this time, the attribute set of the vehicle corps is as follows: (13) in, This indicates the effectiveness of vehicle group reputation assessment; (3) Unsafe state :exist At that moment, the car group Malicious nodes were detected within the vehicle group through a security and reputation management algorithm. At this point, the attributes of the vehicle group are as follows: (14) (4) Expansion state :exist Time, node Apply to join the car group The attributes of the vehicle group are as follows: (15) (5) Reduction state :exist Time, node Whether leaving the group of vehicles voluntarily or involuntarily The attributes of the vehicle group are as follows: (16) (6) Topology optimization state :exist At that moment, the car group Middle node The vehicle group topology was found to be unreasonable, and its relationship with neighboring nodes was also found to be problematic. Establishing more connections increases the effectiveness of vehicle group safety and reputation assessment; the vehicle group's attribute set is as follows: (17) (7) Merged status :exist At that moment, the car group , They met and merged to form a new car group. The size of the vehicle group has increased; the attributes of the vehicle group are as follows: (18) in, Indicates a group of vehicles , The newly established trusted connection between them; (8) Split state :exist At that moment, the car group The vehicle may split into multiple new vehicle groups, either actively or passively, such as vehicle groups. , The attributes of the vehicle group are as follows: (19) (9) Extinction state :exist Time, node Whether leaving the group of vehicles voluntarily or involuntarily At this point, the attributes of the vehicle group are as follows: (20) Step 3.2 Security Collaboration Model Based on the dynamic evolution method of the autonomous vehicle swarm given in step 2, and the changes in the attribute set of the above-mentioned vehicle swarm state transitions, the safety cooperation model of the autonomous vehicle swarm is obtained as follows: (21) in, It represents the combination of different attributes of a vehicle group; the collaborative model consists of a series of changes caused by the vehicle group state and safety reputation assessment, with different vehicle group states... and the utility of security reputation assessment between nodes This will change the characteristics of the vehicle group, leading to vehicle group evolution events. When the trigger condition is met, the vehicle group's attributes are updated to a temporary state. Execute the corresponding dynamic evolution algorithm The vehicle group eventually maintained the new state. Vehicle group blockchain Corresponding changes may also occur.

2. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 1, characterized in that, The security and reputation management algorithm: 。 3. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 1, characterized in that, The node addition algorithm is Algorithm 2, and the specific steps are as follows: (1) Free nodes Towards the car group Nodes within The sending node adds the REQ_ADD request and its own status information; (2) Node After receiving a node join request, calculate the relative distance between the two nodes. and speed similarity It then checks whether the certificate is on the certificate revocation list; if not, it notifies the node. Send a request rejection message RESP_REF; (3) If the relative motion between the two nodes meets the requirements, then further calculate the nodes. right Direct credit assessment utility If the utility of the security reputation assessment is below the threshold If so, the node will be refused entry into the vehicle group; (4) When all the above conditions are met, send to the node Send the RESP_AGREE consent message; node Then a public key pair is generated. and private key It is used for secure communication between nodes and transmits its own user information. and public key Send to vehicle group nodes Further forwarding to the nodes where the core nodes of the vehicle group are concentrated. ; (5) Nodes The received user information is encrypted to generate the distributed identity DID of the newly joined node, and the node's distributed ID, encrypted identity information, node public key, and timestamp are recorded as a transaction ledger for storage on the blockchain. (6) Return the updated information.

4. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 3, characterized in that, The node joining algorithm: 。 5. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 1, characterized in that, The vehicle group merging algorithm is Algorithm 3, and the specific steps are as follows: (1) Group of vehicles Nodes within With a group of cars Nodes within Encounter, Node First, calculate the vehicle group within the communication range. If the relative status information of the node meets the requirements, a vehicle group merging request REQ_MERGE and its own status information are sent to it. (2) Node Upon receiving a node join request, the system calculates the relative state information between the two nodes. If the requirements are not met, it sends a request to the node. Send a rejection message RESP_REF for the merge request; (3) If the two nodes are within the range of neighboring nodes, then the direct safety reputation assessment utility between the two nodes is further calculated. If the direct reputation assessment utility of any one of the neighboring nodes of the two vehicle groups meeting is lower than the threshold, the direct safety reputation assessment utility is calculated. If so, then the vehicle group merging will be refused; (4) When the direct security reputation assessment utility among all nodes meets the conditions, the node is given... Send the RESP_AGREE message to agree to join; the two nodes send each other the reputation tokens owned by each vehicle node in their respective vehicle group, clear historical data records, and use the total number of security reputation tokens as the new starting block; (5) If neither vehicle group has a core node chain, the algorithm ends; if one vehicle group has a core node chain and the other vehicle group does not, the vehicle group with the core node chain will authenticate the nodes of the other vehicle group and store them in the blockchain. If both vehicle groups have core node chains, then the authentication records of the two vehicle groups are merged into a new blockchain ledger, and the Certificate Revocation Lists (CRLs) of the two vehicle groups are also merged.

6. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security reputation management as described in claim 5, characterized in that, The vehicle group merging algorithm: 。 7. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 1, characterized in that, The vehicle group splitting algorithm is Algorithm 4, and the specific steps are as follows: (1) Node Discovered a node in another core node chain If the relative node state is unstable, it broadcasts a vehicle group split request REQ_SPLIT to other nodes in the vehicle group. and Node status information; (2) Nodes in the vehicle group are arranged according to their distance from the node. , The network distance, i.e., the number of hops, is used to sort the vehicles, and the vehicles that are to be added to the two split nodes are judged in order from the nearest to the farthest. (3) If the node to be assigned has a direct or indirect connection with both nodes and the network connection distance is equal, then compare the relative state of the node with the directly connected nodes in the two split vehicle groups and add the vehicle group where the node with the closer relative node state is located. (4) If the node to be assigned is reachable from only one of the nodes, it shall be directly added to the split vehicle group to which the node is located, and its communication connection shall be maintained. (5) When all nodes in the vehicle group have selected the split vehicle group to join, the nodes will broadcast and update their own status information within the vehicle group. The two vehicle groups will update their certificate revocation lists respectively, and the vehicle group split will be completed.

8. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 7, characterized in that, The vehicle group splitting algorithm: 。 9. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security and reputation management as described in claim 1, characterized in that, The vehicle group topology optimization algorithm: 。 10. The method for constructing a collaborative model for autonomous vehicle swarms based on blockchain security reputation management as described in claim 1, characterized in that, The vehicle group dynamic evolution algorithm: 。

Images