An internet of vehicles emergency message trust evaluation method and system
By combining blockchain networks and random forest algorithms with smart contracts to design a trust value query algorithm, the problem of low accuracy in trust assessment of emergency messages under the high proportion of malicious vehicles in the Internet of Vehicles was solved. This achieved higher assessment accuracy and lower storage overhead, improving the robustness and query efficiency of the system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-06-06
- Publication Date
- 2026-03-20
AI Technical Summary
With a high proportion of malicious vehicles, the accuracy of trust assessment for emergency messages in existing vehicle-to-everything (V2X) systems decreases, making it difficult to maintain road traffic safety and reliability in the face of rampant malicious messages.
By employing a blockchain network, vehicle registration center, roadside units, and edge servers, and combining random forest algorithm and smart contracts, a trust value query algorithm is designed. This algorithm improves the efficiency and accuracy of trust queries by predicting the authenticity of emergency messages and utilizing a switchable caching strategy.
With a high proportion of malicious vehicles, the accuracy of emergency message trust assessment was improved, storage space overhead was reduced, and the robustness and query efficiency of the system were enhanced.
Smart Images
Figure CN116828416B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of Internet of Vehicles, and particularly relates to an Internet of Vehicles emergency message trust evaluation method and system. BACKGROUND
[0002] In recent years, as an important part of Intelligent Transportation System (ITS), Vehicular Ad-hoc NETworks (VANETs) has attracted extensive research interest from governments, industries and academia. In VANETs, there may be malicious vehicles that spread a large number of false emergency messages to deceive other vehicles. If these malicious messages are not identified and evaluated, the safety and reliability of road traffic will be severely damaged.
[0003] LIU et al. proposed a VANETs trust management system based on blockchain, which uses Hidden Markov Model (HMM) in machine learning to construct message sequences, thereby evaluating the trust of messages. YANG et al. proposed a distributed VANETs trust management system based on blockchain, which calculates the message credibility by the location of vehicles and uses Bayesian Model (BAYES) to infer the aggregated credibility of message sets. ZHOU et al. proposed a blockchain-assisted VANETs emergency message trust evaluation system (BEMTE), which queries the trust value with low latency through a trust value query algorithm, and introduces the trust value into a multi-reference factor evaluation algorithm to calculate the message credibility, thereby realizing low-latency and accurate message trust evaluation. Although the above systems are based on blockchain technology, they realize the secure distributed storage of vehicle trust values in VANETs, and achieve high message trust evaluation accuracy through the proposed trust evaluation model, but in the case of high proportion of malicious vehicles, the message trust evaluation accuracy decreases significantly. SUMMARY
[0004] The main purpose of the present application is to overcome the shortcomings and deficiencies of the prior art, and to provide an Internet of Vehicles emergency message trust evaluation method and system.
[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows:
[0006] An Internet of Vehicles emergency message trust evaluation method, comprising a blockchain network, a vehicle registration center (VRC), a roadside unit (RSU), an edge server (ES) and a vehicle, and the method comprises the following steps:
[0007] S1, initialization, each part is initialized, the blockchain network is activated, the initialization of the encryption algorithm AES and the digital signature algorithm RSA is completed on the RSU and the VRC, and the training of the machine model is completed on the edge server;
[0008] S2, the vehicle registers in a vehicle registration center, and vehicle data is saved to a block chain;
[0009] S3, the vehicle spreads an emergency message after perceiving an emergency accident;
[0010] S4, a vehicle receiving the emergency message sends a vehicle trust value query request to a roadside unit;
[0011] S5, the roadside unit interacts with a block chain network to query the vehicle trust value;
[0012] S6, the roadside unit interacts with an edge server to analyze the emergency message;
[0013] S7, the vehicle receiving the emergency message performs trust evaluation on the emergency message;
[0014] S8, the vehicle receiving the emergency message feeds back the emergency message.
[0015] The application also comprises a vehicle networking emergency message trust evaluation system, which adopts the vehicle networking emergency message trust evaluation method provided by the application, and is provided with a block chain network, a vehicle registration center VRC, a roadside unit RSU, an edge server ES and a vehicle, and comprises:
[0016] A system initialization module is used for initialization of each component and node of the system;
[0017] A vehicle registration module is used for registration of the vehicle in the vehicle registration center;
[0018] An emergency message spreading module is used for spreading of the emergency message by the vehicle after perceiving the emergency accident;
[0019] A trust value query and evaluation module is used for query of the trust value by the vehicle to the roadside unit, query of the vehicle trust value by the roadside unit interacting with the block chain network, and analysis of the emergency message by the roadside unit interacting with the edge server;
[0020] An emergency message feedback module is used for feedback of the vehicle on the authenticity of the emergency message.
[0021] Compared with the prior art, the application has the following advantages and beneficial effects:
[0022] 1, the application realizes distributed and credible storage of trust data by using a block chain; a trust value query algorithm based on an intelligent contract and a switchable cache strategy is designed, which can effectively improve the RSU trust query efficiency and reduce the storage space overhead.
[0023] 2、The application introduces the random forest algorithm into the emergency message trust evaluation process of the Internet of Vehicles, can greatly improve the message trust evaluation accuracy, can still maintain a high trust evaluation accuracy under a high proportion of malicious vehicles, and shows good robustness. BRIEF DESCRIPTION OF DRAWINGS
[0024] Figure 1 is a flowchart of the method of the application;
[0025] Figure 2 is a schematic diagram of the system of the application;
[0026] Figure 3 is a flowchart of the random forest algorithm;
[0027] Figure 4 is a comparison diagram of the trust value query delay of each cache strategy;
[0028] Figure 5 is a comparison diagram of the roadside unit storage overhead of each cache strategy;
[0029] Figure 6 is a comparison diagram of the message trust evaluation accuracy of each scheme under each proportion of malicious vehicles;
[0030] Figure 7 is a comparison diagram of the message trust evaluation delay of each scheme. DETAILED DESCRIPTION
[0031] The application will be further described in detail below in combination with embodiments and drawings, but the embodiments of the application are not limited thereto.
[0032] Embodiments
[0033] As shown in Figure 1 and Figure 2 , the application is a trust evaluation method for emergency messages of the Internet of Vehicles, which is provided with a blockchain network, a vehicle registration center VRC, a roadside unit RSU, an edge server ES and a vehicle.
[0034] In the embodiment, the blockchain network provides a trusted storage of trust data for vehicles. The vehicle registration center (VRC) provides a network registration service for vehicles. The roadside unit (RSU) provides two services for vehicles, the first service is to respond to the query request of the vehicle by interacting with the blockchain network, and the second service is to evaluate the real probability of the accident message of the vehicle by the machine learning model on the edge server. The edge server (ES) is a remote and processing-capable server, and the machine learning model deployed thereon can extract and analyze relevant features of the emergency message content and output the prediction probability of the message being a real message. The vehicle is equipped with an on-board unit (OBU) with functions such as perception, calculation and storage. The vehicle communicates with other vehicles and RSUs through V2V and V2R respectively (the communication process is encrypted).
[0035] As shown in Figure 1 , the method of the embodiment comprises the following steps:
[0036] S1, initialization, each component is initialized, the blockchain network is activated, the initialization of the encryption algorithm AES and the digital signature algorithm RSA is completed on the RSU and the VRC, and the training of the machine model is completed on the edge server;
[0037] In the embodiment, the machine model is specifically a random forest model.
[0038] The base classifier in the random forest uses the same but independent K CART decision trees, and the Gini index is used as the measurement index for dividing the nodes of the decision tree.
[0039] When constructing the decision tree, the attribute with the smallest Gini index is preferentially selected as the node of the decision tree to obtain higher accuracy, and the calculation formula of the Gini index is:
[0040]
[0041] wherein, C k is the sample subset of D belonging to the kth category, |C k | is the number of samples, and |D| is the total number of samples of the data set D.
[0042] Suppose for each candidate feature A j ,j∈1,2,...,N of a decision tree node t, the corresponding Gini index is calculated as:
[0043]
[0044] wherein, V represents the number of values of the feature A j , D v represents the sample subset of the feature A j with the value v, and |Dv | represents the number of the sample subset; select the feature that minimizes the Gini index as the partition basis of the current node, that is
[0045]
[0046] At this time is the feature that minimizes the Gini index after partition, through which the sample corresponding to the current node is divided into V subsets, each subset corresponds to the feature with value v;
[0047] As Figure 3 shown, the random forest classifies and predicts samples by training multiple decision trees, and adopts voting or averaging to obtain the final result, specifically:
[0048] Repeat random sampling with replacement from the data set D to extract K sub-data sets;
[0049] Based on the K sub-data sets, select the optimal feature to establish K decision trees through the Gini index, randomly select N features from M features as the split feature set of the decision tree node, N < M, calculate the Gini index, select the feature that minimizes the Gini index to establish the node, divide the training data set, and repeat the above division process until the Gini index of the leaf node in the decision tree is 0;
[0050] Then, the random forest model {t1, t2, t3,..., tk} is obtained by combining the K decision trees obtained by training, the test sample x is input into the random forest, and the classification results of each decision tree {t1(x), t2(x), t3(x),..., tk(x)} are obtained, and finally the classification results of each decision tree are averaged and output, that is
[0051]
[0052] S2, the vehicle registers in the vehicle registration center, and the vehicle data is saved to the blockchain; specifically:
[0053] The vehicle registers in the vehicle registration center, and the vehicle registration center saves the vehicle identity ID i , the initial trust value T i , the public key Dp a After string concatenation, it gets Save it to the blockchain, complete registration and return a success result.
[0054] S3, the vehicle spreads the emergency message after perceiving the emergency accident; specifically:
[0055] A vehicle V j, j = 1,..., n, after perceiving the emergency accident e, the accident and related information are converted into a fixed format message by the vehicle-mounted OBU is expressed as:
[0056]
[0057] wherein ID j is the identity of the vehicle V j , is the public key of the vehicle V j , represents the time of the accident, represents the location of the accident, represents the detailed description of the accident e by the vehicle V j , env represents environmental information, v represents vehicle information, and r represents road information.
[0058] Finally, the vehicle V j generates an emergency message and broadcasts it to the remaining vehicles.
[0059] S4, the vehicle receiving the emergency message sends a vehicle trust value query request to the roadside unit
[0060] S5, the roadside unit interacts with the blockchain network to query the vehicle trust value; specifically:
[0061] After receiving the vehicle trust value query request , the RSU first verifies the signature using the public key in , and then according to the identity ID j in it, calls the trust value query algorithm to query the corresponding trust value T j . is expressed as:
[0062]
[0063] wherein Exp time , Ds Msq respectively represent the validity time of the message request, the public key of the vehicle V i , and the signature of the request, and the vehicle V i is a vehicle receiving the emergency message, i = 1,..., n. is expressed as:
[0064]
[0065] wherein, respectively represent the location and time of broadcast message generation;
[0066] Then, the RSU will carry the m env , m v , m r The three types of information closely related to the accident are input into the random forest model to obtain the message authenticity probability
[0067] The RSU compares the message authenticity probability with the switching strategy threshold Thr cache :
[0068]
[0069] wherein the value of the switching strategy threshold Thr cache is 0.5;
[0070] If Cache method = 0, it means that the query-after-caching strategy is applied, under which the RSU waits for the vehicle to initiate the trust value query and caches the query content and query record of the vehicle; if Cache method = 1, it means that the pre-caching strategy is applied, under which the RSU immediately collects the trust values of all vehicles within the range.
[0071] The RSU inputs the value of Cache method and the ID j into the trust value query algorithm to query the trust value T j .
[0072] The trust value query algorithm is as follows:
[0073] First, according to the value of Cache method , the pre-caching strategy or the query-after-caching strategy is started;
[0074] When the pre-caching strategy is started, the RSU will inquire the ID of each vehicle entering its range, then query the trust value corresponding to the vehicle ID in the blockchain, and then cache these trust values locally. When an accident occurs, as soon as a vehicle initiates a trust value query request to the RSU, the RSU directly reads the trust value from the cache to quickly respond to the request;
[0075] When the query-after-caching strategy is started, the RSU will not inquire the vehicle ID, but wait until a vehicle initiates a trust value query request to the RSU for a target vehicle. The RSU first queries the trust value of the target vehicle in the blockchain, then caches it, and when a subsequent vehicle initiates a trust value query request for the same target vehicle, directly reads the trust value from the cache and responds to the request.
[0076] S6. Roadside units and edge servers interact to analyze emergency messages; specifically:
[0077] While the RSU calls the algorithm to query the vehicle's trust value, the RSU also includes accident information in the query. Send to the edge server, the edge server will then send the m in the message. env ,m v ,m r The data is fed into a trained random forest model. After feature extraction and analysis, the random forest model outputs the true probability of the message. Edge servers will It is returned to RSU, which packages the trust value and the message's true probability into a single data set. Return to the vehicle. Represented as:
[0078]
[0079] Among them, Exp time Indicates the message validity period, Pk rsu Ds is the public key of RSU. Msr This indicates the signature of the message.
[0080] S7. Vehicles receiving emergency messages undergo an emergency message trust assessment; specifically:
[0081] When vehicle V i Received RSU response Then, first use the public key Pk rsu Verify the signature, and then perform an emergency message trust assessment; the emergency message trust assessment is conducted in two parts, specifically:
[0082] Phase 1, based on Carrying broadcast vehicle V j Trust value T j The message aggregation credibility P(e / C) is calculated using a Bayesian model. e ), P(e / C e The calculation process for ) is as follows:
[0083]
[0084] in, Calculated using vehicle trust value, message freshness, and vehicle location:
[0085]
[0086] Among them, T t and T drespectively represent the influence of message freshness and message generation location on the environmental trust value; λ, θ, μ are the weights of vehicle trust value, message freshness and message generation location respectively, λ+θ+μ=1; define t Thr and d Thr as the message validity time threshold and the message validity distance threshold; V i obtain the accident occurrence time from the accident occurrence location the broadcast message generation time and the broadcast message generation location Let T t and T d be calculated as follows:
[0087]
[0088]
[0089] wherein γ is a preset parameter, controlling the change rate of T t and T d , Δt-t Thr represents the difference between message freshness and message validity time threshold, Δd-d Thr represents the difference between the distance between message generation location and accident occurrence location and the message validity distance threshold;
[0090] In the second stage, based on the real prediction probability of emergency message carried by V i received n emergency messages about the accident e, vehicle V i also gets n prediction probabilities returned by RSU j∈n, vehicle V i takes the value of n prediction probabilities, the calculation process is as follows:
[0091]
[0092] Then vehicle V i combines P(e / C e ) and P pre (e) to get the final prediction probability P final (e), the process is as follows:
[0093] P final (e)=ψ·P pre (e)+ζ·P(e / C e )
[0094] Wherein, ψ and ζ respectively represent the weight of the probability inferred by the Bayesian according to the message credibility and the probability predicted by the machine learning model according to the accident message characteristics, ψ+ζ=1;
[0095] Finally, P(e) is compared with the trust threshold T Thr If P(e)≥T Thr , the vehicle V i trusts the accident message; if P(e)<T Thr , the vehicle V i does not trust the accident message; the trust threshold T Thr is selected according to the comparison of multiple experiments in actual implementation.
[0096] S8, the vehicle receiving the emergency message feeds back the emergency message; specifically:
[0097] The vehicle receiving the emergency message selects whether to trust the emergency accident message according to the trust evaluation result of the emergency message, if yes, changes the driving route or slows down the speed; if no, continues to drive and feeds back the authenticity of the accident message on the accident road segment, the RSU collects the feedback result, and reevaluates the trust value of the emergency message broadcast vehicle.
[0098] In another embodiment, a vehicle networking emergency message trust evaluation system is also provided, the system adopts the vehicle networking emergency message trust evaluation method of the above-mentioned embodiment, the system is provided with a blockchain network, a vehicle registration center VRC, a roadside unit RSU, an edge server ES and a vehicle, and comprises:
[0099] A system initialization module is configured to initialize each component and node of the system.
[0100] A vehicle registration module is configured to register the vehicle in the vehicle registration center.
[0101] An emergency message dissemination module is configured to disseminate the emergency message by the vehicle after perceiving the emergency accident.
[0102] A trust value query and evaluation module is configured to query the trust value by the vehicle from the roadside unit, interactively query the trust value of the vehicle by the roadside unit and the blockchain network, and interactively analyze the emergency message by the roadside unit and the edge server.
[0103] An emergency message feedback module is configured to feed back the authenticity of the emergency message by the vehicle.
[0104] As shown in FIG. 1, it is a comparison diagram of the average trust value query delay of the trust value query algorithm under the cache strategy of the present application and the prior art (for example, the pre-cache strategy and the post-cache strategy involved in a blockchain-assisted vehicle networking emergency message trust evaluation system proposed by Zhou et al.). Figure 4 As shown in FIG. 1, it is a comparison diagram of the average trust value query delay of the trust value query algorithm under the cache strategy of the present application and the prior art (for example, the pre-cache strategy and the post-cache strategy involved in a blockchain-assisted vehicle networking emergency message trust evaluation system proposed by Zhou et al.).
[0105] As Figure 5 shown, is the average storage overhead of RSU under the trust value query algorithm of the caching strategy of the application and the prior art (the pre-caching strategy and the caching strategy after query involved in a blockchain-assisted trust evaluation system for emergency messages in VANETs proposed by Zhou et al., a VANETs trust management system based on blockchain proposed by Liu et al., and a distributed VANETs trust management system based on blockchain proposed by Yang et al.) under the condition of 10 groups of broadcast messages.
[0106] As Figure 6 shown, is the average storage overhead of RSU under the trust value query algorithm of the caching strategy of the application and the prior art (the pre-caching strategy and the caching strategy after query involved in a blockchain-assisted trust evaluation system for emergency messages in VANETs proposed by Zhou et al., a VANETs trust management system based on blockchain proposed by Liu et al., and a distributed VANETs trust management system based on blockchain proposed by Yang et al.) under the condition of 10 groups of broadcast messages.
[0107]
[0108] Among them, TP and TN represent trusted real accident messages and untrusted false accident messages respectively; FP represents trusted false accident messages, and FN represents untrusted real accident messages. As Figure 6 shown, under the environment of various malicious vehicle ratios, the message trust evaluation accuracy of the application is better than that of the prior art, and shows high robustness.
[0109] As Figure 7 shown, is the average storage overhead of RSU under the trust value query algorithm of the caching strategy of the application and the prior art (the pre-caching strategy and the caching strategy after query involved in a blockchain-assisted trust evaluation system for emergency messages in VANETs proposed by Zhou et al., a VANETs trust management system based on blockchain proposed by Liu et al., and a distributed VANETs trust management system based on blockchain proposed by Yang et al.) under the condition of 10 groups of broadcast messages.
[0110] The application firstly introduces a random forest model in the evaluation process, which assists the roadside unit to analyze the emergency message in the Internet of Vehicles, and outputs the prediction probability of the message being a real message. Then, based on the prediction probability output by the random forest model, a switchable cache strategy is proposed, and a trust value query algorithm is designed combined with the smart contract to balance the conflict between the query efficiency and the storage space overhead of the roadside unit in the existing scheme; at the same time, the prediction probability is introduced as a reference factor into the message trust evaluation algorithm to obtain higher message trust evaluation accuracy. Finally, compared with the existing system, the roadside unit in the system proposed by the application achieves a better balance between the trust value query efficiency and the storage overhead, is more suitable for actual scenarios, and the message trust evaluation accuracy of the system proposed by the application has greatly improved compared with the existing system, and has high robustness under high malicious vehicle proportion.
[0111] It should also be noted that in this specification, terms such as "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed or inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.
[0112] The above description of disclosed embodiments enables a person skilled in the art to implement or use the application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A trust evaluation method for emergency messages in Internet of Vehicles, characterized in that, The method comprises the following steps: S1, initialization, each component is initialized, the blockchain network is activated, the initialization of the encryption algorithm AES and the digital signature algorithm RSA is completed on the RSU and the VRC, and the training of the machine model is completed on the edge server; S2, the vehicle is registered in the vehicle registration center, and vehicle data is saved to the blockchain; S3, the vehicle spreads emergency messages after perceiving an emergency accident; S4, the vehicle receiving the emergency message sends a vehicle trust value query request to the roadside unit; S5, the roadside unit interacts with the blockchain network to query the vehicle trust value; specifically: RSU received a vehicle trust value query request Then, use public key in Verify the signature, and then identify the identity within it. Call the trust value query algorithm to query the corresponding trust value. ; Represented as: wherein, respectively represent the validity time of the message request, the public key of the vehicle and the signature of the request, the vehicle i.e. a vehicle that has received an emergency message, i = 1,..., n; is represented as: wherein , respectively indicate the location and time of broadcast message generation; Then, the RSU will carry the three types of information closely related to the accident into the random forest model to get the message true probability ; The RSU compares the message authenticity probability to a handover policy threshold and compares: wherein the handover policy threshold has a value of 0.5; If , it means that the query-after-caching strategy is applied, under which the RSU waits for the vehicle to initiate the trust value query and caches the query content and query record of the vehicle; if , it means that the pre-caching strategy is applied, under which the RSU collects the trust values of all vehicles in the range at once. The RSU will the values of input to the trust value query algorithm, and query the trust value of the RSU. The trust value query algorithm is specifically: First, according to the value of the , a pre-cache strategy or a post-query cache strategy is initiated; When the pre-caching strategy is started, the RSU will inquire the ID of each vehicle entering its range, then query the trust value corresponding to the vehicle ID in the blockchain, and then cache these trust values locally. When an accident occurs, as soon as a vehicle initiates a trust value query request to the RSU, the RSU directly reads the trust value from the cache to quickly respond to the request; When the query-after-caching strategy is started, the RSU does not inquire the vehicle ID, but waits until a vehicle initiates a trust value query request to the RSU for a target vehicle. The RSU first queries the trust value of the target vehicle in the blockchain, then caches it, and when a subsequent vehicle initiates a trust value query request for the same target vehicle, directly reads the trust value from the cache and responds to the request; S6, the roadside unit interacts with the edge server to analyze the emergency message; specifically: RSU calls the algorithm to query the vehicle trust value at the same time, RSU carries the accident message in the query Sent to the edge server, the edge server will Input into the trained random forest model, the random forest model outputs the message real probability after feature extraction and analysis , the edge server returns To the RSU, the RSU packs the trust value and the message real probability into Return to the vehicle, Expressed as: wherein, denotes a message validity time, is a public key of the RSU, denotes a signature on the message; S7, the vehicle receiving the emergency message performs emergency message trust evaluation; specifically: When the vehicle receives the return of the RSU , first use the public key to verify the signature, and then perform emergency message trust evaluation; the trust evaluation of the emergency message is divided into two parts, which are: In a first stage, based on Broadcasting vehicles of trust values , the message aggregation credibility is calculated by Bayesian model , The calculation process is as follows: wherein, is calculated by the vehicle trust value, the message freshness, and the vehicle location: in, and These represent the impact of message freshness and message generation location on the environment trust value, respectively. , , These are the weights for vehicle trust value, message freshness, and message generation location, respectively. ;definition and These are the message validity time threshold and the message validity distance threshold; from Obtain the time of the accident Location of the accident Broadcast message generation time and the location where broadcast messages are generated ,make , , and The calculations are as follows: wherein, is a preset parameter, controlling and a rate of change of, represents a difference between the message freshness and a message validity time threshold value, represents a difference between the distance of the message generation location and the accident occurrence location and a message validity distance threshold value; The second stage is based on carried about the emergency message of the true prediction probability , if received emergency message about the accident , the vehicle also get RSU returned prediction probability , the vehicle on the prediction probability is valued, the calculation process is: Next vehicle Will And Combined, resulting in the final prediction probability , as follows: wherein, and respectively represent the weight of the probability inferred by the Bayesian according to the credibility of the message and the probability predicted by the machine learning model according to the characteristics of the accident message, ; Finally, the trust threshold value is compared with the , the vehicle trusts the accident message; if , the vehicle does not trust the accident message; the trust threshold value is selected according to multiple experiments in actual implementation. S8, the vehicle receiving the emergency message feeds back the emergency message. 2.The trust evaluation method for emergency messages in V2X according to claim 1, characterized in that, In step S1, the machine model is specifically a random forest model; In random forests, the base classifiers use the same but independent methods. A CART decision tree, which uses the Gini index as a metric for splitting decision tree nodes; When constructing a decision tree, the attribute with the smallest Gini index is preferentially selected as a decision tree node to obtain higher accuracy. The calculation formula of the Gini index is: in, yes The middle belongs to the first A subset of samples from each category, It is the number of its samples. It is a dataset The total number of samples; Assume for each candidate feature of a decision tree node its corresponding Gini index is calculated as: wherein, representing a feature the number of values, representing a feature is equal to a subset of samples, representing the number of the subset of samples; the feature that makes the Gini index minimum is selected as the division basis of the current node, namely At this time is the feature that makes the Gini index minimum after partition, through which the current node is divided into subsets, each corresponding to a feature with a value of part. Random forest classifies and predicts samples by training multiple decision trees, and obtains the final result by voting or averaging. Specifically: repeating the random sampling with replacement from the dataset to extract sub-datasets; Based on one sub-data set, the optimal feature is selected by Gini index to establish one decision tree, from one feature, randomly select one feature, , as the split feature set of the decision tree node, calculate the Gini index, select the feature that makes the Gini index minimum, establish the node, divide the training data set, repeat the above division process until the Gini index of the leaf node in the decision tree is 0; The random forest model is obtained by combining the decision trees The random forest model is obtained by combining the decision trees The random forest model is obtained by combining the decision trees The random forest model is obtained by combining the decision trees The random forest model is obtained by combining the decision trees 。 3.The trust evaluation method for emergency messages in V2X according to claim 1, characterized in that, Step S2 is specifically: The vehicle is registered at the vehicle registration center, which then identifies the vehicle. Initial trust value Public key ,describe After concatenating the strings, we get || || || It saves the data to the blockchain, completes the registration, and returns a success result. 4.The trust evaluation method for emergency messages in V2X according to claim 1, wherein, Step S3 is specifically: A certain vehicle , after sensing an emergency accident , converts the accident and related information into a message in a fixed format through an on-board OBU , and indicates in, For vehicles Corresponding identity identifier, For vehicles public key, Indicates the time of the accident. Indicates the location where the accident occurred. Indicates vehicle For the accident Detailed description, Indicates environmental information, This indicates information about the vehicle itself. Indicates road information; Finally, the vehicle generates an emergency message and broadcasts it to the remaining vehicles. 5.The trust evaluation method for emergency messages in V2X according to claim 1, wherein, Step S8 is specifically: The vehicle receiving the emergency message selects whether to trust the emergency accident message according to the emergency message trust evaluation result. If trusted, the driving route is changed or the vehicle speed is slowed down. If not trusted, the vehicle continues to drive and feeds back the authenticity of the accident message on the accident road segment. The RSU collects the feedback result and reevaluates the trust value of the emergency message broadcasting vehicle.
6. An Internet of Vehicles emergency message trust evaluation system, characterized in that, The system adopts the vehicle-to-everything emergency message trust evaluation method of any one of claims 1-5, and the system comprises a blockchain network, a vehicle registration center VRC, a roadside unit RSU, an edge server ES, and a vehicle, comprising: A system initialization module for initializing each component and node of the system; A vehicle registration module for registering the vehicle in the vehicle registration center; An emergency message spreading module for spreading emergency messages by the vehicle after perceiving an emergency accident; The trust value query and evaluation module is configured to query the trust value by the vehicle to the road side unit, query the trust value of the vehicle by the road side unit interacting with the blockchain network, and analyze the emergency message by the road side unit interacting with the edge server. The emergency message feedback module is configured to feed back the authenticity of the emergency message by the vehicle.