Anti-attack node density adaptive event verification method in vehicle-mounted ad hoc network

Abstract

The invention relates to an anti-attack node density adaptive event verification method in a vehicle-mounted ad hoc network, which is used for solving the problems of lack of evaluation basis for the credibility of event messages, lack of persuasive event authenticity reasoning, excessive dependence on an ideal environment and the like in the existing event verification method. According to the method, the reputation is used as an important evaluation factor for the authenticity of the event message, the credibility of the event message sent by the vehicle node is represented and evaluated in multiple dimensions, various factors related to the authenticity of the event are comprehensively considered, a reasonable evidence fusion mode is used, and finally the authenticity of the event is inferred, so that the reliability of the event message is improved. Therefore, the influence of hostile attacks is reduced, the performance of event verification is improved, and the performance of an event verification scheme in non-ideal environments such as small node density or multiple malicious nodes is guaranteed.

Description

Technical Field

[0001] The invention belongs to the technical field of vehicle-mounted ad hoc networks, and in particular relates to an attack-resistant node density adaptability event verification method. Background Art

[0002] Vehicle-mounted ad hoc networks (VANETs) are wireless ad hoc networking technologies based on vehicle-mounted devices. They realize information interaction and communication connection between Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) through automatic networking, changing the operation mode of traditional transportation. In recent years, with the development of autonomous driving technology and the advancement of smart city construction, traditional traffic management systems have gradually transformed into intelligent traffic management systems (ITS). In this context, VANETs have shown great potential in providing safety applications. Applying them to transportation systems can not only effectively improve the efficiency of traffic management, but also provide strong guarantees for road safety.

[0003] In VANETs, ​​vehicles are usually equipped with various sensors as tools to continuously perceive the surrounding environment and collect effective data, which enables vehicles to detect abnormal situations such as changes in road conditions, thereby reducing their own safety risks. At the same time, V2V and V2I communications can enable the dissemination of information about road conditions, traffic accidents, bad weather, and other related traffic events. This is not limited to sharing between neighbors. For events that last a long time, the powerful dissemination capabilities of the road test unit (RSU) can also help other vehicles with overlapping routes to understand the traffic conditions in a timely manner to consider whether to change routes in advance. In addition, ITS uses RSU to collect traffic event information, which is also helpful for real-time grasp of first-hand traffic conditions, and optimizes traffic flow by controlling traffic lights and other measures, quickly mobilizes rescue forces, coordinates various departments for emergency response, and improves accident response efficiency. At the same time, a large amount of traffic information is also conducive to analyzing traffic development trends and optimizing traffic networks.

[0004] At present, a large number of studies have focused on security issues such as privacy and identity in VANETs. However, due to the high mobility of vehicles and the openness and dynamism of vehicle networks, authenticated legitimate vehicles are not completely trustworthy, and malicious attackers are still a huge security challenge. Attackers may maliciously spread false messages, resulting in uncertainty in traffic event information and causing damage to both traffic safety and management efficiency. Therefore, how to verify and judge the authenticity of event messages and how to deal with attacks on VANETs intelligent traffic safety applications are urgent issues that need to be solved.

[0005] Existing event verification is mainly based on the number of votes. This idea regards reports on the same event sent by different vehicles as votes for the authenticity of the event. All vehicles reporting the event are regarded as witnesses of the event. If the total number of votes is large enough, the event is judged to be true. Although this method is simple, it also has loopholes that cannot be ignored: in areas or time periods with low vehicle density, the total number of vehicles around the event location may be difficult to reach the threshold of voting validity, resulting in misjudgment, so the threshold cannot be too large. However, a smaller threshold is difficult to resist malicious attackers, which leads to a dilemma when determining the threshold. In addition, existing studies often assume that there are more vehicles and fewer malicious attackers, resulting in the event verification scheme being not convincing enough and the judgment of the authenticity of the event not being accurate enough.

[0006] In order to make the interaction between nodes in the network more reliable, many studies have introduced trust management mechanisms in VANETs. Usually, in a vehicle network embedded with a trust management mechanism, each vehicle has a specific reputation to reflect its overall image and performance over a long period of time in the past, which is calculated based on the vehicle's behavior record, safety, degree of cooperation, etc. Reputation breaks the generality of traffic event messages from different vehicles and can more reasonably distinguish potential attackers from other vehicles. By using reputation as an important factor in measuring the credibility of event messages sent by vehicles, the possibility of the event actually occurring can be more reasonably inferred to resist malicious attacks. Summary of the invention

[0007] Due to the characteristics of VANETs, ​​malicious vehicles are inevitable, and they may provide false event information to obtain road privileges or meet other malicious purposes. Therefore, different vehicles may have conflicting opinions on the authenticity of the same event, and researchers have begun to focus on how to judge the authenticity of events. The current event verification scheme still has the following problems: 1) The authenticity of the event depends entirely on the number of votes and thresholds, which has limitations when the vehicle density is low. 2) There is a lack of multi-dimensional utilization of other information related to the authenticity of the event, such as the credibility of the voter itself, the distance between the event location and the voter's location, etc. 3) There is a lack of solutions that are applicable to various node densities and malicious node percentages. Therefore, how to determine a more effective and convincing event verification scheme based on the characteristics of the VANET network is an urgent problem to be solved.

[0008] In order to solve the technical problems existing in the prior art, including the technical problems mentioned above, the present invention provides an attack-resistant node density adaptability event verification method, and the technical solution is as follows:

[0009] Step 1: First, the present invention classifies the nodes existing in the network and constructs an attack model for the event verification scenario, dividing the attacks into two categories: malicious verification attacks and false report attacks, and the two attacks have their own characteristics.

[0010] Step 2: The workflow of the present invention is divided into three stages: reporting, verification, and judgment. In the reporting stage, the event reporter sends a report on the event information to the RSU; in the verification stage, the RSU requests other nodes around the event to submit a verification report as a verifier; in the judgment stage, the RSU uses the information provided by the reporter and all verifiers to judge the authenticity of the event.

[0011] Step 3: Next, a credibility assessment model based on fuzzy logic is proposed to characterize and evaluate the credibility of event messages sent by vehicle nodes through dimensions such as space, time, and reputation, thereby reducing the impact of malicious attacks.

[0012] Step 4: Based on the proposed credibility assessment model, the present invention designs a comprehensive event authenticity reasoning method, which takes into account factors such as credibility, number of votes, event importance, predefined roles, etc., and ultimately makes a reasonable inference on the authenticity of the event.

[0013] The following is a specific description of the method of the present invention.

[0014] Step 1: In the context of collecting and disseminating VANET events, classify the nodes in the network, including vehicle nodes and RSU nodes; each vehicle node in the network has a unique ID V and reputation r. Vehicle nodes are divided into trusted nodes and malicious nodes according to whether they are likely to carry out malicious attacks in the network. Trusted nodes are honest nodes that will never carry out malicious attacks. Trusted nodes are divided into regular nodes and public trust nodes.

[0015] Messages sent by malicious nodes to maliciously sabotage event verification correctness are collectively referred to as malicious messages. Existing malicious attacks include malicious verification attacks and false report attacks;

[0016] Step 2: Reporting and verification of incidents;

[0017] When a traffic incident occurs k When t, the witness node of the event sends an event report to the RSU as the reporter. After receiving the first report about the event, the RSU enters the reporting phase, which lasts for t 1， A reporter set is created, and the event report includes event information and reporter information.

[0018] After the reporting phase, the RSU calculates the average of the node positions based on all the reports received as the actual location of the event. At the same time, the road where the reporter is located is taken as the road where the event occurred, and the earliest timestamp in the report is taken as the time of the event. The importance factor of the event is calculated based on the importance of the event identified by different reporters.

[0019] Next, RSU is the traffic event E k Create a unique ID k , broadcast the verification request for the event, enter the verification phase, and create a verifier set; after receiving the request, other nodes around the event location except the reporter will judge the authenticity of the described event and give authenticity opinions in the verification report as verifiers;

[0020] After receiving the verification message, RSU will check the validity of the verification based on the event ID, node location, road number and timestamp; the verification phase ends after time t2, and RSU checks the validity of the verification based on the ID k Collect all relevant verifications to further determine the authenticity of the incident using the information provided by the reporter and verifier;

[0021] Step 3: Establish a credibility evaluation model based on fuzzy control, with node N i Proximity With reputation i As input, through fuzzification, fuzzy reasoning and defuzzification, the output is the credibility of the assessment

[0022] For an event, N i Is a reporter or a validator, the proximity of the node Calculated by the following formula:

[0023]

[0024] in, and Representing temporal proximity and spatial proximity respectively. Measured by the time difference between the event occurrence time and the time reported by the node, It is measured based on the spatial distance between the event location and the node;

[0025] Fuzzy the proximity and reputation i Divided into three levels: low, medium and high, credibility Divided into five levels: low, lower, medium, higher, and high; the output is obtained according to the set fuzzy control rules, and then defuzzified to obtain the credibility of the evaluation

[0026] Step 4: Reasoning about the authenticity of the event based on the theory of evidence. If the final result does not meet the conditions for determining the authenticity of the event, secondary reasoning is performed;

[0027] For the authenticity of events, there is an identification framework in Representative event E k is true, Representative event E k is false, Indicates that event E cannot be judged k The authenticity of the three is independent of each other; the node credibility evaluated by the credibility evaluation model is used to obtain the basic probability distribution BPA, which is as follows:

[0028] For reporters and those who believe that the event E k For real verifiers:

[0029]

[0030] For the event E k For the fake verifier:

[0031]

[0032] in, is any node N i The basic confidence function of i ∈Ptn represents a public trust node, t f is an intermediate variable;

[0033] Then evidence fusion is performed, and calculation The fusion probability

[0034]

[0035] Represents a fusion operation, as follows:

[0036]

[0037] Where X and Y are arbitrary subsets of the identification framework Ω, and K is an intermediate variable in the calculation process; when Much greater than When , the event is true, otherwise it is a false event; but when When , secondary reasoning is required, and Δ is the threshold value requiring secondary reasoning;

[0038] In the second inference, RSU will report and think that the event E k The true verifier is put into the confirmer set Con kIn the above example, we consider event E k Put the false verifier into the denier set Deny k |Con k | and |Deny k |Equal to and It indicates the number of confirmers and the number of denyers; for Con k Any node N in i and Deny k Any node N in j , calculate their weights β respectively i and γ j :

[0039]

[0040] Then calculate the trust factors of the confirmer set and the denier set respectively and

[0041]

[0042] Set the vote validity threshold to Its importance factor IF k The relationship is as follows:

[0043]

[0044] Among them, n0 is the set value, Δn determines IF k rate of growth;

[0045] Reporters and verifiers are collectively referred to as participants. for and The sum of and In the case of and Event E k is judged to be true; on the contrary, if and Event E k is judged as false, where Thr1 is the set threshold; in addition to these two cases, the comprehensive factor is introduced and Respectively represent event E k True or false, when When , the event is judged to be true, otherwise it is judged to be false, and the calculation is as follows:

[0046]

[0047] Among them, θ1 and θ2 are the trust factors and The coefficient, scaling factor and The calculation method is as follows:

[0048]

[0049] in, It's Con k The number of trusted nodes in the It's Deny k The number of trusted nodes in the network.

[0050] When the vehicle density is small, The threshold Thr2 is used to distinguish true and false events.

[0051] Furthermore, the format of the event report is as follows:

[0052]

[0053] Among them, Msg R It is an incident report, ID i The reporter is N i The identity of Loc i and Road i Represents the reporter's location and the road number, Type k and Used to indicate the type and importance of the reported incident. It is divided into three levels: low, medium and high, represented by -1, 0 and 1. i is the timestamp of the reporter;

[0054] The format of the verification request is as follows:

[0055] Req E ={ID RSU ,ID k ,Type k ,RL k ,RD k ,RT k ,TS k}

[0056] Among them, Req E To verify the request, ID RSU and TS k Represents the RSU identity and the time when the verification phase starts, RL k is the real location of the event, RD kThe road where the reporter is located is the road where the incident occurred, RT k The time when the event occurred;

[0057] The format of the verification report is as follows:

[0058]

[0059] Among them, Msg V To verify the report, ID j Is a validator N j The identity of Loc j and Road j Represents the location of the verifier and the road number, is the verifier’s authenticity opinion, TS j is the timestamp of the validator.

[0060] Furthermore, the importance factor is calculated by the following formula:

[0061]

[0062] Among them, Rep. k Traffic incident E k The set of reporters.

[0063] Furthermore, the time proximity It is calculated in the following way:

[0064]

[0065] Δt max =t1+t2

[0066] Among them, for the reporter, Is the time the message was received RT k For the verifier, Is the time the message was received The average of the timestamps reported by the reporter The difference, Δt max It is the sum of the duration of the reporting and verification phases;

[0067] The spatial proximity It is calculated by:

[0068]

[0069]

[0070] Among them, (lng1, lat1) is the longitude and latitude of the event location, (lng2, lat2) is the longitude and latitude of the node location, R is the radius of the earth, For node N i The spatial distance from the event, d max is the maximum threshold.

[0071] Furthermore, the fuzzy control rules are as follows:

[0072] reputation Proximity Credibility L L L L M RL L H M M L RL M M M M H RH H L RH H M RH H H H

[0073] Among them, L, RL, M, RH, and H represent low, lower, medium, higher, and high levels respectively.

[0074] Furthermore, the defuzzification is specifically obtained by using a centroid method.

[0075] Furthermore, before the evidence is fused, a Bayesian approximation is performed on the basic probability distribution:

[0076]

[0077] in, for Representation after Bayesian approximation.

[0078] Furthermore, the trust factor and The coefficients θ1 and θ2 are calculated by the following formula:

[0079]

[0080] Among them, u and v are set parameters, and e is a natural constant.

[0081] The beneficial effects brought by the technical solution provided by the present invention are as follows:

[0082] 1. A credibility assessment model based on fuzzy logic is designed to evaluate the credibility of vehicle nodes' opinions on the authenticity of events from three aspects: time proximity, spatial proximity and reputation. It can use multi-dimensional effective information to distinguish malicious messages and effectively reduce the impact of malicious verification attacks or false report attacks.

[0083] 2. An event authenticity reasoning algorithm is designed, which does not use the number of votes as the main reasoning basis. It uses evidence theory to reasonably integrate the credibility of nodes. It also considers relevant factors such as event importance, predefined roles, and voting ratio, reducing the possibility of misjudgment.

[0084] 3. Since the judgment of the authenticity of the event does not simply rely on the threshold method and has a weak dependence on the proportion of trusted nodes, the present invention ensures its applicability under various node densities and malicious node percentages. Even when the node density is small or the percentage of malicious nodes is high, it can have good performance and improve the accuracy of event verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is a work flow chart of the present invention;

[0086] Figure 2 It is a general flow chart of event authenticity judgment of the present invention;

[0087] Figure 3 It is a schematic diagram of the membership function designed by the present invention;

[0088] Figure 4 It is the overall flow chart of the event authenticity reasoning of the present invention;

[0089] Figure 5 It is a secondary reasoning flow chart of the authenticity of the events of the present invention. DETAILED DESCRIPTION

[0090] The specific implementation of the present invention will be described in detail below in conjunction with the accompanying drawings and attached tables so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when the detailed description of known functions and designs will dilute the main content of the present invention, these descriptions will be omitted here.

[0091] As a special type of self-organizing network, the main research directions of vehicle-mounted ad hoc networks (VANETs) are: network protocol design, low-latency applications, data sharing and management, security and privacy protection, etc. Among them, the purpose of network protocol design is to develop efficient routing protocols to ensure the rapid transmission of data between vehicles. Low-latency applications are mainly aimed at autonomous driving and vehicle networking applications, studying low-latency communication technologies and optimizing network architecture to support high-frequency data exchange. In addition, research on data sharing and management mainly focuses on how to store and analyze large-scale data, and how to effectively share data between vehicles, such as traffic information, hazard warnings, etc. As for security and privacy protection, research is mainly carried out in data encryption, authentication mechanisms, trust management, identity privacy, and security applications.

[0092] The present invention is mainly applied to the security of VANETs. For the application of combining VANETs with intelligent transportation systems (ITS), an attack-resistant node density adaptive event verification method is proposed to solve the problem of event authenticity judgment. The road test unit (RSU) can use the designed credibility evaluation model to evaluate the credibility of each node's opinion on the authenticity of the event, and based on the evaluation results, reasonably use the credibility of each node's opinion and other related factors to finally infer the authenticity of the event, thereby improving the accuracy of event verification under various node densities and malicious node percentages. The specific implementation steps of the method are as follows:

[0093] Step 1: In the scenario of VANETs event collection and dissemination, it mainly includes the communication between vehicle nodes and RSU nodes. Therefore, the nodes in the network can be mainly divided into two categories: vehicle nodes and RSU nodes.

[0094] (1) Vehicle nodes: In VANETs, ​​all vehicle nodes communicate through the same wireless communication protocol, such as IEEE 802.11p. These nodes have certain limitations in terms of energy consumption, computing processing and data storage capabilities. Each vehicle node in the network has a unique ID. V , whose reputation r has been calculated based on factors such as past behavior records, and all reputations are within the legal range. According to whether the node has the possibility of malicious attacks in the network, it can be divided into trusted nodes and malicious nodes.

[0095] 1) Trusted nodes: honest nodes that will never conduct malicious attacks. Trusted nodes are divided into regular nodes and public trust nodes according to the predefined roles assigned to each node. Public trust nodes are all trusted nodes, with a proportion of μ in all nodes, and are considered to exhibit stronger cooperation and reliability. Trusted nodes do not necessarily contribute to the collection and dissemination of events. Assume that regular nodes and public trust nodes participate in the reporting or verification of events with probabilities p1 and p2 respectively, where p1 <p2。

[0096] 2) Malicious nodes: dishonest nodes that may carry out malicious attacks. All malicious nodes are regular nodes and do not necessarily pose a threat to normal activities in the network. Assume that a malicious node attacks with a probability of p3, and p3 changes with the node's reputation. It is assumed that when the reputation of the malicious node is around 0.6 (medium range), p3 is the largest. The reason is that when the node reputation is low, it will avoid attacks in order not to be isolated from the network; when the node reputation is high, it means that it has not attacked for a long time, and the probability of attacking again is still low.

[0097] (2) RSU node: As a fixed communication node deployed on road infrastructure, it plays a key role in realizing VAENT services, improving network performance and security. RSU has stronger computing power and storage capacity than vehicle nodes, and its coverage is much larger than the communication radius of vehicle nodes, which can achieve wider information dissemination. Assuming that RSU is trustworthy, it will work normally according to the workflow, and it stores the reputation and predefined roles of all vehicle nodes in the network, but cannot distinguish between trusted nodes and malicious nodes. After judging the authenticity of the event, RSU broadcasts the information of the real event and stores it through the cloud, InterPlanetary File System (IPFS), etc.

[0098] Malicious nodes carry out malicious attacks without reason, indiscriminately, and randomly. Specifically, the messages sent by malicious nodes to verify the correctness of malicious sabotage events are collectively referred to as malicious messages. There are two possible malicious attacks:

[0099] (1) Malicious verification attack: When an incident does occur, the attacker may not report the incident to the RSU and perform a false verification. This process can be done alone or in collusion with other attackers. If the RSU judges the real incident as false, it may lead to the inability to handle the accident scene in a timely manner and the inability to provide timely treatment for the injured.

[0100] (2) False report attack: When no event occurs, the attacker may send a forged event report to the RSU. This process can also be done individually or in collusion. If the RSU judges the false event as real, it may cause other nodes to abandon their original optimal route, and even cause the transportation, medical and other related departments to waste money, time, manpower and other resources.

[0101] Step 2: The workflow of this method is divided into three stages: reporting, verification, and judgment. Figure 1 The workflow diagram of this method is as follows:

[0102] Reporting stage: When a traffic incident occurs k When the event is witnessed by node N i Send event report Msg to RSU as reporter R , RSU enters the reporting phase after receiving the first report about the event. The duration of this phase is t1, at which time RSU will create a reporter set Rep k , and wait for more information about the traffic incident E k Msg R The format is as follows:

[0103]

[0104] Where ID i The reporter is N i The identity of Loc i and Road i Respectively represent the reporter's location and the road number. k and Used to indicate the type and importance of the reported incident. It is divided into three levels: low, medium and high, represented by -1, 0 and 1. i is the timestamp of the reporter.

[0105] After the reporting phase, the RSU calculates the average of the node positions based on all the reports received as the true position RL of the event. k At the same time, the road where the reporter is located is taken as the road RD where the incident occurred. k The earliest timestamp in the report is used as the event occurrence time RT k Due to the importance of different nodes to the reported events There are differences in opinions, and the importance factor IF is calculated by the following formula k Can represent the comprehensive importance of the event:

[0106]

[0107] Among them, Rep. k Traffic incident E k The set of reporters.

[0108] Verification phase: Next, RSU is E k Create a unique ID k , broadcast the verification request Req for this event E , enter the verification phase and create a verifier set Ver k . ID RSU and TS k Respectively represent the RSU identity and the time when the verification phase starts, Req E The format is as follows:

[0109] Req E ={ID RSU ,ID k ,Type k ,RL k ,RD k ,RT k ,TS k}(3)

[0110] After receiving the request, other nodes around the event location except the reporter will determine the authenticity of the described event and act as a verifier in verifying the report Msg V The authenticity of the comments is given in Msg V The format is as follows:

[0111]

[0112] Where ID j Is a validator N j The identity of Loc j and Road j They represent the location of the verifier and the road number. is the verifier’s authenticity opinion, TS j is the timestamp of the validator.

[0113] Judgment phase: In order to prevent invalid verification of nodes, RSU will check the validity of verification based on event ID, node location, road number and timestamp after receiving the verification message. The verification phase ends after time t2, and RSU can k Collect all relevant verifications, and use the information provided by the reporter and verifier to further determine the authenticity of the event. It should be noted that there are many studies on node location detection and verification, so this embodiment does not consider the problem of node forgery.

[0114] In addition, if Figure 2 As shown in the figure, the event authenticity judgment consists of two parts: credibility assessment and event authenticity inference. RSU first processes the information collected from reporters and verifiers as the basis for credibility assessment, and then reasonably integrates the credibility of the opinions of each node, taking into account relevant factors such as the number of votes and predefined roles, so as to infer the authenticity of the relevant events. Next, the event authenticity judgment is further explained.

[0115] Step 3: Different nodes may give different opinions on the authenticity of the event. In order to correctly judge whether the event is true or not, it is an indispensable link to evaluate the trustworthiness of the opinions of each node. The reporter or verifier must send information such as location and timestamp to prove the validity of their opinions, which affects the credibility of the node's opinions. In addition, the reputation of the node itself reflects its overall image. The higher it is, the more credible it was in the past, which also affects the current credibility assessment to a certain extent. Fuzzy logic is a generalization of Boolean logic that is either 0 or 1. It removes the strict boundaries between different classifications to characterize the ambiguity of data size. Therefore, fuzzy logic is often used for approximate reasoning rather than precise determination. Due to the incompleteness and uncertainty of the information provided by the node, fuzzy logic can better handle the credibility assessment. For convenience, this method does not emphasize the distinction between the credibility of the node and the credibility of the node's opinion. The credibility of the node in the following text means the credibility of the node's opinion.

[0116] For an event, N i Can be a reporter or a verifier. If N i The position of is closer to the event location, so it can be considered more credible at the spatial level. i The closer the reporting or verification time is to the event, the more reliable it can be considered in terms of time. The proximity of a node to an event can be measured as follows:

[0117]

[0118] in and Representing temporal proximity and spatial proximity respectively. It can be calculated by formula (6). From formulas (7)-(9), we can see that for the reporter, Is the time the message was received RT k For the verifier, Is the time the message was received The average of the timestamps reported by the reporter This distinction can prevent the verifier from Always less than the reporter. Δt max It is the sum of the duration of the two phases of reporting and verification. Therefore, if the node reports or verifies more promptly, Will be bigger.

[0119]

[0120] Δt max=t1+t2(9)

[0121] In addition, according to the longitude and latitude of the event location (lng1, lat1) and the longitude and latitude of the node location (lng2, lat2), as well as the radius of the earth R, the node N can be calculated i Spatial distance from the event As shown in formula (10):

[0122]

[0123] like The smaller it is, the more node N i The more credible it may be, will also be larger, as shown in formula (11). max is the maximum distance threshold, only distances not exceeding this threshold are considered valid.

[0124]

[0125] A credibility evaluation model based on fuzzy logic is established. i Proximity With reputation i As input, the credibility of the assessment As output, the whole process of credibility assessment is based on fuzzy logic and is divided into three steps: fuzzification, fuzzy reasoning and defuzzification.

[0126] Fuzzification: In fuzzy logic, membership is used to measure the degree to which data belongs to different categories. The larger the value, the higher the probability of belonging to the category, and vice versa. and reputation i Divided into three levels: low (L), medium (M) and high (H), credibility It is divided into five levels: low (L), lower (RL), medium (M), higher (RH), and high (H). The membership function can indicate the degree to which a certain value belongs to different levels. Figure 3 The designed proximity membership function, reputation membership function and credibility membership function, wherein the minimum value of legitimate reputation is 0.3 as an example. According to the membership function, when the proximity is 0.4, the probability of the proximity belonging to the low, medium and high levels is 0.5, 0.5 and 0 respectively, which means that the value 0.4 is considered to be the same as the low and medium levels, and is not considered to be a high level at all. The selection of the membership function can be determined based on experience and needs, and the present invention adopts the classic trapezoidal and triangular functions.

[0127] Fuzzy reasoning: After fuzzifying the input, the fuzzy output can be obtained by combining the reasoning method and fuzzy control rules. Table 1 shows the designed fuzzy control rules, which are used to connect the input and output, and describe the output rules corresponding to different inputs. For example, if the reputation is medium and the proximity is low, the credibility is low.

[0128] Table 1 is the fuzzy control rules designed by the present invention.

[0129] Table 1

[0130] Serial number reputation Proximity Credibility 1 L L L 2 L M RL 3 L H M 4 M L RL 5 M M M 6 M H RH 7 H L RH 8 H M RH 9 H H H

[0131] Defuzzification: The output obtained according to the fuzzy control rules is not the final specific value, so defuzzification is still required.

[0132] The designed model uses the centroid method for defuzzification, as shown in formula (12).

[0133]

[0134] Among them, μ(x) is the credibility membership function, ∫·dx represents the integral, and the range of the integral is obtained by combining the specific values ​​of proximity and reputation with the output of the fuzzy control rule. The calculation process and results of this method are easy to understand and explain. It is the most reasonable and accurate method, and the use of trapezoidal and triangular membership functions makes it unnecessary to worry about the complexity of calculation.

[0135] Step 4: The information provided by the nodes often has a certain degree of uncertainty. Starting from these initial evidences, we draw conclusions that have a certain degree of uncertainty but are still reasonable or basically reasonable through reasoning. This process is called uncertainty reasoning. By using the method of uncertainty reasoning, the ultimate goal is to draw a judgment result on the authenticity of the event, so how to reasonably integrate the evidence of each node is the key.

[0136] Evidence theory is a probability-based evidence fusion calculation method. Compared with Bayesian reasoning that requires prior probability as a known condition, evidence theory is more suitable for the problems considered in this invention. For the authenticity of an event, the identification framework Ω is in Representative event E k is true, Representative event E k is false, Indicates that event E cannot be judged k The authenticity of the three is independent of each other. l = 1, 2, 3 is the basic confidence function on the power set P(Ω), representing The probability of occurrence, using the node credibility, node opinion and predefined roles evaluated by the credibility assessment model, can be used to obtain the basic probability assignment (BPA). is any node N i The basic confidence function of E. k For a verifier that is true, its BPA is shown in formula (13). Similarly, for a verifier that believes that event E k For a fake verifier, its BPA is shown in formula (14). i ∈Ptn) is more reliable than regular nodes, and the probability distribution of the two is different, as shown in formula (15), α is a role weight between 0 and 1, and its size can be determined based on experience.

[0137]

[0138] Through the fusion based on evidence theory, we can calculate according to formula (16): The fusion probability The specific calculation method is shown in formula (17), where X and Y are arbitrary subsets of the identification framework Ω, and K is an intermediate variable in the calculation process.

[0139]

[0140] The specific operations are as follows:

[0141]

[0142] In addition, performing Bayesian approximation on BPA before evidence fusion can greatly simplify the calculation and improve the efficiency of fusion, as shown in formula (18).

[0143]

[0144] The overall flow chart of event authenticity reasoning is as follows: Figure 4 As shown, when Much greater than When the event is true, it can be considered as a false event. When , the judgment will be more uncertain, and other valid information must be used for secondary reasoning. Δ is the threshold that requires secondary reasoning.

[0145] In the second inference, RSU will report and think that the event E k The true verifier is put into the confirmer set Con k In the above example, we consider event E k Put the false verifier into the denier set Deny k |Conk | and |Deny k |Equal to and It represents the number of confirmers and the number of denyers. k Any node N in i and Deny k Any node N in j , calculate the weight β according to formula (19) i and γ j Next, calculate the trust factors of the two sets separately and As shown in formula (20).

[0146]

[0147] In addition, reporters and verifiers can be collectively referred to as participants, and the number of participants That is and sum. when When the number of votes is greater than the number of votes persuasive, it is reasonable to use the number of votes as one of the factors for judging the authenticity of the event. is the voting validity threshold, The magnitude of the event will be affected by the event's importance factor IF k The impact of important events and unimportant events has different requirements for the threshold. k <0, the event is not very important, It can be set to a general value n0 that is not too small; when IF k ≥0, IF k The larger the value, the more important the event is in the eyes of more nodes. Highly important events may have a significant impact on traffic safety and driving environment. Therefore, in order to treat such events more carefully, Will follow IF k increases with the increase of , Δn determines IF k The growth rate is shown in formula (21).

[0148]

[0149] Specifically, if Figure 5 As shown in the figure, it is a flowchart of secondary reasoning, and the reasoning result will be used as the final judgment result; and In the case of and Event E k is judged to be true; on the contrary, if and Event E k In addition to these two cases, the present invention takes the proportional factor into consideration to achieve the final judgment, as shown below:

[0150]

[0151] in, and Represents event E k A comprehensive factor that is true or false. When , the event is judged to be true, otherwise it is judged to be false. and The calculation method is as follows:

[0152]

[0153] in, It's Con k The number of trusted nodes in the It's Deny k The number of public trust nodes in θ1 and θ2 are the trust factors and The coefficient of or As shown in formula (24), different values ​​of u and v will affect the correlation between the trust factor and its coefficient, and e is a natural constant.

[0154]

[0155] The reason for this correlation between the two is that and Respectively reflect Con k and Deny k The comprehensive trustworthiness of the nodes in the two sets is expressed as Con k For example, when Con k When the midpoints are not so reliable, more consideration of the scaling factor is needed to prevent misjudgment, which also means that a larger To make the event be judged as true; when Con k When the intermediate nodes are relatively trustworthy, more consideration is given to the overall level of trustworthiness.

[0156] It should also be noted that when the vehicle density is small, there may be a situation where only a few nodes (such as one node) report the event, and no node verifies it. In this case, you can set The threshold Thr2 is used to distinguish true and false events. The value of Thr2 depends on specific requirements.

[0157] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A method for event verification of node density adaptability against attacks in a vehicle-mounted ad hoc network, characterized in that: It includes the following steps: Step 1: In the context of collecting and disseminating VANET events, classify the nodes in the network, including vehicle nodes and RSU nodes; each vehicle node in the network has a unique ID V and reputation r. Vehicle nodes are divided into trusted nodes and malicious nodes according to whether they are likely to carry out malicious attacks in the network. Trusted nodes are honest nodes that will never carry out malicious attacks. Trusted nodes are divided into regular nodes and public trust nodes. The messages sent by malicious nodes for verifying the correctness of malicious destruction events are collectively referred to as malicious messages. Existing malicious attacks include malicious verification attacks and false report attacks. Step 2: Reporting and verification of events; When a traffic incident occurs k When t, the witness node of the event sends an event report to the RSU as the reporter. After receiving the first report about the event, the RSU enters the reporting phase, which lasts for t 1， A reporter set is created, and the event report includes event information and reporter information. After the reporting stage, the RSU calculates the average value of the node positions based on all the received reports as the true position of the event. At the same time, the road where the reporter is located is used as the road where the event occurs, and the earliest timestamp in the report is used as the event occurrence time. Calculate the importance factor of the event according to the importance of the event recognized by different reporters. Next, RSU is the traffic event E k Create a unique ID k , broadcast the verification request for the event, enter the verification phase, and create a verifier set; after receiving the request, other nodes around the event location except the reporter will judge the authenticity of the described event and give authenticity opinions in the verification report as verifiers; After receiving the verification message, RSU will check the validity of the verification based on the event ID, node location, road number and timestamp; the verification phase ends after time t2, and RSU checks the validity of the verification based on the ID k Collect all relevant verifications to further determine the authenticity of the incident using the information provided by the reporter and verifier; Step 3: Establish a credibility evaluation model based on fuzzy logic, with node N i Proximity With reputation i As input, through fuzzification, fuzzy reasoning and defuzzification, the output is the credibility of the assessment For an event, N i Is a reporter or a validator, the proximity of the node Calculated by the following formula: in, and Representing temporal proximity and spatial proximity respectively. Measured by the time difference between the event occurrence time and the time reported by the node, It is measured based on the spatial distance between the event location and the node; Fuzzy the proximity and reputation i Divided into three levels: low, medium and high, credibility Divided into five levels: low, lower, medium, higher, and high; the output is obtained according to the set fuzzy control rules, and then defuzzified to obtain the credibility of the evaluation Step 4: Conduct event authenticity reasoning based on the evidence theory. If the final result does not meet the event authenticity determination condition, secondary reasoning is carried out. For the authenticity of events, there is an identification framework in Representative event E k is true, Representative event E k is false, Indicates that event E cannot be judged k The authenticity of the three is independent of each other; the node credibility evaluated by the credibility evaluation model is used to obtain the basic probability distribution BPA, which is as follows: For reporters and those who believe that the event E k For real verifiers: For event E k For the fake verifier: in, is any node N i The basic confidence function, α is the weight value, N i ∈Ptn represents a public trust node, t f is an intermediate variable; Then evidence fusion is performed, and calculation The fusion probability Indicates the fusion operation, as follows: Where X and Y are arbitrary subsets of the identification framework Ω, and K is an intermediate variable in the calculation process; when Much greater than When , the event is true, otherwise it is a false event; but when When , secondary reasoning is required, and Δ is the threshold value requiring secondary reasoning; In the second inference, RSU will report and think that the event E k The true verifier is put into the confirmer set Con k In the above example, we consider event E k Put the false verifier into the denier set Deny k |Con k | and |Deny k |Equal to and It represents the number of confirmers and the number of deniers; for Con k Any node N in i and Deny k Any node N in j , calculate their weights β respectively i and γ j : Then calculate the trust factors of the confirmer set and the denier set respectively and Set the vote validity threshold to Its importance factor IF k The relationship is as follows: Among them, n0 is the set value, Δn determines IF k rate of growth; Reporters and verifiers are collectively referred to as participants. for and The sum of and In the case of and Event E k is judged to be true; on the contrary, if and Event E k is judged as false, where Thr1 is the set threshold; in addition to these two cases, the comprehensive factor is introduced and Respectively represent event E k True or false, when When , the event is judged to be true, otherwise it is judged to be false, and the calculation is as follows: Among them, θ1 and θ2 are the trust factors and The coefficient, scaling factor and The calculation method is as follows: in, It's Con k The number of trusted nodes in the Deny k The number of trusted nodes in the network.

2. The method for verifying the node density adaptability event of the vehicle-mounted ad hoc network against attacks according to claim 1, characterized in that: All the公信 nodes are trusted nodes, and the proportion in all nodes is μ. Assume that the normal nodes and the公信 nodes participate in the reporting or verification of events with probabilities p1 and p2 respectively, where p1 < p2. Assume that the malicious nodes attack with probability p3, and p3 changes with the node reputation.

3. The method for verifying the node density adaptability event of the vehicle-mounted ad hoc network against attacks according to claim 1, characterized in that: The format of the event report is as follows: Among them, Msg R It is an incident report, ID i The reporter is N i The identity of Loc i and Road i Represents the reporter's location and the road number, Type k and Used to indicate the type and importance of the reported incident. It is divided into three levels: low, medium and high, represented by -1, 0 and 1. i is the timestamp of the reporter; The format of the verification request is as follows: Req E ={ID RSU ,ID k ,Type k ,RL k ,RD k ,RT k ,TS k } Among them, Req E To verify the request, ID RSU and TS k Represents the RSU identity and the time when the verification phase starts, RL k is the real location of the event, RD k The road where the reporter is located is the road where the incident occurred, RT k The time when the event occurred; The format of the verification report is as follows: Among them, Msg V To verify the report, ID j Is a validator N j The identity of Loc j and Road j Represents the location of the verifier and the road number, is the verifier’s authenticity opinion, TS j is the timestamp of the validator.

4. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 3 is characterized in that: The importance factor is calculated by the following formula: Among them, Rep. k Traffic incident E k The set of reporters.

5. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 4 is characterized in that: The temporal proximity It is calculated in the following way: Δt max =t1+t2 Among them, for the reporter, Is the time the message was received RT k For the verifier, Is the time the message was received The average of the timestamps reported by the reporter The difference, Δt max It is the sum of the duration of the reporting and verification phases; The spatial proximity It is calculated by: Among them, (lng1, lat1) is the longitude and latitude of the event location, (lng2, lat2) is the longitude and latitude of the node location, R is the radius of the earth, For node N i The spatial distance from the event, d max is the maximum threshold.

6. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 5 is characterized in that: The specific fuzzy control rules are as follows: Among them, L, RL, M, RH, and H respectively represent low level, relatively low level, medium level, relatively high level, and high level.

7. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 6 is characterized in that: The defuzzification is specifically obtained by the centroid method.

8. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 7, characterized in that: Before the evidence fusion, Bayesian approximation is performed on the basic probability assignment: in, for Representation after Bayesian approximation.

9. The method for verifying the node density adaptability event of the vehicle ad hoc network against attacks according to claim 8, characterized in that: The trust factor and The coefficients θ1 and θ2 are calculated by the following formula: Among them, u and v are set parameters, and e is the natural constant. It should be noted that there is an unclear term "公信节点" in the original text. It might need to be further clarified to ensure more accurate translation. Here it is directly translated as "公信节点" for the sake of following the translation rules.

10. The method for verifying the node density adaptability event against attacks in a vehicle-mounted ad hoc network according to claim 9, characterized in that: When the vehicle density is small, The threshold Thr2 is used to distinguish true and false events.

Images