An autonomous driving network malicious node detection method and system

Abstract

The application provides an automatic driving network malicious node detection method and system, relates to the technical field of automatic driving, and solves the problem of efficient and safe monitoring of malicious nodes. The method comprises the following steps: a trusted fixed device collects positions uploaded by all vehicles passing through a covered road section at the current time, generates a vehicle position topology graph as a real-time position topology graph; the real-time position topology graph and a historical position topology graph are subjected to graph structure detection, and a node with the smallest average distance to other nodes in the real-time position topology graph is selected as a monitoring node; the monitoring node calculates the confidence of other nodes through broadcast messages; when the monitoring node receives a voting request message of other nodes, the monitoring node queries an inspection node corresponding to the voting request message, and if the confidence of the inspection node is lower than a confidence threshold, the monitoring node initiates voting for the inspection node; the monitoring node performs weighted calculation on the voting of other nodes, and performs screening on malicious nodes.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, specifically to a method and system for detecting malicious nodes in an autonomous driving network. Background Technology

[0002] With the rapid development of autonomous driving technology, the connectivity between autonomous vehicles and the internet is increasing. However, the security of autonomous driving networks faces serious challenges. Malicious nodes may impersonate sensors, autonomous vehicles, and edge processing servers to launch targeted attacks on autonomous driving networks, thereby affecting the safety and reliability of autonomous driving. Therefore, efficient and secure monitoring of malicious nodes has become an urgent problem to be solved. Summary of the Invention

[0003] In view of the shortcomings of the aforementioned related technologies, this application provides a method and system for detecting malicious nodes in autonomous driving networks to solve the above-mentioned technical problems.

[0004] Firstly, this application provides a method for detecting malicious nodes in an autonomous driving network, comprising:

[0005] The system relies on fixed equipment to collect the locations of all vehicles passing through the covered road segment at the current moment, generating a vehicle location topology map as a real-time location topology map.

[0006] The fixed equipment receives the vehicle position topology map of the previous time from the equipment of the previous road segment as the historical position topology map;

[0007] The fixed equipment is trusted to perform graphic structure detection on the real-time location topology map and the historical location topology map. If the stability of the graphic structure is greater than the empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node.

[0008] The monitoring node listens to the periodic broadcast messages from nodes within a fixed distance and calculates the confidence level of other nodes based on the broadcast messages;

[0009] When a monitoring node receives a voting request message from another node, it queries the check node corresponding to the voting request message. If the confidence level of the check node is lower than the confidence level threshold, it initiates a vote for the check node.

[0010] In one embodiment of this application, the method further includes:

[0011] The monitoring node performs a weighted calculation of the votes from other nodes to filter out malicious nodes.

[0012] In one embodiment of this application, the monitoring node performs a weighted calculation of the votes of other nodes to filter malicious nodes, including:

[0013] The confidence levels of each node's votes are weighted to calculate the voting results of each node against the malicious node.

[0014] In one embodiment of this application, a trusted fixed device performs graphic structure detection on a real-time location topology map and a historical location topology map. If the stability of the graphic structure is greater than an empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node, including:

[0015] If the stability of the graphical structure is greater than the empirical threshold, monitoring nodes are selected step by step according to the geometric center position of the topological structure.

[0016] In one embodiment of this application, a trusted fixed device performs graphic structure detection on a real-time location topology map and a historical location topology map. If the stability of the graphic structure is greater than an empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node, including:

[0017] Calculate the distance of each node to other nodes in turn, sort them, and select the node with the smallest distance to other nodes.

[0018] In one embodiment of this application, the trusted fixed device performs graphic structure detection on the real-time location topology map and the historical location topology map. If the stability of the graphic structure is greater than an empirical threshold, before selecting the node with the smallest average distance from other nodes in the real-time location topology map as the monitoring node, the following steps may also be performed:

[0019] Determine whether the stability of the graphical structure is greater than an empirical threshold;

[0020] If the stability of the graph structure is less than the empirical threshold, the stability of the nodes should be given priority, and the node with the highest stability should be selected as the monitoring node.

[0021] In one embodiment of this application, determining whether the stability of the graphic structure is greater than an empirical threshold includes:

[0022] Detect nodes in the location topology graph and detect changes in the set of neighbor nodes of a specific node;

[0023] By comparing the graphic composed of sequentially connected nodes with the basic geometry, the basic geometry of the real-time location topology map and the historical location topology map is found. If the number of edits required to transform the basic geometry of the historical location topology map into the basic geometry of the real-time location topology map is less than a specific number of steps, the stability of the graphic structure is determined to be greater than an empirical threshold.

[0024] In one embodiment of this application, a monitoring node listens to periodic broadcast messages from nodes within a fixed distance, and calculates the confidence level of other nodes based on the broadcast messages, including:

[0025] The detection node records the sending timestamp of messages sent by the monitored node and the receiving timestamp of messages received locally;

[0026] Calculate the time difference between the sending timestamp and the receiving timestamp;

[0027] The confidence level of the monitored node is calculated using the time difference.

[0028] In one embodiment of this application, the method further includes:

[0029] If the detection result is that the node being checked is a non-malicious node and the suspicious instruction is not executable based on roadside sensor data, then a check request to determine the sensor data is sent to each roadside device and edge server.

[0030] Roadside equipment and edge servers collect data from roadside sensors for testing. If the test is successful, a vote is initiated on the trusted fixed equipment.

[0031] Secondly, this application provides an autonomous driving network malicious node detection system, comprising:

[0032] Each node is set up in the autonomous driving network; including roadside devices, edge servers, vehicles, and sensors; roadside devices and edge servers vote to obtain trusted fixed devices; trusted fixed devices are used to perform the method of the first aspect to select monitoring nodes in vehicles; monitoring nodes perform the method of the first aspect to monitor vehicle nodes and filter malicious nodes.

[0033] As described above, the method and system for detecting malicious nodes in an autonomous driving network provided in this application have the following beneficial effects:

[0034] In the above process, since the number of vehicles passing through a certain section of road changes constantly, selecting vehicles as monitoring nodes probabilistically avoids the possibility of monitoring nodes being consistently unreliable, thus ensuring the reliability probability of the monitoring nodes. Through analysis of the vehicle node location distribution topology, monitoring nodes with low communication consumption and high reliability are selected. The selection of monitoring nodes considers the node's computing power, distance to the base station, and network topology, choosing nodes with optimal conditions as monitoring nodes.

[0035] Monitoring nodes monitor the status of all vehicles, while surveillance nodes are responsible for monitoring the content and numbering of data packets from sensors and edge servers to ensure data integrity and reliability. When an anomaly occurs at a node, the monitoring nodes initiate an assessment, vote on the issue, and promptly identify and isolate malicious nodes to ensure network security. The voting results are tallied, and a majority vote determines the authenticity of the event, thus preventing malicious uploads and erroneous events from impacting the autonomous driving network.

[0036] In the network system of autonomous vehicles, hardware and software modules based on this invention are deployed; the hardware modules are responsible for data collection and transmission, while the software modules are responsible for monitoring, judgment, and comparison; by combining the synergistic effect of the hardware and software modules, comprehensive monitoring and protection of the autonomous driving network can be achieved. Attached Figure Description

[0037] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0038] Figure 1 This is a flowchart of the steps involved in detecting malicious nodes in an autonomous driving network.

[0039] Figure 2 This is an example diagram illustrating the comparison process between historical and real-time location topology maps in one embodiment.

[0040] Figure 3 This is a flowchart of another method for detecting malicious nodes in an autonomous driving network. Detailed Implementation

[0041] The embodiments of this application will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. This application can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be understood that the preferred embodiments are only for illustrating this application and are not intended to limit the scope of protection of this application.

[0042] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. Therefore, the drawings only show the components related to this application and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0043] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the present application. However, it will be apparent to those skilled in the art that embodiments of the present application may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present application.

[0044] Example 1

[0045] This application proposes a method for detecting malicious nodes in an autonomous driving network. The method is applied to an autonomous driving network, which includes roadside equipment, roadside sensors, edge servers, base stations, vehicles, etc. Figure 1 As shown, Figure 1 This is a flowchart of the steps involved in detecting malicious nodes in an autonomous driving network. The steps include:

[0046] S1: Relying on fixed equipment to collect the locations uploaded by all vehicles passing through the covered road segment at the current moment, generating a vehicle location topology map as the real-time location topology map.

[0047] Trusted fixed devices are roadside devices selected by vehicles and other equipment in the previous time step to perform node monitoring in the next time step. For example, at time T, vehicles and other equipment select a certain edge server as a trusted fixed device, which is responsible for covering the road segment at time T+t and generating a vehicle location topology map.

[0048] Each vehicle in the topology graph can be considered a node.

[0049] Roadside equipment, base stations, and servers are deployed at intervals based on road traffic flow and the positioning and perception accuracy required for autonomous driving. For example, based on the average traffic flow and the required autonomous driving accuracy for a certain road, a set of equipment is deployed every M meters on that road. Each set of equipment includes sensors, roadside devices, and edge servers. The coverage area of ​​one set of equipment is M meters.

[0050] S2: Relying on the fixed equipment to receive the vehicle position topology map of the previous time segment transmitted by the equipment of the previous road segment as the historical position topology map.

[0051] Assuming the vehicle location topology map for the fixed equipment was generated at time T, and the equipment for the previous road segment generated its historical location topology map at time Tt, then time t is based on the vehicle speed on that road.

[0052] The normal distribution is represented by the average or median time taken to travel from one set of roadside equipment coverage to the next set of roadside equipment coverage, denoted as t.

[0053] S3: Relying on fixed equipment to perform graphic structure detection on real-time location topology map and historical location topology map, if the stability of the graphic structure is greater than the empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node.

[0054] The process of selecting monitoring nodes in S3 can be summarized as follows: if the stability of the graphical structure is greater than the empirical threshold, monitoring nodes are selected step by step according to the geometric center position of the topological structure.

[0055] For example, if the topology has a center position, the node at the center position can be selected. The sum of the distances from the center position node to other nodes is L1, and the sum of the distances from the topology vertex node to other nodes is L2, where L1 < L2. If there is no node at the center position of the topology, then the node located in the middle of the lane can be selected.

[0056] Alternatively, you can calculate the distance of each node to other nodes in turn, sort them, and select the node with the smallest distance to other nodes.

[0057] If the overall topology is stable, data transmission efficiency can be considered. Therefore, nodes at or near the geometric center are selected as monitoring nodes.

[0058] Before executing step S3, step S2-3 can be executed first. S2-3: Determine whether the stability of the graph structure is greater than the empirical threshold. If the stability of the graph structure is less than the empirical threshold, then the stability of the nodes should be given priority, and the node with the highest stability should be selected as the monitoring node.

[0059] The process of determining the stability of a graphic structure is as follows:

[0060] Following the predetermined lane sequence, starting from the middle lane and moving to the outer lanes on both sides, the system compares whether vehicles in each lane have changed lanes. If a vehicle that has not changed lanes is found, the system checks whether that vehicle has overtaken the vehicle in front of it, until the first vehicle that has neither changed lanes nor overtaken is identified as the monitoring node. If there are only two lanes, the system starts from the right slow lane and compares whether vehicles in each lane have changed lanes.

[0061] Vehicles that do not perform lane changes or overtaking maneuvers have higher stability and require less temporary computing resources, making them more reliable as monitoring nodes.

[0062] S31: Detect nodes in the location topology graph and detect changes in the set of neighbor nodes of a specific node.

[0063] Nodes that exist in both the real-time location topology map and the historical location topology map can be used as specific nodes.

[0064] S31 can be used to exclude situations where too many new nodes are added, such as vehicles newly entering the lane from a service station or road entrance; or to exclude situations where too many nodes are removed, such as vehicles leaving the road.

[0065] Changes in the set of neighboring nodes of a specific node can be detected by comparing changes in the set of neighboring nodes using Jaccard.

[0066]

[0067] Indicates vehicle iAt any moment t The stability index of the neighbor node set is such that the closer it is to 1, the smaller the change in the neighbor node set. For vehicles i At time T+ t The set of neighboring nodes, where |·| represents the number of elements in the set (cardinality). For the intersection operation of sets, It is the common part of the neighbor node sets at two different times. The union operation of sets, It represents all elements of the neighbor node set at two different times. Assume a vehicle... It is a specific vehicle, vehicle Neighbor set For vehicles within a fixed distance, this can be considered as the road segment covered by the roadside equipment located on the vehicle's section at time t, which is at a fixed distance.

[0068] If step S31 shows that the set of neighboring nodes of a specific node has changed significantly, it can be inferred that the current topology is unstable and step S32 is unnecessary.

[0069] S32: Compare the graphic composed of sequentially connected nodes with the basic geometry. By comparing, find the basic geometry of the real-time location topology and the historical location topology. If the number of edits required to transform the basic geometry of the historical location topology to the basic geometry of the real-time location topology is less than a specific number of steps, determine that the stability of the graphic structure is greater than the empirical threshold.

[0070] Figure 2 This is an example diagram illustrating the comparison process between historical and real-time location topology maps in one embodiment. Figure 2 Taking the illustrated topology as an example, in the first case, connecting nodes sequentially in the order of node 1, node 2, and node 3, the side lengths and angles of the topology graph at time T to T+t change significantly. Each side of the triangle needs to be edited, and it can be seen that the relative positions of nodes 1, 2, and 3 also change significantly. In the second case, connecting nodes sequentially in the order of node 1, node 2, and node 3, the change in the topology graph from time T to T+t is smaller, and the similarity is high. It can be determined that in the second case, the number of edits required to transform the basic geometry of the historical position topology graph to the basic geometry of the real-time position topology graph is less than a specific number of steps. Therefore, the node with the smallest average distance from other nodes in the real-time position topology graph is directly selected as the monitoring node to minimize the resource consumption for data communication between the monitoring node and other nodes.

[0071] Once the fixed equipment identifies the monitoring node, it broadcasts the monitoring node to other nodes.

[0072] S4: The monitoring node listens to the periodic broadcast messages of nodes within a fixed distance and calculates the confidence level of other nodes based on the broadcast messages.

[0073] Broadcast messages can be, for example, beacon messages, heartbeat packets, etc.

[0074] Step S4, calculating the confidence scores of other nodes, can be performed in the following manner:

[0075] S41: The detection node records the sending timestamp of the message sent by the monitored node and the receiving timestamp of the message received locally.

[0076] For example, vehicle A broadcasts a Beacon message containing its location, and the data packet is timestamped. The receive timestamp is the time recorded by the monitoring node when it receives the broadcast message.

[0077] S42: Calculate the time difference between the sending timestamp and the receiving timestamp.

[0078] S43: Calculate the confidence level of vehicle A at the monitored node using the time difference.

[0079] For example, if the time difference is greater than a certain empirical value that indicates network congestion or node anomaly, it can be determined that the confidence level of vehicle A is low.

[0080] Alternatively, step S4, which calculates the confidence scores of other nodes, is as follows:

[0081] S411: The monitoring node records the sending timestamp of messages sent by the monitored node and the receiving timestamp of messages received locally.

[0082] S422: Calculate the time difference between the sending timestamp and the receiving timestamp.

[0083] S433: Monitor the number of messages forwarded by the monitored node per unit time and the sequence number of the broadcast messages.

[0084] S444: Calculate the confidence level of the monitored node vehicle A by using the time difference, the number of messages forwarded per unit time, and the continuity of the sequence number.

[0085] For example, if the forwarding rate of vehicle A is lower than a certain value, or if vehicle A sends a sequence number of 1235 and the monitoring node finds that 4 is missing, the confidence level of vehicle A can be reduced.

[0086] S5: When a monitoring node receives a voting request message from another node, it queries the check node corresponding to the voting request message. If the confidence level of the check node is lower than the confidence level threshold, it initiates a vote for the check node.

[0087] For example, when vehicle A receives a braking command from vehicle B, but vehicle A determines that braking is not necessary based on data collected by sensors and other information (relying on information collected by roadside sensors sent by fixed equipment), vehicle A sends a voting request message to the monitoring node. The voting request message contains: the ID of the node that sent the suspicious command, vehicle A's sensor data, and the suspicious command.

[0088] Before determining whether the confidence level of the checking node is lower than the confidence level threshold, the listening node can also make a secondary judgment on the correctness of the suspicious instruction based on the sensor data contained in the voting request message and the roadside sensor data sent by the trusted fixed equipment. If the monitoring node calculates that the suspicious instruction needs to be executed based on the sensor data of vehicle A, it determines the confidence level of vehicle A, i.e., the instruction sending node. If the confidence level is lower than the confidence level threshold, it initiates a vote for vehicle A.

[0089] S6: The monitoring node performs a weighted calculation of the votes of other nodes to filter out malicious nodes.

[0090] In the weighted calculation of S6, the confidence levels of each node's votes can be weighted to calculate the voting results of each node against the malicious node.

[0091] For example, suppose the voting nodes are vehicle C and vehicle D, vehicle C has an execution degree of Q, vehicle D has an execution degree of P, vehicle C's vote is 1, 1 represents a non-malicious node, -1 represents a non-malicious node, vehicle D's vote is -1, and the voting result is Q*(1)+P*(-1).

[0092] Additionally, monitoring nodes can query trusted fixed devices to request the number of editing steps required to transform the basic geometry of the historical location topology map to the basic geometry of the real-time location topology map. Based on the nodes corresponding to the number of editing steps, the votes are weighted.

[0093] If the voting result exceeds the voting threshold, the detection node is determined to be a malicious node. The voting threshold can change dynamically according to the network status. For example, the more voting request messages a monitoring node receives, the larger α becomes. Voting threshold = ω + α * voting threshold, where ω is a decimal between 0 and 1.

[0094] refer to Figure 2 In the first case, from the topology of T to T+t, the number of editing steps for nodes 1, 2, and 3 is 1, and the number of editing steps for node 4 is 0. Then we can set the weights of nodes 1, 2, and 3 to P, and the weight of node 4 to Q, where Q > P.

[0095] During the execution of step S6, the voting results are weighted and calculated based on the confidence level and the stability of the node itself to ensure the reliability of the voting.

[0096] In the above process, since the number of vehicles passing through a certain section of road changes constantly, selecting vehicles as monitoring nodes probabilistically avoids the possibility of monitoring nodes being consistently unreliable, thus ensuring the reliability probability of the monitoring nodes. Through analysis of the vehicle node location distribution topology, monitoring nodes with low communication consumption and high reliability are selected. The selection of monitoring nodes considers the node's computing power, distance to the base station, and network topology, choosing nodes with optimal conditions as monitoring nodes.

[0097] Monitoring nodes monitor the status of all vehicles, while surveillance nodes are responsible for monitoring the content and numbering of data packets from sensors and edge servers to ensure data integrity and reliability. When an anomaly occurs at a node, the monitoring nodes initiate an assessment, vote on the issue, and promptly identify and isolate malicious nodes to ensure network security. The voting results are tallied, and a majority vote determines the authenticity of the event, thus preventing malicious uploads and erroneous events from impacting the autonomous driving network.

[0098] Example 2

[0099] Figure 3 As shown, Figure 3 This is a flowchart of another method for detecting malicious nodes in an autonomous driving network. The steps include:

[0100] S101: Relying on fixed equipment to collect the locations uploaded by all vehicles passing through the covered road segment at the current time, generating a vehicle location topology map as the real-time location topology map.

[0101] S102: Relying on the fixed equipment to receive the vehicle position topology map of the previous time period transmitted by the equipment of the previous road segment as the historical position topology map.

[0102] S103: The fixed equipment is trusted to perform graphic structure detection on the real-time location topology map and the historical location topology map. If the stability of the graphic structure is greater than the empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node.

[0103] S104: The monitoring node listens to the periodic broadcast messages of nodes within a fixed distance and calculates the confidence level of other nodes based on the broadcast messages.

[0104] S105: When a monitoring node receives a request message from another node, it queries the check node corresponding to the request message. If the confidence level of the check node is lower than the confidence level threshold, it initiates a vote for the check node.

[0105] S106: The monitoring node performs a weighted calculation of the votes of other nodes to filter out malicious nodes.

[0106] S107: If the detection result is that the node being checked is a non-malicious node and the suspicious instruction is not executable based on roadside sensor data, then a check request to determine the sensor data is sent to each roadside device and edge server.

[0107] S108: Roadside equipment and edge servers collect data from roadside sensors for detection. If the detection is correct, a vote is initiated on the trusted fixed equipment.

[0108] In the above process, if each vehicle node checks and finds no malicious nodes through voting, then the instruction sent by vehicle B received by vehicle A should be executable. However, if the instruction received by the node that initiated the voting request is still unexecutable, it may be due to sensor data errors. Therefore, vehicle nodes can initiate checks on roadside devices, edge servers, etc., and the roadside devices and edge servers can perform internal screening and voting to avoid the occurrence of network untrustworthiness caused by attacks on trusted fixed devices.

[0109] This application also provides an autonomous driving network malicious node detection system, comprising: each node is set in the autonomous driving network; including roadside equipment, edge servers, vehicles and sensors; the roadside equipment and edge servers vote to obtain a trusted fixed device; the trusted fixed device is used to perform the method of the first aspect to select monitoring nodes in the vehicles; the monitoring nodes perform the method of the first aspect to monitor vehicle nodes, and monitor and screen malicious nodes.

[0110] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the user rights update method provided in the various embodiments described above.

[0111] In the embodiments of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The terms "comprising" and "including" as used throughout the specification and claims are open-ended terms and should therefore be interpreted as "comprising but not limited to".

[0112] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. A method for detecting malicious nodes in an autonomous driving network, characterized in that, Applications in autonomous driving networks include: The system relies on fixed equipment to collect the locations of all vehicles passing through the covered road segment at the current moment, generating a vehicle location topology map as a real-time location topology map. The fixed equipment receives the vehicle position topology map of the previous time segment from the equipment of the previous road segment as the historical position topology map; The fixed equipment is trusted to perform graphic structure detection on the real-time location topology map and the historical location topology map. If the stability of the graphic structure is greater than the empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node. The monitoring node listens to the periodic broadcast messages from nodes within a fixed distance and calculates the confidence level of other nodes based on the broadcast messages; When a monitoring node receives a voting request message from another node, it queries the check node corresponding to the voting request message. If the confidence level of the check node is lower than the confidence level threshold, it initiates a vote for the check node. Determining whether the stability of a graphical structure exceeds an empirical threshold includes: Detect nodes in the location topology graph and detect changes in the set of neighbor nodes of a specific node; By comparing the graphic composed of sequentially connected nodes with the basic geometry, the basic geometry of the real-time location topology map and the historical location topology map is found. If the number of edits required to transform the basic geometry of the historical location topology map into the basic geometry of the real-time location topology map is less than a specific number of steps, it is determined that the stability of the graphic structure is greater than the empirical threshold. To detect changes in the set of neighboring nodes of a specific node, Jaccard is used to compare changes in the set of neighboring nodes. Indicates vehicle i At any moment t The stability index of the neighbor node set is such that the closer it is to 1, the smaller the change in the neighbor node set. For vehicles i At time T+ t Let |·| be the set of neighboring nodes, and |·| be the number of elements in the set. For the intersection operation of sets, It is the common part of the neighbor node sets at two different times. The union operation of sets, It is the set of all elements of the neighbor nodes at two different times.

2. The method according to claim 1, characterized in that, The method also includes: The monitoring node performs a weighted calculation of the votes from other nodes to filter out malicious nodes.

3. The method according to claim 2, characterized in that, The monitoring node performs a weighted calculation of the votes from other nodes to filter out malicious nodes, including: The confidence levels of each node's votes are weighted to calculate the voting results of each node against the malicious node.

4. The method according to claim 1, characterized in that, The system relies on fixed equipment to perform graphical structure detection on real-time and historical location topology maps. If the stability of the graphical structure exceeds an empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node, including: If the stability of the graphical structure is greater than the empirical threshold, monitoring nodes are selected step by step according to the geometric center position of the topological structure.

5. The method according to claim 1, characterized in that, The system relies on fixed equipment to perform graphical structure detection on real-time and historical location topology maps. If the stability of the graphical structure exceeds an empirical threshold, the node with the smallest average distance from other nodes in the real-time location topology map is selected as the monitoring node, including: Calculate the distance of each node to other nodes in turn, sort them, and select the node with the smallest distance to other nodes.

6. The method according to claim 1, characterized in that, Before selecting the node with the smallest average distance from other nodes in the real-time location topology map as the monitoring node, if the stability of the graphic structure is greater than an empirical threshold, the following steps can be performed: [The text abruptly ends here, so the translation stops as well.] Determine whether the stability of the graphical structure is greater than an empirical threshold; If the stability of the graph structure is less than the empirical threshold, the stability of the nodes should be given priority, and the node with the highest stability should be selected as the monitoring node.

7. The method according to claim 1, characterized in that, The monitoring node listens to periodic broadcast messages from nodes within a fixed distance and calculates the confidence level of other nodes based on these broadcast messages, including: The detection node records the sending timestamp of messages sent by the monitored node and the receiving timestamp of messages received locally; Calculate the time difference between the sending timestamp and the receiving timestamp; The confidence level of the monitored node is calculated using the time difference.

8. The method according to claim 1, characterized in that, The method also includes: If the detection result is that the node being checked is a non-malicious node and the suspicious instruction is not executable based on roadside sensor data, then a check request to determine the sensor data is sent to each roadside device and edge server. Roadside equipment and edge servers collect data from roadside sensors for testing. If the test is successful, a vote is initiated on the trusted fixed equipment.

9. A malicious node detection system for autonomous driving networks, characterized in that, Each node is located within the autonomous driving network; including roadside equipment, edge servers, vehicles, and sensors; Roadside equipment and edge servers vote to gain the trust of fixed equipment; The trusted fixed equipment is used to select a monitoring node in the vehicle as described in any one of claims 1-8; The monitoring node performs the method described in any one of claims 1-8 to monitor vehicle nodes and filter out malicious nodes.

Images