A vehicle fleet trust management system and method based on fuzzy logic and deep learning

Abstract

The application relates to a vehicle team trust management system and method based on fuzzy logic and deep learning, and belongs to the field of vehicle-mounted communication. The system comprises a node authentication module for initializing all node information in a vehicle team, a module for predicting the arrival time of the vehicle team at a station, a computing message credibility module for uploading messages by nodes and processing, and the like. When the vehicle team transports materials, the vehicles in the vehicle team upload traffic information or road condition information to a team leader, the team leader vehicle calculates the uploaded messages to obtain a message credibility score msgcre1, meanwhile, other vehicles in the vehicle team also score the message to obtain a message credibility score msgcre2. Then, the predicted arrival time of the vehicle team at the station is compared with the actual arrival time of the vehicle team at the station to obtain a time interval t. Finally, the data processing results are input into a fuzzy logic model to obtain a trust value of the vehicle, and a malicious vehicle is detected according to the high and low of the trust value. The application effectively solves the problems of safety, privacy and trust management of the Internet of Vehicles.

Description

Technical Field

[0001] This invention relates to a fleet trust management system and method based on fuzzy logic and deep learning, belonging to the field of vehicle communication. Background Technology

[0002] For logistics and transportation, trust management of the fleet is necessary to ensure the timely and safe arrival of transported goods. In this invention, trust management of the fleet primarily addresses the issue of malicious vehicles uploading false information, causing the fleet to deviate from its route and ultimately delaying its arrival at the destination. Summary of the Invention

[0003] The technical problem this invention aims to solve is to prevent convoys from being misled by malicious vehicles, causing them to deviate from the planned route and delaying the transportation of supplies.

[0004] The technical solution of the present invention is: a fleet trust management system based on fuzzy logic and deep learning, including a node authentication module for initializing information of all nodes in the fleet, a node message upload and processing message credibility calculation module, a fleet arrival time prediction module, a vehicle trust value calculation module, and a blockchain network module for storing data. The node authentication module that initializes all node information in the fleet includes: The traffic management authority is used to issue new certificates, public and private keys, to nodes in the fleet, which refer to the vehicles in the fleet and all the stations the fleet passes through along its route. The node upload message and processed message credibility calculation module includes: Mobile nodes refer to all vehicles in the convoy during its journey. The message content includes all traffic and road condition information witnessed by all vehicles within their communication range during the convoy's journey. The module for predicting the arrival time of the convoy at the station includes: Vehicles, all vehicles in the convoy; Stations are all stations along the convoy's route, used to process convoy-related data and predict the convoy's arrival time at stations. The module for calculating vehicle trust values ​​includes: The message credibility score msgcre1 is obtained by the team leader's vehicle using a fuzzy logic algorithm to evaluate the authenticity of the message content uploaded by the current vehicle. The message credibility score msgcre2 is obtained by neighboring vehicles evaluating the authenticity of the broadcast message content of the current vehicle. The time interval t refers to the difference between the predicted arrival time of the convoy and the actual arrival time of the convoy. The blockchain network module includes: The first layer of public blockchain uses the POT consensus algorithm. The team leader vehicle packages and stores the message information uploaded by the vehicle and the calculated message credibility score msgcre1 into this blockchain. The second layer of public blockchain uses the POW consensus algorithm to store the actual arrival time of the convoy at the station and the difference t between the predicted and actual arrival times.

[0005] Specifically, the mobile node has computing power and storage capacity.

[0006] Specifically, the station is a fixed, stationary node on the road.

[0007] Specifically, the first and second layer public blockchains are open and transparent, and all nodes can upload information and access their contents.

[0008] A fleet trust management method based on fuzzy logic and deep learning includes the following steps: When the fleet travels along a fixed route, upon departure from the previous station, the arrival time at the next station is predicted. During the journey, vehicles broadcast message content and upload the message content to the leader vehicle. The leader vehicle calculates the message credibility score msgcre1 of the currently uploaded message. Neighboring vehicles evaluate the currently broadcast message to obtain a message credibility score msgcre2. Upon arrival at the next station, the trust value of the vehicle is calculated based on the predicted arrival time, message credibility score msgcre1, message credibility score msgcre2, and the time interval t between the actual arrival time and the prediction. This allows for the identification of malicious and honest vehicles, with malicious vehicles being penalized and honest vehicles being rewarded.

[0009] The specific steps are as follows: Step 1: When departing from the previous station, use a Long Short-Term Memory (LSTM) neural network algorithm to predict the time it will take for the convoy to arrive at the next station. Step 2: When the message content appears, the vehicles in the convoy broadcast the message content and upload the message content to the team leader's vehicle; Step 3: The team leader's vehicle uses a fuzzy logic algorithm to calculate the credibility of the uploaded message content based on the event type, vehicle location, and message occurrence location, obtaining a message credibility score msgcre1. The message content and message credibility score msgcre1 are then stored in the first-layer public blockchain. At the same time, neighboring vehicles in the convoy also evaluate this message, obtaining a message credibility score msgcre2, which is also stored in the first-layer public blockchain. Step 4: Compare the calculated message credibility score msgcre1 with the credibility threshold. If the credibility score is less than the credibility threshold, the message is untrustworthy and is ignored. If the credibility score is greater than or equal to the credibility threshold, proceed to the next step. Step 5: Upon arrival at the next station, the predicted arrival time is compared with the actual arrival time to obtain the time interval t, which is then stored in the second-layer public blockchain. Step 6: Based on the message credibility score msgcre1, message credibility score msgcre2 and time interval t, use fuzzy logic algorithm to obtain the vehicle trust value and store it in the second-layer blockchain; Step 7: When the vehicle trust value is less than the minimum trust value threshold, the vehicle is considered a malicious vehicle. The traffic management bureau will delete the certificate issued to the vehicle and penalize the vehicle. When the vehicle trust value is greater than or equal to the minimum trust value threshold, the vehicle is considered an honest vehicle and will be awarded a reward.

[0010] Specifically, the long short-term memory neural network algorithm includes: Input layer: vehicle location, convoy departure time, distance between stations, and road congestion level; Output layer: Arrival time of the convoy; Training set: Data sets related to data types.

[0011] Specifically, the training data set comes from the dataset provided by the DataCastle competition platform, totaling 100,000 data points, including relevant data on New York taxis, covering the period from January 1, 2022 to January 3, 2022. This allows us to predict the arrival time at the next station as long as the fleet departs from one station.

[0012] The beneficial effects of this invention are as follows: Compared to traditional vehicle network trust management, this invention breaks away from the traditional approach that only considers a large number of vehicles and uploaded messages. This invention also considers situations where there are fewer vehicles in the convoy, yet it can still accurately determine the accuracy of uploaded messages, thereby detecting malicious vehicles in the area. The Long Short-Term Memory (LSTM) neural network algorithm used in this invention predicts the convoy's arrival time at a station. It only requires training a dataset for predicting convoy arrival times at one station beforehand. Then, when vehicles depart from a new station, the arrival time of the convoy at the next station can be quickly predicted based on this dataset, significantly shortening the time for detecting malicious vehicles in the entire trust management system. Finally, the two-layer public blockchain technology used in this invention effectively protects vehicle privacy and security, preventing malicious vehicles from querying the trust values ​​of legitimate vehicles while ensuring that all vehicles can access road condition information without restriction. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2 This is a schematic diagram of the communication method of the present invention; Figure 3 This is an overall block diagram of the present invention; Figure 4 This is a schematic diagram of the fuzzy logic algorithm structure of the present invention; Figure 5 This is a schematic diagram of the structure of the long short-term memory network of this invention; Figure 6 This is a schematic diagram of the public blockchain structure of this invention. Detailed Implementation

[0014] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0015] Example 1, such as Figure 1-6 As shown, a fleet trust management system based on fuzzy logic and deep learning includes a node authentication module for initializing information of all nodes in the fleet, a module for calculating the credibility of messages uploaded and processed by nodes, a module for predicting the arrival time of the fleet at the station, a module for calculating vehicle trust values, and a blockchain network module for storing data. The node authentication module that initializes all node information in the fleet includes: The traffic management authority is used to issue new certificates, public and private keys, to nodes in the fleet, which refer to the vehicles in the fleet and all the stations the fleet passes through along its route. The node upload message and processed message credibility calculation module includes: Mobile nodes refer to all vehicles in the convoy during its journey. The message content includes all traffic and road condition information witnessed by all vehicles within their communication range during the convoy's journey. The module for predicting the arrival time of the convoy at the station includes: Vehicles, all vehicles in the convoy; Stations are all stations along the convoy's route, used to process convoy-related data and predict the convoy's arrival time at stations. The module for calculating vehicle trust values ​​includes: The message credibility score msgcre1 is obtained by the team leader's vehicle using a fuzzy logic algorithm to evaluate the authenticity of the message content uploaded by the current vehicle. The message credibility score msgcre2 is obtained by neighboring vehicles evaluating the authenticity of the current vehicle's broadcast message content, based on methods such as voting. The time interval t refers to the difference between the predicted arrival time of the convoy and the actual arrival time of the convoy using the Long Short-Term Memory Neural Network algorithm. The blockchain network module includes: The first layer of public blockchain uses the POT consensus algorithm. The team leader vehicle packages and stores the message information uploaded by the vehicle and the calculated message credibility score msgcre1 into this blockchain. The second layer of public blockchain uses the POW consensus algorithm to store the actual arrival time of the convoy at the station and the difference t between the predicted and actual arrival times.

[0016] Furthermore, the mobile node possesses computing and storage capabilities.

[0017] Furthermore, the station is a fixed, stationary node on the road.

[0018] Furthermore, the first and second layer public blockchains are open and transparent, allowing all nodes to upload information and access their contents.

[0019] A fleet trust management method based on fuzzy logic and deep learning includes the following steps: When the fleet travels along a fixed route, upon departure from the previous station, the arrival time at the next station is predicted. During the journey, vehicles broadcast message content and upload the message content to the leader vehicle. The leader vehicle calculates the message credibility score msgcre1 of the currently uploaded message. Neighboring vehicles evaluate the currently broadcast message to obtain a message credibility score msgcre2. Upon arrival at the next station, the trust value of the vehicle is calculated based on the predicted arrival time, message credibility score msgcre1, message credibility score msgcre2, and the time interval t between the actual arrival time and the prediction. This allows for the identification of malicious and honest vehicles, with malicious vehicles being punished and honest vehicles being rewarded, thereby achieving trust management.

[0020] The specific steps are as follows: Step 1: When departing from the previous station, use a Long Short-Term Memory (LSTM) neural network algorithm to predict the time it will take for the convoy to arrive at the next station. Step 2: When the message content appears, the vehicles in the convoy broadcast the message content and upload the message content to the team leader's vehicle; Step 3: The team leader's vehicle uses a fuzzy logic algorithm to calculate the credibility of the uploaded message content based on the event type, vehicle location, and message occurrence location, obtaining a message credibility score msgcre1. The message content and message credibility score msgcre1 are then stored in the first-layer public blockchain. At the same time, neighboring vehicles in the convoy also evaluate this message, obtaining a message credibility score msgcre2, which is also stored in the first-layer public blockchain. Step 4: Compare the calculated message credibility score msgcre1 with the credibility threshold. If the credibility score is less than the credibility threshold, the message is untrustworthy and is ignored. If the credibility score is greater than or equal to the credibility threshold, proceed to the next step. Step 5: Upon arrival at the next station, the predicted arrival time is compared with the actual arrival time to obtain the time interval t, which is then stored in the second-layer public blockchain. Step 6: Based on the message credibility score msgcre1, message credibility score msgcre2 and time interval t, use fuzzy logic algorithm to obtain the vehicle trust value and store it in the second-layer blockchain; Step 7: When the vehicle trust value is less than the minimum trust value threshold, the vehicle is considered a malicious vehicle. The traffic management bureau will delete the certificate issued to the vehicle and penalize the vehicle. When the vehicle trust value is greater than or equal to the minimum trust value threshold, the vehicle is considered an honest vehicle and will be awarded a reward.

[0021] Furthermore, the Long Short-Term Memory Neural Network algorithm includes: Input layer: vehicle location, convoy departure time, distance between stations, and road congestion level; Output layer: Arrival time of the convoy; Training set: Data sets related to data types.

[0022] Furthermore, the training data set comes from the dataset provided by the DataCastle competition platform, totaling 100,000 data points, including relevant data on New York taxis, covering the period from January 1, 2022 to January 3, 2022. This allows us to predict the arrival time at the next station as long as the fleet departs from one station.

[0023] Furthermore, the vehicle trust value is calculated using a fuzzy logic algorithm. The input features are: message credibility score msgcre1 and message credibility score msgcre2 obtained by fuzzy logic calculation, and the time interval t between the actual time and the predicted time of the convoy's arrival at the station.

[0024] The output feature is the vehicle's trust value.

[0025] like Figure 2 As shown, mobile nodes within the communication range collect data through V2V and V2I communication. When a vehicle wants to upload traffic information, it can obtain the desired information from other vehicles via V2V communication, or it can package the facts of what it witnesses within its own field of vision into a data packet. Finally, this data packet is uploaded to the RSU (Roadside Unit) of the data processing section using V2I communication.

[0026] like Figure 3 As shown, within the witness range of a vehicle, when useful traffic information appears, mobile node vehicles within this range collect the message, form a data packet, and upload it to the team leader. Then, by calculating the credibility of the message uploaded by the vehicle and the evaluation of neighboring vehicles, the arrival time of the convoy is predicted, and the trust value of the vehicle is calculated accordingly. Malicious vehicles are detected based on the level of trust value, thereby realizing trust management.

[0027] like Figure 6 As shown, both layers of the public blockchain consist of a block header and a block body. The block header comprises five parts: the block index, the hash value of the current block, the hash value of the previous block, a timestamp, and a Merkle root. The block body consists of transaction data, which is then hashed, ensuring that even minor operations on the block content will change the hash value.

[0028] This invention adds blockchain to the traditional Internet of Vehicles (IoV), improving data processing speed, effectively preventing uploaded data from being tampered with, and effectively solving the security, privacy, and trust management issues of IoV.

[0029] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A fuzzy logic and deep learning based vehicle fleet trust management system, characterized in that: It includes a node authentication module that initializes information for all nodes in the fleet, a module for calculating the credibility of messages uploaded and processed by nodes, a module for predicting the arrival time of the fleet at stations, a module for calculating vehicle trust values, and a blockchain network module for storing data. The node authentication module that initializes all node information in the fleet includes: The traffic management authority is used to issue new certificates, public and private keys, to nodes in the fleet, which refer to the vehicles in the fleet and all the stations the fleet passes through along its route. The node upload message and processed message credibility calculation module includes: Mobile nodes refer to all vehicles in the convoy during its journey. The message content includes all traffic and road condition information witnessed by all vehicles within their communication range during the convoy's journey. The module for predicting the arrival time of the convoy at the station includes: Vehicles, all vehicles in the convoy; Stations, all stations along the convoy's route; The module for calculating vehicle trust values ​​includes: The message credibility score msgcre1 is obtained by the team leader's vehicle using a fuzzy logic algorithm to evaluate the authenticity of the message content uploaded by the current vehicle. The message credibility score msgcre2 is obtained by neighboring vehicles evaluating the authenticity of the broadcast message content of the current vehicle. The time interval t refers to the difference between the predicted arrival time of the convoy and the actual arrival time of the convoy; The blockchain network module includes: The first layer of public blockchain uses the POT consensus algorithm. The team leader vehicle packages and stores the message information uploaded by the vehicle and the calculated message credibility score msgcre1 into this blockchain. The second layer of public blockchain uses the POW consensus algorithm to store the actual arrival time of the convoy at the station and the difference t between the predicted and actual arrival times. The first and second layers of public blockchains are open and transparent, allowing all nodes to upload information and access their contents.

2. The fuzzy logic and deep learning based fleet trust management system of claim 1, wherein: The mobile node has computing and storage capabilities.

3. The fuzzy logic and deep learning based fleet trust management system of claim 1, wherein: The station is a fixed, stationary node on the road.

4. A method for managing trust in a vehicle fleet based on fuzzy logic and deep learning, characterized in that: The process includes the following steps: When the convoy travels along a fixed route, it predicts the arrival time at the next station when it departs from the previous station. During the journey, vehicles broadcast messages and upload them to the leader vehicle. The leader vehicle calculates the message credibility score msgcre1 for the currently uploaded message. Neighboring vehicles evaluate the broadcast messages of the currently uploaded vehicles to obtain message credibility scores msgcre2. Upon arrival at the next station, the trust value of the vehicle is calculated based on the predicted arrival time, message credibility scores msgcre1 and msgcre2, and the time interval t between the actual arrival time and the prediction. This allows the identification of malicious and honest vehicles, with malicious vehicles being punished and honest vehicles being rewarded. The specific steps are as follows: Step 1: When departing from the previous station, use a Long Short-Term Memory (LSTM) neural network algorithm to predict the time it will take for the convoy to arrive at the next station. Step 2: When the message content appears, the vehicles in the convoy broadcast the message content and upload the message content to the team leader's vehicle; Step 3: The team leader's vehicle uses a fuzzy logic algorithm to calculate the credibility of the uploaded message content based on the event type, vehicle location, and message occurrence location, obtaining a message credibility score msgcre1. The message content and message credibility score msgcre1 are then stored in the first-layer public blockchain. At the same time, neighboring vehicles in the convoy also evaluate this message, obtaining a message credibility score msgcre2, which is also stored in the first-layer public blockchain. Step 4: Compare the calculated message credibility score msgcre1 with the credibility threshold. When the credibility score is less than the credibility threshold, the message is not credible and is ignored. When the credibility level is greater than or equal to the credibility threshold, proceed to the next step; Step 5: Upon arrival at the next station, compare the predicted arrival time with the actual arrival time to obtain the time interval t, and store it in the second-layer public blockchain; Step 6: Based on the message credibility score msgcre1, message credibility score msgcre2 and time interval t, the vehicle trust value is obtained using a fuzzy logic algorithm and stored in the second-layer blockchain; Step 7: When the vehicle trust value is less than the minimum trust value threshold, the vehicle is considered a malicious vehicle. The traffic management bureau will delete the certificate issued to the vehicle and penalize the vehicle. When the vehicle trust value is greater than or equal to the minimum trust value threshold, the vehicle is considered an honest vehicle and will be awarded a reward.

5. The method of claim 4, wherein: The Long Short-Term Memory Neural Network algorithm includes: Input layer: vehicle location, convoy departure time, distance between stations, and road congestion level; Output layer: Arrival time of the convoy; Training set: Data sets related to data types.

6. The method of claim 5, wherein: The training dataset comes from the dataset provided by the DataCastle competition platform, totaling 100,000 data points, including relevant data on New York taxis from January 1, 2022 to January 3, 2022. This allows us to predict the arrival time of a taxi fleet as long as it departs from one station.

Images