A blockchain and load balancing combined vehicle-mounted edge mesh data scheduling system

Abstract

The application discloses a vehicle-mounted edge meshing data scheduling system based on combination of a blockchain and load balancing, comprising a dynamic resource allocation module, an intelligent scheduling module, a data management module, a data security and privacy protection module; the system can dynamically adjust allocation of computing and storage resources, and schedules data according to real-time traffic and vehicle state information, so that efficient vehicle-mounted edge computing is realized.

Description

TECHNICAL FIELD

[0001] The application relates to the fields of vehicle-mounted edge computing, data sharing, load balancing and encryption authentication, and specifically relates to a vehicle-mounted edge grid data scheduling system based on the combination of a blockchain and load balancing. BACKGROUND

[0002] With the rapid development of Internet of Things (IoT), Artificial Intelligence (AI) and edge computing technology, intelligent transportation systems are gradually becoming an important part of future urban traffic management. Vehicle-mounted edge computing distributes computing resources on the edge nodes of vehicles and transportation infrastructure, providing low-latency, high-efficiency data processing capabilities to support applications such as autonomous driving, real-time traffic management and intelligent navigation. However, with the development of Vehicle-to-Everything (V2X) technology, how to effectively schedule data and manage resources has become a problem to be solved.

[0003] Traditional vehicle-mounted edge computing systems often face problems such as load imbalance, high data transmission delay and low resource utilization when facing large amounts of real-time data processing requirements. These problems not only affect the performance of the system, but also may lead to a decrease in the reliability and stability of the system. In order to solve these problems, in recent years, researchers have proposed various data scheduling and load balancing strategies, but these methods mostly fail to fully consider the security and privacy protection of data.

[0004] Blockchain technology, as a distributed ledger technology, has the characteristics of decentralization, non-tamperability and high security, which can effectively solve the security problems of data in the process of transmission and storage. Applying blockchain technology to vehicle-mounted edge computing systems can ensure the integrity and security of data and prevent data from being tampered with and illegally accessed. However, blockchain technology still faces challenges in scalability and performance when dealing with high-frequency and large-scale data. SUMMARY

[0005] The purpose of the present application is to provide a vehicle-mounted edge grid data scheduling system based on the combination of a blockchain and load balancing, which is used to realize data scheduling of a vehicle-mounted network.

[0006] The vehicle-mounted network includes a plurality of central server nodes, a plurality of vehicle-mounted edge nodes and a plurality of vehicle-mounted terminal device nodes; some or all of these nodes are located in a blockchain;

[0007] The vehicle-mounted edge grid data scheduling system includes a dynamic resource allocation module, an intelligent scheduling module, a data management module, a data security and privacy protection module;

[0008] The dynamic resource allocation module predicts the resource demand of nodes in the future t time period based on node monitoring indicators, generates and executes resource allocation strategy S(t);

[0009] The intelligent scheduling module generates task allocation strategies and dynamically adjusts them to minimize task execution time and resource consumption;

[0010] The grid data management module divides the data of the vehicle-mounted edge grid data scheduling system into grid units for management, achieving distributed storage and fast access of data;

[0011] The data security and privacy protection module protects user privacy and data security through encryption technology, access control mechanism and data anonymization processing.

[0012] Further, the dynamic resource allocation module includes a resource monitoring unit, a resource demand prediction and analysis unit, and a resource scheduling and allocation execution unit;

[0013] The resource monitoring unit collects node monitoring indicators R i =(CPU i ,MEM i ,BW i ,STOR i ) to analyze the resource usage of each node and upload the node monitoring indicators to the central database for storage; CPU i ,MEM i ,BW i ,STOR i represent CPU usage, memory usage, bandwidth usage and storage utilization, respectively;

[0014] The resource demand prediction and analysis unit takes node monitoring indicators as input, uses a resource demand prediction model to predict the resource demand of nodes in the future t time period, and generates resource allocation strategy S(t) according to the prediction results; resource allocation strategy S(t) includes resource migration, load balancing and priority adjustment strategies;

[0015] The resource demand prediction model includes a time series analysis model and a deep learning model;

[0016] The resource scheduling and allocation execution unit performs dynamic resource scheduling and allocation according to resource allocation strategy S(t).

[0017] Further, the intelligent scheduling module includes a task initial allocation unit, a task migration unit, and an adaptive adjustment unit;

[0018] The task initial allocation unit sorts the tasks according to the importance and urgency of the tasks, and allocates the tasks based on the current load and resource capacity of the nodes, so as to allocate each task t i to a node n i that can meet the task resource requirement R j .

[0019] The task migration unit monitors the resource usage and task execution state of the nodes in real time during task execution, and if it is detected that the resource usage of a node n j exceeds a preset threshold θ, the task migration mechanism is started to migrate part of the tasks to a node with a load less than a preset value.

[0020] The adaptive adjustment unit uses a machine learning algorithm to predict the resource usage and task load changes of each node in the future t period, and continuously adjusts and optimizes the task allocation strategy according to the task execution results and system performance indicators.

[0021] Further, the grid data management module includes a data grid division unit, a data storage location and access strategy optimization unit;

[0022] The data grid division unit divides the entire data space D into multiple grid units according to geographical location and data type, denoted as G = {g1, g2, …, g M}; g M is the Mth grid unit; M is the number of grid units;

[0023] The data grid division unit stores the data distribution of each grid unit in the corresponding vehicle edge node or cloud storage node;

[0024] The data storage location and access strategy optimization unit dynamically adjusts the data storage location and access strategy during data access according to the access frequency and node load;

[0025] The step of adjusting the data storage location and access strategy includes: if the data access frequency growth rate of the grid unit is greater than a preset value, migrating these data to a more efficient storage node to reduce access delay and improve response speed.

[0026] Further, the data security and privacy protection module includes a data encryption and transmission protection unit, an access control and permission management unit, and a data anonymization and privacy protection unit;

[0027] The data encryption and transmission protection unit uses a TLS or IPSec encrypted channel for data transmission, and uses E(D t , K t ) as the encryption method for transmitting data D t ; Kt to transmit the encryption key;

[0028] The data encryption and transmission protection unit encrypts the stored data using an encryption algorithm; the encryption algorithm includes AES, RSA encryption algorithm;

[0029] The access control and permission management unit ensures that only authorized users and devices can access and operate data through access control and permission management mechanisms;

[0030] The data anonymization and privacy protection uses data anonymization and privacy protection technology to protect user privacy involved in data processing;

[0031] The privacy protection technology is differential privacy technology.

[0032] Further, in the vehicle-mounted network, the center server node and the vehicle-mounted edge node are connected one-to-many, and the vehicle-mounted edge node is connected one-to-many with the vehicle-mounted device node.

[0033] The number of access devices of each vehicle-mounted edge node is m is the number of vehicle-mounted edge nodes; n is the number of vehicle-mounted device nodes;

[0034] The number of access devices of each center server node is k is the number of center server nodes.

[0035] Further, the center server node, the vehicle-mounted edge node, and the vehicle-mounted device all have unique identifiers.

[0036] Further, any node performs self-registration when it is started for the first time.

[0037] The registration process is: the node sends a registration request of basic information to the central management system or the blockchain network; the central system verifies and records the basic information to ensure the legality and security of the node; the basic information includes unique identifier, geographic location, hardware configuration;

[0038] Any node periodically reports the running state to the central system or the blockchain network, including the current load, available resources and network state.

[0039] Further, the set of nodes in the blockchain is denoted as N={n1,n2,…,nk}; the computing power C k of each node n i is C i ≥ C min and the storage capacity S i ≥ S min ; C min , S min are the lower limits of computing power and storage capacity; i=1,2,…,k;

[0040] The genesis block of the blockchain is denoted as G0; the genesis block stores the initial state of the blockchain and information of all participating nodes;

[0041] The functions of the blockchain are implemented through smart contracts; a set of smart contracts is denoted as SC={sc1,sc2,…,sc m}, which are used to manage node identities, record transactions, and allocate resources.

[0042] Further, a vehicle-mounted network updating module is further included;

[0043] The vehicle-mounted network updating module periodically updates the vehicle-mounted network based on the computing power of the nodes;

[0044] The updating step includes:

[0045] 1) The vehicle-mounted network updating module divides the vehicle-mounted network into multiple grid areas and calculates the computing power of each grid area; if there is a grid unit with computing power less than a preset threshold C threshold , an edge node is added in the grid area to make the computing power of the grid area greater than or equal to C threshold ;

[0046] 2) After adding the nodes, the DBSCAN clustering algorithm is used to divide the nodes in the vehicle-mounted network into different network clusters; in each network cluster, the communication delay between any two nodes is less than a communication delay threshold T max , and the distance between any two nodes is greater than a node density threshold D min ;

[0047] The distance d ij between any two nodes is as follows:

[0048]

[0049] wherein, is the difference in latitude between node i and node j, and Δλ=λ j -λ i is the difference in longitude, and r is the radius of the earth; is the latitude of node j and node i; λ j , λ i is the longitude of node j and node i;

[0050] 3) All identified nodes with load capacity are obtained, and all nodes are mapped to a geographic information system map, and the network connection between devices is drawn on the map to show the communication relationship between nodes, and an updated vehicle-mounted network is obtained; the device data includes the unique identifier of the device, the geographic location, the load capacity, the adjacent node information, and the current network state.

[0051] The load capacity of the node is as follows:

[0052]

[0053] In the formula, D is the load capacity of the node; S cpu is the available CPU resources of the node; S cpu_total is the total CPU resources of the node; S m is the available memory of the node; S m_total is the total memory of the node; S s is the available storage of the node; S s_total is the total storage of the node; S B is the available bandwidth of the node; S B_total is the total bandwidth of the node; a1, a2, a3, a4 are weight coefficients.

[0054] The technical effect of the present application is self-evident. The present application proposes a vehicle-mounted edge meshing data scheduling system based on the combination of blockchain and load balancing. The system uses blockchain technology to ensure the security of data transmission and storage, and optimizes the efficiency and resource utilization of data processing through load balancing strategy.

[0055] The present application applies blockchain technology to the vehicle-mounted edge computing environment, establishes a decentralized and secure data transmission and storage mechanism, and realizes load balancing through intelligent scheduling algorithm to improve the overall performance and reliability of the system.

[0056] The system of the present application can dynamically adjust the allocation of computing and storage resources, schedule data according to real-time traffic and vehicle state information, and realize efficient vehicle-mounted edge computing.

[0057] At the same time, the present application combines the security and transparency of blockchain technology to ensure the security and credibility of data in the process of transmission and processing. The present application not only improves the performance of the vehicle-mounted edge computing system, but also enhances the security and reliability of the system, providing a new technical solution for the development of intelligent transportation system. BRIEF DESCRIPTION OF DRAWINGS

[0058] Fig. 1 is the architecture diagram of the vehicle-mounted edge scheduling system;

[0059] Fig. 2 is the structure diagram of the vehicle-mounted edge meshing scheduling system;

[0060] Fig. 3 is the flowchart of the blockchain and load balancing scheduling algorithm. DETAILED DESCRIPTION

[0061] The application will be further described in conjunction with the following examples, but should not be understood as limiting the above-mentioned subject matter of the application to the following examples. Various substitutions and modifications can be made according to ordinary technical knowledge and conventional means in the art without departing from the technical idea of the application, and all such substitutions and modifications shall be included in the protection scope of the application.

[0062] Example 1

[0063] Referring to Figs. 1-3 A vehicle-mounted edge meshed data scheduling system based on the combination of blockchain and load balancing, which is used to realize data scheduling of vehicle-mounted network;

[0064] The vehicle-mounted network includes a plurality of central server nodes, a plurality of vehicle-mounted edge nodes, and a plurality of vehicle-mounted terminal device nodes; part or all of these nodes are located in a blockchain;

[0065] The vehicle-mounted edge meshed data scheduling system includes a dynamic resource allocation module, an intelligent scheduling module, a data management module, a data security and privacy protection module;

[0066] The dynamic resource allocation module predicts the resource demand of nodes in a future t time period based on node monitoring indicators, generates and executes a resource allocation strategy S(t);

[0067] The intelligent scheduling module generates a task allocation strategy and dynamically adjusts it with the goal of minimizing task execution time and resource consumption;

[0068] The meshed data management module divides the data of the vehicle-mounted edge meshed data scheduling system into grid units for management, realizing distributed storage and fast access of data;

[0069] The data security and privacy protection module protects user privacy and data security through encryption technology, access control mechanism, and data anonymization processing.

[0070] Example 2

[0071] A vehicle-mounted edge meshed data scheduling system based on the combination of blockchain and load balancing, the technical content is the same as that of example 1, further, the dynamic resource allocation module includes a resource monitoring unit, a resource demand prediction and analysis unit, and a resource scheduling and allocation execution unit;

[0072] The resource monitoring unit collects node monitoring indicators R i =(CPU i ,MEM i ,BW i ,STOR i), to analyze the resource usage of each node and upload the node monitoring indicators to the central database for storage; CPU i , MEM i , BW i , STOR i respectively represent CPU usage, memory occupation, bandwidth usage and storage utilization;

[0073] The resource demand prediction and analysis unit takes the node monitoring indicators as input, uses the resource demand prediction model to predict the resource demand of the node in the future t time period, and generates a resource allocation strategy S(t) according to the prediction result; the resource allocation strategy S(t) includes resource migration, load balancing and priority adjustment strategies;

[0074] The resource demand prediction model includes a time series analysis model and a deep learning model.

[0075] The resource scheduling and allocation execution unit performs dynamic resource scheduling and allocation according to the resource allocation strategy S(t).

[0076] Embodiment 3:

[0077] A vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing, the technical content of any one of embodiments 1-2, further, the intelligent scheduling module includes a task initial allocation unit, a task migration unit, and an adaptive adjustment unit.

[0078] The task initial allocation unit sorts tasks according to their importance and urgency, and allocates tasks based on the current load and resource capacity of the nodes, so that each task t i is assigned to a node n i that can meet the task resource demand R j ;

[0079] The task migration unit monitors the resource usage and task execution status of the nodes in real time during task execution, and if it detects that the resource usage of a node n j exceeds a preset threshold θ, it starts a task migration mechanism to migrate part of the tasks to a node with a load less than a preset value.

[0080] The adaptive adjustment unit uses machine learning algorithms to predict the resource usage and task load changes of each node in the future t time period, and continuously adjusts and optimizes the task allocation strategy according to the task execution results and system performance indicators.

[0081] Embodiment 4:

[0082] A vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing, the technical content of any one of embodiments 1-3, further, the mesh data management module includes a data grid division unit, a data storage location and access strategy optimization unit;

[0083] The data grid division unit divides the entire data space D into multiple grid units according to geographical location and data type, denoted as G={g1, g2, …, g M};g M is the Mth grid unit; M is the number of grid units;

[0084] The data grid division unit stores the data distribution of each grid unit in the corresponding vehicle-mounted edge node or cloud storage node;

[0085] The data storage location and access strategy optimization unit dynamically adjusts the storage location and access strategy of the data during data access according to the access frequency and the load of the node;

[0086] The step of adjusting the storage location and access strategy of the data includes: if the data access frequency growth rate of the grid unit is greater than a preset value, then migrating these data to a more efficient storage node to reduce access delay and improve response speed.

[0087] Embodiment 5:

[0088] A vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing, the technical content of any one of embodiments 1-4, further, the data security and privacy protection module includes a data encryption and transmission protection unit, an access control and permission management unit, a data anonymization and privacy protection unit;

[0089] The data encryption and transmission protection unit uses a TLS or IPSec encrypted channel for data transmission, and uses E(D t ,K t as the encryption method for transmitting data D t ; K t is the transmission encryption key;

[0090] The data encryption and transmission protection unit uses an encryption algorithm to encrypt the stored data; the encryption algorithm includes AES and RSA encryption algorithms;

[0091] The access control and permission management unit ensures that only authorized users and devices can access and operate data through the access control and permission management mechanism;

[0092] The data anonymization and privacy protection unit protects user privacy involved in the data processing process using data anonymization and privacy protection technology;

[0093] The privacy protection technology mentioned is differential privacy technology.

[0094] Example 6:

[0095] A vehicle-mounted edge gridded data scheduling system based on blockchain and load balancing, with the same technical content as any one of embodiments 1-5, further wherein, in the vehicle network, the central server node and the vehicle edge node are connected in a one-to-many manner, and the vehicle edge node and the vehicle terminal device node are connected in a one-to-many manner.

[0096] The number of connected devices per vehicle edge node is m represents the number of vehicle-mounted edge nodes; n represents the number of vehicle-mounted terminal device nodes.

[0097] The number of devices connected to each central server node is: k represents the number of central server nodes.

[0098] Example 7:

[0099] A vehicle-mounted edge gridded data scheduling system based on the combination of blockchain and load balancing, with the same technical content as any one of embodiments 1-6, further wherein the central server node, vehicle-mounted edge node, and vehicle-mounted terminal device all have unique identifiers.

[0100] Example 8:

[0101] A vehicle-mounted edge grid-based data scheduling system based on blockchain and load balancing, with the same technical content as any one of embodiments 1-7, further wherein any node performs self-registration upon first startup;

[0102] The registration process is as follows: a node sends a registration request with basic information to the central management system or blockchain network; the central system verifies and records the basic information to ensure the legitimacy and security of the node; the basic information includes a unique identifier, geographical location, and hardware configuration;

[0103] Any node periodically reports its operational status to the central system or blockchain network, including its current load, available resources, and network status.

[0104] Example 9:

[0105] A vehicle-mounted edge gridded data scheduling system based on blockchain and load balancing, with technical content identical to any one of embodiments 1-8, further wherein the set of nodes in the blockchain is denoted as N = {n1, n2, ..., n}. k}; Each node n i Computing power C i ≥C min And storage capacity S i ≥S min Cmin , S min is the lower limit of computing power and storage capacity; i = 1, 2, …, k;

[0106] The genesis block of the blockchain is denoted as G0; the genesis block stores the initial state of the blockchain and information of all participating nodes;

[0107] The functions of the blockchain are implemented through smart contracts; a set of smart contracts is denoted as SC = {sc1, sc2, …, sc m}, which are used to manage node identities, record transactions, and allocate resources.

[0108] Embodiment 10:

[0109] A vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing, the technical content of which is the same as any one of embodiments 1-9, further comprising a vehicle-mounted network updating module;

[0110] The vehicle-mounted network updating module periodically updates the vehicle-mounted network based on the computing power of the nodes;

[0111] The updating step includes:

[0112] 1) The vehicle-mounted network updating module divides the vehicle-mounted network into multiple grid areas and calculates the computing power of each grid area; if there is a grid unit with computing power less than a preset threshold C threshold , a vehicle-mounted edge node is added in the grid area to make the computing power of the grid area greater than or equal to C thresho;d ;

[0113] 2) After adding the node, the DBSCAN clustering algorithm is used to divide the nodes in the vehicle-mounted network into different network clusters; in each network cluster, the communication delay between any two nodes is less than a communication delay threshold T max , and the distance between any two nodes is greater than a node density threshold D min ;

[0114] The distance d ij between any two nodes is as follows:

[0115]

[0116] wherein, is the difference in latitude between node i and node j, Δλ = λ j - λ i is the difference in longitude, and r is the radius of the earth; is the latitude of node j, node i; λ j , λ i is the longitude of node j, node i;

[0117] 3) Obtain all identified and loadable nodes, and map all nodes to a geographic information system map, and draw the network connection between devices on the map to show the communication relationship between nodes, and obtain the updated vehicle-mounted network; the device data includes the unique identifier of the device, the geographic position, the load capacity, the adjacent node information and the current network state;

[0118] The load capacity of the node is as follows:

[0119]

[0120] In the formula, D is the load capacity of the node; S cpu is the available CPU resource of the node; S cpu_total is the total CPU resource of the node; S m is the available memory of the node; S m_total is the total memory of the node; S s is the available storage of the node; S s_total is the total storage of the node; S B is the available bandwidth of the node; S B_total is the total bandwidth of the node; a1, a2, a3, a4 are weight coefficients.

[0121] Example 11:

[0122] A vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing, by deploying vehicle-mounted edge nodes and constructing a blockchain network, combining dynamic resource allocation and intelligent scheduling algorithm, real-time scheduling of mesh data processing tasks, and protecting data security through data encryption and access control mechanism, realizing an efficient and secure vehicle-mounted edge computing system. All the devices involved in the present application can connect to the network and have completed time synchronization, the communication nodes are composed of lightweight vehicle-mounted nodes in the automobile network environment, road facility computing nodes in the road network system, edge side devices responsible for network communication transfer in the road network system and full nodes of data centers processing vehicle networking business, different nodes realize the safe sharing of data through two kinds of cooperative working modes, so as to prevent network attacks and protect data security, the specific scheme includes the following steps:

[0123] Step 1, deploying vehicle-mounted edge nodes

[0124] Step 1.1 mesh device network

[0125] Due to the large number of devices in the vehicle networking system, and the existence of mobile devices similar to the physical location and online state of the vehicle-mounted device, if there are a large number of such devices at the same time, if the entire vehicle networking system is used as a computing power cluster, and the computing power in the entire computing power cluster is used, the centralized network structure cannot meet the demand for a large number of task scheduling in the entire cluster, at the same time, it may cause a large number of vehicle-mounted devices to upload data to the same center server, resulting in a decrease in server device computing power and low data sharing efficiency. The present application introduces a vehicle-mounted edge node specially used to establish a connection with the vehicle-mounted edge node to balance the load pressure of the center server, thereby reducing the load pressure of the center server, and improving the load coordination capability of the entire vehicle network computing power cluster. Assuming that the number of vehicle-mounted devices is n, denoted as T1, T2,..., T n , assuming that the number of edge nodes is m, E1, E2,..., E m , then the number of devices accessed by each edge node is Assuming that the number of center servers is k, C1, C2,..., C k , then the number of devices accessed by each center node is All nodes are arranged in a network plane, and the center node and the edge node are connected one-to-many, and the edge node and the vehicle-mounted node are connected one-to-many, so as to form a network topology graph of the vehicle-mounted node, the edge node and the center node.

[0126] Based on the network topology structure formed above, it can be converted into a two-dimensional plane network topology graph, and through reasonable grid division, all devices in the grid can be arranged in order, thereby forming a two-dimensional grid graph. In this way, the vehicle-mounted edge node can form a structured layout in geography and logic to optimize data processing and resource management, and improve the overall efficiency and reliability of the system.

[0127] Step 1.2 Calculate the load capacity of the node

[0128] In the two-dimensional grid graph mentioned above, actively identifying the device and calculating its load capacity is a key step to ensure efficient operation of the system and reasonable allocation of resources. This process can be described in detail through the following steps.

[0129] Firstly, each device needs to be actively identified and assigned a unique identifier. Each on-board edge node needs to be assigned a unique identifier (ID), which can be achieved by combining hardware characteristics (such as MAC address) and software characteristics (such as a unique ID generated when the node is registered). The allocation of a unique identifier not only ensures the uniqueness of the node, but also facilitates subsequent management and tracking. At the same time, when the node is first started, it performs self-registration. The registration process includes the node sending a registration request containing its unique identifier, geographical location, hardware configuration, and other basic information to the central management system or blockchain network. The central system verifies and records these information to ensure the legality and security of the node. The node needs to periodically report its running status to the central system or blockchain network, including the current load situation, available resources, and network status. This periodic status report ensures real-time monitoring of the system for each node and enables timely response to any abnormal situations.

[0130] Then, all nodes in the grid need to be evaluated for their load capacity to determine the size of the computing load they can carry. Each on-board edge node needs to monitor its load in real time, which includes CPU usage, memory usage, storage space usage, network bandwidth utilization, and battery power. The monitoring program inside the node can use system calls or monitoring tools provided by the operating system to obtain these data. Based on real-time monitoring data, the node needs to evaluate its load capacity. The evaluation can be achieved by defining a set of standard calculation formulas or models. For example, the load capacity of a node can be evaluated by the following simplified formula:

[0131]

[0132] Where the weight of each resource can be adjusted according to system requirements. The result of this evaluation formula can be used as a measure of the node's load capacity.

[0133] The system needs to dynamically adjust task allocation and resource usage based on the results of the node's load capacity evaluation. Nodes with higher load capacity can be assigned more tasks, while nodes with lower load capacity need to reduce the amount of tasks or be prioritized for resource scheduling. This dynamic load balancing ensures the overall efficient operation of the system, avoiding overloading or idling of some nodes. The load capacity evaluation results of all nodes need to be recorded in the central system or blockchain network to facilitate global resource management and optimization. The system should regularly update these records to reflect the latest node status and load situation, ensuring the accuracy and timeliness of scheduling decisions.

[0134] Through the above steps, the device is actively identified and its load capacity is calculated, which not only helps the efficient operation of the vehicle-mounted edge computing system, but also realizes the reasonable allocation and dynamic adjustment of resources, improves the overall performance and reliability of the system. The real-time state and load capacity information of each node provides a comprehensive perspective for the system, ensuring the optimization of data scheduling and resource management.

[0135] Step 1.3 Detection of regional computing capacity

[0136] First, a comprehensive computing capacity detection is performed on each grid region. This step requires the deployment of a real-time monitoring system that can continuously collect performance data of each vehicle-mounted edge node. Assume that the performance indicators of each node n include CPU usage C n , memory occupancy R n , storage space utilization S n , network bandwidth usage BW n , and average response time RT n . Through these performance indicators, the system can evaluate the load and processing capacity of each node. We define the comprehensive computing capacity C g of each grid unit as follows:

[0137]

[0138] where ω1, ω2, ω3, ω4, ω5 are the weight coefficients of each performance indicator, reflecting the relative importance of each indicator to the comprehensive computing capacity. For grid units with a comprehensive computing capacity C g lower than the preset threshold C threshold , they can be identified as regions with poor computing capacity. After completing the detection, the system generates a detailed report listing the grid regions that need to be optimized and their specific performance bottlenecks.

[0139] Step 1.4 Adjusting the number of nodes

[0140] After identifying the regions with poor computing capacity, the next step is to add new vehicle-mounted edge nodes in these regions. The selection of new nodes should be based on the results of the previous detection, and hardware devices with high computing and communication capabilities should be selected. Assume that the computing capacity of the new node is C NEW , and the appropriate geographical location L is selected for node deployment to ensure that the new node can effectively cover the entire weak region and provide necessary computing and storage resources. After the installation of the new node, configuration and initialization are performed, including operating system installation, edge computing platform deployment, and network connection testing. After configuration is completed, the new node needs to be integrated with the existing system through a self-registration mechanism, and the node identifier ID NEW, hardware information and location data are sent to the central management system for verification and authentication. After authentication, the new node can participate in actual computation and data processing tasks. Through real-time monitoring and dynamic load balancing, the system transfers part of the tasks to the new node to optimize the overall grid area's computing power. The new comprehensive computing power C g ′ is:

[0141] C g ′ = C g + C NEW

[0142] Ensure load balancing and efficient operation. In this way, gradually improve the overall performance of the area with poor computing power, and achieve balanced development and resource optimization of the system.

[0143] Step 1.5: Divide network sub-clusters

[0144] First, the criteria for dividing network nodes need to be determined, which are mainly based on the geographical location of nodes, communication delay, node density and physical distance. By using geographic information system (GIS) data combined with actual traffic and communication network conditions, the relative positions and communication paths between nodes are identified. Set a maximum communication delay threshold T max and a minimum node density threshold D min to ensure that nodes within each network cluster can communicate efficiently and have sufficient resources. To accurately divide network clusters, the physical distance d ij and communication delay t ij between nodes need to be calculated. The physical distance d ij can be calculated using the Haversine formula based on the geographical coordinates (longitude and latitude) of the nodes:

[0145]

[0146] where Δφ = φj - φi is the difference in latitude between nodes i and j, Δλ = λ j - λ i is the difference in longitude, and r is the radius of the Earth. Based on these distance data, a distance matrix D is constructed, where D ij represents the distance between nodes i and j.

[0147] Using the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm, nodes can be divided into different network clusters. The DBSCAN algorithm identifies high-density areas and divides clusters by setting ∈ (the maximum distance between nodes) and minPts (the minimum number of nodes required to form a cluster). First, a randomly selected unvisited node is checked for all nodes within its ∈ neighborhood. If the number of nodes in the neighborhood is greater than or equal to minPts, a cluster is formed, and all unvisited nodes within the cluster are marked as visited, and the process continues to check the nodes in its neighborhood. Repeat the above steps until all nodes are visited and classified. After the initial division, the cluster results need to be optimized and verified. Through simulation and actual measurement, the communication delay and data transmission efficiency of the nodes in the cluster are verified to ensure efficient communication in each cluster. Finally, the divided network cluster information is recorded in the central management system or blockchain network, and these records are updated and maintained regularly to reflect the latest node status and cluster structure. Through these steps, the structure of the vehicle-mounted edge computing system is optimized, and the data transmission efficiency and resource utilization are improved.

[0148] Step 1.6 Reconstruct the grid map

[0149] First, collect all identified and load-capable device data, including device unique identifier, geographic location (longitude and latitude), load capacity, adjacent node information, and current network status. These data need to be updated through regular status reports to ensure their accuracy and timeliness. Then, based on these geographic location data, all devices are mapped to a geographic information system (GIS) map, using software such as ArcGIS or QGIS to convert the longitude and latitude of the device to coordinate points on the map. To improve visualization, different colors and sizes can be used to identify nodes, with larger green dots representing higher load capacity nodes and smaller red dots representing lower load capacity nodes.

[0150] Next, draw the network connection between devices on the map to show the communication relationship between nodes. Construct the adjacency matrix AAA, where A ij =1 indicates that there is a direct communication connection between node i and node j, and A ij=0 indicates no direct connection. Adjacency matrices are used to draw connections between devices, reflecting actual communication links, with different colors or thicknesses representing connection quality or bandwidth. Based on cluster partitioning results, nodes are divided into different network clusters, and each cluster is displayed on the map with different colors or bounding boxes to clearly distinguish the scope of each cluster. Ensure that connections within a cluster are visually distinct; solid lines represent connections within the same cluster, while dashed lines represent connections between different clusters. The system should have dynamic update and interactive functions, automatically updating node status and connectivity through real-time monitoring data. Dynamic data visualization tools (such as D3.js, Leaflet, etc.) are used to achieve real-time updates and interactive operations of nodes and connections. Users can click on nodes to view detailed information or select specific areas for zooming in and detailed browsing, ensuring the efficient operation of the vehicle-mounted edge computing system.

[0151] Step 2: Build a blockchain network

[0152] Step 2.1 Node Deployment and Configuration

[0153] First, select appropriate devices as nodes for the blockchain network. These devices can be vehicle-mounted edge nodes, roadside infrastructure, or cloud servers. Define the node set as N = {n1, n2, ..., n}. k}, each node n i You need to install blockchain software, such as Hyperledger Fabric or Ethereum, and perform the necessary configurations. This includes setting the node's identity ID. i Distribute encryption key K i Configure network parameters and chaincode (smart contracts). Ensure all nodes have sufficient computing and storage capacity to support blockchain operation. Simultaneously, secure communication channels need to be established between nodes, using TLS (Transport Layer Security) or VPN (Virtual Private Network) to protect data transmission security. (The last sentence appears to be incomplete and possibly refers to a separate step.) i and S i Representing node n respectively i The computing and storage capabilities ensure that C i ≥C min And S i ≥S min .

[0154] Step 2.2 Initialize the blockchain

[0155] After the nodes are deployed and configured, the blockchain network is created. The initial configuration of the blockchain is defined, including the setup of the genesis block (GenesisBlock), which is set as G0. The genesis block contains the initial state of the blockchain network and information about all participating nodes. Then, the blockchain software is run on each node, starting the blockchain network and performing initialization operations to synchronize all nodes to the genesis block. During the initialization process, the consensus mechanism is set, such as PoW (Proof of Work), PoS (Proof of Stake), or PBFT (Practical Byzantine Fault Tolerance), ensuring that the network can reach consensus and process transactions. The choice of consensus mechanism M affects the security and efficiency of the blockchain, for example:

[0156] M = PoW or M = PoS or M = PBFT

[0157] Step 2.3 Deploying smart contracts and testing

[0158] Finally, smart contracts are deployed to implement specific functions of the blockchain network. Set the collection of smart contracts as SC = {sc1, sc2,..., sc m}, which can be used to manage node identities, record transactions, allocate resources, etc. After deploying the smart contracts, comprehensive functional testing and performance testing are conducted. The test content includes verifying the correctness of the smart contract V(sc i ), detecting the efficiency of transaction processing E(sc i ), and evaluating the stability and security of the network. By simulating real scenarios, it is ensured that the smart contracts can operate normally and the blockchain network can efficiently process data. After testing is completed, the blockchain network is officially enabled, and routine maintenance and monitoring are carried out to ensure its continuous and stable operation.

[0159] Step 3, Dynamic resource allocation

[0160] Step 3.1 Node deployment and configuration

[0161] First, establish a real-time monitoring system to collect and analyze the resource usage of each node. The monitoring indicators include CPU usage, memory occupancy, bandwidth usage, and storage utilization, etc. The resource usage of each node n i can be represented as a vector R i = (CPU i , MEM i , BW i , STOR i ). Through distributed monitoring tools such as Prometheus or Zabbix, these data are collected regularly (e.g. every second) and stored in a central database. Ensure the real-time and accuracy of data collection to reflect the resource usage in the network in a timely manner.

[0162] Step 3.2 Resource demand prediction and analysis

[0163] Based on the collected real-time data, use machine learning algorithms for resource demand prediction.

[0164] Assume the resource demand of each node is D i (t) = (CPU d,i (t), MEM d,i (t), BW d,i (t), STOR d,i (t)), predict the resource demand in the future period. Common prediction models include time series analysis (such as ARIMA model) and deep learning model (such as LSTM network). Through prediction, determine the resource demand of each node at time t in the future, and evaluate whether there is a resource shortage or excess. According to the prediction result, generate resource allocation strategy S(t) for adjusting the resource allocation of each node.

[0165] Step 3.3 Resource scheduling and allocation execution

[0166] According to the resource allocation strategy S(t) generated by prediction, perform dynamic resource scheduling and allocation. The scheduling strategy can include resource migration, load balancing and priority adjustment, etc. Specifically, use container technology (such as Docker) or virtualization technology (such as KVM) to migrate resources from lightly loaded nodes to heavily loaded nodes. For example, assume that from node n i to node n j migrate CPU resources ΔCPU, then the adjusted resource allocation can be represented as:

[0167] R' i (t) = (CPU i - ΔCPU, MEM i , BW i , STOR i )

[0168] R' j (t) = (CPU j + ΔCPU, MEM j , BW j , STOR j )

[0169] By performing resource migration and adjustments, the resource utilization of each node is ensured to remain within a reasonable range, avoiding resource waste and performance bottlenecks. Simultaneously, latency and overhead during resource scheduling are monitored to ensure efficient scheduling and minimal impact on system performance. The execution of resource allocation strategies should include automation and intelligence, utilizing automated operation and maintenance tools (such as Kubernetes) to achieve dynamic scheduling and allocation of resources to meet the high-efficiency operation requirements of the vehicular edge network.

[0170] Step 4: Intelligent Scheduling Algorithm

[0171] The intelligent scheduling algorithm aims to optimize the allocation and execution efficiency of tasks in vehicular edge networks. By combining load balancing, real-time data analysis, and machine learning techniques, the intelligent scheduling algorithm can dynamically adjust task allocation strategies to improve resource utilization and network performance. This algorithm needs to consider multiple factors, including node computing power, current load, network bandwidth, task priority, and latency requirements. The main objective is to minimize task execution time and resource consumption while ensuring system stability and reliability.

[0172] Step 4.1 Task Priority Classification and Initial Assignment

[0173] First, categorize the tasks based on their importance and urgency. Let the task set be T = {t1, t2, ..., t...} m}, each task t i Having priority p i and resource requirements R i =(CPU) i MEM i BW i STOR i According to priority p) i The tasks are sorted, and high-priority tasks are assigned to nodes with sufficient resources. The initial allocation strategy is based on the current load and resource capabilities of the nodes, assuming the node set is N = {n1, n2, ..., n}. k}, for each task t i Choose the most suitable node n j , so that n j The resource requirements R that can meet the task i .

[0174] Step 4.2 Real-time monitoring and task migration

[0175] During task execution, the resource usage and task execution status of nodes are monitored in real time. Data is collected through a distributed monitoring system, including node CPU utilization, memory usage, network bandwidth, and storage utilization. Node n is defined. j The resource status is R j =(CPU)j ,MEM j ,BW j ,STOR j If the resource usage of a certain node n j is detected to exceed the preset threshold θ, the task migration mechanism is started, and part of the tasks are migrated to the lightly loaded node n l . During the migration process, the state and data integrity of the tasks need to be ensured, and the delay and overhead during the migration process need to be reduced.

[0176] Step 4.3 Machine learning prediction and optimization

[0177] Use machine learning algorithms to predict the resource usage and task load changes of each node in the future period. Assume that the resource requirement of each node n j is

[0178] D j (t)=(CPU d,j (t),MEM d,j (t),BW d,j (t),STOR d,j (t)),

[0179] Make predictions through time series analysis or deep learning models such as LSTM. According to the prediction results, adjust the task allocation strategy and perform resource scheduling and optimization in advance. The goal of the algorithm is to minimize the task execution time T i and resource consumption C i , while improving the overall performance and stability of the system.

[0180] Step 4.4 Feedback mechanism and adaptive adjustment

[0181] Introduce a feedback mechanism to continuously adjust and optimize the scheduling algorithm based on task execution results and system performance indicators. Collect data such as task completion time, resource usage efficiency, and network delay as the basis for algorithm optimization. Use reinforcement learning methods to find the optimal scheduling strategy through continuous testing and adjustment. Set a reward function R(t) to give positive or negative feedback based on the improvement of system performance indicators, adjust algorithm parameters, and achieve adaptive optimization. Ultimately, the intelligent scheduling algorithm can dynamically adapt to changes in the network environment, improving the task execution efficiency and resource utilization of the vehicle edge network.

[0182] Step 5, Grid data management

[0183] Grid-based data management aims to optimize data storage and transmission efficiency in vehicular edge networks by dividing data into grid cells for management, enabling distributed storage and fast access. Grid-based data management needs to consider the geographical location of data, access frequency, and resource conditions of storage nodes to ensure efficient use and reliable transmission of data.

[0184] Step 5.1: Data grid division and distributed storage

[0185] First, the entire data space is divided into multiple grid cells according to geographical location and data type. Set the data space D, which is divided into G = {g1, g2, …, g n} grid cells. Each grid cell g i contains data within a specific area, allocated according to the geographical location and type of data. Use geographic information system (GIS) tools to map data into corresponding grid cells.

[0186] Then, the data of each grid cell is stored in the corresponding vehicle edge node or cloud storage node. Set the node set N = {n1, n2, …, n k}, according to the storage capacity and geographical location of the nodes, the data of the grid cell g i is stored in the most suitable node n j . Ensure that the storage distribution of data is balanced to avoid overloading some nodes. At the same time, to improve the reliability and access speed of data, data redundancy and replication technology can be used to store critical data on multiple nodes.

[0187] Step 5.2: Data access optimization and dynamic adjustment

[0188] During data access, according to the access frequency and node load, dynamically adjust the storage location and access strategy of data. Set the data access set A = {a1, a2, …, a m}, each data access a i contains the accessed grid cell g i and data item d i . By monitoring the data access frequency f(a i ) and the resource usage R j = (CPU j , MEM j , BW j , STOR j ) of the nodes, determine whether to adjust the storage location of the data.

[0189] If the data access frequency of certain grid cells significantly increases, these data are migrated to more efficient storage nodes to reduce access latency and improve response speed. Machine learning algorithms are used to predict future data access patterns, and pre-scheduling and optimization are performed based on the prediction results. For example, assuming the access frequency of grid cell g i is predicted as f(g i ,t), the data storage strategy can be adjusted according to the prediction results. Through these optimization measures, efficient management and dynamic adjustment of grid data are realized, ensuring efficient and reliable data transmission and storage in the vehicle edge network.

[0190] Step 6: Data security and privacy protection

[0191] Data security and privacy protection are crucial in the vehicle edge network, ensuring the confidentiality, integrity, and availability of data during transmission and storage. By adopting encryption technology, access control mechanisms, and data anonymization processing, user privacy and data security can be effectively protected.

[0192] Step 6.1: Data encryption and transmission protection

[0193] First, encryption technology is used to protect data transmission and storage. For data transmission, use TLS (Transport Layer Security) or IPSec (Internet Protocol Security) to encrypt the channel, ensuring that data is not eavesdropped and tampered with during network transmission. Set the encryption method for transmitting data D t as E(D t ,K t ), where K t is the transmission encryption key. In terms of data storage, use AES (Advanced Encryption Standard) or RSA (RSA encryption algorithm) to encrypt the stored data, and set the encryption method for storing data D s as E(D s ,K s ), where K s is the storage encryption key. Through encryption technology, the confidentiality and integrity of data during transmission and storage are ensured.

[0194] Step 6.2: Access control and permission management

[0195] Second, implement strict access control and permission management mechanisms to ensure that only authorized users and devices can access and operate data. Use the Role-Based Access Control (RBAC) or Attribute-Based Access Control (ABAC) model to assign different access permissions based on user roles or attributes. Set the user set U = {u1, u2, …, u m}, and each user u i is assigned a permission set P i = {pi1 ,p i2 ,…,p in}. Access control lists (ACLs) are implemented on nodes to record access permissions for each user or device, and to verify and authorize each data access request. Through strict access control, unauthorized access and data leakage are prevented.

[0196] Step 6.3: Data anonymization and privacy protection

[0197] Finally, data anonymization and privacy protection techniques are adopted to ensure effective protection of user privacy during data processing. Techniques such as data de-identification (e.g., de-identification, pseudo-anonymization), and differential privacy are used to remove or obscure sensitive information before processing user data. Set the original data D r and the processed data D a , generate anonymized data D r through the anonymization function A(D a ), to ensure that the processed data cannot be directly linked to individuals. Apply differential privacy technology to add noise ∈ to the data to ensure that the query result does not leak individual privacy. The differential privacy model can be represented as D dp = D + ∈. Through these privacy protection measures, the privacy and security of user data during processing and analysis are ensured.

[0198] Through the three sub-steps of data encryption and transmission protection, access control and permission management, data anonymization and privacy protection, data security and privacy protection in the vehicle edge network are realized. Advanced encryption technology is used to protect data transmission and storage, strict access control mechanisms prevent unauthorized access, and data anonymization and privacy protection techniques ensure that user privacy is not leaked during data processing. The comprehensive application of the above measures can effectively protect the data security and user privacy in the vehicle edge network.

[0199] Embodiment 12:

[0200] An application of the vehicle edge mesh data scheduling system based on the combination of blockchain and load balancing according to any one of embodiments 1-11, comprising the following steps:

[0201] Step 1, deploy vehicle edge nodes

[0202] Deploying computing nodes on vehicles and transportation infrastructure is the foundation of the entire system. Each vehicle edge node n i is equipped with necessary sensors, processors and storage devices for real-time data collection, processing and storage. In order to ensure coverage of the entire target area, all devices are meshed and divided into grid cells G = {g1, g2, …, g n}. In the grid, the areas with poor computing power are added with additional vehicle-mounted edge nodes to make up for the lack of computing resources. Each node is labeled according to its geographical location and computing power, and its computing power C i and storage capacity S i are set i to ensure that C min ≥ C i and S min ≥ S k . At the same time, the network nodes in the adjacent area are divided into different network clusters, and through active identification of devices and recording of their load capacity, a regional network diagram is finally drawn, showing the node distribution and load situation, as shown in Fig. 1 The system architecture diagram before the grid is shown in Fig. 2 , and the system structure diagram after the grid is shown in

[0203] Step 2, building a blockchain network

[0204] Building a blockchain network involves organizing multiple vehicle-mounted edge nodes N = {n1, n2, …, n k} into a decentralized distributed system. Each node installs blockchain software such as Hyperledger Fabric or Ethereum, and configures its identity ID i , encryption key K i and chain code. The initial configuration includes setting the genesis block G0, which contains the initial state and node information of the network. After starting the blockchain, all nodes synchronize to the genesis block and run the selected consensus mechanism, such as PoW, PoS or PBFT. Through these mechanisms, all data transactions are verified and recorded to ensure data integrity and security. Nodes establish secure communication channels through TLS or VPN to prevent data from being eavesdropped or tampered with during transmission.

[0205] Step 3, dynamic resource allocation

[0206] Dynamic resource allocation ensures system load balancing by real-time monitoring and predictive adjustment of computing and storage resources of each node. First, establish a real-time monitoring system to regularly collect node resource usage

[0207] R i = (CPU i , MEM i , BW i , STOR i )

[0208] Then, use machine learning algorithms to predict future resource requirements

[0209] MEM_{d,i}(t),BW_{d,i}(t),STOR_{d,i}(t)),

[0210] Generate a resource allocation strategy S(t). Based on the strategy, dynamically adjust resource allocation using container or virtualization technologies, for example, from node n. i To node n j Migrate resource ΔCPU, and the adjusted resource status is as follows:

[0211] R′ i (t)=(CPU i -ΔCPU,MEM i BW i STOR i )

[0212] and R j ′(t)=(CPU j +ΔCPU,MEM j BW j STOR j ).

[0213] Step 4: Intelligent Scheduling Algorithm

[0214] Intelligent scheduling algorithms utilize machine learning to analyze historical data and real-time status, predicting data traffic and computational demands to achieve efficient scheduling. The task set T = {t1, t2, ..., t...} m} By priority p i Sort and assign tasks to nodes with sufficient resources. Monitor task execution status and node load in real time. If node resource utilization exceeds a threshold θ, initiate task migration to move tasks to nodes with lighter loads. Predict node resource requirements using time series analysis or LSTM models.

[0215] D j (t)=(CPU d,j (t),MEM d,j (t),BW d,j (t),STOR d,j (t)),

[0216] The task allocation strategy is adjusted based on the prediction results. A feedback mechanism is introduced to continuously optimize the scheduling strategy through reinforcement learning, thereby improving the overall system performance and resource utilization.

[0217] Step 5: Grid-based data management

[0218] Grid-based data management divides vehicle-mounted edge nodes into multiple grid cells, enabling distributed data storage and collaborative processing. The data space D is divided into G = {g1, g2, ..., g...}. n There are} grid cells, and the nodes within each cell collaboratively process and store data. The data is distributed and stored on corresponding on-board edge nodes or cloud storage nodes. N = {n1, n2, ..., nn}k} are allocated based on their storage capacity and geographical location, ensuring balanced data storage and efficient access. During data access, the access frequency f(a i ) and node resource status R j are dynamically adjusted, using machine learning algorithms to predict future access patterns f(g i ,t) and optimize data storage strategies, ensuring efficient and reliable data management.

[0219] Step 6, Data Security and Privacy Protection

[0220] Data security and privacy protection are achieved through encryption technology, access control, and data anonymization. Data transmission uses TLS or IPSec encrypted channels to ensure data is not intercepted and tampered with, and the encryption method for transmitting data D t is E(D t ,K t ). Data storage uses AES or RSA encryption algorithms, and the encryption method for storing data D s is E(D s ,K s ). Access control uses RBAC or ABAC models to assign permissions P i based on user roles or attributes, and implements access control list (ACL) verification and authorization on nodes. Data anonymization uses desensitization and differential privacy technology to remove or obscure sensitive information before processing user data, and the anonymization function A(D r ) generates data D a , and the differential privacy model is represented as D dp =D+∈. Through these measures, the security of data transmission and storage and the protection of user privacy are ensured.

Claims

1. A blockchain-based and load-balanced vehicular edge meshed data scheduling system, characterized in that: The vehicle-mounted edge mesh data scheduling system is used for realizing data scheduling of a vehicle-mounted network; The vehicle-mounted network comprises a plurality of central server nodes, a plurality of vehicle-mounted edge nodes and a plurality of vehicle-mounted terminal device nodes; part or all of the nodes are located in a blockchain; The vehicle-mounted edge mesh data scheduling system comprises a dynamic resource allocation module, an intelligent scheduling module, a data management module, a data security and privacy protection module; The dynamic resource allocation module predicts resource demand of the nodes in a future t time period based on node monitoring indicators, and generates and executes a resource allocation strategy S(t); The intelligent scheduling module generates a task allocation strategy and dynamically adjusts the task allocation strategy with the objective of minimizing task execution time and resource consumption; The data management module divides data of the vehicle-mounted edge mesh data scheduling system into grid units for management, and realizes distributed storage and fast access of the data; The data security and privacy protection module protects user privacy and data security through encryption technology, access control mechanisms and data anonymization processing; The dynamic resource allocation module comprises a resource monitoring unit, a resource demand prediction and analysis unit and a resource scheduling and allocation execution unit; The resource monitoring unit collects node monitoring indicators R i =(CPU i ,MEM i ,BW i ,STOR i ) to analyze the resource usage of each node and upload the node monitoring indicators to a central database for storage; CPU i ,MEM i ,BW i ,STOR i respectively represent CPU usage, memory occupation, bandwidth usage, and storage utilization. The resource demand prediction and analysis unit takes node monitoring indicators as input, predicts resource demand of the nodes in a future t time period by using a resource demand prediction model, and generates a resource allocation strategy S(t) according to a prediction result; the resource allocation strategy S(t) comprises resource migration, load balancing and priority adjustment strategies; The resource demand prediction model comprises a time series analysis model and a deep learning model; The resource scheduling and allocation execution unit performs dynamic resource scheduling and allocation according to the resource allocation strategy S(t); The intelligent scheduling module comprises a task initial allocation unit, a task migration unit and an adaptive adjustment unit; The initial task allocation unit sorts tasks according to their importance and urgency, and allocates tasks based on the current load and resource capabilities of the nodes, thereby allocating tasks t to each task. i Select R that meets the task's resource requirements. i node n j ; The task migration unit monitors resource usage and task execution state of the nodes in real time during task execution, and if it detects that the resource usage of a node n j exceeds a preset threshold θ, it starts a task migration mechanism to migrate part of the tasks to a node with a load less than a preset value. The adaptive adjustment unit predicts resource usage and task load changes of the nodes in a future t time period by using a machine learning algorithm, and continuously adjusts and optimizes the task allocation strategy according to task execution results and system performance indicators.

2. The vehicle-mounted edge mesh data scheduling system based on the combination of blockchain and load balancing according to claim 1, characterized in that: The data management module comprises a data grid division unit and a data storage location and access strategy optimization unit; The data grid division unit divides the entire data space D into a plurality of grid units according to geographical positions and data types, denoted as G={g1, g2, …, gM}; gM is the Mth grid unit; M is the number of grid units. M};g M M is the number of grid units. The data grid division unit stores data of each grid unit in a corresponding vehicle-mounted edge node or a cloud storage node; The data storage location and access strategy optimization unit dynamically adjusts the storage location and access strategy of the data according to access frequency and node load conditions in a data access process; The steps of adjusting the storage location and access strategy of the data comprise: if the data access frequency growth rate of the grid units is greater than a preset value, the data is migrated to a more efficient storage node to reduce access delay and improve response speed. 3.The blockchain-based and load-balanced vehicle-mounted edge mesh data scheduling system according to claim 1, characterized in that: The data security and privacy protection module comprises a data encryption and transmission protection unit, an access control and permission management unit and a data anonymization and privacy protection unit; The data encryption and transmission protection unit uses a TLS or IPSec encrypted channel for data transmission, and uses E(D t ,K t ) as the encryption method for transmission data D t ; K t is a transmission encryption key. The data encryption and transmission protection unit encrypts stored data by using an encryption algorithm; the encryption algorithm comprises AES and RSA encryption algorithms; The access control and permission management unit ensures that only authorized users and devices can access and operate data through an access control and permission management mechanism; The data anonymization and privacy protection uses data anonymization and privacy protection technology to protect user privacy involved in data processing; The privacy protection technology is differential privacy technology.

4. The blockchain-based and load balancing combined vehicle-mounted edge mesh data scheduling system according to claim 1, characterized in that: In the vehicle-mounted network, the center server node and the vehicle-mounted edge node are connected one-to-many, and the vehicle-mounted edge node is connected one-to-many with the vehicle-mounted device node; The number of each vehicle-mounted edge node access equipment is m is the number of vehicle-mounted edge nodes; n is the number of vehicle-mounted end device nodes; The number of access devices per central server node is k is the number of central server nodes. 5.The blockchain-based and load-balanced vehicle-mounted edge mesh data scheduling system according to claim 1, characterized in that: The center server node, the vehicle-mounted edge node and the vehicle-mounted device all have unique identifiers. 6.The blockchain-based and load-balanced vehicle-mounted edge mesh data scheduling system according to claim 1, characterized in that: When any node starts for the first time, it performs self-registration; The registration process is: the node sends a registration request of basic information to the central management system or the blockchain network; the central system verifies and records the basic information to ensure the legality and security of the node; the basic information includes unique identifier, geographic location, hardware configuration; Any node periodically reports the running state to the central system or the blockchain network, including the current load, available resources and network state. 7.The blockchain-based and load-balanced vehicle-mounted edge mesh data scheduling system according to claim 1, wherein: The set of nodes in a blockchain is denoted as N = {n1, n2, ..., n}. k }; Each node n i Computing power C i ≥C min And storage capacity S i ≥S min C min S min These are the lower limits of computing and storage capacity; i = 1, 2, ..., k; The genesis block of the blockchain is recorded as G0; the genesis block stores the initial state of the blockchain and the information of all participating nodes; The functions of the blockchain are implemented through smart contracts; the set of smart contracts is denoted as SC = {sci, sc2,..., sc m}, which are used to manage node identities, record transactions, and allocate resources. 8.The blockchain-based and load balancing combined vehicle-mounted edge mesh data scheduling system according to claim 1, wherein, It also includes a vehicle-mounted network updating module; The vehicle-mounted network updating module periodically updates the vehicle-mounted network based on the node computing power; The updating steps include: 1) The vehicle-mounted network updating module divides the vehicle-mounted network into multiple grid areas and calculates the computing capacity of each grid area; if there is a grid unit with a computing capacity less than a preset threshold C threshold , a vehicle-mounted edge node is added in the grid area to make the computing capacity of the grid area greater than or equal to C threshold ; 2) After adding the node, the nodes in the in-vehicle network are divided into different network clusters by the DBSCAN clustering algorithm; in each network cluster, the communication delay between any two nodes is less than the communication delay threshold T max , and the distance between any two nodes is greater than the node density threshold D min ; The distance d between any two nodes ij As follows: wherein, is the difference in latitude of node i and node j, Δλ = λ j -λ i is the difference in longitude, r is the earth radius; is the latitude of node j, node i; λ j , λ i is the longitude of node j, node i; 3) Get all identified nodes with load capacity, and map all nodes to a geographic information system map, and draw the network connection between devices on the map to show the communication relationship between nodes, and get the updated vehicle-mounted network; the device data includes the unique identifier, geographic location, load capacity, adjacent node information and current network state of the device; The load capacity of the node is as follows: In the formula, D is the load capacity of the node; S cpu is the available CPU resource of the node; S cpu_total is the total CPU resource of the node; S m is the available memory of the node; S m_total is the total memory of the node; S s is the available storage of the node; S s_total is the total storage of the node; S B is the available bandwidth of the node; S B_total is the total bandwidth of the node; a1, a2, a3, a4 are weight coefficients.

Images