An intelligent networked vehicle computing power resource sharing scheduling method and system

Abstract

The application provides an intelligent networked vehicle computing resource sharing scheduling method and system, which solves the scheduling problem of sharing computing resources of intelligent networked vehicles. The genetic algorithm is adopted, and the number of iterations of the genetic algorithm is improved. Through virtualization, containerization, resource management and other technologies, the safety isolation and scheduling of vehicle-mounted computing power are realized. A large number of idle computing power of intelligent networked vehicles is connected to a dynamic network, and large raw data information that needs to be processed is distributed to the dynamic computing power network node. According to the attributes of different computing tasks and the differences of each computing power node, the genetic algorithm is used to match the computing tasks and the computing power nodes, reduce the computing delay in the computing power scheduling process, and improve the computing power scheduling efficiency.

Description

TECHNICAL FIELD

[0001] The application is used in the technical field of intelligent networked vehicles, and particularly relates to a computing power sharing and computing power resource scheduling method and system for intelligent networked vehicles. BACKGROUND

[0002] With the development of intelligent networked vehicles, the requirements for computing power and low latency in the field of autonomous driving are increasingly high, especially when computing tasks such as virtual reality (VR), augmented vehicle reality (AVR) and image recognition need to be performed, which brings great challenges to autonomous driving vehicles with limited computing resources. When autonomous driving vehicles need to process these raw data in real time and meet the low latency and high reliability performance of data processing and transmission. The intelligent networked vehicle computing power sharing method can reduce the high latency problem caused by the shortage of on-board computing resources, and through V2X communication technology, the computing tasks of the vehicle are sent to the vehicle with idle computing power, and the computing tasks of the vehicle are processed in real time, thereby reducing the overall latency of the computing tasks.

[0003] However, different application scenarios require different scheduling designs of computing power sharing resources. The design of the vehicle networking cooperative computing resource scheduling system using the particle swarm algorithm has achieved some results. In the patent document CN202210367435.4 "A vehicle networking cooperative computing resource scheduling design method based on a particle swarm algorithm", the computing resource scheduling problem is converted into an optimization function solving problem, and then a heuristic algorithm is applied and designed to equivalent the complex mixed integer nonlinear programming problem into the variables in the particle swarm evolution mechanism. In the patent document CN202110857102.5 "A computing task offloading method for an edge vehicle networking system", the traditional particle swarm algorithm is innovated, the priority attribute is set according to the characteristics of the task, so that the tasks with higher priority can be processed first, and the computing time and waiting time of tasks with different attributes are balanced, and the total average weighted computing time of all tasks in the system is the shortest.

[0004] The above two technologies use the particle swarm algorithm for optimization design of the computing power resource scheduling system, but the particle swarm algorithm has the problem of falling into a local optimal solution, and cannot obtain the optimal task allocation scheme. SUMMARY

[0005] The application provides an intelligent networked vehicle computing power resource sharing scheduling method and system based on an improved genetic algorithm, which aims to solve the problem of sharing computing power resources for intelligent networked vehicles. According to the attributes of different computing tasks and the differences between each computing power node, the genetic algorithm is used to match the computing tasks and the computing power nodes, to reduce the computing delay in the computing power scheduling process and improve the computing power scheduling efficiency.

[0006] The general idea of the present application is to design an intelligent connected vehicle computing power sharing cloud platform to perform scheduling of computing power resources, to count the residual computing power of intelligent connected vehicles within a predetermined range around the platform by the platform, to realize safe isolation and scheduling of vehicle-mounted computing power through virtualization, containerization, and resource management, to establish a dynamic network connection with a large amount of idle computing power of intelligent connected vehicles, and to distribute large raw data information to be processed to dynamic computing power network nodes.

[0007] The intelligent connected vehicle computing power sharing cloud platform allows new energy intelligent connected vehicle owners to join their vehicles to the computing power sharing cluster. Without affecting the functions and performance of the vehicles, the idle computing power of intelligent connected vehicles is provided, and the corresponding benefits of the vehicle owners are fed back through the calculation time or the calculation data volume.

[0008] The computing task is the raw data processing of virtual reality (VR), augmented vehicle reality (AVR), or image recognition, etc. during the movement of the intelligent connected vehicle. When the computing capacity of the vehicle is insufficient to meet the performance requirements of the computing task, the vehicle will send the computing task and the raw data to the computing power sharing cloud platform through V2X communication technology.

[0009] The technical solution of the present application is as follows:

[0010] According to one embodiment of the present application, the present application proposes a method for sharing and scheduling computing power resources of intelligent connected vehicles, comprising the following steps:

[0011] Step 1, initialize shared computing power information and computing task information.

[0012] Step 2, obtain the CPU computing power, bandwidth, and maximum number of simultaneous processing tasks of the vehicle local computing node in the computing power sharing scheduling area, and count the computing tasks uploaded by the vehicles in the target area, i.e. the computing power sharing scheduling area, as initial task parameters.

[0013] Step 3, apply a genetic algorithm to construct a chromosome containing the computing power scheduling problem, wherein each gene on the chromosome is an element of a feasible solution, and the set Encode the chromosome, wherein is an integer variable, representing the selected offloading vehicle number, represents the number of computing tasks.

[0014] Step 4, input the initial task parameters and the constructed chromosome into the objective function to obtain the initial computing task offloading output result.

[0015] Step 5, input the initial computing task offloading output result, and introduce a genetic algorithm to calculate the optimization result.

[0016] Step 6, the fitness of the chromosome is calculated by a genetic algorithm to perform crossover, mutation and replication on the chromosome

[0017] Step 7, dynamic selection of the number of iterations of the genetic algorithm: according to the number of computing power sharing nodes and the number of computing tasks, dynamically select whether to limit the number of evolutions or to limit the allowed range.

[0018] Step 8, judge whether the allowed condition of evolution is reached; if not, return to step 6.

[0019] Step 9, if yes, output the optimal calculation result, and get the vehicle serial number based on the optimal calculation result, end the iterative calculation, and complete the computing power scheduling task.

[0020] Here, the computing power sharing scheduling area refers to a preset range around the periphery, and the preset range is determined according to the size of the computing task transmission delay. Since the total delay target is composed of two parts, one part is composed of the delay of the computing power sharing vehicle computing and issuing the task, and the other part is composed of the delay of the computing task transmission to the shared computing power node. The preset range is determined according to the size of the transmission delay. When the transmission delay of the computing power node is greater than a set threshold, this computing power node is no longer selected as the node for unloading the computing task by the computing power sharing cloud platform.

[0021] According to an embodiment of the present application, the genetic algorithm is, in the initial stage of the algorithm, a roulette method is used to randomly generate a set of computing power scheduling feasible solutions, that is, the first generation of chromosomes; then the fitness function is used to calculate the fitness of each chromosome, and the probability of each chromosome being selected in the next evolution is calculated according to the fitness; the above is the preparation process, and the following is the formal "genetic algorithm evolution" process. N-M chromosomes are generated by "crossover", and then the N-M chromosomes generated after the crossover are subjected to "mutation" operation, and then M chromosomes are generated by "replication". At this point, N chromosomes are generated, and then the fitness and the probability of being selected next time are calculated, until the limited number of evolutions is completed or the limited allowed range is reached.

[0022] The gene and the chromosome are in the genetic algorithm. We first need to map the shared computing power scheduling problem into a mathematical problem, that is, the so-called "mathematical modeling". A feasible solution of this problem is called a "chromosome". A feasible solution is generally composed of multiple elements, and each element is called a "gene" on the chromosome. Each factor affecting the optimization target of the total delay can be represented as a "gene". The number of genes and the type of gene are determined by the factors affecting the optimization target in the actual problem. The set of genes is denoted as G = {g1, g2, g3, …, gn}, where n is the number of genes. The chromosome is encoded, wherein ​is an integer variable, representing the selected unloading vehicle serial number, represents the number of computing tasks.

[0023] The fitness function can measure the advantages and disadvantages of the chromosomes in the genetic algorithm. The genetic algorithm iterates N times in the running process, and generates several chromosomes in each iteration. The fitness function gives a score to all the chromosomes generated in this iteration, generates a fitness matrix to evaluate the fitness of these chromosomes, and then uses the selection probability matrix to eliminate the chromosomes with low fitness and only keeps the chromosomes with high fitness, so that the quality of the chromosomes will be better and better after several iterations.

[0024] The crossover is that the genetic algorithm generates N chromosomes in each iteration, which is called "evolution" in the genetic algorithm. The newly generated chromosomes in each evolution are generated by "crossover". The process of crossover needs to find two chromosomes from the last generation of computing power scheduling feasible solution chromosomes, one is the father and the other is the mother. Then cut off a certain position of the two chromosomes and splice them together to generate a new chromosome. This new chromosome contains a certain number of father's genes and a certain number of mother's genes. The selection of the last generation of father and mother chromosomes can be completed by the roulette algorithm. After each evolution, the fitness of each chromosome is calculated, and then the fitness probability of each chromosome is calculated by the formula represents the fitness of the chromosome selected, represents the fitness of the chromosome , and represents the sum of the fitness of all chromosomes. When the crossover process is performed, the parent chromosomes need to be selected according to this probability. The probability of being selected by the chromosome with higher fitness is higher.

[0025] The mutation and replication of the chromosome, mutation is that when we generate a new chromosome by crossover, we need to randomly select several genes on the new chromosome and randomly modify the value of the gene, thereby introducing new genes to the existing chromosome, breaking through the current search limit, and more conducive to the algorithm to find the global optimal solution; Replication is that in each evolution, in order to preserve the excellent chromosomes of the last generation, the highest fitness chromosomes of the last generation are directly copied to the next generation. Assuming that N chromosomes are generated in each evolution, then in each evolution, N-M chromosomes are generated by crossover, and the remaining M chromosomes are generated by copying the M chromosomes with the highest fitness of the last generation.

[0026] ​In one embodiment of the present application, there is also provided an intelligent connected vehicle computing resource sharing scheduling system for implementing the method of the aforementioned embodiments, comprising: a computing task or shared computing resource acquisition base station, a cloud server of the base station, a remote intelligent connected vehicle computing node, and a computing task uploading vehicle.

[0027] The computing task or shared computing resource acquisition base station and the cloud server of the base station serve as a computing resource sharing cloud platform, which comprises a computing task uploading module, a computing resource sharing node module, and a computing resource scheduling module.

[0028] Specifically, the computing task uploading module comprises a sensor raw data acquisition unit, a task timer, a task monitor, a computing processing unit, and a task queue generation unit.

[0029] The computing resource scheduling module comprises an algorithm unit and a task and computing resource node mapping unit.

[0030] The computing resource sharing node module comprises a total task queue unit, a task monitor, a computing processing unit, and a computing resource node information generation unit.

[0031] According to one embodiment of the present application, in the computing task uploading module, the raw data acquisition unit mainly receives the sensors on the intelligent connected vehicle for collection through V2X communication technology, including cameras, laser radars, millimeter wave radars, etc.; the task timer mainly calculates the time for each collected raw data computing task to prevent exceeding the maximum delay time of each computing task; the task monitor mainly monitors each uploaded computing task to ensure that the computing task does not miss or have data errors, etc.; the computing processing unit mainly classifies and packages the data in the computing task, adds the properties and requirements of each computing task; and the task queue generation unit mainly sorts according to the properties, size, time sequence, etc. of each task to prepare for the generation of a task and computing resource node mapping graph by the subsequent computing resource scheduling module.

[0032] According to one embodiment of the application, in the computing power sharing node module, the total task queue is mainly to count the computing task queue that has been issued to each computing node, to provide key information and data indicators for subsequent monitoring and management of computing power nodes; the task monitor is to supervise the calculation of each computing task in each computing power node, to ensure efficient and low-delay completion of each computing task, and to provide key information for subsequent computing power nodes on how much computing power they can provide and how long the computing delay will be; the computing processing unit is mainly to count the computing power usage of each computing power node and the key information of each computing power node, to provide information and data collection for subsequent formation of the attributes of each computing power node; and the computing power node information generation unit is mainly to generate real-time updated computing power node attribute information for each computing power node according to the key information of each computing power provided by the computing processing unit, to provide key information for the computing power scheduling module to allocate tasks to each computing power node.

[0033] According to one embodiment of the application, in the computing power scheduling module, the algorithm unit adopts a genetic algorithm to calculate the computing power scheduling strategy, mainly according to the task queue information generated by the computing task module and the computing power node information provided by the computing power sharing node, to use the genetic algorithm to perform task and computing power node scheduling matching work according to the attributes, requirements, and other information of each task, in order to efficiently, low-delay, and accurately calculate the results of each computing task; and the task and computing power node mapping unit is mainly to match the computing task with the computing power node, to determine the specific computing task to be issued to a certain computing power node or several computing power nodes, for subsequent computing task issuing work.

[0034] The advantages of the application are as follows:

[0035] The application uses a cloud platform computing power scheduling system to count the residual computing power of intelligent connected vehicles within a predetermined range, adopts a genetic algorithm to design the scheduling of computing power resources, and realizes the safe isolation and scheduling of vehicle-mounted computing power through virtualization, containerization, and resource management technologies. A dynamic network connection is established for a large amount of idle computing power of intelligent connected vehicles, and large raw data information that needs to be processed is issued to dynamic computing power network nodes. By allowing new energy intelligent connected vehicle owners to join their vehicles into a computing power sharing cluster, the idle computing power of intelligent connected vehicles is provided without affecting the functions and performance of the vehicles. The vehicle owners' corresponding benefits are fed back through the calculation time or the amount of calculation data. The genetic algorithm has the characteristics of group search and can avoid the problem of falling into a local optimal solution.

[0036] In addition, the number of iterations of the genetic algorithm is improved, mainly according to the attributes of different computing tasks, including the size, delay, distance and other information of the computing task, and the key data information of the computing node, including the size of the remaining computing power of the computing node, the distance of the computing node, the energy remaining condition of the computing node, etc., the number of iterations of the genetic algorithm is adaptively adjusted, the computing delay in the computing power scheduling process is reduced, and the computing power scheduling efficiency is improved, which is more valuable in application. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 An embodiment of an intelligent connected vehicle computing power sharing scheduling system;

[0038] Figure 2 A computing power sharing cloud platform component of an intelligent connected vehicle computing power sharing scheduling system;

[0039] Figure 3 An implementation flowchart of an intelligent connected vehicle computing power sharing scheduling method. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application. The described embodiments are only a part of the embodiments of the present application.

[0041] Figure 1 An embodiment of an intelligent connected vehicle computing power sharing scheduling system;

[0042] The computing task or shared computing power acquisition base station includes 130, 131, …, the cloud server 140, 141, … of the base station, the remote intelligent connected vehicle computing node 120, 121, 122, 123, 124, …, the computing task uploading vehicle 110, 111, 112, 113, …, in the process of computing power sharing, the intelligent connected vehicle is the provider of shared computing power, and is also the user of shared computing power. The base station and the cloud server of the base station can be used as the intelligent connected vehicle computing power sharing cloud platform, and the computing power sharing cloud platform can be coupled to the mobile core network cloud platform 150 through communication, and the mobile core network 150 can provide services for the computing power sharing cloud platform through the 5G network. The intelligent connected vehicles within the signal range of the base station can be used as parallel computing nodes, and the computing power is scheduled through the computing power sharing cloud platform.

[0043] When a large number of intelligent connected vehicles drive on the road, the vehicle-mounted camera collects a large amount of video image information around the vehicle, the vehicle-mounted RTK inertial integrated navigation collects the vehicle's own positioning information, and through the vehicle-mounted communication unit OBU, the 5G communication technology is used to send a large amount of raw data information collected in real time to the intelligent connected vehicle computing power sharing platform.

[0044] When the intelligent connected vehicle computing power sharing cloud platform receives a computing task, the computing power is scheduled through the computing power scheduling module, and then the task is issued to each computing power node. Each computing power node needs to process the issued computing task, for example, the road video data is processed by the vehicle-mounted computing power resource data transmission module, the vehicle-mounted state data transmission module and the computing task management module.

[0045] The intelligent connected vehicle computing power sharing scheduling system realizes the computing task offloading mode by modeling the computing power nodes and computing tasks in the intelligent connected vehicle system, converting the complex computing power scheduling event into an optimal function solving problem, applying and designing a heuristic algorithm, equivalent the complex mixed integer nonlinear programming problem into the variable in the chromosome evolution mechanism in the genetic algorithm, and eliminating the factors that have no influence on the algorithm. The task offloading model can be greatly simplified.

[0046] The specific modules involved in the computing power sharing cloud platform are as shown in Figure 2 As shown, it includes a computing task uploading module, a computing power sharing node module and a computing power scheduling module. The relationship between each module and the functional units it contains is as follows:

[0047] The sensor raw data acquisition unit in the computing task module obtains the computing task data uploaded by each vehicle-end sensor through V2X communication technology. The task timer counts the time spent on the computing task according to the uploading time of each task. The task monitor monitors the uploading of each computing task. When the uploading and monitoring of the computing task are completed, the computing task is packaged, and then the sequence of the task queue is generated and transmitted to the computing power scheduling module for matching calculation of the computing power node. On the other hand, the computing power sharing node module also calculates the computing task queue in the computing power node, and the task monitor also monitors the completion of each computing task. The task queue unit and the task monitor unit upload key information to the computing processing unit to calculate the properties of each computing power node, and finally generate the key information of the computing power node and transmit it to the computing power scheduling module. When the computing power scheduling module receives the key information of the computing task and the key information of the computing power node at the current time, it can use the genetic algorithm to perform the matching and scheduling process of the computing task and the computing power node, and finally generate a mapping diagram of the specific computing task issued to the computing power node.

[0048] From the above, the specific idea of computing power sharing scheduling is to only keep the position of the vehicle, the local computing power resource of the vehicle, and the computing power monitor. When a task is generated, the key information such as the size and generation time of the task is transmitted to the computing power scheduling module. The computing power scheduling module regards all the tasks as a chromosome, and each task is regarded as a gene in a single chromosome. All the chromosomes constitute a feasible solution set. In the iteration process, the result generated by each iteration is fed back to the computing power scheduling module. Based on this, the computing power scheduling module updates the unloading node of each task and updates the position of the next iteration of the task. When the number of iterations is reached or the result of the iteration tends to the limited allowed range, the computing power scheduling module returns the optimal result of the algorithm to the computing power sharing node module, and the computing power sharing node performs the unloading work of the task based on this.

[0049] Figure 3 An embodiment flowchart of the intelligent networked vehicle computing power sharing platform scheduling method, the method comprising the following steps:

[0050] Step 1, first, the shared computing power information and the computing task information need to be initialized.

[0051] Step 2, then the CPU computing power, bandwidth and maximum number of simultaneous processing tasks of the vehicle local computing node in the computing power sharing scheduling area are obtained, and the uploaded computing tasks of the target area and the shared computing power node information are counted.

[0052] Step 3, when the computing power scheduling module receives the computing task upload module uploaded to the computing task and the computing power sharing node information, the genetic algorithm is used to map the computing power scheduling problem into a mathematical model, and a "chromosome" containing the computing power scheduling problem is constructed. Each gene on the chromosome is an element of a feasible solution. The feasible solution is which computing power node each computing task corresponds to, so that each computing task can be matched to the corresponding computing power node.

[0053] Step 4, the initial task parameters, including the shared computing power information and the computing task information, and the chromosome constructed by the feasible solution are brought into the optimal solution objective function to obtain the initial computing task unloading output result.

[0054] Step 5, the initial output result is input, and the genetic algorithm is introduced to calculate the optimization result.

[0055] Suppose there are computing power tasks, and the computing power scheduling system needs to allocate shared computing power nodes to process. The task length , the maximum delay time allowed for computing tasks , the processing speed of each computing power node , and the bandwidth of the computing power node Given that the task allocation method is established, the total processing time of all tasks is minimized.

[0056] Create a task length matrix, representing the task length of all tasks using the matrix `tasks`: ,So In This indicates the task number, while Indicates task The task length; Establish a maximum task latency matrix, representing the maximum latency of all tasks using the matrix `times`: ,So In This indicates the task number, while Indicates task The maximum latency time. Establish a computing node processing speed matrix, representing the processing speed of all shared computing nodes using the matrix `nodes`: ,So In This indicates the number of the shared computing power node, while Represents a node The processing speed; establish a computing power node bandwidth matrix, and represent the bandwidth of all shared computing power nodes using the matrix bend: ,So In This indicates the number of the shared computing power node, while Represents a node The bandwidth.

[0057] We establish a task processing time matrix. Once the task matrix (tasks) and node matrix (nodes) are determined, the processing time for all tasks assigned to all nodes can be determined. We represent this using the matrix timeMatrix, which is a two-dimensional array.

[0058] Indicates the task Assigned to node The processing time is calculated using the following formula: .

[0059] A chromosome matrix is ​​established. Each evolution generates N chromosomes, each representing a feasible solution to the current problem. A feasible solution consists of multiple elements, each called a gene of the chromosome. A chromosome matrix is ​​used to record the feasible solutions in each evolutionary process of the algorithm. The structure of a chromosome is as follows: A chromosome is represented by a one-dimensional array, where the index of the array represents the task number, and the value represents the node number. Therefore... The meaning is: the task is assigned to the node .

[0060] The fitness matrix is established, and the fitness of each chromosome is evaluated by the fitness function, and then the fitness matrix is established, and the chromosomes with high fitness are retained and the chromosomes with low fitness are eliminated. In the process of implementing the computing power scheduling algorithm, a fitness matrix needs to be established to record the fitness of the current N chromosomes, as shown below: . The subscript of the array represents the number of chromosomes, and represents the fitness of the chromosome with the number . In the computing power scheduling process, the total execution time of N tasks is taken as the standard for fitness evaluation. If the total execution time is long, the fitness is poor; if the total execution time is short, the fitness is high. The calculation formula of the fitness of each chromosome is: , the total calculation delay of each feasible solution , and the total transmission delay of each feasible solution ;

[0061] A selection probability matrix is established. In each evolution process, the probability of each chromosome being selected in the next evolution needs to be calculated according to the fitness matrix. The matrix is as follows: The subscript of the matrix represents the number of chromosomes, and the value in the matrix represents the selection probability of the chromosome. The calculation formula is as follows: .

[0062] Step 6, through the fitness calculation of the chromosomes of the genetic algorithm, the chromosomes are crossed, mutated and copied.

[0063] Step 7, dynamic selection of the number of iterations of the genetic algorithm. According to the number of computing power sharing nodes and the number of computing tasks, it is determined whether to limit the number of evolution times or to limit the allowed range,

[0064] When the number of computing power nodes is less than the set threshold and the number of computing tasks exceeds the set threshold, the number of iterations is limited to reduce the calculation delay, and in other cases, the allowed precision range is limited for iteration calculation,

[0065] Step 8, judge whether the allowed condition of evolution is reached, if not, return to step 6;

[0066] Step 9, if yes, output the optimal calculation result, and get the unloading vehicle serial number based on the optimal calculation result, end the iteration calculation, and complete the computing power scheduling task.

[0067] ​It should be noted that the term "comprising" and its variations as used in this application are open-ended, meaning "including but not limited to". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; and the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below. The modifications of "a" and "a plurality" mentioned in this application are illustrative and not restrictive, and those skilled in the art should understand that, unless explicitly indicated in the context, they should be understood as "one or more".

[0068] The names of the messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0069] It should be understood that those skilled in the art can use the various forms of processes shown above to rearrange, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this application does not impose any limitations on this.

[0070] The above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and are not intended to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for sharing and scheduling computing resources in intelligent connected vehicles, characterized in that, include: Step 1: Initialize shared computing power information and computing task information; Step 2: Obtain the CPU computing power, bandwidth, and maximum number of tasks that can be processed simultaneously for the local computing node of the vehicle in the computing power sharing and scheduling area, as well as the computing tasks uploaded by the vehicles in the computing power sharing and scheduling area, as the initial task parameters. Step 3: Apply a genetic algorithm to construct a chromosome containing the computational scheduling problem, where each gene on the chromosome represents an element of a feasible solution, using a set... Encoding chromosomes, where This is an integer variable representing the selected unloading vehicle sequence number. Indicates the number of computational tasks; Step 4: Substitute the initial task parameters and the constructed chromosome into the objective function aimed at minimizing the total task processing latency to obtain the initial computation task unloading output. Step 5: Input the initial computation task unloading output results and introduce the genetic algorithm to calculate the optimization results; Step 6: Calculate chromosome fitness using a genetic algorithm to determine the chromosomes. Perform crossover, mutation, and replication; Step 7, Dynamically select the number of iterations for the genetic algorithm: Based on the number of computing power sharing nodes and the number of computing tasks, dynamically select whether to limit the number of iterations or limit the allowed precision range for the number of iterations; when the number of computing power nodes is less than the set threshold and the number of computing tasks exceeds the set threshold, limit the number of iterations to reduce computing latency; when the number of computing power nodes is greater than the set threshold and the number of computing tasks does not exceed the set threshold, limit the allowed precision range for the number of iterations. Step 8: Determine if the conditions for evolution have been met; if not, return to step 6. Step 9: If yes, output the optimal calculation result, obtain the unloading vehicle sequence number based on the optimal calculation result, end the iterative calculation, and complete the computing power scheduling task.

2. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 1, characterized in that, The computing power sharing scheduling area is a preset range around the computing task, which is determined by the magnitude of the transmission delay of the computing task. The total delay target consists of two parts: one part is the delay of the computing power sharing vehicle calculating and issuing the task, and the other part is the delay of the computing task being transmitted to the shared computing power node. The preset range is determined by the magnitude of the transmission delay. When the transmission delay of the computing power node exceeds the set threshold, this computing power node will no longer be selected by the computing power sharing cloud platform to unload computing tasks.

3. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 1 or 2, characterized in that, The genetic algorithm is as follows: In the initial stage of the algorithm, a set of feasible solutions for computational scheduling is randomly generated using the roulette wheel method, which is the first generation of chromosomes; then, the fitness function is used to calculate the fitness of each chromosome, and the probability of each chromosome being selected in the next evolution is calculated based on the fitness; then the genetic algorithm evolution process begins: through crossover, NM chromosomes are generated, and then mutation operations are performed on the NM chromosomes generated after crossover, and then M chromosomes are generated by replication, until N chromosomes are generated. Then, the fitness and the probability of being selected in the next evolution are calculated for each of the N chromosomes, until the limited number of evolutions is completed or the limited allowable range is reached.

4. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 3, characterized in that, The genetic algorithm includes establishing a task length matrix, a computing node processing speed matrix, a computing node bandwidth matrix, a task processing time matrix, a chromosome matrix, a fitness matrix, and a selection probability matrix.

5. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Create a task length matrix to represent the task length of all computing power tasks using the matrix `tasks`: , Indicates task Task length, Indicates the task number; Establish a matrix representing the maximum latency of all tasks, using the matrix `times`: , Indicates task The maximum delay time.

6. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establish a computing power node processing speed matrix: Represent the processing speed of all shared computing power nodes using a matrix `nodes`. , Indicates the number of the shared computing power node. Represents a node Processing speed.

7. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establish a computing power node bandwidth matrix: Represent the bandwidth of all shared computing power nodes using the matrix bend. , Indicates the number of the shared computing power node. Represents a node The bandwidth.

8. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establish a task processing time matrix: The task processing time of all tasks assigned to all nodes is represented by the matrix timeMatrix. , Indicates the task Assigned to node The processing time is calculated using the following formula: .

9. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establishing a chromosome matrix: The structure of a single chromosome is as follows: A chromosome is represented by a one-dimensional array, where the array index represents the task number and the array value represents the node number. Indicates the task Assigned to node .

10. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establish the fitness matrix: , The array index represents the chromosome number. This indicates that the number is The fitness of chromosomes; the formula for calculating the fitness of each chromosome: ,in It is the total computational delay for each feasible solution. It is the total transmission delay for each feasible solution.

11. The intelligent connected vehicle computing resource sharing and scheduling method according to claim 4, characterized in that, Establish the selection probability matrix: The subscripts in the matrix represent chromosome numbers, and the values ​​in the matrix represent the selection probability corresponding to that chromosome. The calculation formula is as follows: , Indicates the number is The fitness of chromosomes.

12. An intelligent connected vehicle computing resource sharing and scheduling system implementing the method of any one of claims 1-11, characterized in that, The system includes: a base station for acquiring computing tasks or shared computing power, a cloud server for the base station, a remote intelligent connected vehicle computing power node, and a vehicle for uploading computing tasks; The base station and its cloud server serve as a cloud platform for sharing computing power for intelligent connected vehicles. Intelligent connected vehicles within the signal range of the base station all function as parallel computing nodes, and computing power is scheduled through the cloud platform. The computing power sharing cloud platform includes a computing task upload module, a computing power sharing node module, and a computing power scheduling module. The computing task upload module includes a sensor raw data acquisition unit, a task timer, a task monitor, a computing processing unit, and a task queue generation unit. The computing power scheduling module includes an algorithm unit and a task-to-computing-node mapping unit; The computing power sharing node module includes a general task queue unit, a task monitor, a computing processing unit, and a computing power node information generation unit.

13. The intelligent connected vehicle computing resource sharing and scheduling system according to claim 12, characterized in that: In the computing task upload module, the raw data acquisition unit receives raw data collected by sensors on the intelligent connected vehicle via V2X communication; the task timer times the computing task for each collected raw data to prevent exceeding the maximum delay time for each computing task. The task monitor monitors each uploaded computing task to ensure that no computing tasks are missed or data errors occur; the computing processing unit classifies and packages the data in the computing tasks, and adds the attributes and requirements of each computing task. The task queue generation unit sorts each task according to its attributes, size, and time order, preparing for the subsequent computing power scheduling module to generate a task-to-computing-node mapping diagram.

14. The intelligent connected vehicle computing resource sharing and scheduling system according to claim 12, characterized in that: In the computing power scheduling module, the algorithm unit uses a genetic algorithm to calculate the computing power scheduling strategy. That is, based on the task queue information generated by the computing task module and the computing power node information provided by the computing power sharing node, the genetic algorithm performs scheduling matching work between tasks and computing power nodes according to the attributes and requirements of each task. The task and computing power node mapping unit matches computing tasks with computing power nodes, and determines that each computing task is specifically distributed to one or more computing power nodes in preparation for the subsequent distribution of computing power tasks.

15. The intelligent connected vehicle computing resource sharing and scheduling system according to claim 12, characterized in that: In the computing power sharing node module, the total task queue unit counts the computing task queues that have been distributed to each computing node; the task monitor supervises the computing status of each computing task in each computing node; and the computing processing unit counts the computing power usage of each computing node and the key information of each computing node, so as to collect and calculate information data for the subsequent formation of the attributes of each computing node. The computing node information generation unit generates real-time updated computing node attribute information for each computing node based on the key computing information provided by the computing power processing unit, providing key information for the computing power scheduling module to allocate tasks to each computing node.

Images