A network slice task clustering and resource allocation method and system in a vehicle networking scene

Abstract

The application discloses a network slice task clustering and resource allocation method and system in a vehicle networking scene, comprising: acquiring characteristic values of current time service demands and historical service demands of all vehicle users in a target area; clustering according to the characteristic values of the historical service demands of all vehicle users, dividing the vehicle users into three clustering clusters, and mapping service demands of vehicle users in the same clustering cluster to the corresponding same network slice to obtain vehicle users associated with each network slice; wherein the network slice comprises three network slices of URLLC, eMBB and MMTC; a resource allocation optimization model is solved using an adaptive inertia weighted particle swarm optimization algorithm to obtain a resource allocation scheme; wherein the resource allocation optimization model aims to maximize network utility; and wireless resource allocation of each vehicle user at the current time is performed according to the resource allocation scheme.

Description

Technical Field

[0001] This invention belongs to the field of 5G communication technology, specifically to the field of 5G wireless access network slicing, and more specifically to a method and system for network slicing task clustering and resource allocation in a vehicle-to-everything (V2X) scenario. Background Technology

[0002] With the development of artificial intelligence, vehicle intelligence has become a new direction for the development of the automotive and intelligent industries. Meanwhile, the Internet of Vehicles (IoV) is a crucial component of future intelligent transportation. Vehicles can use IoV technology to communicate with other road users and infrastructure. Network slicing, as a key technology of 5G networks, can be tailored to the increasing number of IoV service types, thereby providing better services.

[0003] In recent years, the 5G ecosystem has primarily targeted various vertical industry users, especially the automotive industry. With the explosive growth of mobile traffic, the automotive vertical market is undergoing rapid technological transformation. For example, the emergence of more and more connected and autonomous vehicles makes the realization of a V2X communication ecosystem readily achievable. However, this technological transformation has also brought many challenges to researchers in the automotive and communications industries, such as how to accurately provide low-latency, high-reliability information transmission for various types of V2X services. Network slicing technology is a key and feasible technology for addressing this challenge. It divides the existing physical network into multiple logically independent networks and provides differentiated services for various services, flexibly allocating resources according to the different QoS requirements of different types of services. As an effective resource allocation method, when facing network needs with diverse vertical industries such as V2X, it can customize dedicated network environments and provide better services. Currently, research on applying 5G network slicing technology for service matching and resource allocation in V2X scenarios is still in the development stage.

[0004] The In-Vehicle Information Services Industry Application Alliance (TIAA) categorizes connected vehicle services into three main application scenarios: information services, traffic efficiency, and traffic safety. Traffic safety includes 33 application scenarios such as forward collision warning, blind spot warning, and forward accident alert; traffic efficiency includes 16 application scenarios such as speed limit reminders, adaptive cruise control, and speed reduction sign reminders; and information services include 23 application scenarios such as media downloads, personal data synchronization, and vehicle and RSU data calibration. The TIAA has also defined the related requirements for each scenario. Given such a diverse range of connected vehicle services, a major challenge is how to classify V2X communication services according to service requirements and map them to corresponding video slices, while ensuring that the service requirements of each slice are guaranteed. Summary of the Invention

[0005] K-means algorithm, also known as k-means algorithm, is a commonly used clustering algorithm. However, the original K-means algorithm has two drawbacks: First, the original K-means algorithm randomly selects K points in the data set as cluster centers initially. The selection of initial cluster centers will affect the clustering results, so the finally obtained cluster division is related to the randomly selected "cluster centers", that is, multiple division results may be obtained. This is because the K-means algorithm converges to a local minimum instead of a global minimum. Second, in each iteration, it is necessary to calculate the distances from all sample points to all centroids, which is time-consuming. K-means++ is designed to address the problem of initial cluster center selection in the traditional algorithm. Different from the traditional K-means algorithm, this optimized algorithm only randomly selects one initial cluster center, then calculates the distance between each sample and its current nearest cluster center. The larger the distance, the higher the probability that the sample is selected as the next cluster center. Finally, the roulette wheel method is used to select the next cluster center. Elkan K-means algorithm is designed to solve the distance calculation problem of the traditional algorithm. Elkan K-means utilizes the triangle properties that the sum of two sides of a triangle is greater than or equal to the third side, and the difference between two sides is less than the third side, to reduce the number of distance calculations. For a sample point x and two centroids A and B, the distance D(A,B) between the two centroids is calculated in advance. If 2D(x,A) < D(A,B), it can be determined that D(x,A) < D(x,B), so there is no need to additionally calculate D(x,B).

[0006] Therefore, the research objective of the present invention is to perform cluster analysis based on K-means++ and Elkan K-means algorithms according to the similarity of service requirements of Internet of Vehicles services, create the same number of network slices as the number of clusters, and map the Internet of Vehicles services of the same cluster to the corresponding slices, so that services in the same cluster can serve users in the same slice. Aiming at the management and allocation of wireless resources between slices, the present invention uses a proportional fair algorithm. After obtaining the number of services associated with each slice based on the results of K-means++ and Elkan K-means algorithms, with the objective of maximizing network utility, an Adaptive Inertia Weight Particle Swarm Optimization (AIW-PSO) algorithm is used to perform proportional allocation of wireless resources.

[0007] In order to solve the above problems, the present invention provides a method and system for network slice task clustering and resource allocation in Internet of Vehicles scenarios.

[0008] Technical solution: To solve the above technical problems, the technical solution adopted by the present invention is:

[0009] In a first aspect, the present invention provides a method for network slice task clustering and resource allocation in Internet of Vehicles scenarios, comprising:

[0010] Obtain the feature values ​​of the current and historical business needs of all vehicle users within the target area;

[0011] Based on the characteristic values ​​of the historical service requirements of all vehicle users, the vehicle users are clustered into three clusters, and the service requirements of vehicle users in the same cluster are mapped to the corresponding network slice to obtain the vehicle users associated with each network slice; wherein the network slices include three types of network slices: URLLC, eMBB, and MMTC.

[0012] Based on the current service needs of all vehicle users and the vehicle users associated with each network slice, an adaptive inertial weighted particle swarm optimization algorithm is used to solve the resource allocation optimization model to obtain a resource allocation scheme; wherein the resource allocation optimization model aims to maximize network utility.

[0013] The wireless resources for each vehicle user are allocated according to the resource allocation scheme at the current moment.

[0014] Obtain the feature values ​​of the current and historical business needs of all vehicle users within the target area;

[0015] Based on the characteristic values ​​of the historical service requirements of all vehicle users, the vehicle users are clustered into three clusters, and the service requirements of vehicle users in the same cluster are mapped to the corresponding network slice to obtain the vehicle users associated with each network slice; wherein the network slices include three types of network slices: URLLC, eMBB, and MMTC.

[0016] Based on the current service needs of all vehicle users and the vehicle users associated with each network slice, an adaptive inertial weighted particle swarm optimization algorithm is used to solve the resource allocation optimization model to obtain a resource allocation scheme; wherein the resource allocation optimization model aims to maximize network utility.

[0017] The wireless resources for each vehicle user are allocated according to the resource allocation scheme at the current moment.

[0018] In some embodiments, the characteristics of a vehicle user's business requirements are data rate and latency.

[0019] In some embodiments, K-means++ and Elkan K-means clustering algorithms are used to cluster based on the feature values ​​of the historical business needs of all vehicle users, specifically including the following steps:

[0020] S21. Input dataset X, number of clusters k; wherein dataset X includes feature values ​​of historical business needs of all vehicle users;

[0021] S22. Select any data in the set and set it as the cluster center β1;

[0022] S23. Calculate the distance d(β) from other data points to the cluster center, and solve for the probability P(β);

[0023]

[0024] S24. Select all k cluster centers by repeating step S23;

[0025] S25. Use the triangle relation to select the cluster center closest to the data point, and assign the data point to the cluster closest to the cluster center to obtain k clusters {S1,S2,...,Sk};

[0026] S26. Recalculate the cluster center βk;

[0027] xi represents the i-th sample point in the dataset;

[0028] S27. Repeat steps S25 and S26 until the cluster centers no longer change, resulting in k clusters {S1,S2,...,Sk}.

[0029] In some embodiments, the data rate of the vehicle user us connected to the road test unit r for:

[0030]

[0031] in, This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates that vehicle user us communicates with road test unit r and connects to network slice s; the road test unit set R = {1,2,...,r} and the network slice set S = {1,2,...,s}, and SINR represents the signal-to-noise ratio of vehicle user us communicating with road test unit r.

[0032] In some embodiments, the signal-to-noise ratio (SINR) for communication between the vehicle user us and the roadside unit r is:

[0033]

[0034] in, The transmit power of the road test unit r, It is path loss. Indicates channel gain; δ 2 This represents noise power.

[0035] In some embodiments, the resource allocation optimization model includes an objective function and constraints;

[0036] The objective function is:

[0037]

[0038] The constraints are:

[0039]

[0040]

[0041]

[0042]

[0043] in, This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates vehicle user u s It communicates with the drive test unit r and connects to the network slice s; For road test unit r vehicle user u s Data rate; Road test unit set R = {1,2,...,r}, Vehicle user set U S ={1,2,...,u s} and the network slice set S = {1, 2, ..., s};

[0044] This indicates the vehicle user u connected to the road test unit r. s The amount of resources allocated, a s W represents the proportion of resources allocated by the base station to network slice s; total Indicates the total amount of resources; This indicates the vehicle user u connected to the road test unit r. s The actual bandwidth obtained, SINR represents the bandwidth of vehicle user u. s The signal-to-noise ratio (SNR) of communication with the road test unit r, V min This represents the minimum data rate required to meet the needs of connected vehicle users; q u Indicates the size of the transmitted data packet, d max This indicates the maximum tolerable latency for vehicle users.

[0045] In some embodiments, an adaptive inertial weighted particle swarm optimization algorithm is used to solve the resource allocation optimization model to obtain a resource allocation scheme, including:

[0046] Define a constraint function to select particles based on whether the constraint conditions are met, where the constraint function constraint(i) is:

[0047]

[0048] Among them, g i (x) represents all inequality constraints in the resource allocation optimization model;

[0049] Adaptive inertial weights are set to improve performance, wherein the adaptive inertial weights w(k) are defined as:

[0050]

[0051] Among them, w start ,w end The values ​​represent the maximum and minimum values ​​of the inertia weight, k represents the current iteration number, and T represents the maximum and minimum values ​​of the inertia weight. max This indicates the maximum number of iterations.

[0052] In some embodiments, the resource allocation scheme is obtained by solving the resource allocation optimization model using an adaptive inertial weighted particle swarm optimization algorithm, and further includes:

[0053] The resource allocation optimization model is rewritten in an unconstrained form:

[0054]

[0055]

[0056]

[0057]

[0058]

[0059]

[0060] Where λ is the penalty factor; This represents the optimal objective function after processing using the penalty function method. The term represents the penalty function, which consists of all the inequality constraints; g1, g2, g3, and g4 are the inequality constraints in the resource allocation optimization model.

[0061] Secondly, the present invention provides a network slicing task clustering and resource allocation system in the context of vehicle networking, including a processor and a storage medium;

[0062] The storage medium is used to store instructions;

[0063] The processor is configured to operate according to the instructions to perform the steps of the method according to the first aspect.

[0064] Thirdly, the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0065] Beneficial Effects: This invention provides a method for network slicing task clustering and resource allocation in a vehicle-to-everything (V2X) scenario. First, for different types of V2X services, based on K-means++ and Elkan K-means algorithms, clustering analysis is performed on different types of V2X services according to demand similarity, and these services are mapped to corresponding slices. Second, in terms of resource allocation, an Adaptive Inertial Weighted Particle Swarm Optimization (AIW-PSO) algorithm with the goal of maximizing network utility is adopted. Simulation results show that the improved algorithm increases running time and the speed of approaching the optimal value, and also improves network utilization compared to traditional resource allocation methods. Attached Figure Description

[0066] Figure 1 This is a schematic diagram of a vehicle-to-everything (V2X) system based on network slicing according to an embodiment of the present invention;

[0067] Figure 2 This is a schematic diagram of a method flow according to an embodiment of the present invention;

[0068] Figure 3 This is a schematic diagram of vehicle user clustering results according to an embodiment of the present invention;

[0069] Figure 4 This is a schematic diagram comparing network utility according to an embodiment of the present invention. Detailed Implementation

[0070] The present invention will be further described below with reference to the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and should not be used to limit the scope of protection of the present invention.

[0071] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0072] In the description of this invention, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0073] In one possible application scenario, this application considers a road area covered by a single base station. The road area is deployed using several roadside units (RSUs), where R = {1, 2, ..., r}, and vehicle user U... S ={1,2,...,u s The network slices are composed of S = {1, 2, ..., s}. Vehicle users within the area connect to the Road Testing Unit (RSU) according to their service needs, such as... Figure 1 As shown, vehicle users, roadside units (RSUs), network slices, and base stations constitute a vehicle-to-everything (V2X) system based on network slices.

[0074] Example 1

[0075] like Figure 2 As shown, a method for network slicing task clustering and resource allocation in a vehicle-to-everything (V2X) scenario includes:

[0076] S1. Obtain the feature values ​​of the current and historical business needs of all vehicle users within the target area;

[0077] In this embodiment, the vehicle networking system specifically includes:

[0078] Vehicle users communicate with the road test unit and connect via the corresponding network slice, using... It means if and only if The time indicates vehicle user u s It communicates with the road test unit r and connects to the network slice s. Vehicle users communicate with the road test unit RSU that has the highest signal-to-noise ratio (SINR). SINR is abbreviated as:

[0079]

[0080] in The transmit power of the Roadside Unit (RSU) It is path loss. Indicates channel gain. Noise power δ 2 The path loss is set to -174dBm / Hz and is 128.1 + 37.6lg(d), where d is in km and path loss is in dB.

[0081] Based on the importance of the service, three types of network slices—URLLC, eMBB, and MMTC—were configured, and resources were allocated according to the number of current vehicle users. The amount of resources allocated to a network slice is related to the number of vehicle users connected to that network slice. s | represents the number of vehicle users connected to network slice s, and U represents the total number of vehicle users. The amount of resources obtained by network slice s is Therefore, we can obtain the road test unit r, the vehicle user u. s transmission rate for:

[0082]

[0083] in, This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates vehicle user u s It communicates with the road test unit r and connects to the network slice s; the road test unit set R = {1,2,...,r} and the network slice set S = {1,2,...,s}, SINR represents the vehicle user u. s The signal-to-noise ratio during communication with the road test unit r.

[0084] S2. Based on the feature values ​​of the historical service requirements of all vehicle users, cluster the vehicle users into three clusters, and map the service requirements of vehicle users in the same cluster to the corresponding network slice to obtain the vehicle users associated with each network slice; wherein the network slices include three types of network slices: URLLC, eMBB, and MMTC.

[0085] The Industrial Application Alliance for Vehicle Information Services (TIAA) categorizes connected vehicle services into three main application scenarios: information services, traffic efficiency, and traffic safety.

[0086] In some embodiments, K-means++ and Elkan K-means clustering algorithms are used to cluster based on the feature values ​​of the historical business needs of all vehicle users, specifically including the following steps:

[0087] S21. Input dataset X, number of clusters k; wherein dataset X includes feature values ​​of historical business needs of all vehicle users;

[0088] S22. Select any data in the set and set it as the cluster center β1;

[0089] S23. Calculate the distance d(β) from other data points to the cluster center, and solve for the probability P(β);

[0090]

[0091] S24. Select all k cluster centers by repeating step S23;

[0092] S25. Using the triangle relation, select the cluster center closest to the data point, and assign the data point to the cluster closest to the cluster center, resulting in k clusters {S1, S2, ..., S...}. k};

[0093] S26. Recalculate cluster center β k ;

[0094] x i This represents the i-th sample point in the dataset;

[0095] S27. Repeat steps S25 and S26 until the cluster centers no longer change, resulting in k clusters {S1, S2, ..., S...}. k}

[0096] In this embodiment, data rate and latency are extracted as feature values ​​from the vehicle network service requirements. All service requirements at a certain moment are clustered, and vehicle network services in the same cluster are mapped to the corresponding network slices to ensure that vehicle network services in the same cluster serve vehicle users in the same network slice.

[0097] like Figure 3 The diagram shown is a clustering result obtained by clustering based on the feature values ​​of the historical business needs of all vehicle users in this embodiment. Vehicle users are clustered into three major categories. The business needs of vehicle users in the same cluster are mapped to the corresponding network slices to obtain the vehicle users associated with each network slice.

[0098] S3. Based on the current service needs of all vehicle users and the vehicle users associated with each network slice, the resource allocation optimization model is solved using the Adaptive Inertial Weighted Particle Swarm Optimization (AIW-PSO) algorithm to obtain the resource allocation scheme; wherein the resource allocation optimization model aims to maximize network utility.

[0099] In some embodiments, this step specifically includes:

[0100] First, wireless resources will be allocated with the goal of maximizing network utility. In order to meet the minimum service requirements of vehicle users as much as possible, the indicators in the previous task clustering process, namely the rate requirements and latency requirements of vehicle networking services, will be used as constraints to ensure that while maximizing network utility, the number of vehicle users that meet the QoS requirements of the entire system is increased as much as possible, and a resource allocation optimization model will be constructed.

[0101] The resource allocation optimization model includes an objective function and constraints;

[0102] The objective function is as follows:

[0103]

[0104] in This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates vehicle user u s It communicates with the drive test unit r and connects to the network slice s; Road test unit r vehicle user u s Data rate; Road test unit set R = {1,2,...,r}, Vehicle user set U S ={1,2,...,u s} and the network slice set S = {1, 2, ..., s}.

[0105] The constraints are as follows:

[0106] (1) The radio resources allocated to a vehicle user cannot exceed the total resources contained in the connected network slice:

[0107]

[0108] in, This indicates the vehicle user u connected to the road test unit r. s The amount of resources allocated, W total Indicates the total amount of resources.

[0109] (2) The vehicle service data rate must at least meet the minimum service requirement rate:

[0110]

[0111] in, This indicates the vehicle user u connected to the road test unit r. s Actual bandwidth obtained, V min This indicates the minimum data rate required to meet the needs of vehicle users.

[0112] (3) This indicates that the vehicle service delay at least meets the maximum tolerable delay:

[0113]

[0114] Where, q u Indicates the size of the transmitted data packet, d max This indicates the maximum tolerable latency for vehicle users.

[0115] (4) A vehicle user can only communicate with one network slice at a time through one Roadside Unit (RSU) unit:

[0116]

[0117] One approach to solving the aforementioned optimization problem is to use the Particle Swarm Optimization (PSO) algorithm. The traditional PSO implementation mainly involves: 1. Initializing information for all particles (particle position, particle velocity). 2. Setting relevant parameters (maximum number of iterations, Imax, etc.). 3. Determining the current fitness value of each particle; comparing the fitness values ​​of each particle across all its experiences to determine its optimal position. 4. Comparing the particle's fitness value with the global optimal fitness value across all its experiences to determine the global optimal position of the swarm. 5. Updating the particle's position and velocity according to relevant criteria. 6. Iterating; if the iteration termination condition is met, the optimal solution is obtained and the iteration terminates; otherwise, iterating and updating the particle's position and velocity continues.

[0118] However, the basic particle swarm optimization algorithm does not have particle selection. Some bad particles will reduce the convergence speed of the algorithm and affect the overall performance of the algorithm. Furthermore, the inertia factor w has a very direct impact on the optimization performance of the algorithm. A smaller w value is beneficial to the convergence of the algorithm, while a larger w value is beneficial to escaping local optima.

[0119] Therefore, this application improves the algorithm's convergence speed by setting a particle constraint mechanism for particle selection. A linearly decreasing inertia weight is introduced to dynamically change the inertia weight value during iteration, thereby increasing the global iterative optimization capability of the initial particle swarm and strengthening the final particle swarm's ability to escape local optima. This application defines a constraint function to select particles based on whether the constraint conditions are met. The constraint function is defined as follows:

[0120]

[0121] In the formula g i (x) represents all inequality constraints in the optimization function. The constraint function contains all the constraints of the optimization problem and consists of all the constraints. The value of the constraint function reflects the distance between the particle and the boundary. The larger the value of the constraint function, the more particles need to be eliminated.

[0122] It is pointed out that reasonable inertia weights can significantly improve algorithm performance, while traditional particle swarm optimization algorithms use uniform inertia weights during the calculation process, ignoring the differences between different generations of particles.

[0123] Therefore, this application adopts a linearly decreasing parameter setting, starting with a larger inertia weight. This is beneficial for improving the global iterative optimization capability of the initial particle swarm. Ending with a smaller inertia weight helps enhance the ability of the final particle swarm to escape local optima. The adaptive inertia weight is defined as follows:

[0124]

[0125] Where w start ,w end The values ​​represent the maximum and minimum inertia weights. Generally, the algorithm performs best when the initial inertia weight is 0.9 and the minimum inertia weight is 0.4. k represents the current iteration number, and T... max This represents the maximum number of iterations. When the weights are larger, the algorithm is better able to perform a good global search; when the weights are smaller, the algorithm has good local search capabilities. Therefore, this adaptive weighting approach ensures that the algorithm does not get stuck in local minima in the early stages and can find the optimal value more quickly in the later stages.

[0126] For the resource allocation optimization model, this application uses the penalty function method to transform the constrained problem into an unconstrained problem, and then uses an improved particle swarm optimization algorithm to solve the unconstrained optimization problem. The penalty function method does not require estimating the obtained solution; it imposes a large penalty on infeasible solutions, with the aim of discarding them. The resource allocation optimization model of this application is rewritten in an unconstrained form:

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133] Here, λ is the penalty factor, and a large value is given. The main idea of ​​the penalty function method is to assign a large function value to infeasible solutions, thereby forcing the iterative particles to approach the feasible region during the solution process, and finally obtaining an optimal solution close to the original objective function after multiple iterations.

[0134] S4. Allocate wireless resources to each vehicle user at the current time according to the resource allocation scheme.

[0135] To further illustrate the algorithm proposed in this patent, simulation verification was conducted.

[0136] Figure 4 The diagram shows a comparison of network utility using the AIW-PSO algorithm (as described in this application) with traditional Round-Robin Scheduling (RRS) and Proportional Fairness (PFA) algorithms. It can be seen that the AIW-PSO algorithm used in this invention achieves higher network utility in vehicular network slice resource allocation. This indicates that in the initial search phase, AIW-PSO improves the algorithm's global search capability through inertial weight adjustment. This allows the algorithm to not only cover a wider range of resource allocation schemes but also effectively identify multiple potential excellent solution regions. This global exploration capability is crucial for addressing the diverse user needs in the vehicular network environment, ensuring that the algorithm can quickly locate promising solution sets within the vast solution space. In the mid-to-late search phase, the AIW-PSO algorithm demonstrates its ability to fine-tune resource allocation. At this stage, the algorithm further explores the previously identified excellent solution regions, refining resource allocation strategies to adapt to the specific needs of different users, thereby improving network utility.

[0137] Simulation results show that the improved algorithm increases the running time and the speed of approaching the optimal value, and also improves the network utilization compared with the traditional resource allocation method.

[0138] Example 2

[0139] Secondly, this embodiment provides a network slicing task clustering and resource allocation system in a vehicle-to-everything (V2X) scenario, including a processor and a storage medium;

[0140] The storage medium is used to store instructions;

[0141] The processor is configured to operate according to the instructions to perform the steps of the method according to Embodiment 1.

[0142] Example 3

[0143] Thirdly, this embodiment provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0144] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0145] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0146] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0147] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0148] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for network slicing task clustering and resource allocation in a vehicle-to-everything (V2X) scenario, characterized in that, include: Obtain the feature values ​​of the current and historical business needs of all vehicle users within the target area; Based on the feature values ​​of the historical business needs of all vehicle users, the vehicle users are clustered into three clusters, and the business needs of vehicle users in the same cluster are mapped to the corresponding network slice to obtain the vehicle users associated with each network slice. The network slices mentioned above include three types: URLLC, eMBB, and MMTC. Based on the current business needs of all vehicle users and the vehicle users associated with each network slice, the resource allocation optimization model is solved using the adaptive inertial weighted particle swarm optimization algorithm to obtain the resource allocation scheme. The resource allocation optimization model described herein aims to maximize network utility. The wireless resources for each vehicle user at the current moment are allocated according to the resource allocation scheme. The resource allocation optimization model includes an objective function and constraints. The objective function is: ； The constraints are: ； ； ； ； in, This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates vehicle user It communicates with the drive test unit r and connects to the network slice s; For road test unit r vehicle users Data rate; Drive test unit set Vehicle User Group and network slice sets ; This indicates the vehicle user connected to the road test unit r. The amount of resources allocated. This indicates that the base station provides network slicing. The proportion of resources allocated; Indicates the total amount of resources; This indicates the vehicle user connected to the road test unit r. Actual bandwidth obtained Indicates vehicle user The signal-to-noise ratio for communication with the road test unit r. This indicates the minimum data rate required to meet the needs of connected vehicle users. Indicates the size of the transmitted data packet. This indicates the maximum tolerable latency for vehicle users.

2. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 1, characterized in that, The key characteristics of vehicle users' business needs are data rate and latency.

3. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 1, characterized in that, The K-means++ and Elkan K-means clustering algorithms are used to cluster all vehicle users based on the feature values ​​of their historical business needs. The specific steps include: S21. Input dataset X, number of clusters k; wherein dataset X includes feature values ​​of historical business needs of all vehicle users; S22. Select any data in the set and set it as the cluster center. ; S23. Calculate the distance from other data points to the cluster center. Solve for the probability ; ； S24. Select all k cluster centers by repeating step S23; S25. Use the triangle relation to select the cluster center closest to the data point, and assign the data point to the cluster closest to the cluster center to obtain k clusters. ; S26. Recalculate cluster centers ; , This represents the i-th sample point in the dataset; S27. Repeat steps S25 and S26 until the cluster centers no longer change, resulting in k clusters. .

4. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 2, characterized in that, The vehicle user connected to the road test unit r data rate for: ； in, This indicates the state in which a vehicle user communicates with a road test unit and is connected via the corresponding network slice, if and only if The time indicates vehicle user Communicates with drive unit r and connects to network slice s; Drive unit set and network slice sets , This indicates the vehicle user connected to the road test unit r. Actual bandwidth obtained Indicates vehicle user The signal-to-noise ratio during communication with the drive unit r.

5. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 4, characterized in that, vehicle users Signal-to-noise ratio for communication with drive test unit r for: ； in, The transmit power of the road test unit r, It is path loss. Indicates channel gain; This represents noise power.

6. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 1, characterized in that, The resource allocation optimization model is solved using an adaptive inertial weighted particle swarm optimization algorithm to obtain resource allocation schemes, including: Define a constraint function to select particles based on whether the constraint conditions are met, wherein the constraint function... for: ； in, This represents all inequality constraints in the resource allocation optimization model; Setting adaptive inertial weights improves performance, wherein the adaptive inertial weights Defined as: ； in, This represents the maximum and minimum values ​​of the inertia weight. Indicates the current iteration number. This indicates the maximum number of iterations.

7. The method for network slicing task clustering and resource allocation in the vehicle-to-everything (V2X) scenario according to claim 1 or 6, characterized in that, The resource allocation scheme is obtained by solving the resource allocation optimization model using the adaptive inertial weighted particle swarm optimization algorithm, and also includes: The resource allocation optimization model is rewritten in an unconstrained form: ； ； ； ； ； ； Where λ is the penalty factor; This represents the optimal objective function after processing using the penalty function method. The penalty function term is composed of all the inequality constraints. These are the constraints of the inequalities in the resource allocation optimization model.

8. A network slicing task clustering and resource allocation system for a vehicle-to-everything (V2X) scenario, characterized in that, Including processor and storage media; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1 to 7.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images