Cloud game scheduling method, system and computer-readable storage medium

Abstract

The present invention discloses a cloud game scheduling method, system, and computer-readable storage medium. The method includes: obtaining at least one original scheduling plan for a cloud game; calculating the adaptation value of each original scheduling plan and clustering all original scheduling plans; generating a pre-scheduling plan based on the adaptation value and clustering results; determining a target scheduling plan based on the pre-scheduling plan, and scheduling the cloud game based on the target scheduling plan. The present invention generates a pre-scheduling plan based on the adaptation value of each original scheduling plan and the clustering results of clustering all original scheduling plans, optimizes the pre-scheduling plan to obtain a target scheduling plan, and schedules the cloud game based on the target scheduling plan, so that the target scheduling plan can reasonably schedule the cloud game, thereby improving the practicality of cloud game scheduling.

Description

[0001] Technology Neighborhood

[0002] The present invention relates to the field of artificial intelligence technology, and in particular to a cloud game scheduling method, system, and computer-readable storage medium. Background Art

[0003] With the development of cloud computing, various cloud games have emerged. Cloud games need to be scheduled through cloud servers. However, the existing cloud game scheduling methods only consider part of the resource requirements of cloud games, which can easily lead to unreasonable cloud game deployment, high cloud server procurement costs, resource waste, and load imbalance resulting in slow server response speed and network congestion. As a result, cloud game scheduling is not practical and seriously affects the user experience.

[0004] Therefore, how to improve the practicality of cloud game scheduling is an urgent problem that needs to be solved. Summary of the Invention

[0005] The main purpose of the present invention is to propose a cloud game scheduling method, system and computer-readable storage medium, aiming to solve the problem of how to improve the practicality of cloud game scheduling.

[0006] To achieve the above objectives, the present invention provides a cloud game scheduling method, which includes the following steps:

[0007] Obtain at least one original scheduling plan for the cloud game;

[0008] Calculate the adaptation value of each original scheduling scheme and cluster all original scheduling schemes;

[0009] generating a pre-scheduling plan according to the adaptation value and the clustering result;

[0010] A target scheduling plan is determined based on the pre-scheduling plan, and cloud gaming scheduling is performed based on the target scheduling plan.

[0011] Optionally, the step of clustering all original scheduling solutions includes:

[0012] Determine the probability value of each original scheduling plan belonging to each cluster according to the adaptation value of each original scheduling plan and the preset number of clusters;

[0013] generating a first membership matrix according to the probability value, all original scheduling schemes and the preset number of clusters;

[0014] Calculating the target cluster center of each cluster according to the first membership matrix and the adaptation value of each original scheduling scheme;

[0015] Calculating a second membership matrix based on the target cluster center;

[0016] Calculating the variance between the first membership matrix and the second membership matrix;

[0017] If the variance is not greater than the variance threshold, clustering all original scheduling solutions according to the first membership matrix and the target cluster center;

[0018] If the variance is greater than the variance threshold, the second membership matrix is ​​used as the first membership matrix, and the step is re-executed: the target cluster center of each cluster is calculated according to the first membership matrix and the adaptation value of each original scheduling scheme.

[0019] Optionally, the step of determining the probability value of all original scheduling solutions belonging to each cluster according to the adaptation value of each original scheduling solution and the preset number of clusters includes:

[0020] Randomly select several preset clustering original scheduling plans as the pre-clustering centers of each cluster;

[0021] According to the adaptation value of each original scheduling plan and the adaptation value of each pre-clustering center, the distance from each original scheduling plan to each cluster is calculated;

[0022] An original scheduling scheme is selected in turn, and the probability value of the selected original scheduling scheme belonging to each cluster is calculated based on the distance from the selected original scheduling scheme to each cluster.

[0023] Optionally, the step of calculating the probability value of the selected original scheduling solution belonging to each cluster according to the distance from the selected original scheduling solution to each cluster includes:

[0024] One cluster is selected in turn, and the sum of the distance from the selected original scheduling solution to the selected cluster and the quotient of the distance from the selected original scheduling solution to each cluster is taken as the probability value that the selected original scheduling solution belongs to the selected cluster.

[0025] Optionally, the step of generating a first membership matrix according to the probability value, all original scheduling schemes and the preset number of clusters includes:

[0026] Determining the number of matrix rows according to the number of all original scheduling solutions, and determining the number of matrix columns according to the preset number of clusters;

[0027] A first membership matrix is ​​generated according to the probability value, the number of matrix rows, and the number of matrix columns.

[0028] Optionally, the step of calculating the target cluster center of each cluster according to the first membership matrix and the adaptation value of each original scheduling scheme includes:

[0029] The probability value of each original scheduling scheme belonging to each cluster in the first membership matrix is ​​obtained, and the target cluster center of each cluster is calculated based on the probability value and the adaptation value of each original scheduling scheme.

[0030] Optionally, the step of determining the target scheduling scheme according to the pre-scheduling scheme includes:

[0031] Randomly perturbing the optimal pre-scheduling scheme among the pre-scheduling schemes to generate a neighborhood pre-scheduling scheme;

[0032] Calculating a neighborhood transformation coefficient according to a preset number of neighborhood transformations, and comparing the neighborhood transformation coefficient with a preset neighborhood transformation probability parameter to obtain a comparison result;

[0033] A target scheduling solution is determined according to the comparison result, the optimal pre-scheduling solution and the neighborhood pre-scheduling solution.

[0034] Optionally, the step of determining a target scheduling solution according to the comparison result, the optimal pre-scheduling solution, and the neighborhood pre-scheduling solution includes:

[0035] If the comparison result is that the neighborhood transformation coefficient is not greater than the preset neighborhood transformation probability parameter, determining a target scheduling scheme according to the optimal pre-scheduling scheme and the neighborhood pre-scheduling scheme;

[0036] If the comparison result is that the neighborhood transformation coefficient is greater than the preset neighborhood transformation probability parameter, a local search is performed on the neighborhood pre-scheduling scheme; if the local search result meets the preset conditions, a target scheduling scheme is determined based on the optimal pre-scheduling scheme and the local pre-scheduling scheme; if the local search result does not meet the preset conditions, the step is re-executed: randomly perturb the optimal pre-scheduling scheme to generate a neighborhood pre-scheduling scheme.

[0037] In addition, to achieve the above-mentioned purpose, the present invention also provides a cloud game scheduling system, which includes: a memory, a processor, and a cloud game scheduling program stored on the memory and runnable on the processor. When the cloud game scheduling program is executed by the processor, the steps of the cloud game scheduling method described above are implemented.

[0038] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a cloud game scheduling program is stored. When the cloud game scheduling program is executed by a processor, the steps of the cloud game scheduling method described above are implemented.

[0039] The cloud game scheduling method proposed in the present invention obtains at least one original scheduling plan for a cloud game; calculates the adaptation value of each original scheduling plan and clusters all the original scheduling plans; generates a pre-scheduling plan based on the adaptation value and clustering results; determines a target scheduling plan based on the pre-scheduling plan, and performs cloud game scheduling based on the target scheduling plan. The present invention generates a pre-scheduling plan based on the adaptation value of each original scheduling plan and the clustering results of clustering all the original scheduling plans, optimizes the pre-scheduling plan to obtain a target scheduling plan, and performs cloud game scheduling based on the target scheduling plan, so that the target scheduling plan can reasonably schedule cloud games, thereby improving the practicality of cloud game scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention;

[0041] Figure 2 This is a flow chart of the first embodiment of the cloud game scheduling method of the present invention;

[0042] Figure 3 This is a flow chart of the second embodiment of the cloud game scheduling method of the present invention;

[0043] Figure 4 2 is a flow chart of the third embodiment of the cloud game scheduling method of the present invention.

[0044] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0045] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0046] like Figure 1 As shown, Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0047] The device in the embodiment of the present invention may be a PC or a server device.

[0048] like Figure 1As shown, the device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0049] Those skilled in the art will understand that Figure 1 The device structure shown in the figure does not constitute a limitation of the device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0050] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a cloud game scheduling program.

[0051] Among them, the operating system is a program that manages and controls portable storage devices and software resources, and supports the operation of network communication modules, user interface modules, cloud game schedulers and other programs or software; the network communication module is used to manage and control the network interface 1002; the user interface module is used to manage and control the user interface 1003.

[0052] exist Figure 1 In the storage device shown, the storage device calls the cloud game scheduling program stored in the memory 1005 through the processor 1001, and executes the operations in each embodiment of the following cloud game scheduling method.

[0053] Based on the above hardware structure, an embodiment of the cloud game scheduling method of the present invention is proposed.

[0054] Reference Figure 2 , Figure 2 This is a flow chart of a first embodiment of a cloud game scheduling method according to the present invention, which includes:

[0055] The cloud game scheduling method of this embodiment can be used in the cloud game scheduling system of a communication service organization. The cloud game scheduling system can run in terminal devices such as PC systems, cloud servers, and smart terminals. For the convenience of description, the cloud game scheduling system is used as an example for explanation.

[0056] Step S10, obtaining at least one original scheduling plan for the cloud game;

[0057] In this embodiment, the cloud game scheduling system obtains cloud games that require scheduling and randomly generates multiple original scheduling plans based on the cloud games. Each original scheduling plan includes a scheduling plan for each cloud game that requires scheduling. The cloud game scheduling system randomly selects at least one of the multiple randomly generated original scheduling plans. It should be noted that cloud gaming, also known as gaming on demand, is an online gaming technology based on cloud computing. Cloud gaming can include, but is not limited to, competitive games, parkour games, shooting games, and board games.

[0058] Furthermore, the cloud game scheduling system obtains the current cloud server information and the cloud game information corresponding to the cloud game to be scheduled. Based on the current cloud server information and the cloud game information, it randomly generates multiple original scheduling plans. The cloud game scheduling system pre-sets the number of original scheduling plans to be obtained. After randomly generating multiple original scheduling plans, the cloud game scheduling system randomly obtains a preset number of original scheduling plans from the randomly generated multiple original scheduling plans. The current cloud server information includes, but is not limited to: the cost of the cloud server, CPU memory capacity, GPU memory capacity, disk capacity, and bandwidth size; the cloud game information includes, but is not limited to: the CPU, GPU, bandwidth, disk resources, and configuration requirements required for the normal operation of the cloud game.

[0059] Step S20, calculating the adaptation value of each original scheduling scheme and clustering all the original scheduling schemes;

[0060] In this embodiment, after obtaining the original scheduling plan, the cloud game scheduling system obtains the preset clustering number and the pre-created scheduling model, calculates the adaptation value of each original scheduling plan based on the scheduling model, and clusters all the original scheduling plans according to the preset clustering number to obtain the clustering results.

[0061] Furthermore, the creation process of the scheduling model is as follows:

[0062] The cloud game scheduling system obtains cloud server information, cloud game information, and the number of cloud games, and performs data preprocessing. Cloud server information includes: cloud server cost, CPU memory capacity, GPU memory capacity, disk capacity, and bandwidth size; cloud game information includes: the CPU, GPU, bandwidth, and disk resources required for the normal operation of the cloud game, as well as the configuration requirements. The specific process of data preprocessing is as follows: 1) Cloud server resource capacity and cost are dimensionless, and cloud server configuration information is simplified to cloud server configuration level, which is labeled 1, 2, 3, ... 2) Cloud servers are sorted from low to high according to cloud server configuration level. 3) The resource size required for the cloud game is dimensionless, and the required configuration information is simplified to match the cloud server configuration level.

[0063] The cloud game scheduling system sets the total number of cloud games N. The resources required by each cloud game are: CPU demand c i 、GPU requirements i , bandwidth requirements bw i , disk requirements disk i , Configuration level required for cloud gaming i , where i = 1, 2, 3…, N; set the number of cloud servers to M, and the server configuration levels to R j , and their corresponding resource capacities are: CPU capacity C j , GPU capacity G j , bandwidth size BW j , disk capacity DISK j , cloud server cost j , where j = 1, 2, 3…, M.

[0064] The cloud game scheduling system obtains constraint information: (1) The resource capacity of cloud servers of the same configuration level is the same, and the cloud game configuration level is not higher than the cloud server configuration level; (2) Each cloud game can only be deployed on one cloud server; (3) The scheduling order of cloud games on the same cloud server does not affect the final scheduling result; (4) The final resource consumption and load balancing rate of each server should be as equal or close as possible.

[0065] The cloud game scheduling system obtains and determines the cloud server cost, penalty cost for not deploying cloud games, cost of deviation from cloud game configuration requirements, remaining cloud server resources, and penalty cost for cloud server load balancing variance based on cloud server information, cloud game information, and constraint information. It then generates a scheduling model based on the cloud server cost, penalty cost for not deploying cloud games, cost of deviation from cloud game configuration requirements, remaining cloud server resources, and penalty cost for cloud server load balancing variance. The scheduling model is as follows:

[0066]

[0067] Where i represents cloud gaming; j represents cloud server; x represents cloud server; ij Indicates whether cloud server j can deploy cloud game i and can deploy x ij If it is 1, x cannot be deployed. ij is 0; r′ i Indicates the actual configuration requirement level of cloud game i; left j represents the sum of remaining resources on server j after M cloud games are deployed; σ 2 represents the variance of the cloud server's load balancing; N1 represents the number of undeployed cloud games; w1, w2, w3, w4, and w5 represent the weighted coefficients of each part respectively; the five parts in the scheduling model are respectively the cloud server cost, the penalty cost of undeployed cloud games, the cloud game configuration demand deviation cost, the cloud server's remaining resources, and the cloud server's load balancing variance penalty cost.

[0068] Furthermore, the process of calculating the adaptation value of each original scheduling scheme is as follows:

[0069] The cloud game scheduling system obtains a set of cloud servers, arranges each cloud server in the set of cloud servers from low to high according to its configuration level, and initializes the cloud game deployment list of each cloud server; the cloud game scheduling system selects a cloud server according to the arrangement order of the cloud servers, and selects a cloud game in turn, and determines whether the configuration requirement level of the selected cloud game meets the configuration level requirements of the selected cloud server; determines whether the remaining CPU capacity of the selected cloud server meets the CPU resources required by the selected cloud game; determines whether the remaining GPU capacity of the selected cloud server meets the GPU resources required by the selected cloud game; determines whether the remaining disk capacity of the selected cloud server meets the disk resources required by the selected cloud game; determines whether the remaining bandwidth size of the selected cloud server meets the requirements of the selected cloud game Required bandwidth resources; if one of the above conditions is not met, then reselect a cloud server according to the arrangement order of the cloud servers, and repeat the judgment of the above conditions; if all the above conditions are met, then pair the selected cloud game with the cloud server that meets all the above conditions, and update the remaining CPU capacity, remaining GPU capacity, remaining disk capacity and remaining bandwidth size of the cloud server, and repeat the steps of selecting cloud servers and cloud games until each cloud game is paired with the cloud server to obtain a matching result; among them, there may be some cloud games in the matching results that cannot be paired with any cloud server in the cloud server set. The cloud game scheduling system will record the number of cloud games that cannot be paired in each original scheduling plan.

[0070] The cloud game scheduling system deploys each cloud game on its paired cloud server based on the matching results, while cloud games that cannot be paired are not deployed on any cloud server, obtaining the deployment results corresponding to each original scheduling plan.

[0071] Based on the deployment results corresponding to each original scheduling plan, the cloud gaming scheduling system calculates the cloud server cost, penalty cost for undeployed cloud games, cost of deviation from cloud gaming configuration requirements, remaining cloud server resources, and penalty cost for cloud server load balancing variance for each original scheduling plan. The system then inputs these cost, penalty cost, deviation from cloud gaming configuration requirements, remaining cloud server resources, and penalty cost for cloud server load balancing variance into a pre-created scheduling model, which then calculates the adaptation value for each original scheduling plan. It should be noted that a smaller adaptation value indicates a more reasonable original scheduling plan and a higher degree of adaptation to the scheduling model.

[0072] Furthermore, the step of clustering all original scheduling solutions includes:

[0073] Step a: determining the probability value of each original scheduling solution belonging to each cluster according to the adaptation value of each original scheduling solution and the preset number of clusters;

[0074] Step b: generating a first membership matrix according to the probability value, all original scheduling schemes and the preset number of clusters;

[0075] Step c, calculating the target cluster center of each cluster based on the first membership matrix and the adaptation value of each original scheduling scheme;

[0076] Step d, calculating a second membership matrix based on the target cluster center;

[0077] Step e: calculating the variance between the first membership matrix and the second membership matrix;

[0078] Step f: If the variance is not greater than the variance threshold, clustering all original scheduling solutions according to the first membership matrix and the target cluster center;

[0079] Step g: If the variance is greater than the variance threshold, the second membership matrix is ​​used as the first membership matrix, and the step is re-executed: the target cluster center of each cluster is calculated based on the first membership matrix and the adaptation value of each original scheduling scheme.

[0080] In steps a to g, the cloud game scheduling system determines the probability value of each original scheduling plan belonging to each cluster based on the adaptation value of each original scheduling plan and the preset number of clusters, and generates a first membership matrix based on the probability value, the preset number of clusters, and all the original scheduling plans. The first membership matrix is ​​a fuzzy classification matrix, and the first membership matrix is ​​a matrix with N rows and m columns: Among them, r ij is the probability value that the i-th original scheduling solution belongs to the j-th cluster, N is the number of original scheduling solutions, and m is the preset number of clusters.

[0081] After calculating the first membership matrix, the cloud game scheduling system calculates the target cluster center of each cluster based on the first membership matrix and the adaptation value of each original scheduling plan, and calculates the second membership matrix based on the target cluster center;

[0082] Specifically, the step of calculating the target cluster center of each cluster according to the first membership matrix and the adaptation value of each original scheduling scheme includes:

[0083] The probability value of each original scheduling scheme belonging to each cluster in the first membership matrix is ​​obtained, and the target cluster center of each cluster is calculated based on the probability value and the adaptation value of each original scheduling scheme.

[0084] The cloud game scheduling system obtains the probability value of each original scheduling plan belonging to each cluster in the first membership matrix, and calculates the target cluster center of each cluster based on the probability value and the adaptation value of each original scheduling plan using the following formula:

[0085]

[0086] Among them, C j is the target cluster center, N is the number of original scheduling solutions, m is the preset number of clusters, x i is the adaptation value corresponding to the i-th original scheduling scheme, u ij is the probability value that the i-th original scheduling solution contained in the first membership matrix belongs to the j-th cluster.

[0087] The cloud game scheduling system calculates the second membership matrix based on the target cluster center using the following formula:

[0088]

[0089] in, is the probability value of the i-th original scheduling scheme contained in the second membership matrix belonging to the j-th cluster, m is the preset number of clusters, x i is the adaptation value corresponding to the i-th original scheduling scheme, C j and Ck is the target cluster center.

[0090] After calculating the second membership matrix, the cloud game scheduling system calculates the variance between the first membership matrix and the second membership matrix. The formula for calculating the variance between the first membership matrix and the second membership matrix is:

[0091]

[0092] Among them, σ 2 is the variance, N is the number of cloud games, N1 is the number of undeployed cloud games, m is the preset number of clusters, is the probability value of the i-th original scheduling scheme contained in the first membership matrix belonging to the j-th cluster, is the probability value that the i-th original scheduling solution contained in the second membership matrix belongs to the j-th cluster.

[0093] After calculating the variance between the first membership matrix and the second membership matrix, the cloud game scheduling system compares the variance with the variance threshold to obtain a comparison result. If the comparison result shows that the variance between the first membership matrix and the second membership matrix is ​​not greater than the variance threshold, the original scheduling plan set is clustered according to the first membership matrix and the target cluster center; if the comparison result shows that the variance between the first membership matrix and the second membership matrix is ​​greater than the variance threshold, the second membership matrix is ​​used as the first membership matrix, and the clustering according to the first membership matrix and the original scheduling plan set is re-executed. The steps and subsequent steps of calculating the target cluster center of each cluster based on the adaptation value of the initial scheduling scheme and updating the number of cycles until the variance between the first membership matrix and the second membership matrix is ​​calculated to be no greater than the variance threshold, and the original scheduling scheme set is clustered according to the first membership matrix and the cluster center; or until the number of cycles reaches the maximum number of cycles, the variance between the first membership matrix and the second membership matrix is ​​not calculated to be no greater than the variance threshold, and the original scheduling scheme set is clustered according to the second membership matrix and the corresponding target cluster center. It can be understood that clustering is to divide all the original scheduling schemes into a preset number of clusters according to the probability value of each original scheduling scheme belonging to the target cluster center of each cluster.

[0094] Step S30, generating a pre-scheduling plan according to the adaptation value and the clustering result;

[0095] In this embodiment, the cloud game scheduling system obtains a preset parameter set and generates a pre-scheduling plan based on the original scheduling plan according to the preset parameter set, the adaptation value, and the clustering results. It should be noted that the preset parameter set is pre-set in the cloud game scheduling system based on cloud server information and cloud game information. The preset parameter set includes: mutation probability, crossover probability, first selection probability, and second selection probability.

[0096] Specifically, the cloud game scheduling system randomly generates a first random value and compares the first random value with the mutation probability in the preset parameter set. If the first random value is less than the mutation probability, a cluster is randomly selected in the clustering result, and a random scheduling plan is randomly generated to replace the target cluster center of the selected cluster. If the first random value is not less than the mutation probability, the target cluster center of the selected cluster is not changed.

[0097] After randomly generating a first random value, the cloud game scheduling system randomly generates a second random value and compares the second random value with the crossover probability in the preset parameter set. If the second random value is less than the crossover probability, a cluster is randomly selected in the clustering result, and a third random value is randomly generated. The third random value is compared with the first selection probability in the preset parameter set. If the third random value is less than the first selection probability, the original scheduling plan corresponding to the target cluster center of the selected cluster is obtained, and the original scheduling plan is reversed to generate a corresponding pre-scheduling plan; if the second random value is not less than the crossover probability, two clusters are randomly selected in the clustering result, and a fourth random value is randomly generated. The fourth random value is compared with the second selection probability in the preset parameter set. If the fourth random value is less than the second selection probability, the original scheduling plans corresponding to the target cluster centers of the two selected clusters are obtained, and the two original scheduling plans are sequentially crossed to generate two corresponding pre-scheduling plans.

[0098] After generating a pre-scheduling plan, the cloud game scheduling system calculates the adaptation value of the pre-scheduling plan through the scheduling model, and compares the adaptation value of the pre-scheduling plan with the adaptation value of the original scheduling plan used to generate the pre-scheduling plan. If the adaptation value of the pre-scheduling plan is less than the adaptation value of the original scheduling plan used to generate the pre-scheduling plan, the pre-scheduling plan is used to replace the original scheduling plan used to generate the pre-scheduling plan.

[0099] After generating the pre-scheduling scheme, the cloud game scheduling system counts the number of pre-scheduling schemes. If the number of pre-scheduling schemes is lower than the number of original scheduling schemes, the above steps of generating the pre-scheduling scheme are re-executed until the number of pre-scheduling schemes is not lower than the number of original scheduling schemes in the original scheduling scheme set, i.e., the pre-scheduling scheme is obtained; or when the cloud game scheduling system cyclically executes the steps of generating the pre-scheduling scheme and reaches a preset maximum number of cycles, the cycle is stopped to obtain the pre-scheduling scheme. It can be understood that since the adaptation value of the pre-scheduling scheme is lower than the adaptation value of the original scheduling scheme that generates the pre-scheduling scheme, the pre-scheduling scheme is used to replace the original scheduling scheme that generates the pre-scheduling scheme. Therefore, the pre-scheduling scheme includes both part of the original scheduling scheme and part of the pre-scheduling scheme.

[0100] It should be noted that the process of inversion is as follows: assuming that the original scheduling plan corresponding to the target cluster center in a randomly selected cluster is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, where 1-10 represent the cloud games included in the original scheduling plan, and the arrangement order is the scheduling order of the cloud games. The cloud game scheduling system randomly generates two random numbers and marks the subsequence between the two random numbers. Assuming that the two random numbers are 4 and 7, the subsequence between the two random numbers is {4, 5, 6, 7}, and then the marked subsequence is re-ordered. New reversal, such as reversing {4, 5, 6, 7} to {7, 6, 5, 4}, and finally reinserting the reversed subsequence into the corresponding position of the original scheduling plan to generate a pre-scheduling plan, such as {1, 2, 3, 7, 6, 5, 4, 8, 9, 10}; the sequential crossover process is as follows: assuming that the original scheduling plans corresponding to the target cluster centers in the two randomly selected clusters are: Plan 1 is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10} and Plan 2 is {7, 1, 8, 6, 2, 3, 10, 4, 9, 5}, where 1-10 represent the cloud games included in the original scheduling plan, and the arrangement order is the scheduling order of the cloud games. The cloud game scheduling system randomly generates two random numbers and marks the subsequence 1 between the two random numbers in plan 1. For example, if the two random numbers are 4 and 7, then the subsequence 1 between the two random numbers marked in plan 1 is {4, 5, 6, 7}. Then, plan 2 is marked according to the sequence value of subsequence 1, and the unmarked sequence value in plan 2 is selected to determine subsequence 2. For example, if subsequence 1 is {4, 5, 6, 7}, then subsequence 2 is {*, 1, 8, *, 2, 3, 10, *, 9, *}; finally, subsequence 1 and subsequence 2 are sequentially crossed to form a pre-scheduling plan, such as {1, 8, 2, 4, 5, 6, 7, 3, 10, 9}; after generating a pre-scheduling plan, the cloud game scheduling system swaps plan 1 and plan 2, that is, plan 1 is {7, 1, 8, 6, 2, 3, 10, 4, 9, 5}, and plan 2 is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, and repeats the above sequential crossing process to generate another pre-scheduling plan.

[0101] Step S40: determine a target scheduling plan based on the pre-scheduling plan, and perform cloud game scheduling according to the target scheduling plan.

[0102] In this embodiment, after generating a pre-scheduling plan, the cloud game scheduling system selects the optimal pre-scheduling plan from all pre-scheduling plans, determines the target scheduling plan based on the optimal pre-scheduling plan, and then schedules all clouds in an orderly manner according to the target scheduling plan.

[0103] Specifically, the step of determining the target scheduling scheme according to the pre-scheduling scheme includes:

[0104] Step h. Randomly perturb the optimal pre-scheduling scheme among the pre-scheduling schemes to generate a neighborhood pre-scheduling scheme;

[0105] In this step, the cloud game scheduling system selects the pre-scheduling scheme corresponding to the minimum adaptation value as the optimal pre-scheduling scheme according to the adaptation values ​​corresponding to all pre-scheduling schemes. After determining the optimal pre-scheduling scheme, the cloud game scheduling system randomly perturbs the optimal pre-scheduling scheme to generate a neighborhood pre-scheduling scheme. Specifically, the random perturbation process is as follows: Assume that the optimal pre-scheduling scheme is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, where 1-10 represent the cloud games included in the original scheduling scheme, and the order of arrangement is cloud games. To determine the scheduling order, the cloud game scheduling system randomly generates two random numbers between 1 and N (N: the total number of cloud games); then, using these two random numbers as the starting and ending points, the original scheduling plan is divided into three subsequences. For example, if the two random numbers are 4 and 7, the optimal pre-scheduling plan is divided into {1, 2, 3}, {4, 5, 6, 7}, and {8, 9, 10}; finally, the three subsequences are randomly recombined to obtain the neighborhood pre-scheduling plan after random perturbation, such as: {4, 5, 6, 7, 1, 2, 3, 8, 9, 10}.

[0106] Step i, calculating a neighborhood transformation coefficient according to a preset number of neighborhood transformations, and comparing the neighborhood transformation coefficient with a preset neighborhood transformation probability parameter to obtain a comparison result;

[0107] In this step, the cloud game scheduling system obtains the neighborhood transformation parameters of the adaptive neighborhood transformation strategy, calculates the neighborhood transformation coefficient according to the preset number of neighborhood transformations in the neighborhood transformation parameters, and compares the neighborhood transformation coefficient with the adaptive neighborhood transformation probability parameter in the neighborhood transformation parameters to obtain a comparison result; it should be noted that the adaptive neighborhood transformation probability parameter is set in advance in the cloud game scheduling system; the calculation formula for calculating the neighborhood transformation coefficient according to the preset number of neighborhood transformations is:

[0108]

[0109] Among them, r5 is the neighborhood transformation coefficient, n is the number of cycles, N iter is the maximum number of cycles.

[0110] Step j: determining a target scheduling solution based on the comparison result, the optimal pre-scheduling solution and the neighborhood pre-scheduling solution.

[0111] In this step, the cloud game scheduling system determines the target scheduling plan based on the comparison results, the adaptation value of the optimal pre-scheduling plan, and the adaptation value of the neighboring pre-scheduling plan.

[0112] Specifically, step j includes:

[0113] Step j1: If the comparison result shows that the neighborhood transformation coefficient is not greater than the preset neighborhood transformation probability parameter, determining a target scheduling solution based on the optimal pre-scheduling solution and the neighborhood pre-scheduling solution;

[0114] In this step, if the cloud game scheduling system obtains a comparison result that the neighborhood transformation coefficient is not greater than the adaptive neighborhood transformation probability parameter, the adaptation value of the optimal pre-scheduling scheme is compared with the adaptation value of the neighborhood pre-scheduling scheme. If the adaptation value of the optimal pre-scheduling scheme is less than the adaptation value of the neighborhood pre-scheduling scheme, the optimal pre-scheduling scheme is used as the target scheduling scheme. If the adaptation value of the optimal pre-scheduling scheme is not less than the adaptation value of the neighborhood pre-scheduling scheme, the neighborhood pre-scheduling scheme is used as the target scheduling scheme.

[0115] Step j2: If the comparison result is that the neighborhood transformation coefficient is greater than the preset neighborhood transformation probability parameter, a local search is performed on the neighborhood pre-scheduling scheme. If the local search result meets the preset conditions, a target scheduling scheme is determined based on the optimal pre-scheduling scheme and the local pre-scheduling scheme. If the local search result does not meet the preset conditions, the step is re-executed: randomly perturb the optimal pre-scheduling scheme to generate a neighborhood pre-scheduling scheme.

[0116] In this step, if the cloud game scheduling system obtains a comparison result that the neighborhood transformation coefficient is greater than the adaptive neighborhood transformation probability parameter, it performs a local search on the neighborhood pre-scheduling scheme to obtain a local pre-scheduling scheme, wherein the local search process is: assuming that the neighborhood pre-scheduling scheme obtained after random perturbation is {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, where 1-10 represents the cloud games included in the domain scheduling scheme, and the arrangement order is the scheduling order of the cloud games, and randomly generates 2 random numbers between 1 and N (N: the total number of cloud games); then the cloud games corresponding to the two random numbers are swapped to obtain a local pre-scheduling scheme, such as the two random numbers are 4 and 7 respectively, then the local pre-scheduling scheme is {1, 2, 3, 7, 5, 6, 4, 8, 9, 10}.

[0117] After obtaining the local pre-scheduling scheme, the cloud game scheduling system calculates the adaptation value of the local cloud game scheme through the scheduling model, and compares the adaptation value of the optimal pre-scheduling scheme with the adaptation value of the local pre-scheduling scheme. When the adaptation value of the local pre-scheduling scheme is less than the adaptation value of the optimal pre-scheduling scheme, the neighborhood pre-scheduling scheme is replaced by the local pre-scheduling scheme. When the adaptation value of the local pre-scheduling scheme is not less than the adaptation value of the optimal pre-scheduling scheme, it is not necessary to replace the neighborhood pre-scheduling scheme with the local pre-scheduling scheme; and a local search is performed on the neighborhood pre-scheduling scheme to obtain a new local pre-scheduling scheme to make the neighborhood become The number of changes is increased by 1, and the above steps are repeated until the number of neighborhood changes reaches the maximum number of neighborhood changes in the neighborhood change parameter. The adaptation value of the optimal pre-scheduling scheme is compared with the adaptation value of the local pre-scheduling scheme finally determined after the cycle. When the adaptation value of the local pre-scheduling scheme finally determined after the cycle is less than the adaptation value of the optimal pre-scheduling scheme, the local pre-scheduling scheme finally determined after the cycle is used as the target scheduling scheme. When the adaptation value of the optimal pre-scheduling scheme is less than the adaptation value of the local pre-scheduling scheme finally determined after the cycle, the optimal pre-scheduling scheme after the cycle is used as the target scheduling scheme.

[0118] When the adaptation value of the local pre-scheduling scheme is not less than the adaptation value of the optimal pre-scheduling scheme, the steps of randomly perturbing the optimal pre-scheduling scheme, generating a neighborhood pre-scheduling scheme, and calculating the adaptation value of the neighborhood pre-scheduling scheme through the scheduling model and subsequent steps are re-executed to finally determine the target scheduling scheme.

[0119] The cloud game scheduling method of this embodiment obtains at least one original scheduling plan for the cloud game; calculates the adaptation value of each original scheduling plan and clusters all the original scheduling plans; generates a pre-scheduling plan based on the adaptation value and clustering results; determines a target scheduling plan based on the pre-scheduling plan, and performs cloud game scheduling based on the target scheduling plan. The present invention generates a pre-scheduling plan based on the adaptation value of each original scheduling plan and the clustering results of clustering all the original scheduling plans, optimizes the pre-scheduling plan to obtain a target scheduling plan, and performs cloud game scheduling based on the target scheduling plan, so that the target scheduling plan can reasonably schedule cloud games, thereby improving the practicality of cloud game scheduling.

[0120] Further, refer to Figure 3 Based on the first embodiment of the present application, in the second embodiment of the present application, the step of determining the probability value of all original scheduling solutions belonging to each cluster according to the adaptation value of each original scheduling solution and the preset number of clusters includes:

[0121] Step k, randomly selecting a number of preset clustering original scheduling schemes as the pre-clustering centers of each cluster;

[0122] Step 1: Calculate the distance from each original scheduling plan to each cluster based on the adaptation value of each original scheduling plan and the adaptation value of each pre-clustering center;

[0123] Step n: select an original scheduling plan in turn, and calculate the probability value of the selected original scheduling plan belonging to each cluster based on the distance from the selected original scheduling plan to each cluster.

[0124] In this embodiment, the cloud gaming scheduling system randomly selects several preset clusters of original scheduling plans from all original scheduling plans as the pre-cluster centers of each cluster. Based on the adaptation value of each original scheduling plan and the adaptation value of the pre-cluster center corresponding to each cluster, the system calculates the distance from each original scheduling plan to each cluster. Then, the system selects an original scheduling plan from all original scheduling plans in turn. Based on the distance from the selected original scheduling plan to each cluster, the system calculates the probability that the selected original scheduling plan belongs to each cluster. It should be noted that the distance from each original scheduling plan to each cluster can be Euclidean distance, Manhattan distance, Chebyshev distance, normalized Euclidean distance, Mahalanobis distance, etc.

[0125] Furthermore, the step of calculating the probability value of the selected original scheduling solution belonging to each cluster according to the distance from the selected original scheduling solution to each cluster includes:

[0126] One cluster is selected in turn, and the sum of the distance from the selected original scheduling solution to the selected cluster and the quotient of the distance from the selected original scheduling solution to each cluster is taken as the probability value that the selected original scheduling solution belongs to the selected cluster.

[0127] In this step, after selecting an original scheduling plan, the cloud game scheduling system selects a cluster from all clusters in turn, takes the distance from the selected original scheduling plan to the selected cluster as the numerator, and takes the distance from the selected original scheduling plan to each cluster as the denominator, and calculates the quotient of the distance from the selected original scheduling plan to the selected cluster and the distance from the selected original scheduling plan to each cluster respectively, and then adds all the quotients to obtain the sum of the distance from the selected original scheduling plan to the selected cluster and the distance from the selected original scheduling plan to each cluster, and uses the sum value as the probability value that the selected original scheduling plan belongs to the selected cluster.

[0128] The formula for calculating the probability that the selected original scheduling solution belongs to the selected cluster is as follows:

[0129]

[0130] Among them, p is the probability value of the selected original scheduling scheme belonging to the selected cluster, m is the preset number of clusters, d0 is the distance from the selected original scheduling scheme to the selected cluster, d iis the distance from the selected original scheduling plan to the i-th cluster, i = 1, 2, 3…m.

[0131] The formula for calculating d0 is:

[0132] d0=|x0-x j |

[0133] Among them, x0 is the adaptation value of the original scheduling scheme selected, x j is the fitness value of the pre-cluster center of the selected cluster.

[0134] Calculate d i The formula is:

[0135] d i =|X0-X i |

[0136] Among them, x0 is the adaptation value of the original scheduling scheme selected, x i is the adaptation value of the pre-clustering center of the i-th cluster, i = 1, 2, 3…m.

[0137] The cloud game scheduling system marks the original scheduling plans that have completed calculations, and repeatedly selects an unmarked original scheduling plan from all the original scheduling plans for calculation until all the original scheduling plans are calculated, and obtains the probability value of each original scheduling plan belonging to each cluster.

[0138] In a feasible embodiment: assuming that there are a total of 10 original scheduling schemes and 5 clusters; the cloud game scheduling system selects an original scheduling scheme from the 10 original scheduling schemes and selects a cluster from the 5 clusters.

[0139] The cloud game scheduling system executes step 1: taking the distance from the selected original scheduling plan to the selected cluster as the numerator, and the distance from the selected original scheduling plan to each cluster as the denominator, calculates the quotient of the distance from the selected original scheduling plan to the selected cluster and the distance from the selected original scheduling plan to each cluster, and then adds all the quotients to obtain the sum of the distance from the selected original scheduling plan to the selected cluster and the distance quotient from the selected original scheduling plan to each cluster, and uses the sum value as the probability value that the selected original scheduling plan belongs to the selected cluster.

[0140] After the cloud gaming scheduling system completes step 1, it proceeds to step 2: marking the clusters calculated in step 1, selecting an unmarked cluster from the five clusters, and repeating step 1. This continues until the probability of the selected original scheduling solution belonging to each of the five clusters is calculated.

[0141] After the cloud gaming scheduling system calculates the probability that the selected original scheduling plan belongs to each of the five clusters, it executes step 3: marking the calculated original scheduling plans, removing the mark from each of the five clusters, selecting an unmarked original scheduling plan from the 10 original scheduling plans, selecting a cluster from the five clusters, and repeating steps 1 and 2. After calculating the probability that the selected original scheduling plan belongs to each of the five clusters, the cloud gaming scheduling system repeats step 3 until it calculates the probability that each of the 10 selected original scheduling plans belongs to each of the five clusters.

[0142] The cloud game scheduling system of this embodiment calculates the distance from each original scheduling plan to each cluster based on the adaptation value of each original scheduling plan and the adaptation value of each pre-clustering center, and calculates the probability value of each original scheduling plan belonging to each cluster based on the distance, thereby improving the accuracy of the calculated probability value, helping to improve the accuracy of clustering the original scheduling plans, and thus improving the practicality of cloud game scheduling.

[0143] Further, refer to Figure 4 Based on the first and second embodiments of the present application, in the third embodiment of the present application, the step of generating a first membership matrix according to the probability value, all original scheduling schemes, and the preset number of clusters includes:

[0144] Step m, determining the number of matrix rows according to the number of all original scheduling solutions, and determining the number of matrix columns according to the preset number of clusters;

[0145] Step o: generating a first membership matrix according to the probability value, the number of matrix rows and the number of matrix columns.

[0146] In this embodiment, the cloud game scheduling system determines the number of matrix rows according to the number of all original scheduling schemes, and determines the number of matrix columns according to the preset number of clusters, and then generates a first membership matrix according to the probability value of each original scheduling scheme belonging to each cluster, the number of matrix rows and the number of matrix columns; for example: assuming that in the scheduling process of a certain cloud game, the number of all original scheduling schemes is 10, and the preset number of clusters is 5, the cloud game scheduling system determines the number of matrix rows to be 10 according to the number of all original scheduling schemes, and determines the number of matrix columns to be 5 according to the preset number of clusters, and then uses the probability value of the first original scheduling scheme belonging to the first cluster as the element value of the first row and first column in the matrix, and uses the probability value of the first original scheduling scheme belonging to the second cluster as the element value of the first row and second column in the matrix, and so on, and fills the probability values ​​of all original scheduling schemes belonging to each cluster into the matrix of 10 rows and 5 columns in sequence to generate a first membership matrix.

[0147] The cloud game scheduling system of this embodiment determines the number of matrix rows based on the number of all original scheduling schemes, and determines the number of matrix columns based on the preset number of clusters, and then generates a first membership matrix based on the probability value, the number of matrix rows and the number of matrix columns, thereby improving the accuracy of generating the first membership matrix, helping to improve the accuracy of clustering the original scheduling schemes, and thus improving the practicality of cloud game scheduling.

[0148] The present invention also provides a cloud game scheduling system.

[0149] The cloud game scheduling system includes: a memory, a processor, and a cloud game scheduling program stored in the memory and executable on the processor. When the cloud game scheduling program is executed by the processor, the steps of the cloud game scheduling method described above are implemented.

[0150] Among them, the method implemented when the cloud game scheduling program running on the processor is executed can refer to the various embodiments of the cloud game scheduling method of the present invention, and will not be repeated here.

[0151] The present invention also provides a computer-readable storage medium.

[0152] The computer-readable storage medium stores a cloud game scheduling program, which, when executed by a processor, implements the steps of the cloud game scheduling method described above.

[0153] Among them, the method implemented when the cloud game scheduling program running on the processor is executed can refer to the various embodiments of the cloud game scheduling method of the present invention, and will not be repeated here.

[0154] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0155] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0156] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0157] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A cloud game scheduling method, characterized in that: The cloud game scheduling method includes the following steps: Obtain at least one original scheduling plan for the cloud game; Calculate the adaptation value of each original scheduling plan through the pre-created scheduling model and cluster all the original scheduling plans; Before the step of calculating the adaptation value of each original scheduling solution by using the pre-created scheduling model and clustering all the original scheduling solutions, the method further includes: Obtain cloud server information, cloud game information, and the number of cloud games. The cloud server information includes the cloud server's cost, CPU memory capacity, GPU memory capacity, disk capacity, and bandwidth size. The cloud game information includes the CPU, GPU, bandwidth, disk resources, and configuration requirements required for the normal operation of the cloud game. Preprocessing the cloud server information, cloud game information, and the number of cloud games, and creating a scheduling model based on the preprocessed cloud server information, cloud game information, and the number of cloud games; The step of pre-processing the cloud server information, cloud game information, and cloud game quantity includes: The resource capacity and cost of cloud servers are dimensionlessly processed. Cloud server configuration information is simplified into cloud server configuration levels and labeled. Sort the cloud servers from low to high according to their configuration levels; Dimensionless processing of the resource size required for cloud gaming simplifies the configuration information required for cloud gaming and matches it with the configuration level of the cloud server. generating a pre-scheduling plan according to the adaptation value and the clustering result; A target scheduling plan is determined based on the pre-scheduling plan, and cloud gaming scheduling is performed based on the target scheduling plan.

2. The cloud game scheduling method according to claim 1, wherein: The step of clustering all original scheduling schemes includes: Determine the probability value of each original scheduling plan belonging to each cluster according to the adaptation value of each original scheduling plan and the preset number of clusters; generating a first membership matrix according to the probability value, all original scheduling schemes and the preset number of clusters; Calculating the target cluster center of each cluster according to the first membership matrix and the adaptation value of each original scheduling scheme; Calculating a second membership matrix based on the target cluster center; Calculating the variance between the first membership matrix and the second membership matrix; If the variance is not greater than the variance threshold, clustering all original scheduling solutions according to the first membership matrix and the target cluster center; If the variance is greater than the variance threshold, the second membership matrix is ​​used as the first membership matrix, and the step is re-executed: the target cluster center of each cluster is calculated according to the first membership matrix and the adaptation value of each original scheduling scheme.

3. The cloud game scheduling method according to claim 2, wherein: The step of determining the probability value of all original scheduling schemes belonging to each cluster according to the adaptation value of each original scheduling scheme and the preset number of clusters includes: Randomly select several preset clustering original scheduling plans as the pre-clustering centers of each cluster; According to the adaptation value of each original scheduling plan and the adaptation value of each pre-clustering center, the distance from each original scheduling plan to each cluster is calculated; An original scheduling scheme is selected in turn, and the probability value of the selected original scheduling scheme belonging to each cluster is calculated based on the distance from the selected original scheduling scheme to each cluster.

4. The cloud game scheduling method according to claim 3, wherein: The step of calculating the probability value of the selected original scheduling solution belonging to each cluster based on the distance from the selected original scheduling solution to each cluster includes: One cluster is selected in turn, and the sum of the distance from the selected original scheduling solution to the selected cluster and the quotient of the distance from the selected original scheduling solution to each cluster is taken as the probability value that the selected original scheduling solution belongs to the selected cluster.

5. The cloud game scheduling method according to claim 2, wherein: The step of generating a first membership matrix according to the probability value, all original scheduling schemes and the preset number of clusters includes: Determining the number of matrix rows according to the number of all original scheduling solutions, and determining the number of matrix columns according to the preset number of clusters; A first membership matrix is ​​generated according to the probability value, the number of matrix rows, and the number of matrix columns.

6. The cloud game scheduling method according to claim 2, wherein: The step of calculating the target cluster center of each cluster according to the first membership matrix and the adaptation value of each original scheduling scheme includes: The probability value of each original scheduling scheme belonging to each cluster in the first membership matrix is ​​obtained, and the target cluster center of each cluster is calculated based on the probability value and the adaptation value of each original scheduling scheme.

7. The cloud game scheduling method as described in claim 2, characterized in that: The step of determining the target scheduling scheme according to the pre-scheduling scheme comprises: Randomly perturbing the optimal pre-scheduling scheme among the pre-scheduling schemes to generate a neighborhood pre-scheduling scheme; Calculating a neighborhood transformation coefficient according to a preset number of neighborhood transformations, and comparing the neighborhood transformation coefficient with a preset neighborhood transformation probability parameter to obtain a comparison result; A target scheduling solution is determined according to the comparison result, the optimal pre-scheduling solution and the neighborhood pre-scheduling solution.

8. The cloud game scheduling method according to claim 7, wherein: The step of determining a target scheduling solution according to the comparison result, the optimal pre-scheduling solution, and the neighborhood pre-scheduling solution includes: If the comparison result is that the neighborhood transformation coefficient is not greater than the preset neighborhood transformation probability parameter, determining a target scheduling scheme according to the optimal pre-scheduling scheme and the neighborhood pre-scheduling scheme; If the comparison result is that the neighborhood transformation coefficient is greater than the preset neighborhood transformation probability parameter, a local search is performed on the neighborhood pre-scheduling scheme. If the local search result meets the preset conditions, a target scheduling scheme is determined based on the optimal pre-scheduling scheme and the local pre-scheduling scheme; if the local search result does not meet the preset conditions, the step is re-executed: randomly perturb the optimal pre-scheduling scheme to generate a neighborhood pre-scheduling scheme.

9. A cloud game scheduling system, characterized in that: The cloud game scheduling system includes: a memory, a processor, and a cloud game scheduling program stored in the memory and executable on the processor. When the cloud game scheduling program is executed by the processor, the steps of the cloud game scheduling method according to any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a cloud game scheduling program, which, when executed by a processor, implements the steps of the cloud game scheduling method according to any one of claims 1 to 8.

Images