A network modal game evolutionary decision-making method and device

Abstract

This invention provides a method and apparatus for network modal game evolutionary decision-making. The method includes: acquiring real service information at a target time; inputting the real service information at the target time into a pre-trained network modal game evolutionary decision-making model to obtain network modal strategies; wherein the network modal game evolutionary decision-making model is trained based on historical service data and network resource change data; the network modal strategies include: resource allocation evolution strategies for each network modality on a network element and network modality addition / deletion evolution strategies in the global network environment. Through the method provided by this invention, for new service requests and network resource changes at a certain time, resource allocation evolution strategies for each network modality on a network element are generated, and network modality addition / deletion evolution strategies in the global network environment are given at a longer time scale. This achieves spatiotemporal integrated perception and adjustment of network modal evolution strategies, ultimately ensuring performance and improving resource utilization.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for evolutionary decision-making in network modal games. Background Technology

[0002] Multimodal intelligent networks (MIMs) are technologies that integrate multiple network technologies into a multimodal network through programmable techniques, aiming to combine the advantages of various networks and enhance the network's ability to support diverse services. The network multimodal symbiosis and evolution target model, based on the scalability of the infrastructure throughout its entire lifecycle, introduces the network modality diversity (V) indicator in addition to the traditional network Quality of Service (S) and Resource Reusability (M) metrics. It proposes a three-dimensional architecture of network modalities: "evolution, generation, and symbiosis." This is reflected in the sharing and collaborative utilization of underlying resources by network modalities within the same infrastructure, forming a symbiotic state that continuously evolves (expand / shrink, generate / optimize / die out, etc.) according to service and resource conditions. The task of the model's evolution aspect involves, at the micro level, determining the timing and strategy of evolution through the deduction of cooperative game theory among network modalities at different times; and at the macro level, quantitatively evaluating and regulating the evolution path throughout the entire lifecycle with the goal of achieving optimal SMV in the spatiotemporal joint dimension. The combination of these two approaches provides a strong guarantee for achieving complete SMV intersection of multimodal networks throughout their entire lifecycle.

[0003] Existing technologies are typically developed based on single-mode network scheduling algorithms. This approach usually focuses on the allocation and management of resources in a fixed network mode under specific scenarios, using heuristic methods or some AI algorithms to solve the problem. Currently, multimodal networks are still in the exploratory stage. Based on the initial concept of multimodal networks, the entire plane is divided into a data plane, a control plane, and a service plane, and the algorithm is optimized under multi-dimensional definitions.

[0004] How to adjust the network modality evolution strategy to ultimately ensure performance and improve resource utilization is a technical problem that needs to be solved. Summary of the Invention

[0005] This invention provides a method and apparatus for evolutionary decision-making in network modal games, in order to overcome the deficiencies in the prior art.

[0006] This invention provides a network modal game evolutionary decision-making method, comprising the following steps:

[0007] Obtain real-time business information at the target moment;

[0008] The real business information at the target time is input into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0009] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0010] According to the present invention, a network modal game evolution decision-making method is provided, wherein the network modal game evolution decision-making model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module;

[0011] The step of inputting the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy includes:

[0012] The actual business information at the target time is input into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the actual business information and the predicted business information;

[0013] The mixed service information is input into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality;

[0014] The first mapping relationship is input into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and the second mapping relationship is input into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment.

[0015] According to the network modal game evolution decision-making method provided by the present invention, the step of inputting the real service information at the target time into the service situation prediction module to obtain mixed service information includes:

[0016] The actual business information at the target time is input into the business situation prediction module to obtain the predicted business information;

[0017] The real business information at the target time and the predicted business information are fused according to a preset ratio to obtain the hybrid business information.

[0018] According to the network modal game evolution decision-making method provided by the present invention, the service clustering and modality adaptation module includes: a service clustering submodule and a modality adaptation submodule;

[0019] The step of inputting the hybrid service information into the service clustering and modality adaptation module to obtain the first mapping relationship and the second mapping relationship includes:

[0020] The mixed business information is input into the business clustering submodule, and business clustering is performed through meta-learning and adversarial learning algorithms to obtain multiple business clusters;

[0021] The multiple service clusters are input into the modality adaptation submodule, and the service and network modality are adapted through metric learning method to obtain the first mapping relationship and the second mapping relationship.

[0022] According to the present invention, a network modal game evolution decision-making method is provided, wherein inputting the first mapping relationship into the network modal agile control module to obtain the resource allocation evolution strategy of each network modality on the network element includes:

[0023] The first mapping relationship and the first network configuration information are input into the network modality agile control module, and the network configuration information for the next moment is generated through cooperative game between different network modalities; wherein, the network configuration information for the next moment is: the allocation ratio of each network modality resource on each network element.

[0024] According to the present invention, a network modal game evolution decision-making method is provided, wherein inputting the second mapping relationship into the network modal long-term evolution module to obtain the network modal addition and deletion evolution strategy in the global network environment includes:

[0025] The second mapping relationship and the second network configuration information are input into the network modality long-term evolution module, and the network modality configuration at the next time step is obtained through reinforcement learning algorithm; wherein, the network modality configuration at the next time step is: the network modality addition and deletion evolution strategy in the global network environment and the feature information of the new modality.

[0026] The present invention also provides a network modal game evolution decision-making device, comprising the following modules:

[0027] The acquisition unit is used to acquire real business information at the target time.

[0028] The decision-making unit is used to input the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0029] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0030] According to the present invention, a network modal game evolution decision-making device is provided, wherein the network modal game evolution decision-making model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module;

[0031] The decision-making unit is specifically used for:

[0032] The actual business information at the target time is input into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the actual business information and the predicted business information;

[0033] The mixed service information is input into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality;

[0034] The first mapping relationship is input into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and the second mapping relationship is input into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the network modal game evolution decision method as described above.

[0036] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the network modal game evolution decision-making method as described above.

[0037] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the network modal game evolution decision method as described above.

[0038] This invention provides a network modal game evolutionary decision-making method and apparatus. It acquires real-time business information at a target moment and inputs this information into a pre-trained network modal game evolutionary decision-making model to obtain network modal strategies. The network modal game evolutionary decision-making model is trained based on historical business data and network resource change data. The network modal strategies include resource allocation evolution strategies for each network modality on a network element and network modality addition / deletion evolution strategies in the global network environment. Therefore, this invention, through training, obtains a network modal game evolutionary decision-making model. For new business requests and network resource changes at a certain moment, it generates resource allocation evolution strategies for each network modality on a network element and provides network modality addition / deletion evolution strategies in the global network environment over a longer time scale. This achieves spatiotemporal integrated perception and adjustment of network modal evolution strategies, ultimately ensuring performance and improving resource utilization. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating the network modal game evolution decision-making method provided by the present invention.

[0041] Figure 2 This is a complete flowchart of the network modal game evolution decision-making method provided by the present invention.

[0042] Figure 3 This is a schematic diagram illustrating the principle of the network modal game evolutionary decision-making model provided by the present invention.

[0043] Figure 4 This is a schematic diagram of the network modal game evolution decision-making device provided by the present invention.

[0044] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0046] The following is combined with Figures 1-5 This invention describes a network modal game evolution decision-making method and apparatus.

[0047] It should be noted that in the existing technologies, (1) a single network modality scheduling algorithm is developed. This approach usually focuses on the allocation and management of resources in a fixed network modality under a specific scenario, using heuristic methods or some AI algorithms to solve the problem. (2) An attempt based on the concept of multimodal networks. Currently, multimodal networks are still in the exploratory stage. According to the initial concept of multimodal networks, the entire plane is divided into a data plane, a control plane, and a service plane, and the algorithm is optimized under a multi-dimensional definition. Based on this, the present invention provides a network modality game evolution decision-making method to solve at least one of the above problems.

[0048] Figure 1 This is a flowchart illustrating the network modal game evolutionary decision-making method provided by the present invention, as shown below. Figure 1 As shown, the method includes the following:

[0049] Step 100: Obtain the actual business information at the target time.

[0050] Figure 2 This is a complete flowchart of the network modal game evolutionary decision-making method provided by the present invention. The following is a combination of... Figure 2 The evolutionary decision-making method for network modal games provided by this invention will be described.

[0051] It should be noted that the products or projects applying this invention can be any environment that supports multimodal networks. For example, by embedding the model framework of this invention into the top-level controller of a network environment composed of network elements with multimodal processing capabilities, such as communication networks, social networks, and the Internet of Things, and utilizing spatiotemporal integrated sensing technology to perceive changes in services and network resources, and providing network modality evolution strategies at appropriate times, the overall service quality and resource utilization of the multimodal network can be improved. Network operators can achieve precise control and long-term evolution management of network modes, improve network performance and resource utilization efficiency, and reduce operating costs; users can also obtain better service quality assurance.

[0052] Specifically, see Figure 2 The actual business information at the target time can be real-time business information composed of multiple business requirements.

[0053] Step 200: Input the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0054] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0055] It should be noted that, please continue to refer to Figure 2 The network modal game evolution decision model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module.

[0056] This embodiment proposes a novel network modal game evolution mechanism for multimodal network environments. For example... Figure 2 As shown, its core is a decision-making model composed of an agile control module and a long-term evolution module. It uses a situation prediction module and a business clustering and modality adaptation module to achieve spatiotemporal integrated perception as inputs to the two core modules. With the help of AI technology, especially reinforcement learning technology, which can dynamically generate strategies according to the environmental state, and data-driven situation prediction technology, it can control the allocation of network modal resources at the micro level and control the number of network modalities and feature output strategies at the macro level, thereby completing the network modality evolution.

[0057] Step 200 involves inputting the real business information at the target time into a pre-trained network modal game evolutionary decision model to obtain the network modal strategy, including:

[0058] Step 210: Input the real business information at the target time into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the real business information and the predicted business information.

[0059] Step 210 specifically includes:

[0060] Step 211: Input the actual business information at the target time into the business situation prediction module to obtain the predicted business information.

[0061] Step 212: Merge the real business information at the target time and the predicted business information according to a preset ratio to obtain the mixed business information.

[0062] In one embodiment, Figure 3 This is a schematic diagram illustrating the principle of the network modal game evolutionary decision-making model provided by the present invention, as shown below. Figure 3 As shown, at a certain moment For example, the specific explanation is as follows.

[0063] 1) The data-driven situation prediction module outputs mixed business information. .

[0064] Considering that the spontaneity of business operations (irregular changes, non-linear growth, etc.) may affect system stability and strategy smoothness, a data-driven situation prediction module is used. This module utilizes historical business information to predict the current moment, and then a time-series module within the module merges the actual business input with the predicted business information according to a certain ratio to obtain hybrid business information. .

[0065] Step 220: Input the mixed service information into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality.

[0066] It should be noted that the business clustering and modality adaptation module includes a business clustering submodule and a modality adaptation submodule.

[0067] Step 220 specifically includes:

[0068] Step 221: Input the mixed business information into the business clustering submodule, and perform business clustering through meta-learning and adversarial learning algorithms to obtain multiple business clusters.

[0069] Step 222: Input the multiple service clusters into the modality adaptation submodule, and use the metric learning method to adapt the service and network modality to obtain the first mapping relationship and the second mapping relationship.

[0070] In one embodiment, see continue to see Figure 3 The specific details are as follows.

[0071] 2) Business clustering yields clusters. .

[0072] Based on the S, M, and V features of the business, and considering their feature space and temporal distribution, meta-learning and adversarial learning algorithms are used to perform adaptive clustering of the business, resulting in different business clusters. .

[0073] 3) Network modality adaptation determines the compatibility between services and network modality. .

[0074] Metric learning methods can be used, such as mapping business cluster features and network modality features to a unified metric space through learnable parameters A. Ultimately, the adaptation between business and network modes is completed, and their mapping relationship is obtained. Together with the situation prediction and business clustering modules, it enables integrated spatiotemporal perception of business, serving as an important component of the input for the subsequent two core modules.

[0075] Step 230: Input the first mapping relationship into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and input the second mapping relationship into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment.

[0076] Step 230 specifically includes:

[0077] Step 231: Input the first mapping relationship and the first network configuration information into the network modality agile control module, and generate the network configuration information for the next moment through cooperative game between different network modalities; wherein, the network configuration information for the next moment is: the allocation ratio of each network modality resource on each network element.

[0078] In one embodiment, see continue to see Figure 3 The specific details are as follows.

[0079] 4) The network modality agile control module obtains the network configuration for the next time step. .

[0080] On a smaller time scale Based on the input mixed service information and first network configuration information Network configuration information for the next time step is generated through cooperative game theory among different network modalities. This network configuration information specifies the resource allocation ratio for each network modality on each network element. By using reinforcement learning algorithms and a multi-agent structure, different network elements are modeled as tuples of states, actions, and rewards. ,in Represents a set of states. Represents intelligent agents A set of actions and rewards. The time input is Through rewarded observations Generate action strategy Finally, the network configuration for the next moment is obtained. This section's action space allocates a proportion of network modal resources on each network element, with rewards based on resource utilization and service quality (QoS factors such as latency). This module can generate configuration policies for resources on network elements, safeguarding network modal resource utilization and service quality.

[0081] Step 232: Input the second mapping relationship and the second network configuration information into the network modality long-term evolution module, and obtain the network modality configuration at the next time step through the reinforcement learning algorithm; wherein, the network modality configuration at the next time step is: the network modality addition and deletion evolution strategy in the global network environment and the feature information of the new modality.

[0082] In one embodiment, see continue to see Figure 3 The specific details are as follows.

[0083] 5) The network modality long-term evolution module obtains the network modality configuration for the next time step. .

[0084] On a larger time scale Above, predict business information based on input. Second network configuration information The network modality configuration for the next time step is generated through reinforcement learning algorithms. In terms of reinforcement learning settings, the input is... Through rewarded observations Generate the configuration, i.e., action policy, of each mode in the network at the next time step. Finally, the network state at the next moment is obtained. In this section, the algorithm's action space comprises the number and feature configuration of network modalities, with the optimization objective being to improve the average task service quality and overall network resource utilization. This module is triggered under three conditions: ① when the previous network modality agile control module fails to continuously optimize task service quality and resource utilization; ② when existing network modalities cannot meet newly arriving business demands; ③ when a certain preset time is reached, this module is executed, providing network modality addition / deletion strategies and new modality feature information by fully weighing the deployment costs and adjustment benefits of network modalities.

[0085] Based on the above embodiments, in this embodiment, the model training includes three parts: training of four modules: a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module. Specifically, when training the network modality agile control module, the input is a mixture of real and predicted business information, aiming to prevent business changes from impacting system stability. Simultaneously, when training the network modality long-term evolution module, the input is only predicted business information, aiming to enhance policy smoothness and reduce the impact of sudden business changes on policy quality.

[0086] In a network system, a large amount of existing business data and network resource change data can be used to train the network modality agile control module and the network modality long-term evolution module according to the methods described in the above embodiments. For a new business request and network resource change at a certain moment, the previously trained network modality agile control model is used to generate resource allocation evolution strategies for each network modality on the network element, and the network modality long-term evolution model is used to provide network modality addition and deletion evolution strategies in the global network environment at a longer time scale.

[0087] At a smaller time scale, the number and characteristics of existing network modes are considered quantitative. Business data obtained by mixing real business information with predicted business information obtained through time series modules in a certain proportion, along with information matched to the corresponding network modes, can be injected into the network mode agile control model. This can meet the network resource demands generated by business changes and improve resource utilization and network service quality.

[0088] On a larger time scale, the business information and network resource change information obtained after using the situation prediction of the time series module are fully utilized and injected into the long-term evolution model of network modes. The cost of adding or deleting network modes and the gain of existing network modes are fully considered, so as to provide a guarantee for stable network operation and long-term resource utilization.

[0089] The network modality game evolution decision-making method provided in this invention, at the micro level, utilizes a spatiotemporal sequence prediction model and historical network data to accurately predict future business and resource situations, providing a basis for network situation inference. It models network modality competition for local network resources as a micro-game decision-making problem, constructs an agile network modality control mechanism, and enables multi-agent cooperative game inference of multiple network modalities in local network areas, providing accurate timing and strategies for network modality evolution. At the macro level, based on long-term SMV quantitative index evaluation and feedback, it utilizes a deep learning model tailored to the characteristics of network modality data to construct a long-term network modality evolution mechanism, adjusting the evolution path of network modalities and providing network modality evolution strategies. The problem it solves is to ensure performance and improve resource utilization in a multi-network modality ring competition and cooperation environment by sensing and adjusting network modality evolution strategies through spatiotemporal integration.

[0090] The above describes the steps of the network modal game evolution decision-making method provided by this invention. As can be seen from the above description, according to the network modal game evolution decision-making method provided by this invention, real business information at a target time is obtained; this real business information at the target time is input into a pre-trained network modal game evolution decision-making model to obtain network modal strategies; wherein, the network modal game evolution decision-making model is obtained by training based on historical business data and network resource change data; the network modal strategies include: resource allocation evolution strategies for each network modality on a network element and network modality addition / deletion evolution strategies in the global network environment. Therefore, this invention, through training, obtains a network modal game evolution decision-making model, generates resource allocation evolution strategies for each network modality on a network element for new business requests and network resource changes at a certain time, and provides network modality addition / deletion evolution strategies in the global network environment at a longer time scale, achieving spatiotemporal integrated perception and adjustment of network modal evolution strategies, ultimately ensuring performance and improving resource utilization.

[0091] The following describes the network modal game evolution decision-making device provided by the present invention. The network modal game evolution decision-making device described below can be referred to in correspondence with the network modal game evolution decision-making method described above.

[0092] Figure 4 This is a schematic diagram of the network modal game evolution decision-making device provided by the present invention, as shown below. Figure 4 As shown, the network modal game evolution decision-making device provided by the present invention includes:

[0093] Acquisition unit 401 is used to acquire real business information at the target time.

[0094] Decision unit 402 is used to input the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0095] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0096] The network modal game evolution decision-making device provided by this invention acquires real-time business information at a target moment; inputs this real-time business information into a pre-trained network modal game evolution decision-making model to obtain network modal strategies; wherein, the network modal game evolution decision-making model is trained based on historical business data and network resource change data; the network modal strategies include: resource allocation evolution strategies for each network modality on a network element and network modality addition / deletion evolution strategies in the global network environment. Therefore, this invention, through training the network modal game evolution decision-making model, generates resource allocation evolution strategies for each network modality on a network element for new business requests and network resource changes at a certain moment, and provides network modality addition / deletion evolution strategies in the global network environment over a longer time scale, achieving spatiotemporal integrated perception and adjustment of network modal evolution strategies, ultimately ensuring performance and improving resource utilization.

[0097] Based on the above embodiments, in this embodiment, the network modal game evolution decision model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module;

[0098] The decision-making unit 402 is specifically used for:

[0099] The actual business information at the target time is input into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the actual business information and the predicted business information;

[0100] The mixed service information is input into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality;

[0101] The first mapping relationship is input into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and the second mapping relationship is input into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment.

[0102] Based on the above embodiments, in this embodiment, the device further includes a fusion unit, specifically used for:

[0103] The actual business information at the target time is input into the business situation prediction module to obtain the predicted business information;

[0104] The real business information at the target time and the predicted business information are fused according to a preset ratio to obtain the hybrid business information.

[0105] Based on the above embodiments, in this embodiment, the business clustering and modality adaptation module includes: a business clustering submodule and a modality adaptation submodule;

[0106] The device further includes a mapping unit, specifically used for:

[0107] The mixed business information is input into the business clustering submodule, and business clustering is performed through meta-learning and adversarial learning algorithms to obtain multiple business clusters;

[0108] The multiple service clusters are input into the modality adaptation submodule, and the service and network modality are adapted through metric learning method to obtain the first mapping relationship and the second mapping relationship.

[0109] Based on the above embodiments, in this embodiment, the decision unit 402 is specifically used for:

[0110] The first mapping relationship and the first network configuration information are input into the network modality agile control module, and the network configuration information for the next moment is generated through cooperative game between different network modalities; wherein, the network configuration information for the next moment is: the allocation ratio of each network modality resource on each network element.

[0111] Based on the above embodiments, in this embodiment, the decision unit 402 is specifically used for:

[0112] The second mapping relationship and the second network configuration information are input into the network modality long-term evolution module, and the network modality configuration at the next time step is obtained through reinforcement learning algorithm; wherein, the network modality configuration at the next time step is: the network modality addition and deletion evolution strategy in the global network environment and the feature information of the new modality.

[0113] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device can be a robot or other electronic device. This electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540. The processor 510, communication interface 520, and memory 530 communicate with each other via the communication bus 540. The processor 510 can call logical instructions from the memory 530 to execute network modal game evolutionary decision-making methods, including:

[0114] Obtain real-time business information at the target moment;

[0115] The real business information at the target time is input into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0116] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0117] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being able to be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, the computer being able to execute the network modal game evolution decision-making method provided by the above methods, including:

[0119] Obtain real-time business information at the target moment;

[0120] The real business information at the target time is input into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0121] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0122] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the network modal game evolution decision-making methods provided by the above methods, including:

[0123] Obtain real-time business information at the target moment;

[0124] The real business information at the target time is input into the pre-trained network modal game evolution decision model to obtain the network modal strategy;

[0125] The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment.

[0126] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0127] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A network modal game evolutionary decision-making method, characterized in that, include: Obtain real-time business information at the target moment; The real business information at the target time is input into the pre-trained network modal game evolution decision model to obtain the network modal strategy; The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment; The network modal game evolution decision model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module; The step of inputting the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy includes: The actual business information at the target time is input into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the actual business information and the predicted business information; The mixed service information is input into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality; The first mapping relationship is input into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and the second mapping relationship is input into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment; wherein, the network modality agile control module is used to generate network configuration information for the next moment based on the first mapping relationship through cooperative game among different network modalities; the network modality long-term evolution module is used to obtain the network modality configuration for the next moment based on the second mapping relationship through a reinforcement learning algorithm; The business clustering and modal adaptation module includes: a business clustering submodule and a modal adaptation submodule; The step of inputting the hybrid service information into the service clustering and modality adaptation module to obtain the first mapping relationship and the second mapping relationship includes: The mixed business information is input into the business clustering submodule, and business clustering is performed through meta-learning and adversarial learning algorithms to obtain multiple business clusters; The multiple service clusters are input into the modality adaptation submodule, and the service and network modality are adapted through metric learning method to obtain the first mapping relationship and the second mapping relationship.

2. The network modal game evolutionary decision-making method according to claim 1, characterized in that, The step of inputting the real business information at the target time into the business situation prediction module to obtain mixed business information includes: The actual business information at the target time is input into the business situation prediction module to obtain the predicted business information; The real business information at the target time and the predicted business information are fused according to a preset ratio to obtain the hybrid business information.

3. The network modal game evolutionary decision-making method according to claim 1, characterized in that, The step of inputting the first mapping relationship into the network modality agile control module to obtain the resource allocation evolution strategy for each network modality on the network element includes: The first mapping relationship and the first network configuration information are input into the network modality agile control module, and the network configuration information for the next moment is generated through cooperative game between different network modalities; wherein, the network configuration information for the next moment is: the allocation ratio of each network modality resource on each network element.

4. The network modal game evolutionary decision-making method according to claim 1, characterized in that, The step of inputting the second mapping relationship into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment includes: The second mapping relationship and the second network configuration information are input into the network modality long-term evolution module, and the network modality configuration at the next time step is obtained through reinforcement learning algorithm; wherein, the network modality configuration at the next time step is: the network modality addition and deletion evolution strategy in the global network environment and the feature information of the new modality.

5. A network modal game evolution decision-making device, characterized in that, include: The acquisition unit is used to acquire real business information at the target time. The decision-making unit is used to input the real business information at the target time into the pre-trained network modal game evolution decision model to obtain the network modal strategy; The network modal game evolution decision model is obtained by training based on historical business data and network resource change data; the network modal strategy includes: resource allocation evolution strategy for each network modality on the network element and network modality addition and deletion evolution strategy in the global network environment; The network modal game evolution decision model includes: a business situation prediction module, a business clustering and modality adaptation module, a network modality agile control module, and a network modality long-term evolution module; The decision-making unit is specifically used for: The actual business information at the target time is input into the business situation prediction module to obtain mixed business information; wherein, the mixed business information includes: the actual business information and the predicted business information; The mixed service information is input into the service clustering and modality adaptation module to obtain a first mapping relationship and a second mapping relationship; wherein, the first mapping relationship is: the matching relationship between mixed services and network modality, and the second mapping relationship is: the matching relationship between predicted services and network modality; The first mapping relationship is input into the network modality agile control module to obtain the resource allocation evolution strategy of each network modality on the network element, and the second mapping relationship is input into the network modality long-term evolution module to obtain the network modality addition and deletion evolution strategy in the global network environment; wherein, the network modality agile control module is used to generate network configuration information for the next moment based on the first mapping relationship through cooperative game among different network modalities; the network modality long-term evolution module is used to obtain the network modality configuration for the next moment based on the second mapping relationship through a reinforcement learning algorithm; The business clustering and modal adaptation module includes: a business clustering submodule and a modal adaptation submodule; The device further includes a mapping unit, specifically used for: The mixed business information is input into the business clustering submodule, and business clustering is performed through meta-learning and adversarial learning algorithms to obtain multiple business clusters; The multiple service clusters are input into the modality adaptation submodule, and the service and network modality are adapted through metric learning method to obtain the first mapping relationship and the second mapping relationship.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the network modal game evolution decision-making method as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the network modal game evolution decision method as described in any one of claims 1 to 4.

Images