A communication method and apparatus

Abstract

The embodiment of the application provides a kind of communication method and device, it is related to communication technical field.The method comprises: the first network element obtains service request message, service request message includes intention, intention is used to determine the task of request execution.Again according to intention, the class of generative network is provided for task.Determines the experience information corresponding to the class of generative network for the class of generative network.Again, intention and the class of generative network are sent to the second network element.Correspondingly, the intention and the class of generative network from the first network element are received by the second network element.Through the core network architecture based on agent, the class of generative network can expand the service class range of core network, it is helpful to provide more flexible, customized and differentiated service for user based on the core network of agent.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a communication method and apparatus. Background Technology

[0002] Currently, 5G networks can achieve a leap in communication performance through enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable low-latency communications (uRLLC) technologies. At the same time, they can provide differentiated services through network slicing mechanisms supported by single network slice selection assistance information (S-NSSAI).

[0003] However, with the development of communication technology, large language model (LLM) driven agents, with their autonomous decision-making, multi-task collaboration, and semantic understanding capabilities, have provided a new path for network intelligence. This allows future communication technologies to support new services such as computing, sensing, and artificial intelligence (AI), no longer limited to connectivity services. Therefore, the limitations of existing network slicing mechanisms are becoming increasingly apparent: traditional slice classifications (eMBB, uRLLC, mMTC) primarily satisfy connectivity services and struggle to meet the diverse needs of computationally intensive, sensing-driven, and generative AI services. However, facing diverse user network demands, improving the efficiency of network generation and management is a challenge for next-generation networks based on intelligent agents. Summary of the Invention

[0004] This application provides a communication method and apparatus. The method defines generative network categories, which can assist in generating networks that meet business requirements and improve the generation efficiency of generative networks.

[0005] In a first aspect, embodiments of this application provide a communication method applied to a communication device. For example, the communication device may be a first network element, or a component within the communication device (e.g., a processor, circuit, chip, or chip system responsible for communication functions, including but not limited to a modem chip, a baseband chip, a system-on-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip, etc.). Alternatively, the communication device may also be a logic module or software capable of implementing all or part of the functions of a communication device. The method includes: the first network element acquiring a service request message, the service request message including an intent used to determine a requested task. The first network element determines the category of a generative network providing services for the task based on the intent. The category of the generative network is used to determine empirical information corresponding to the category of the generative network. The first network element sends the intent and the category of the generative network to a second network element. Correspondingly, the second network element receives the intent and the category of the generative network from the first network element.

[0006] In this application, a first network element receives a service request message initiated by a user. The first network element can dynamically map the intent to a generative network category. The first network element then sends the generative network category to a second network element. This facilitates the second network element in determining the experiential information related to the generative network category. By determining the generative network category, this application enables network agents to more efficiently retrieve experiential information related to the generative network category, thereby generating the required generative network based on the experiential information. This improves the generation efficiency of the generative network and ultimately achieves the goal of efficiently providing users with differentiated services beyond connectivity.

[0007] In one possible implementation, the first network element receives first information from the second network element. This first information is determined by empirical information corresponding to the generative network's category. By incorporating the generative network's category, the empirical information is made more targeted, improving the adaptability and efficiency of task processing.

[0008] In one possible implementation, the initial information is determined by experiential information that matches the intent. This approach enhances the accuracy of task decomposition, reduces irrelevant data processing, and minimizes resource waste.

[0009] In one possible implementation, the method provided in this application further includes: a first network element determining information about a first sub-task based on first information, wherein the information about the first sub-task is used to indicate multiple sub-tasks after task decomposition. The first network element then sends the information about the first sub-task to a third network element. This method achieves modular task decomposition, clarifies the division of labor among sub-tasks, and improves the systematicness and scalability of multi-network element collaborative work.

[0010] In one possible implementation, the method provided in this application embodiment further includes: a first network element sending the generative network category to a third network element. Transmitting the generative network category between network elements enables the third network element to optimize subtask execution strategies based on the category, thereby enhancing overall task adaptability.

[0011] In one possible implementation, the method provided in this application embodiment further includes: a first network element acquiring rule information. The rule information describes the rules corresponding to the processing tasks under the generative network category. By constraining the task processing flow through the rule information, the operation is ensured to conform to preset standards, reducing execution risks.

[0012] In one possible implementation, the rule information includes: content that is allowed and / or prohibited for the first network element to perform under the category of generative network.

[0013] In one possible implementation, the method provided in this application further includes: a first network element generating a first prompt word based on rule information. The first prompt word is used to convert the content of the rule information into executable instructions. This method transforms abstract rules into concrete instructions, simplifying execution complexity and improving human-computer interaction efficiency and operational accuracy.

[0014] In one possible implementation, the first network element obtains rule information by receiving rule information from the second network element. This approach enables flexible policy updates, adapts to different scenario requirements, and enhances system maintainability.

[0015] In one possible implementation, the generative network corresponds to multiple categories. The first network element acquires rule information by: determining first rule information based on at least one of the multiple categories corresponding to the generative network; the first rule information being used to process the rules corresponding to the task under at least one of the multiple categories of the generative network; the first network element sending the first rule information to a second network element; the first rule information being used to determine rule information; and the first network element receiving the rule information from the second network element. This approach supports multi-category rule fusion, provides composite task processing strategies, and expands the system's applicable scenarios.

[0016] In one possible implementation, the method provided in this application further includes: a first network element determining information about a second subtask based on first information and rule information. The information about the second subtask is used to indicate multiple subtasks after the task is decomposed according to the rule information. The first network element sends the information about the second subtask to a third network element. This approach integrates the dual constraints of empirical information and rule information, optimizes the subtask decomposition logic, and improves the reliability and quality of task execution results.

[0017] In one possible implementation, the method provided in this application further includes: the first network element sending rule information and the type of generative network to the third network element. This method can assist the third network element in making global optimization decisions, avoid information silos, and improve collaborative efficiency.

[0018] Secondly, embodiments of this application provide a communication method applied to a communication device. For example, the communication device may be a second network element, or a component of the communication device (e.g., a processor, circuit, chip, or chip system responsible for communication functions, including but not limited to a modem chip, a baseband chip, a system-on-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip, etc.). Alternatively, the communication device may also be a logic module or software capable of implementing all or part of the functions of a communication device. The method includes: the second network element receiving a generative network category from a first network element. The generative network is used to provide services for tasks corresponding to an intent. The second network element obtains empirical information corresponding to the category of the generative network based on that category.

[0019] In one possible implementation, the second network element obtains the experience information corresponding to the generative network category by retrieving the experience information corresponding to the generative network category from the storage unit. This method allows for precise querying of pre-stored experience information in the storage unit based on the generative network category, significantly improving the retrieval efficiency of experience information.

[0020] In one possible implementation, the second network element obtaining experience information corresponding to the generative network category from the storage unit includes: the second network element sending a first request to the storage unit. The first request includes the generative network category and is used to request experience information corresponding to that category. The second network element receives the experience information corresponding to the generative network category from the storage unit. This method reduces redundant data transmission and improves resource utilization.

[0021] In one possible implementation, the storage unit stores multiple categories of generative networks, along with the corresponding experience information for each category. For example, the storage unit could be a shared memory for the networks. This approach allows for pre-stored classification experience information, supporting rapid retrieval and matching, improving task processing efficiency and adaptability, and reducing real-time computational overhead.

[0022] In one possible implementation, the method provided in this application further includes: a second network element receiving a service request message from a first network element, the service request message including an intent. This method enables dynamic transmission of task requirements, supports collaborative division of labor among multiple network elements, and enhances system flexibility and responsiveness.

[0023] In one possible implementation, the method provided in this application further includes: a second network element determining first information based on experience information corresponding to the category of the generative network, wherein the first information is determined by experience information matching the intent. The second network element then sends the first information to the first network element. This method combines intent and experience information to accurately filter effective information, avoids redundant data processing, and improves decision-making quality.

[0024] In one possible implementation, the method provided in this application further includes: a second network element determining rule information based on experience information corresponding to the category of generative network. The rule information describes the rules corresponding to the task processing under the category of generative network. The second network element sends the rule information to the first network element. This method unifies rule constraints on task processes, ensures operational compliance and security, and reduces execution risks.

[0025] In one possible implementation, the generative network corresponds to multiple categories. The method provided in this application embodiment further includes: a second network element receiving first rule information from a first network element, the first rule information describing the rules corresponding to processing tasks under at least one category among the multiple categories of the generative network. The second network element generates second rule information based on the categories of the generative network. The second rule information describes the rules corresponding to processing tasks under other categories among the multiple categories of the generative network besides at least one category. The second network element generates rule information based on the first rule information and the second rule information. The second network element sends the rule information to the first network element. This method supports cross-category rule integration, provides composite strategies, and expands the system's adaptability to complex scenarios.

[0026] Thirdly, embodiments of this application provide a communication method applied to a communication device. For example, the communication device may be a third network element, or a component within the communication device (e.g., a processor, circuit, chip, or chip system responsible for communication functions, including but not limited to a modem chip, a baseband chip, a system-on-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip, etc.). Alternatively, the communication device may also be a logic module or software capable of implementing all or part of the functions of a communication device. The method includes: the third network element receiving a generative network category from a first network element. The generative network is used to provide services for tasks corresponding to an intent. The third network element obtains empirical information corresponding to the category of the generative network based on that category.

[0027] In one possible implementation, the third network element obtains the experience information corresponding to the generative network category by retrieving the experience information corresponding to the generative network category from the storage unit. This method allows for precise querying of pre-stored experience information in the storage unit based on the generative network category, significantly improving the retrieval efficiency of experience information.

[0028] In one possible implementation, the third network element retrieving experience information corresponding to the generative network category from the storage unit includes: the third network element sending a first request to the storage unit. The first request includes the generative network category and is used to request experience information corresponding to that category. The third network element then receives the experience information corresponding to the generative network category from the storage unit. This method reduces redundant data transmission and improves resource utilization.

[0029] In one possible implementation, the storage unit stores multiple generative network categories and corresponding experience information for each category. For example, the storage unit could be the private memory of a third network element. This approach allows for pre-stored classification experience information, supporting rapid retrieval and matching, improving task processing efficiency and adaptability, and reducing real-time computational overhead.

[0030] In one possible implementation, the method provided in this application further includes: a third network element receiving information about a first subtask from a first network element, wherein the information about the first subtask is used to indicate multiple subtasks after task decomposition. This method clarifies the division of labor among subtasks, providing structured input for subsequent resource allocation and tool invocation, thereby improving execution efficiency.

[0031] In one possible implementation, the method provided in this application further includes: a third network element determining a first toolset based on experience information corresponding to the category of the generative network, the first toolset being used to provide services for subtasks. This approach optimizes resource allocation by matching the best tools based on experience information, accelerates task execution, and reduces trial-and-error costs.

[0032] In one possible implementation, the method provided in this application further includes: a third network element receiving rule information from a first network element. This achieves dynamic synchronization of rule information, ensures policy consistency among multiple network elements, and avoids rule conflicts or information silos.

[0033] In one possible implementation, the method provided in this application further includes: a third network element determining third rule information corresponding to the category of the generative network in a storage unit based on the category of the generative network. The third network element then determines fourth rule information based on the rule information and the third rule information. This approach integrates global and local rules, balancing uniformity with scenario specificity, and enhances the flexibility and accuracy of rule application.

[0034] In one possible implementation, the method provided in this application further includes: a third network element determining a second toolset based on fourth rule information and empirical information corresponding to the generative network category. The second toolset is used to provide services for subtasks under the fourth rule information. In this way, tool selection is dynamically adjusted under the constraints of rule information, balancing compliance and efficiency, and improving the quality of task completion.

[0035] In one possible implementation, the method provided in this application further includes: a third network element generating a second prompt word based on the fourth rule information, the second prompt word being used to convert the content of the fourth rule information into executable instructions. This method can transform complex rules into concise operation instructions, lowering the execution threshold and improving the efficiency and accuracy of human-machine collaboration.

[0036] Fourthly, embodiments of this application provide a communication device. When the device is a first network element, it can implement the methods in the first aspect or any possible implementation of the first aspect, and therefore can also achieve the beneficial effects of the first aspect or any possible implementation of the first aspect. When the device is a second network element, it can also implement the methods in the second aspect or any possible implementation of the second aspect, and therefore can also achieve the beneficial effects of the second aspect or any possible implementation of the second aspect. When the device is a third network element, it can also implement the methods in the third aspect or any possible implementation of the third aspect, and therefore can also achieve the beneficial effects of the third aspect or any possible implementation of the third aspect.

[0037] Fifthly, embodiments of this application provide a computer-readable storage medium storing a computer program or instructions that, when executed on a computer, cause the computer to perform a communication method as described in the first aspect or any possible implementation thereof, or cause the computer to perform a communication method as described in the second aspect or any possible implementation thereof, or cause the computer to perform a communication method as described in the third aspect or any possible implementation thereof.

[0038] Sixthly, embodiments of this application provide a computer program product including instructions that, when executed on a computer, cause the computer to perform a communication method described in the first aspect or various possible implementations of the first aspect, or cause the computer to perform a communication method described in the second aspect or various possible implementations of the second aspect, or cause the computer to perform a communication method described in the third aspect or various possible implementations of the third aspect.

[0039] In a seventh aspect, embodiments of this application provide a communication device for implementing various methods in various possible designs of any of the first, second, or third aspects described above. The communication device may be the first network element, or a device containing the first network element, or a component (e.g., a chip) applied to the first network element. Alternatively, the communication device may be the second network element, or a device containing the second network element, or a component (e.g., a chip) applied to the second network element. Alternatively, the communication device may be the third network element, or a device containing the third network element, or a component (e.g., a chip) applied to the third network element. The communication device includes modules and units corresponding to the above methods. These modules and units may be implemented in hardware, software, or by hardware executing corresponding software. The hardware or software includes one or more modules or units corresponding to the above functions.

[0040] It should be understood that the communication device described in the seventh aspect above may further include: a bus and a memory, the memory being used to store code and data. Optionally, at least one processor communication interface and the memory are coupled to each other.

[0041] Eighthly, embodiments of this application provide a communication device comprising: at least one processor. The at least one processor is coupled to a memory. When the communication device is in operation, the processor executes computer-executable instructions or programs stored in the memory to cause the communication device to perform a method as described in the first aspect or any of the various possible designs of the first aspect. For example, the communication device may be a first network element or a chip applied to a first network element. Alternatively, the processor executes computer-executable instructions or programs stored in the memory to cause the communication device to perform a method as described in the second aspect or any of the various possible designs of the second aspect. For example, the communication device may be a second network element or a chip applied to a second network element. Alternatively, the processor executes computer-executable instructions or programs stored in the memory to cause the communication device to perform a method as described in the third aspect or any of the various possible designs of the third aspect. For example, the communication device may be a third network element or a chip applied to a third network element.

[0042] It should be understood that the memory described in the eighth aspect can also be replaced by a storage medium, and the embodiments of this application do not limit this.

[0043] In one possible implementation, the memory described in the eighth aspect can be internal to the communication device. Of course, the memory can also be located external to the communication device, but at least one processor can still execute computer execution instructions or programs stored in the memory.

[0044] Ninthly, embodiments of this application provide a communication device, which includes one or more modules for implementing the methods of any one of the first, second, and third aspects described above. The one or more modules may correspond to the various steps in the methods of any one of the first, second, and third aspects described above.

[0045] In a tenth aspect, embodiments of this application provide a chip system including a processor. The processor is configured to read and execute a computer program stored in a memory to execute the methods of the first aspect and any possible implementation thereof, or to execute the methods of the second aspect and any possible implementation thereof, or to execute the methods of the third aspect and any possible implementation thereof. Optionally, the chip system may be a single chip or a chip module composed of multiple chips. Optionally, the chip system may also include a memory, which is connected to the processor via a circuit or wire. Further optionally, the chip system may also include a communication interface. The communication interface is used to communicate with other modules outside the chip.

[0046] Eleventhly, embodiments of this application provide a communication system comprising: a first network element, a second network element, and a third network element. The first network element is used to implement the method in the first aspect or any possible implementation thereof. The second network element is used to implement the method in the second aspect or any possible implementation thereof. The third network element is used to implement the method in the third aspect or any possible implementation thereof.

[0047] Any of the devices, computer storage media, computer program products, chips, or communication systems provided above are used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects of the corresponding solutions in the corresponding methods provided above, and will not be repeated here. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the architecture of a communication system provided in an embodiment of this application;

[0049] Figure 2 This is a schematic diagram of a system architecture for a next-generation core network based on intelligent agents, provided in an embodiment of this application.

[0050] Figure 3 This is a flowchart illustrating a communication method provided in an embodiment of this application;

[0051] Figure 4 This is a schematic diagram illustrating a specific implementation method of a large network model for retrieving data based on the category of the received generative network, as provided in an embodiment of this application.

[0052] Figure 5 This is a schematic diagram illustrating an implementation method for transmitting messages with generative network categories between intelligent agents, as provided in an embodiment of this application.

[0053] Figure 6 This is a schematic diagram illustrating a specific implementation method of a network planning agent generating a subtask list based on rule information, as provided in an embodiment of this application.

[0054] Figure 7 This is a schematic diagram illustrating a specific implementation of a generative network corresponding to multiple categories, as provided in an embodiment of this application.

[0055] Figure 8 This is a schematic diagram of a communication device provided in an embodiment of this application;

[0056] Figure 9 This is a schematic diagram of the hardware structure of a communication device provided in an embodiment of this application;

[0057] Figure 10 This is a schematic diagram of a chip structure provided in an embodiment of this application. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; "and / or" in this text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0059] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0060] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0061] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0062] The character " / " generally indicates that the preceding and following objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any single or multiple items. For example, "at least one of a, b, or c" can be expressed as: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0063] Both "...when" and "if" indicate that a corresponding action will be taken under certain objective circumstances. They are not time limits, nor do they require a judgment action to be taken when the action is taken, nor do they imply any other limitations.

[0064] The steps involved in the communication method provided in this application embodiment are merely examples. Not all steps are mandatory, nor are all contents of each piece of information or message mandatory. They can be added or removed as needed during use.

[0065] In this application, the same step or a step or message with the same function can be referenced and learned from each other in different embodiments.

[0066] The system architecture and business scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new business scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0067] To facilitate understanding of the technical solution of this application, the relevant terms involved in this application are introduced below.

[0068] 1. An intelligent agent is an entity capable of sensing its surrounding environment and making autonomous decisions based on changes in that environment; examples include software programs and physical devices. The specific form of an intelligent agent is not limited in the embodiments of this application.

[0069] 2. Agent-based next-generation core network (A-Core) is a core network architecture that deeply integrates agents with various core network functions. A-Core allows for the flexible orchestration, deployment, execution, and recycling of end-to-end network slices to support emerging services. For example, in A-Core, agents collaborate through intent-driven communication.

[0070] 3. Generative network (GN): A highly efficient and customized service created and managed by multiple agents in the core network, which can serve as a lightweight application layer network instance.

[0071] 4. Agent-based interface (ABI), also known as intelligent interface, is used to represent the declarative interface between an agent and a common component.

[0072] In this system, the intelligent agent only needs to declare the desired goal through the ABI, and the receiver will intelligently understand and execute it without specifying the specific operational instructions for each step required to achieve the goal. The data transmitted through the ABI can be unstructured data containing semantic elements (e.g., described using natural language), or semi-structured data containing relevant field tags used to segment semantic elements.

[0073] like Figure 1 As shown, Figure 1 This is a schematic diagram of the architecture of a communication system provided in an embodiment of this application. The communication system includes: a terminal device, a first network element, a second network element, and a third network element. The first network element, the second network element, and the third network element are connected through an intelligent interface. The terminal device can communicate with the first network element, the second network element, and the third network element through an actor.

[0074] The terminal device is used to provide the user's intent to the first network element; the first network element is used to understand the user's intent and decompose the intent into a series of sub-tasks; the second network element is used to infer or generate a generative network that can provide services for the tasks based on the user's intent; and the third network element is used to select tools that can provide services for the sub-tasks based on the description of the sub-tasks.

[0075] In one possible embodiment, the first network element can be a network planning agent, the second network element can be a network generative pre-trained transformer (NetGPT), and the third network element can be a network assembling agent.

[0076] For example, such as Figure 2 The A-Core system architecture shown is Figure 1 This is a specific implementation of the communication system shown. The A-Core system includes a network planning agent, a large network model, and a network assembly agent. Terminal devices, the radio access network (RAN), the access gateway (A-GW), and the actors are connected, enabling communication between the terminal devices or access network devices and various agents. The large network model communicates with various agents through intelligent interfaces.

[0077] The RAN can establish a radio connection between the terminal equipment and the core network. The A-GW can handle the access requests from the terminal equipment and manage the connection between the terminal equipment and the core network.

[0078] Optional, Figure 2The A-Core system architecture shown may also include: a network connection agent, a network execution agent, a network toolbox, a network sandbox, network public memory, and hosting resources.

[0079] Network connectivity agents are responsible for intelligent connectivity management. For example, intelligent connectivity management includes configuring terminal devices or base station functions, and establishing and updating end-to-end network topology and connections according to service requirements.

[0080] The network-running intelligent agent is responsible for managing computing resources. For example, computing resource management can include the deployment, updating, and deletion of functional instances, as well as the dynamic scheduling of computing resources.

[0081] In addition to communicating with various intelligent agents through intelligent interfaces, the network common memory also connects to terminal devices, access network devices, and host resources. The network common memory is used to collect and store network data, knowledge, etc., for intelligent agents to query and use.

[0082] In this embodiment of the application, each intelligent agent may also have a corresponding private memory, such as... Figure 2 As shown. Both public and private memory in a network can collect and store network data and knowledge. However, private memory refers to the memory within a specific intelligent entity, which typically stores the network data and knowledge collected and stored by that entity. Private memory may or may not be located within the intelligent entity.

[0083] The network toolkit communicates with various intelligent agents through intelligent interfaces. It stores a large amount of information about network and application functions. These functions can be flexibly arranged and assembled by intelligent agents to generate personalized application networks, etc.

[0084] Network sandboxes communicate with various intelligent agents through intelligent interfaces. They are used to verify the reasoning results of intelligent agents and to simulate the operational status and performance of the application network before deployment.

[0085] Host resources include physical or virtual resources that support the operation of the entire network architecture, such as servers, storage devices, and computing resources. Host resources provide fundamental support for the operation of various intelligent agents, other network elements, common components, and generative networks.

[0086] In this application embodiment, the terminal device is a user-side device with wireless transceiver capabilities. It can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted, and can also be a sensor-type device. It can also be deployed on water (such as on ships). Furthermore, it can be deployed in the air (e.g., on airplanes, balloons, and satellites). The terminal device can also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile equipment, user terminal, wireless telecom equipment, user agent, user equipment, or user device. The terminal can be a station (STA) in a wireless local area network (WLAN), a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, an in-vehicle device, a wearable device, or a terminal device in a next-generation communication system (e.g., a fifth-generation (5G) communication network) or a terminal device in a future wireless network, etc.

[0087] In addition, terminal devices can also be wearable devices, which are portable devices worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not just hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are feature-rich, large in size, and can perform complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses; and those that focus on a specific application function and require the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring. Examples include smartwatches, smart bracelets, and pedometers. Wireless terminals in vehicles (e.g., automobiles, bicycles, electric vehicles, airplanes, ships, trains, high-speed trains, etc.), virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, smart home devices (e.g., refrigerators, televisions, air conditioners, electricity meters, etc.), intelligent robots, workshop equipment, wireless terminals in self-driving vehicles, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, or wireless terminals in smart homes, and flying devices (e.g., intelligent robots, hot air balloons, drones, airplanes), etc. In this application, for ease of description, the chip deployed in the above-mentioned devices, such as a system-on-a-chip (SOC), baseband chip, or other chip with communication functions, may also be referred to as a terminal device.

[0088] In the embodiments of this application, the functions of the terminal device can also be performed by modules (such as chips or modems) in the terminal device, or by a device that includes the functions of the terminal device.

[0089] Among current related technologies, the three key technical requirements of 5G networks include: enhanced mobile broadband (eMBB), primarily serving services with high requirements for transmission rates and bandwidth, such as high-definition video streaming and virtual reality applications; massive machine-type communications (mMTC), suitable for scenarios requiring support for a large number of device accesses and frequent transmission of small data packets, such as smart agriculture and smart metering; and ultra-reliable low-latency communications (uRLLC), requiring millisecond-level low latency and high reliability, suitable for demanding scenarios such as industrial automation and telemedicine. Simultaneously, 5G networks, through network slicing technology, transform the traditional "one-size-fits-all" network into a flexible, on-demand service system.

[0090] Network slicing technology primarily provides multiple dedicated, virtualized, and isolated logical networks on the same shared physical infrastructure. Each logical network serves a specific business type or industry user. Each network slice can flexibly define its own logical topology, service level agreement (SLA), reliability, and security levels to meet the differentiated needs of different services or users. Through network slicing technology, operators can allocate different slices according to different business needs. Even for the same service, different slices can be assigned to different users, thereby achieving personalized and efficient services.

[0091] As an example, network slices are identified by Single Network Slice Selection Assistance Information (S-NSSAI). S-NSSAI consists of two parts: Slice / Service Type (SST) and Slice Differentiator (SD). SST characterizes the expected network slice behavior in terms of features and services. For example, the standard SST value ranges from 1 to 3, where 1 represents eMBB, 2 represents uRLLC, and 3 represents massive internet of things (MIoT). SD is supplementary to the slice type and is optional information used to further differentiate multiple network slices sharing the same SST. Through S-NSSAI, the network can accurately identify and manage each slice, ensuring that the resource allocation and quality of service for each slice meet its specific SLA requirements. This fine-grained management and identification mechanism enables network slices to operate efficiently in complex multi-service environments, providing users with customized network services.

[0092] However, with the development of communication technology, especially the deep integration of intelligent agents and large language models (LLMs), a breakthrough upgrade has been achieved for the next generation of agent-based network systems. By embedding LLMs into dynamic network environments, LLM-agents possess four core capabilities: environmental awareness and memory storage, multimodal tool invocation, autonomous task planning, and real-time decision execution. This enables them to independently complete complex tasks in intricate network scenarios and achieve efficient cross-agent collaboration. This collaborative mechanism, through the interaction between LLM-agents with different roles, drives the network system to dynamically generate customized services based on user intent, accurately execute end-to-end processes, and intelligently reclaim idle resources. This technological breakthrough not only significantly improves the autonomous decision-making and adaptive capabilities of communication networks but also achieves highly personalized and elastically scalable services through semantic interaction and rule-driven mechanisms. Currently, 5G network slicing technology mainly meets the needs of high-speed transmission, low latency, and large-scale connectivity services. For example, S-NSSAI is used to manage and identify these connectivity services. Therefore, the service categories targeted by 5G network slicing technology are limited and cannot match the new services such as computing, sensing, and AI provided by next-generation network systems based on intelligent agents, resulting in relatively low efficiency in network generation and management for providing new services.

[0093] Therefore, in the face of diverse network and business needs of users, how to improve the efficiency of network generation and management is a key issue facing the next generation of intelligent agents.

[0094] In this application embodiment, the specific structure of the execution subject of a communication method is not particularly limited, as long as communication can be performed according to the communication method of this application by running a program that records the code of a communication method of this application embodiment. For example, the execution subject of a communication method provided in this application embodiment may be a functional module in a first network element that can call and execute a program, or a communication unit applied in the first network element, such as a chip, chip system, integrated circuit, etc. These chips, chip systems, and integrated circuits may be located inside the first network element or may be independent of the first network element, and this application embodiment does not impose any restrictions. Alternatively, the execution subject of a communication method provided in this application embodiment may be a functional module in a second network element that can call and execute a program, or a communication unit applied in the second network element, such as a chip, chip system, integrated circuit, etc. These chips, chip systems, and integrated circuits may be located inside the second network element or may be independent of the second network element, and this application embodiment does not impose any restrictions. Alternatively, the execution subject of a communication method provided in this application embodiment may be a functional module in a third network element that can call and execute a program, or a communication unit applied in the third network element, such as a chip, chip system, integrated circuit, etc. These chips, chip systems, and integrated circuits can be located inside the third network element or can be independent of the third network element; this application does not impose any restrictions on these aspects.

[0095] like Figure 3 As shown, Figure 3 This illustration shows a flowchart of a communication method provided in an embodiment of this application. The method includes:

[0096] Step 301: The first network element obtains a service request message. The service request message includes an intent, which is used to determine the task to be performed in the request.

[0097] In one possible embodiment, the service request message is a description of a user-initiated network service request, including intent in natural language or structured data form.

[0098] As an example, service request messages include intents in natural language. For instance, a service request message can describe the service a user needs in natural language or other forms.

[0099] For example, a service request message might read, "I want to monitor the number of vehicles at Shanghai Century Park No.1 gate from 8:00 to 17:00 daily." This service request message indicates that the user's intention is to collect statistics on the number of vehicles in a specific area during a specific time period.

[0100] As another example, service request messages may include intents in the form of structured data. Structured data refers to data types with a defined format and fixed fields, whose content is organized according to a predefined model (such as tables, key-value pairs, or database schemas). For instance, a service request message may explicitly indicate the content of the request, the type of request, the status of the request, etc.

[0101] For example, a service request message might include the following: "Request content: Detect traffic flow in a specific area during a specific time period; Request category: Smart City; Request status: In progress." From the request content in this message, we can determine that the user's intent is to collect statistics on traffic flow in a specific area during a specific time period.

[0102] In one possible implementation, the first network element can be a network planning agent, which can receive service request messages from terminal devices through an intelligent interface. For example, a user can send a service request message to the network planning agent through a mobile phone or other terminal device.

[0103] Step 302: The first network element determines the category of the generative network that provides services for the task based on the intent. The category of the generative network is used to determine the empirical information corresponding to the category of the generative network.

[0104] In one possible embodiment, after receiving a service request message, the first network element performs intent parsing to determine the type of generative network.

[0105] As an example, the first network element is a network planning agent. When the network planning agent receives a service request message, it can use natural language processing (NLP) or a predefined template to extract the intent in the service request message.

[0106] For example, a service request message might read, "I want to monitor the number of vehicles at Shanghai Century Park No.1 gate from 8:00 to 17:00 daily." The network planning agent, using NLP, can parse the intent as "to count the traffic flow in a specific area during a specific time period." Alternatively, a service request message might read, "Task ID: 1, Category: Smart City, Request Content: Detect traffic flow in a specific area during a specific time period, Status: In Progress." The network planning agent, based on a predefined template, extracts the intent from the request content, namely, "to count the traffic flow in a specific area during a specific time period."

[0107] Generative networks are categorized using a dynamic classification system based on business scenarios or user intent, using natural language tags. The classification criteria for generative networks are not limited to communication connectivity but integrate resource requirements from communication, computing, sensing, and AI.

[0108] In one possible implementation, the category of a generative network can be represented by a category field. Different category fields are used to represent different categories of generative networks.

[0109] For example, the category field can be directly described using natural language. For instance, a category field of "cat: Smart City" indicates that the generative network belongs to the category of "Smart City".

[0110] For example, generative networks include, but are not limited to: smart cities, smart healthcare, low-altitude economy, online games, AI phones, smart robots, autonomous vehicles, unmanned aerial vehicles, maintenance, and debugging.

[0111] In one possible implementation, different generative networks can be classified in multiple ways. For example, they can be classified based on the different users within each category.

[0112] For example, over-the-top (OTT) industries could be categorized as Smart City, Smart Healthcare, Low-altitude Economy, and Online Games. Subscribers could be categorized as AI Phones, Smart Robots, Autonomous Vehicles, and Unmanned Aerial Vehicles. Operators could be categorized as Maintenance and Debugging.

[0113] The experience information comprises historical data on the generative network's task execution, including: information about the generative network and information about service request messages. The generative network information can be used to indicate the category of the generative network, and the service request message information can be used to indicate the specific content of the service request message corresponding to the category of the generative network.

[0114] Optionally, the experience information may also include one or more of the following: information on task execution, information related to the intelligent interface, information about the agent, information about subtasks, information about functions, and information about the user. Specifically, information on task execution can indicate the tasks performed by the generative network category; information related to the intelligent interface can indicate the messages received or sent by the agent; information about the agent can indicate the specific agent involved in the generation of the generative network; information about subtasks can indicate the specific content of the subtasks after task decomposition; information about functions can indicate the various tools used to execute the tasks; and information about the user can indicate the user's feedback to the generative network.

[0115] As an example, a piece of experience information may include: the category field (cat) of the generative network and the service request message (query) corresponding to the category field of the generative network.

[0116] Optionally, experience information may also include one or more of the following: task ID, intelligent interface information (ABImessages), agent ID, subtask list, service function list (SF list), generative network ID, generative network feedback, and user mark.

[0117] The information includes: a task identifier (a string or numeric code identifying network operations); information from the intelligent interface (messages sent by each agent, each containing the task ID, sendfrom, and next action); an agent identifier (a string or numeric code identifying the agent participating in the task execution); a subtask list (an ordered set of independently executable subtasks); a service function list (a set of service function modules selected and combined according to the task, used to construct the generative network); a generative network identifier (a string or numeric code identifying a generative network); generative network feedback (indicating whether the generative network was successfully generated, with the content being "success" or "failure"); and user ratings (indicating user satisfaction with the services provided by the generative network, presented numerically, such as using a score of 0-100 to reflect user satisfaction, with scores below 60 indicating dissatisfaction and above 90 indicating satisfaction, or simply "satisfied" or "unsatisfied").

[0118] In one possible implementation, empirical information corresponding to the categories of the generative network is stored in the network's common memory.

[0119] It is understandable that the network common memory can store experiential information corresponding to one or more generative network categories.

[0120] As an example, the empirical information corresponding to the categories of a generative network is stored in the network's common memory in the form of a table.

[0121] For example, as shown in Table 1, each row in the table represents one piece of experience information. Taking the first piece of experience information as an example, in this experience information, the task identifier is "101"; the category field of the generative network is "Smart City"; the service request message is "Please help me generate a 3D visualization model of Shanghai Century Park"; the information of the intelligent interface includes the messages sent by each agent, such as one message being "Task ID: xxx, Message Source: xxx, Next Action: xxx, xxx"; agent identifiers include "001, 002, 008, 009"; the subtask list is "subtask list 1"; the service function list is "Func list 1"; the generative network identifier is "001"; the feedback of the generative network is "success"; and the user score is "80".

[0122] Table 1

[0123]

[0124]

[0125] In one possible implementation of this application, the first network element can determine the keywords in the intent based on the intent, and determine the category of the generative network by matching the keywords with predefined category classification rules of generative networks.

[0126] As an example, the predefined category classification rules for generative networks can be a mapping table, which includes the category of each generative network and the keyword corresponding to each category.

[0127] For example, the first network element predefines a mapping table between generative network categories and keywords. For instance, keywords corresponding to "smart city" include "traffic flow," "traffic monitoring," "traffic light optimization," and "navigation service." When the first network element parses the intent of a service request message as "to count the traffic flow in a certain area during a certain time period," it can extract the keyword "traffic flow" from the intent. Therefore, the first network element can determine the category of the generative network providing services for the task as "smart city" based on the mapping table.

[0128] In one possible implementation of this application, the first network element may determine, based on its intent, the category of the generative network providing services for the task, which may be one or more.

[0129] For example, the service request message obtained by First Network Element is: "In the Zhangjiang Software Park, Pudong Shanghai, during the time period from 11:00 to 13:00 daily, multiple drones are needed for food delivery. Please provide reliable wireless connections, real-time environmental information, and navigation assistance services for the drones." Based on this service request message, First Network Element can determine that the user's intention is to use multiple drones for food delivery in Zhangjiang Software Park, Pudong New Area, Shanghai, from 11:00 to 13:00, and request that the drones be provided with reliable wireless connections, real-time environmental information, and navigation assistance services. The first network element can extract keywords from the intent such as "navigation assistance service", "environmental information", "drone", and "food delivery". Based on the mapping relationship table, the first network element determines the category of the generative network that provides services for the task as "smart city" and "drone".

[0130] Step 303: The first network element sends its intent and generative network category to the second network element. Correspondingly, the second network element receives the intent and generative network category from the first network element.

[0131] In one possible implementation, the first network element sends an intelligent interface message to the second network element, the intelligent interface message containing a category field corresponding to the generative network category. Optionally, the intelligent interface message may also include a service request message.

[0132] For example, the first network element is a network planning agent, and the second network element is a large network model. The network planning agent sends intelligent interface messages to the large model. The intelligent interface messages include the field "cat: smart city" corresponding to the generative network category.

[0133] Optionally, the intelligent interface message may also include a service request message such as "I want to monitor the number of vehicles at Shanghai Century Park No.1 gate from 8:00 to 17:00 daily."

[0134] In this application, a first network element receives a service request message initiated by a user. The first network element can dynamically map the intent to a generative network category. The first network element then sends the generative network category to a second network element. This facilitates the second network element in determining the experiential information related to the generative network category. By determining the generative network category in this application, the network agent can more efficiently retrieve experiential information related to the generative network category, thereby generating the required generative network based on the experiential information. This improves the generation efficiency of the generative network and ultimately provides users with differentiated services beyond connectivity.

[0135] In one possible embodiment of this application, the method provided in this application further includes: the second network element obtaining empirical information corresponding to the category of the generative network based on the category of the generative network.

[0136] In one possible embodiment of this application, the second network element obtains empirical information corresponding to the category of the generative network from the first storage unit according to the category of the generative network.

[0137] The first storage unit can be the network common memory.

[0138] In one possible embodiment of this application, the second network element sends a first request to the first storage unit. Correspondingly, the first storage unit receives the first request from the second network element. The first request includes a category of the generative network, and is used to request experience information corresponding to that category. After retrieving the experience information corresponding to the category of the generative network based on the category, the first storage unit sends the experience information corresponding to the category of the generative network to the second network element. Correspondingly, the second network element receives the experience information corresponding to the category of the generative network from the first storage unit.

[0139] For example, the second network element is the network big model. The network big model receives a generative network of the category "Smart City". The network big model sends a first request to the network common memory, which includes the category of the generative network. Since multiple pieces of experience information are stored in the network common memory, referring to Table 1, the network common memory retrieves experience information that meets the criteria of "Smart City", such as the first and second pieces of experience information in Table 1. The network common memory sends experience information related to "Smart City" to the network big model, such as the first and second pieces of experience information in Table 1.

[0140] Optionally, the method provided in this application embodiment may further include: the second network element determining the first information based on empirical information corresponding to the category of the generative network.

[0141] In one possible embodiment of this application, the method provided in this application further includes: the second network element sending first information to the first network element. Correspondingly, the first network element receives the first information from the second network element.

[0142] The first piece of information is determined by empirical information corresponding to the category of the generative network.

[0143] In one possible embodiment, the second network element determines that there are multiple pieces of empirical information corresponding to the category of the generative network, and then the second network element summarizes the multiple pieces of empirical information to generate the first information.

[0144] For example, the second network element is the network big model. The network big model receives a generative network of the category "smart city" and retrieves experience information corresponding to "smart city" from the network common memory. Referring to Table 1 in the above embodiment, there are two pieces of experience information corresponding to "smart city," namely, two pieces of experience information corresponding to the category "smart city" for task identifier 101 and task identifier 103, respectively. The network big model summarizes the two pieces of experience information to generate the first piece of information.

[0145] As an example, a large network model will extract the common features of two pieces of empirical information to generate the first piece of information.

[0146] For example, the content of the first piece of experience information is described in the above embodiment and will not be repeated here. The second piece of experience information includes: the task identifier is "103", the category field of the generative network is "smart city", the service request message is "In Zhangjiang Software Park, Pudong, Shanghai, from 11:00 to 13:00, multiple drones are needed to deliver food. Please provide the drones with reliable wireless connection, real-time environmental information and navigation assistance services.", the intelligent interface information includes "task ID: xxx, message source: xxx, next action: xxx, xxx", the intelligent agent identifier includes "001, 003, 006, 010", the subtask list is "subtask list 3", the service function list is "service function list 3", the generative network identifier is "003", the generative network feedback is "success", and the user score is "98". After summarizing two pieces of empirical information, the first piece of information generated by the large network model includes: the category field of the generative network is "smart city", the keyword in the service request message is "location information", the ABI information includes "the starting position of the task identifier, etc.", the agent identifier includes "001, 002, 003, 006, 008, 009, 010", the subtask list is "list a" (the subtasks in list a are the common subtasks in list 1 and list 3), and the service function list is "service function list a" (the functions in service function list a are the common functions in service function list 1 and service function list 3), etc.

[0147] Optionally, the first piece of information may also include all the content from multiple pieces of experience information.

[0148] In one possible embodiment of this application, the first information is determined by empirical information that matches the intent.

[0149] Among them, the experience information that matches the intent refers to the service request message in the experience information whose similarity to the service request message corresponding to the intent is greater than or equal to a preset threshold.

[0150] For example, in experience information 1, the service request message is "Statistics on traffic flow in area A on April 12, 2025", and the service request message corresponding to the intent is "Statistics on traffic flow in area B from 9:00 to 12:00". Both service request messages are related to traffic statistics, so it can be determined that the experience information matches the intent.

[0151] In one possible embodiment, there are multiple pieces of experience information matching the generative network category. Among these multiple pieces of experience information, some service request messages do not match the service request message corresponding to the intent received by the second network element. Therefore, after determining multiple pieces of experience information based on the generative network category, the second network element further determines experience information matching the intent from these multiple pieces of experience information. The second network element determines the first information based on the experience information matching the intent.

[0152] For example, the second network element is the network big model. The network big model receives a generative network category of "smart city" from the network planning agent. Based on "smart city," the network big model searches the network common memory and identifies three pieces of experience information: experience information 1, experience information 2, and experience information 3. Based on the service request messages in these three pieces of experience information, the network big model determines that the similarity between the service request messages in experience information 1 and experience information 3 and the service request message corresponding to the intent received by the network big model reaches a preset threshold. The network big model summarizes experience information 1 and experience information 3 to generate the first piece of information.

[0153] In one possible embodiment of this application, the method provided further includes: a first network element determining information about a first subtask based on first information. The information about the first subtask is used to indicate multiple subtasks after task decomposition. The first network element sends the information about the first subtask to a third network element. Correspondingly, the third network element receives the information about the first subtask from the first network element.

[0154] The information for the first subtask includes the content of the first subtask.

[0155] In one possible implementation, the information for the first subtask can be in the form of a list of subtasks. For example, the list might include identifiers for multiple first subtasks, as well as the content of each first subtask. A list of subtasks can also be called a subtask chain.

[0156] In one possible implementation, the first network element determines the information of the first subtask based on the intent and in conjunction with the first information.

[0157] As an example, the first network element determines the task to be completed based on the intent and breaks down the task into multiple sub-tasks, which can be referenced from the sub-task list in the first information.

[0158] For example, the first network element is a network planning agent. The service request message received by the network planning agent is "Statistics on traffic flow at Gate 1 of Shanghai Century Park between 8:00 and 17:00". After interpreting the intent based on the service request message, the task to be completed is "Generate a statistical report on traffic flow at Gate 1 of Shanghai Century Park between 8:00 and 17:00". The network planning agent decomposes this task into multiple sub-tasks and generates a list of sub-tasks. The sub-tasks can refer to the sub-task list in the first information. For example, the sub-task list in the first information includes: data collection and traffic statistics. Therefore, the sub-tasks generated by the network planning agent can also include data collection and traffic statistics. The subtask list generated by the network planning agent may include: obtaining the geographical location (i.e., Gate 1 of Shanghai Century Park), obtaining the device identifiers for the collected data (i.e., cameras near Gate 1 of Shanghai Century Park, etc.), data collection (i.e., obtaining camera and sensor data from Gate 1 of Shanghai Century Park between 8:00 and 17:00), data cleaning (i.e., filtering out invalid data from the acquired data), traffic statistics (i.e., calculating the real-time number of vehicles), and report generation (i.e., generating a visual statistical report and pushing it to the management platform).

[0159] In one possible embodiment of this application, the method provided by this application further includes: a first network element sending the generative network category to a third network element. Correspondingly, the third network element receives the generative network category from the first network element.

[0160] In one possible implementation of this application, the first network element sends an intelligent interface message to the third network element. The intelligent interface message includes information about the first subtask and the type of generative network.

[0161] For example, the first network element is a network planning agent, and the third network element is a network assembly agent. The network planning agent sends an intelligent interface message to the network assembly agent. The intelligent interface message includes: a subtask list X (including multiple subtasks) and a category field "cat: Smart City" corresponding to the generative network category.

[0162] In one possible embodiment of this application, after the third network element receives the generative network category from the first network element, the method provided in this application embodiment further includes:

[0163] Step 1: The third network element obtains the empirical information corresponding to the category of the generative network.

[0164] In one possible embodiment of this application, the third network element obtains empirical information corresponding to the category of the generative network from the second storage unit according to the category of the generative network.

[0165] In one possible embodiment of this application, a third network element sends a first request to a second storage unit. Correspondingly, the second storage unit receives the first request from the third network element. The first request includes a category of the generative network, and is used to request experience information corresponding to that category. After retrieving the experience information corresponding to the category of the generative network, the second storage unit sends the experience information corresponding to the category of the generative network to the third network element. Correspondingly, the third network element receives the experience information corresponding to the category of the generative network from the second storage unit.

[0166] In one possible implementation, the second storage unit is memory within the third network element, such as private memory. The private memory stores empirical information corresponding to one or more generative network categories. It is understood that the second storage unit can also be located outside the third network element.

[0167] In one possible implementation of this application, the specific implementation method of the third network element determining the empirical information according to the category of the generative network is the same as the specific implementation method of the second network element determining the empirical information according to the category of the generative network, and will not be repeated here.

[0168] The content of the experiential information stored in the private memory and the network public memory can be the same or different.

[0169] For example, the experience information stored in the network public memory includes: task ID, generative network category field (cat), service request message (query), intelligent interface information (ABI messages), agent ID, subtask list, generative network ID (GN ID), generative network feedback (GNfeedback), and user mark, as shown in Table 1 of the above embodiments. The experience information stored in the private memory includes: task ID, generative network category field (cat), ABI information (ABImessages), subtask list, and service function list (SF list).

[0170] For example, as shown in Table 2, each row in the table represents one piece of experience information in the private memory. Taking the first piece of experience information as an example, in this experience information, the task identifier is "101"; the category field of the generative network is "Smart City"; the information of the intelligent interface includes the messages sent by each intelligent agent, such as one message being "Task ID: xxx, Message Source: xxx, Next Action: xxx, xxx"; the subtask list is "Subtask List 1"; and the service function list is "Service Function List 1".

[0171] Table 2

[0172]

[0173]

[0174] It is worth noting that the experience information stored in the network's public memory may differ from the experience information stored in the private memory.

[0175] As an example, the experience information stored in the private memory comes from intelligent interface messages sent and received by the agent to which the private memory belongs, while the experience information stored in the network public memory comes from intelligent interface messages received by multiple agents.

[0176] For example, after a terminal device sends an intent to the first network element, four agents participate in the generation of a generative network to provide customized services for the task indicated by the intent. These four agents are Agent 1, Agent 2, Agent 3, and Agent 4. The execution order among the four agents is Agent 1 to Agent 2, Agent 2 to Agent 3, and Agent 3 to Agent 4. Therefore, the intelligent interface messages are also sent by Agent 1 to Agent 2, Agent 2 to Agent 3, and Agent 3 to Agent 4. The network common memory can store the intelligent interface messages sent by each of the four agents and store them as experience information. The private memory inside each agent can receive the intelligent interface messages sent by the previous agent. For example, Agent 2 only receives the intelligent interface messages sent by Agent 1. Therefore, the private memory inside Agent 2 stores the intelligent interface messages sent by Agent 1 and the intelligent interface messages that Agent 2 is about to send.

[0177] Step 2: The third network element determines the information of the first tool based on the empirical information corresponding to the category of the generative network.

[0178] The first tool is a reusable functional module or service component used to provide services for subtasks. The information of the first tool includes its location, identifier, etc.

[0179] In one possible implementation of this application, the first tool may be stored in a network toolbox.

[0180] In one possible implementation of this application, the third network element selects a first tool from the network toolbox based on the information of the first subtask and combined with empirical information corresponding to the category of the generative network.

[0181] For example, the third network element is the network assembly agent. Based on subtask list 1, the network assembly agent determines that subtask 1 is to obtain the geographic location, subtask 2 is to obtain the terminal devices and base station identifiers near the geographic location, subtask 3 is data collection, subtask 4 is data cleaning, subtask 5 is traffic statistics, and subtask 6 is report generation. Based on retrieved experience information corresponding to the generative network category, the network assembly agent determines the tools to use, including GPS tools, map application programming interfaces (APIs), operator network data interfaces, camera APIs, sensor data collectors, data filtering tools, real-time analytics models, basic counting algorithms, general tools (report generator), and a municipal platform-specific interface (city dashboard API). Based on this, the network assembly agent can select the map application programming interface in the network toolbox to obtain the geographic location; select the operator network data interface to obtain the terminal devices and base station identifiers near the geographic location; select the camera interface (camera_API) to collect data; select the data filtering tool to filter the collected data; select the real-time analysis model (traffic_analytics_model) to perform statistics on the filtered data; and select the general tool (report_generator) to generate a report based on the statistical data.

[0182] In one possible embodiment of this application, the method provided by this application embodiment further includes: a first network element acquiring rule information.

[0183] The rule information describes the rules for processing tasks under the category of generative networks.

[0184] In one possible embodiment of this application, the rule information includes: content that is allowed and / or prohibited for the first network element to perform under the category of generative network.

[0185] As an example, when the generative network is categorized as "Smart City," the rule information may include: 1) If the relationship between the requesting user and the ownership of the requested content cannot be confirmed, you need to call the authentication module to query this relationship; 2) Obtain the coordinates of the relevant area and select the device used for sensing information; 3) It should not have an adverse impact on other users or the environment; etc.

[0186] As another example, when the category of generative networks is "operation and maintenance", the rule information may include: 1) providing task chains and solutions related to network operation, maintenance and debugging; 2) not disrupting network infrastructure, such as causing service interruptions, quality degradation or endangering network security; etc.

[0187] In one possible embodiment of this application, the method provided by this application further includes: a first network element generating a first prompt word based on rule information. The first prompt word is used to convert the content of the rule information into executable instructions.

[0188] In one possible implementation of this application, the first network element generates a first prompt word based on rule information, and the first prompt word is used as input into the larger model within the first network element. When the first network element generates information for the first sub-task according to the intent, the first network element can generate the information for the first sub-task by following the rule information.

[0189] For example, the first network element is a network planning agent. Based on the user's intent, the network planning agent determines the generative network category to be "Smart City," and the user's intent is "to collect traffic flow statistics at a certain location." When the network planning agent receives the rule information corresponding to "Smart City," based on the first rule—"If the relationship between the requesting user and the ownership of the requested content cannot be confirmed, you need to call the identity verification module to query this relationship"—it needs to authenticate the user's identity. If the user has permission to access the monitored data, a generative network is determined to meet the user's request; if the user does not have permission to access the monitored data, the user's request is rejected. The first subtask generated by the network planning agent based on the rule information is "to authenticate the user's identity."

[0190] In one possible embodiment of this application, the first network element obtaining rule information includes: the first network element receiving rule information from the second network element. Correspondingly, the second network element sends rule information to the first network element.

[0191] In one possible embodiment, the second network element retrieves experience information based on the category of generative network, and then generates rule information based on the experience information.

[0192] For example, the second network element is a network assembly agent. The network assembly agent receives a generative network classified as "smart city." Based on "smart city," the agent retrieves multiple pieces of experience information, each containing rule information corresponding to its respective service request message. The agent then generates rule information that matches the current service request message based on the rule information in each piece of experience information.

[0193] In one possible embodiment of this application, the generative network corresponds to multiple categories, and the rule information obtained by the first network element includes:

[0194] Step 1: The first network element determines the first rule information based on at least one of the multiple categories corresponding to the generative network. The first rule information is used to process the rules corresponding to the task under at least one of the multiple categories of the generative network.

[0195] In one possible embodiment, the first network element retrieves experience information from its private memory according to the type of generative network, and generates first rule information based on the retrieved experience information.

[0196] As an example, the first network element determines the generative network category, including Category 1 and Category 2, based on its intent. The first network element retrieves experience information corresponding to Category 1 and Category 2 respectively from its private memory. When the first network element retrieves experience information for Category 1, it generates first rule information based on that experience information; or, when the first network element retrieves experience information for Category 2, it generates first rule information based on that experience information.

[0197] It is worth noting that when the first network element retrieves experience information of category 1 and category 2 in its private memory, it can generate first rule information corresponding to category 1 and first rule information corresponding to category 2 respectively, and then summarize and merge the two first rule information.

[0198] For example, if the first network element is a network planning agent, and the received service request message is "In Zhangjiang Software Park, Pudong New Area, Shanghai, from 11:00 to 13:00, multiple drones are needed for food delivery. Please provide the drones with reliable wireless connectivity, real-time environmental information, and navigation assistance services," then the network planning agent determines the generative network category as "Smart City" and "Intelligent Robot." Based on "Smart City" and "Intelligent Robot," the network planning agent retrieves experience information from its private memory. After retrieving the experience information corresponding to "Smart City," the network planning agent generates first rule information based on this experience information.

[0199] Step 2: The first network element sends the first rule information to the second network element. Correspondingly, the second network element receives the first rule information from the first network element.

[0200] The first rule information is used to determine the rule information.

[0201] In one possible implementation, the first network element sends a request message to the second network element. This request message requests rule information for other categories among multiple categories of the generative network from the second network element. The request message includes the first rule information and the category of the generative network.

[0202] For example, the first network element is a network planning agent, and the second network element is a network assembly agent. The network planning agent sends a request message to the network assembly agent. The request message includes the first rule information corresponding to "Smart City" and the category field "cat: Smart City & Smart Robots" corresponding to the generative network category.

[0203] Step 3: The second network element generates second rule information based on the type of generative network.

[0204] The second rule information is used to process the rules corresponding to the task in the categories other than at least one of the multiple categories of the generative network.

[0205] As an example, when the second network element receives the first rule information corresponding to category 1, the second network element retrieves the experience information corresponding to category 2 from the network common memory according to the categories of the generative network, namely category 1 and category 2. The second network element then generates the second rule information based on the experience information corresponding to category 2.

[0206] For example, the first network element is a network planning agent, and the second network element is a network assembly agent. The network assembly agent receives the first rule information corresponding to "Smart City", and according to the category field "cat: Smart City & Smart Robots" corresponding to the generative network category, retrieves the experience information corresponding to "Smart Robots", and generates the second rule information.

[0207] It is worth noting that the second network element can also simultaneously retrieve the experience information corresponding to Category 1 and Category 2, generate the first rule information corresponding to Category 1 and the first rule information corresponding to Category 2, and summarize and merge the two first rule information.

[0208] Step 4: The second network element generates rule information based on the first rule information and the second rule information.

[0209] In one possible implementation, the first rule information is the rule information corresponding to at least one of the multiple categories corresponding to the generative network, and the second rule information is the rule information corresponding to the other categories of the multiple categories of the generative network besides at least one category.

[0210] In another possible implementation, the first rule information is the rule information corresponding to all categories of the multiple categories corresponding to the generative network, and the second rule information is also the rule information corresponding to all categories of the multiple categories of the generative network. Alternatively, the first rule information is the rule information corresponding to at least one category of the multiple categories corresponding to the generative network, and the second rule information is the rule information corresponding to all categories of the multiple categories of the generative network.

[0211] Step 5: The second network element sends rule information to the first network element. Correspondingly, the first network element receives the rule information from the second network element.

[0212] In one possible embodiment of this application, the method provided further includes: a first network element determining information about a second subtask based on first information and rule information; the first network element sending the information about the second subtask to a third network element; and correspondingly, the third network element receiving the information about the second subtask from the first network element.

[0213] The information for the second subtask is used to indicate the multiple subtasks after the task is decomposed according to the rule information, such as a subtask list.

[0214] In one possible implementation, the first network element generates a preliminary subtask list based on the subtask list summarized in the first information, and then further constrains the subtask list according to the rule information to finally generate a subtask list.

[0215] As an example, the first network element generates subtask list 1 based on the subtask list summarized in the first information. Subtask list 1 includes: subtask 1, subtask 2, subtask 3, and subtask 4. The network planning agent then further constrains subtask list 1 based on rule information, finally obtaining subtask list 2, which includes: subtask 0, subtask 1, subtask 2, subtask 3, and subtask 4.

[0216] For example, the first network element is the network planning agent, and the third network element is the network assembly agent. Taking the service request message "Statistics on traffic flow at Gate 1 of Shanghai Century Park between 8:00 and 17:00" as an example, the network planning agent can generate Subtask List 1 based on the subtask list summarized in the first information. The network planning agent then further constrains Subtask List 1 based on rule information. Based on the first rule information, "If the relationship between the requesting user and the ownership of the requested content cannot be confirmed, you need to call the identity verification module to query this relationship," the network planning agent needs to identify and authenticate the user's identity. Therefore, the first subtask generated by the network planning agent based on the rule information is "Identify and authenticate the user's identity." If the user has access rights, subsequent subtasks include: obtaining the geographical coordinates "Gate 1 of Shanghai Century Park"; obtaining the identifiers of nearby terminal devices and base stations near these coordinates; configuring the terminal devices and base stations with specified identifiers to collect sensing data according to the start and end times input by the user; analyzing the sensing data and counting the number of vehicles in the area, etc. If the user does not have access rights, subsequent subtasks include: Please send me a "No permission" prompt message.

[0217] In one possible embodiment of this application, the method provided by this application further includes: a first network element sending rule information and the category of the generative network to a third network element. Correspondingly, the third network element receives the rule information and the category of the generative network from the first network element.

[0218] In one possible implementation of this application, the first network element sends an intelligent interface message to the third network element, the intelligent interface message including rule information and the category of generative network.

[0219] For example, the first network element is a network planning agent, and the third network element is a network assembly agent. The network planning agent sends an intelligent interface message to the network assembly agent. The intelligent interface message includes information about the second subtask (e.g., a list of subtasks), rule information (e.g., multiple rules), and a category field corresponding to the generative network category (e.g., "cat: Smart City & Smart Robots").

[0220] In one possible embodiment of this application, after the third network element receives the rule information and the category of the generative network from the first network element, the method provided in this application embodiment further includes:

[0221] Step 1: The third network element retrieves the third rule information corresponding to the generative network category from the second storage unit according to the generative network category.

[0222] The second storage unit can be a private memory within the third network element.

[0223] In one possible embodiment, the third network element can retrieve the empirical information corresponding to the category of the generative network from the private memory corresponding to the third network element, and generate third rule information corresponding to the category of the generative network.

[0224] Step 2: The third network element determines the fourth rule information based on the rule information and the third rule information.

[0225] Understandably, the third network element can retrieve the corresponding experience information of the generative network category from its private memory, and generate third rule information based on the experience information to further improve the received rule information.

[0226] In one possible embodiment of this application, the method provided by this application further includes: a third network element determining a second toolset based on fourth rule information and empirical information corresponding to the category of generative network, the second toolset being used to provide services for subtasks under the fourth rule information.

[0227] In one possible implementation of this application, the third network element selects the second tool from the network toolbox based on the fourth rule information and the information of the second subtask, combined with the experience information corresponding to the category of the generative network.

[0228] For example, the third network element is a network assembly agent. Based on subtask list 2, the network assembly agent determines the following subtasks: subtask 1 is user identification and authentication; subtask 2 is obtaining geographical location; subtask 3 is obtaining the identifiers of nearby terminal devices and base stations; subtask 4 is data collection; subtask 5 is data cleaning; subtask 6 is traffic statistics; and subtask 7 is report generation. Based on retrieved experience information corresponding to the generative network category, the network assembly agent determines the tool to use is "authentication protocols and standards" to implement third-party authorization and single sign-on. For example, users log in via account to authenticate whether they have permissions. The tool selection for the remaining subtasks is the same as in the above embodiment and will not be repeated here.

[0229] In one possible embodiment of this application, the method provided by this application further includes: a third network element generating a second prompt word based on fourth rule information. The second prompt word is used to convert the content of the fourth rule information into executable instructions.

[0230] For specific implementation details, please refer to the implementation method of the first prompt word in the above embodiments, which will not be repeated here.

[0231] The following describes the specific implementation of the communication method provided in the embodiments of this application, taking the first network element as the network planning agent, the second network element as the network big model, and the third network element as the network assembly agent as an example.

[0232] like Figure 4 As shown, Figure 4 This illustration shows a specific implementation of a network large model retrieval method based on the category of a received generative network, according to an embodiment of this application. The method includes:

[0233] Step 401: The terminal device sends a service request message to the network planning agent. Correspondingly, the network planning agent receives the service request message from the terminal device.

[0234] The service request message includes an intent, which is used to determine the task that the user requests to perform.

[0235] For example, the content of the service request message query is "I want to monitor the number of vehicles at Shanghai Century Park No.1 gate from 8:00 to 17:00 daily.", which includes the intent that the user wants to count the number of vehicles at Shanghai Century Park No.1 gate between 8:00 and 17:00.

[0236] Step 402: The network planning agent determines the type of generative network based on the service request message.

[0237] In one possible implementation, the network planning agent, based on LLM, parses the intent in the service request message to obtain the intent's feature information. The network planning agent then matches this feature information with the category of the generative network.

[0238] As an example, a network planning agent can pre-define a category library for generative networks. For instance, the category library could include categories such as smart cities and smart healthcare. Specifically, the characteristic information corresponding to smart cities includes traffic monitoring and real-time data, while the characteristic information corresponding to smart healthcare includes remote diagnosis and health monitoring.

[0239] For example, by analyzing the "traffic flow statistics at Gate 1 of Shanghai Century Park from 8:00 to 17:00," the network planning agent can obtain characteristic information such as traffic flow, public areas, and real-time data. The agent then matches this characteristic information with feature information in the generative network's category library, thus classifying the generative network as a smart city.

[0240] Step 403: The network planning agent sends a service request message and a category field corresponding to the category of the generative network to the large network model. Correspondingly, the large network model receives the service request message and the category field corresponding to the category of the generative network from the network planning agent.

[0241] For example, the category field corresponding to the category of the generative network contains the category of the generative network as described in natural language.

[0242] In one possible implementation, the network planning agent sends intelligent interface messages to the large network model via an intelligent interface. These intelligent interface messages include a service request message and a category field corresponding to the category of the generative network.

[0243] For example, the network planning agent sends a service request message query to the large network model along with the category field "cat: Smart City" corresponding to the category of the generative network.

[0244] Step 404: The network big model retrieves experience information from the network common memory based on the category field corresponding to the service request message and the category of the generative network.

[0245] For example, the network big model receives the category field "cat: SmartCity" corresponding to the generative network's category. Based on the empirical information stored in the network's common memory as shown in Table 1, the network big model can determine that the two empirical information entries with task identifiers 101 and 103 belong to the same category, and summarize based on these two empirical information entries. For instance, the network big model uses these two empirical information entries to determine which agents and subtasks participate in the generative network's generation.

[0246] Optionally, the large network model can also store the received service request messages and the category fields corresponding to the generative network's categories in the network's common memory to form new empirical information.

[0247] Step 405: The large network model sends the first information to the network planning agent. Correspondingly, the network planning agent receives the first information from the large network model.

[0248] The first piece of information is determined by empirical information. For example, a large network model summarizes multiple pieces of empirical information retrieved to generate the first piece of information.

[0249] In this embodiment, the intelligent interface message carries a category field, enabling agents participating in the generation or inference of the generative network to explicitly identify the category of the current generative network. Furthermore, the intelligent interface message is stored in the network's common memory, which can be categorized and managed according to the content of the category field corresponding to the category of the generative network. This allows the large network model to retrieve only relevant experience information based on the current generative network's category during retrieval, thereby narrowing the search scope and improving efficiency.

[0250] like Figure 5 As shown, Figure 5 This is a schematic diagram illustrating an implementation method for transmitting messages with generative network categories between intelligent agents, as provided in an embodiment of this application. The method includes:

[0251] Steps 501 to 505 are the same as steps 401 to 405 above, and will not be repeated here.

[0252] Step 506: The network planning agent generates a list of subtasks based on the first information.

[0253] For example, the service request message requests a task to count the traffic flow at Gate 1 of Shanghai Century Park between 8:00 and 17:00. The network planning agent breaks this task down into multiple sub-tasks and generates a list of sub-tasks. The list of sub-tasks includes: data collection (acquiring camera and sensor data from Gate 1 of Shanghai Century Park between 8:00 and 17:00), data cleaning (filtering invalid data), traffic flow statistics (calculating the real-time number of vehicles), and report generation (generating a visual statistical report and pushing it to the management platform).

[0254] Step 507: The network planning agent sends a list of subtasks and a category field corresponding to the category of the generative network to the network assembly agent. Correspondingly, the network assembly agent receives the subtask chain and the category field corresponding to the category of the generative network from the network planning agent.

[0255] In one possible implementation, the network planning agent sends an intelligent interface message to the network assembly agent, which contains a category field corresponding to the category of the generative network.

[0256] For example, the intelligent interface message includes a list of subtasks and a category field `cat` corresponding to the generative network category. The subtask list includes: Subtask 1: Obtain geographic location; Subtask 2: Obtain terminal device and base station identifiers near the geographic location; Subtask 3: Perform data collection; Subtask 4: Perform data cleaning; Subtask 5: Perform traffic statistics; Subtask 6: Generate a traffic statistics report. The category field is "cat: Smart City".

[0257] Step 508: The network assembly agent retrieves experience information from its private memory based on the subtask list and the category field corresponding to the generative network category, and selects tools based on the experience information.

[0258] In one possible embodiment, the private memory stores multiple pieces of experience information, each of which may include a task ID, a category field (cat), a service request message (query), information from the intelligent interface (ABImessages), a subtask list, and a service function list (SF list).

[0259] For example, as shown in Table 2, each row represents one piece of experience information. The specific content is described in the above embodiment and will not be repeated here.

[0260] In one possible implementation, the network assembling agent can select tools from the network toolbox to complete subtasks.

[0261] For example, subtask 1 is to obtain the geographic location. The network assembly agent can select the map application programming interface (API) from the network toolbox to obtain the geographic location. Subtask 2 is to obtain the identifiers of terminal devices and base stations near the geographic location. The network assembly agent can select the operator network data interface from the network toolbox to obtain the identifiers of terminal devices and base stations. Subtask 3 is data collection. The network assembly agent can select the camera interface or sensor data collector from the network toolbox to collect data. Subtask 4 is data cleaning. The network assembly agent can select the data filtering tool from the network toolbox to filter the collected data. Subtask 5 is traffic statistics. The network assembly agent can select the real-time analysis model or basic algorithm from the network toolbox to perform statistics on the filtered data. Subtask 6 is to generate a traffic report. The network assembly agent can select a general tool or a municipal platform-specific interface from the network toolbox to generate a report based on the statistical data.

[0262] In this embodiment, the agent explicitly transmits the categories of the generative network during interaction, and the agent's private memory stores and retrieves information according to these categories. This classification and storage mechanism accelerates the retrieval of experiential information, optimizes the agent's decision-making process, and ensures more efficient resource sharing and task execution.

[0263] like Figure 6 As shown, Figure 6 This is a schematic diagram illustrating a specific implementation method for a network planning agent to generate a subtask list based on rule information, as provided in an embodiment of this application. The specific method includes:

[0264] Steps 601 to 603 are the same as steps 401 to 403 in the above embodiments, and will not be repeated here.

[0265] Step 604: The large network model retrieves empirical information from the network common memory based on the category field corresponding to the category of the generative network, and generates rule information corresponding to the category of the generative network based on the empirical information.

[0266] For example, the rule information generated by the network big data model corresponding to "smart city" includes: a) If the relationship between the requesting user and the ownership of the requested content cannot be confirmed, you need to call the authentication module to query this relationship; b) Obtain the coordinates of the relevant area and select the device used for sensing information; c) It should not have an adverse impact on other users or the environment, etc.

[0267] Step 605: The large network model sends the first information and rule information to the network planning agent. Correspondingly, the network planning agent receives the first information and rule information from the large network model.

[0268] In one possible implementation, the large network model sends feedback information to the network planning agent, which includes both experiential and rule-based information.

[0269] Step 606: The network planning agent generates a list of subtasks based on the first information and the rule information.

[0270] In one possible implementation, the network planning agent generates prompts based on rule information and inputs these prompts into its own large model, so that the subtask list generated by the network planning agent conforms to the rule information.

[0271] For example, the network planning agent determines Subtask List 1 based on the list of used subtasks summarized in the first information. The agent then adds or removes subtasks from Subtask List 1 based on rule information. For instance, the first rule might be: "If the relationship between the requesting user and the ownership of the requested content cannot be confirmed, you need to call the authentication module to query this relationship." If a user hasn't identified themselves but wants to collect traffic flow data from Gate 1 of Shanghai Century Park from 8:00 AM to 5:00 PM, then based on the first rule, if the user's identity is uncertain, authentication is required. Therefore, the network planning agent needs to add a subtask to Subtask List 1: "Identify and authenticate the user's identity." If the user's authentication is successful (i.e., the user has permission to access the monitoring data), a generative network is determined to satisfy the user's request; if the user's authentication fails (i.e., the user does not have permission to access the monitoring data), the network planning agent rejects the user's request.

[0272] Step 607: The network planning agent sends a list of subtasks, the category field corresponding to the category of the generative network, and rule information to the network assembly agent. Correspondingly, the network assembly agent receives the list of subtasks, the category field corresponding to the category of the generative network, and rule information from the network planning agent.

[0273] In one possible implementation, the network planning agent sends instruction information to the network assembly agent, which includes a list of subtasks, a category field corresponding to the category of the generative network, and rule information.

[0274] Step 608: The network assembly agent selects tools based on the subtask list and rule information.

[0275] For example, the first subtask in the subtask list is to identify and authenticate the user's identity. The network assembly agent selects the identity verification module in the network toolbox to authenticate the user's identity.

[0276] In this embodiment, the large network model generates corresponding rule information based on the different generative network categories and transmits it between agents. The agents combine the received rule information to generate prompts and follow these specific rules during reasoning, thereby standardizing the agents' reasoning process. This mechanism ensures the standardization and reliability of the agents when performing tasks, improving the overall system's stability and security.

[0277] like Figure 7 As shown, Figure 7 This is a schematic diagram illustrating a specific implementation of a generative network corresponding to multiple categories, as provided in an embodiment of this application. The specific method includes:

[0278] Step 701: The terminal device sends a service request message to the network planning agent. Correspondingly, the network planning agent receives the service request message from the terminal device.

[0279] For example, the service request message Query reads: "In the Zhangjiang Software Park, Pudong Shanghai, during the time period from 11:00 to 13:00 daily, multiple drones are needed for food delivery. Please provide reliable wireless connections, real-time environmental information, and navigation assistance services for the drones." This means the intention is to provide reliable wireless connections, real-time environmental information, and navigation assistance services for the drones in the Zhangjiang Software Park, Pudong New Area, Shanghai, between 11:00 and 13:00.

[0280] Step 702: The network planning agent determines the category of the generative network based on the service request message and generates the first rule information.

[0281] The generative network corresponds to multiple categories. The first rule information is the rule information corresponding to at least one of the multiple categories that the generative network corresponds to.

[0282] For example, by analyzing the scenario "In Zhangjiang Software Park, Pudong New Area, Shanghai, multiple drones are needed for food delivery between 11:00 and 13:00. Please provide the drones with reliable wireless connectivity, real-time environmental information, and navigation assistance services," the network planning agent can obtain characteristic information such as: public place, drones, real-time environmental information, and navigation assistance services. The network planning agent then matches this characteristic information with feature information in the generative network's category library, resulting in the generative network being categorized as "smart city" and "intelligent robot."

[0283] In one possible implementation, the network planning agent can determine experiential information through private memory and generate first rule information based on the experiential information.

[0284] For example, the private memory of the network planning agent only stores experience information about "smart cities," and the network planning agent generates first rule information based on the experience information about "smart cities."

[0285] Step 703: The network planning agent sends a service request message, a category field corresponding to the generative network's category, and first rule information to the network big model. Correspondingly, the network big model receives the service request message, the category field corresponding to the generative network's category, and first rule information from the network planning agent.

[0286] In one possible implementation, the network planning agent sends a request message to the large network model, which requests second rule information corresponding to the "intelligent robot".

[0287] For example, the request message includes a service request message query, a category field "cat: Smart City & Smart Robot", and the first rule information guideline1.

[0288] Step 704: The large network model generates second rule information based on the category field corresponding to the category of the generative network, and merges the second rule information with the first rule information to generate rule information.

[0289] The second rule information is the rule information for all categories except at least one of the multiple categories corresponding to the generative network.

[0290] In one possible implementation, the large network model retrieves empirical information from the network's common memory based on the category field corresponding to the category of the generative network, generates second rule information based on the empirical information, and merges it with the first rule information to generate rule information.

[0291] For example, the category field "cat: Smart City & Smart Robot" corresponds to the category in the generative network. The first rule information is the rule information corresponding to "Smart City". The large network model retrieves experience information from the network's common memory based on "Smart City" and "Smart Robot", and generates a second rule information guideline2 corresponding to "Smart Robot". The second rule information guideline2 is then merged with the first rule information guideline1 to generate the rule information guideline_new.

[0292] Step 705: The large network model sends the first information and rule information to the network planning agent. Correspondingly, the network planning agent receives the first information and rule information from the large network model.

[0293] The rule information is determined by the first rule information and the second rule information.

[0294] In one possible implementation, the large network model sends feedback information to the network planning agent, which includes initial information and rule information.

[0295] Step 706: The network planning agent generates a list of subtasks based on the first information and the rule information.

[0296] In one possible implementation, the network planning agent generates a preliminary sub-task list based on the first information. The network planning agent generates prompts based on rule information and inputs these prompts into the large model within the network planning agent. This allows the large model to determine whether the preliminary sub-task list needs updating based on the rule information, and then update it to ensure that the sub-task list generated by the network planning agent conforms to the rule information. Step 707: The network planning agent sends the sub-task list, the category field corresponding to the generative network's category, and the rule information to the network assembly agent. Correspondingly, the network assembly agent receives the sub-task list, the category field corresponding to the generative network's category, and the rule information from the network planning agent.

[0297] In one possible implementation, the network planning agent sends instruction information to the network assembly agent, which includes a list of subtasks, a category field corresponding to the category of the generative network, and rule information.

[0298] For example, the instruction information sent by the network planning agent to the network assembly agent includes a subtask list 2, a category field "cat: Smart City & Smart Robot" corresponding to the category of the generative network, and the rule information guideline_new.

[0299] Step 708: The network assembly agent retrieves experience information from its private memory and generates third rule information based on the subtask list, the category field corresponding to the category of the generative network, and the rule information, and then selects a tool.

[0300] In one possible implementation, the network assembling agent retrieves experiential information from its private memory and generates third rule information based on the category field corresponding to the category of the generative network. The third rule information is then merged with the rule information to generate fourth rule information. Based on the fourth rule information and the subtask list, the agent selects a tool.

[0301] In this embodiment, each agent participating in the generative network design generates its own rule information based on the category of the received generative network, and merges it with other received rule information to improve the generation rules of the generative network, thereby guiding and standardizing its own decision-making process. In this way, the agent can comprehensively consider the requirements of different categories, ensuring that the generated generative network is both compliant and efficient, thus improving the reliability and security of the overall system.

[0302] The above mainly describes the solution of the embodiments of this application from the perspective of the interaction between various network elements. It can be understood that each network element, such as network equipment and terminal equipment, includes the corresponding structure and / or software module to perform the above functions in order to achieve the above functions.

[0303] This application embodiment can divide the network device and terminal device according to the above-described method example into functional units. For example, each function can be divided into a separate functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0304] The above implementation method is a combination of Figures 3-7 The methods described in the embodiments of this application have been explained. The communication apparatus provided in the embodiments of this application for executing the above methods is described below. Those skilled in the art will understand that the methods and apparatus can be combined with and referenced in each other. The communication apparatus provided in the embodiments of this application can execute the steps performed by the first network element, the second network element, or the third network element in the above analysis method.

[0305] When using integrated units Figure 8 The communication device involved in the above embodiments is shown. The communication device 80 may include a communication module 801 and a processing module 802.

[0306] In an alternative implementation, the communication device 80 may further include a storage module 803 for storing the program code and data of the communication device.

[0307] On one hand, the communication device 80 is a first network element, or a chip applied in a first network element. In this case, the communication module 801 is used to support communication between the communication device and an external device (e.g., a terminal device). For example, the communication module 801 is used to perform the transmit and receive operations of the first network element in the above method embodiment. The processing module 802 is used to perform the processing operations of the first network element in the above method embodiment.

[0308] The processing module 802 can be a processor or controller, such as a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field-programmable gate array, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, etc. The communication module can be a transceiver, transceiver circuitry, or communication interface, etc. The storage module can be a memory.

[0309] When the processing module 802 is a processor 901 or a processor 905, the communication module 801 is a transceiver 903, and the storage module 803 is a memory 902, the communication device involved in this application can be... Figure 9 The communication device shown.

[0310] Figure 9 This illustration shows a schematic diagram of the hardware structure of a communication device according to an embodiment of this application. The hardware structure of the first network element, second network element, or third network element in this embodiment can be referenced as follows: Figure 9 The structure shown is that the communication device includes a processor 901, a communication line 904, and at least one transceiver. Figure 9 (The illustration is merely exemplary, using transceiver 903 as an example only).

[0311] The processor 901 may be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present application.

[0312] Communication line 904 is used to transmit information between the aforementioned components.

[0313] Transceiver 903 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0314] Optionally, the communication device may also include a memory 902.

[0315] Memory 902 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 902 may exist independently and be connected to processor 901 via communication line 904. Memory 902 may also be integrated with processor 901.

[0316] The memory 902 stores computer execution instructions for implementing the scheme of this application, and its execution is controlled by the processor 901. The processor 901 executes the computer execution instructions stored in the memory 902, thereby implementing the communication method provided in the following embodiments of this application.

[0317] Optionally, the computer execution instructions in the embodiments of this application may also be referred to as application code, and the embodiments of this application do not specifically limit this.

[0318] In a specific implementation, as one example, the processor 901 may include one or more CPUs, for example... Figure 9 CPU0 and CPU1 in the CPU.

[0319] In a specific implementation, as one example, the communication device may include multiple processors, for example... Figure 9 Processors 901 and 902 are shown in the diagram. Each of these processors can be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor here can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0320] Figure 10 This is a schematic diagram of the structure of the chip 100 provided in the embodiments of this application. The chip 100 includes one or more (including two) processors 1010 and a communication interface 1030.

[0321] Optionally, the chip 100 also includes a memory 1040, which may include read-only memory and random access memory, and provides operation instructions and data to the processor 1010. A portion of the memory 1040 may also include non-volatile random access memory (NVRAM).

[0322] In some implementations, memory 1040 stores elements such as execution modules or data structures, or subsets thereof, or extended sets thereof.

[0323] In this embodiment of the application, the corresponding operation is executed by calling the operation instructions stored in the memory 1040 (the operation instructions can be stored in the operating system).

[0324] One possible implementation is that the first network element, the second network element, and the third network element have similar structures, and different devices can use different chips to achieve their respective functions.

[0325] The processor 1010 controls the processing operations of any one of the first network element, the second network element, and the third network element. The processor 1010 can also be called a central processing unit (CPU).

[0326] Memory 1040 may include read-only memory and random access memory, and provides instructions and data to processor 1010. A portion of memory 1040 may also include NVRAM. For example, in an application, memory 1040, communication interface 1030, and memory 1040 are coupled together via bus system 1020, which may include, in addition to data bus, power bus, control bus, and status signal bus, etc. However, for clarity, in... Figure 10 The general labeled all buses as Bus System 1020.

[0327] Various aspects or features of this application can be implemented as methods, apparatus, or articles of manufacture using standard programming and / or engineering techniques. As used herein, the term "article of manufacture" encompasses a computer program accessible from any computer-readable device, carrier, or medium. For example, computer-readable media may include, but are not limited to: magnetic storage devices (e.g., hard disks, floppy disks, or magnetic tapes), optical discs (e.g., compact discs (CDs), digital versatile discs (DVDs), etc.), smart cards, and flash memory devices (e.g., erasable programmable read-only memory (EPROMs), cards, sticks, or key drives, etc.). Additionally, the various storage media described herein may represent one or more devices and / or other machine-readable media for storing information. The term "machine-readable medium" may include, but is not limited to, wireless channels and various other media capable of storing, containing, and / or carrying instructions and / or data.

[0328] It should be understood that the processor mentioned in the embodiments of this application can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0329] It should also be understood that the memory mentioned in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0330] It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA, or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, the memory (storage module) is integrated into the processor.

[0331] It should be noted that the memories described herein are intended to include, but are not limited to, these and any other suitable types of memories.

[0332] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0333] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0334] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0335] 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0336] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0337] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion 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 this application. 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.

[0338] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, The method includes: Obtain a service request message, the service request message including an intent, the intent being used to determine the task to be performed in the request; The category of generative network that provides services for the task is determined based on the stated intent; The intent and the category of the generative network are sent to the second network element, wherein the category of the generative network is used to determine empirical information corresponding to the category of the generative network.

2. The method according to claim 1, characterized in that, The method further includes: The system receives first information from the second network element, the first information being determined by empirical information corresponding to the category of the generative network.

3. The method according to claim 2, characterized in that, The first information is determined by experience information that matches the intent.

4. The method according to claim 2 or 3, characterized in that, The method further includes: Based on the first information, the information of the first subtask is determined, and the information of the first subtask is used to indicate the multiple subtasks after the task is decomposed. Send the information of the first subtask to the third network element.

5. The method according to claim 4, characterized in that, The method further includes: The category of the generative network is sent to the third network element.

6. The method according to any one of claims 2 to 5, characterized in that, The method further includes: Obtain rule information, which describes the rules for processing the task under the category of the generative network.

7. The method according to claim 6, characterized in that, The rule information includes: content that is allowed and / or prohibited from being executed by the first network element under the category of the generative network.

8. The method according to claim 6 or 7, characterized in that, The obtained rule information includes: Receive rule information from the second network element.

9. The method according to claim 6 or 7, characterized in that, The generative network corresponds to multiple categories, and the rule information acquisition includes: Based on at least one of the multiple categories corresponding to the generative network, first rule information is determined, and the first rule information is used to process the rules corresponding to the task under at least one of the multiple categories of the generative network; Send first rule information to the second network element, wherein the first rule information is used to determine the rule information; Receive the rule information from the second network element.

10. The method according to any one of claims 6 to 9, characterized in that, The method further includes: Based on the first information and the rule information, the information of the second subtask is determined, and the information of the second subtask is used to indicate the multiple subtasks after the task is decomposed according to the rule information; Send the information of the second subtask to the third network element.

11. The method according to claim 10, characterized in that, The method further includes: The rule information and the category of the generative network are sent to the third network element.

12. A communication method, characterized in that, The method includes: The category of the generative network is received from the first network element, the generative network being used to provide services for the task corresponding to the intent; Based on the category of the generative network, obtain the empirical information corresponding to the category of the generative network.

13. The method according to claim 12, characterized in that, The acquisition of empirical information corresponding to the category of the generative network includes: Based on the category of the generative network, retrieve the empirical information corresponding to the category of the generative network from the storage unit.

14. The method according to claim 13, characterized in that, The step of retrieving the empirical information corresponding to the category of the generative network from the storage unit according to the category of the generative network includes: A first request is sent to the storage unit, the first request including the category of the generative network, the first request being used to request empirical information corresponding to the category of the generative network; Receive empirical information from the storage unit corresponding to the category of the generative network.

15. The method according to claim 13 or 14, characterized in that, The storage unit stores multiple categories of the generative networks, as well as empirical information corresponding to each category of the generative networks.

16. The method according to any one of claims 12 to 15, characterized in that, The method further includes: Receive a service request message from the first network element, the service request message including the intent.

17. The method according to claim 16, characterized in that, The method further includes: First information is determined based on empirical information corresponding to the category of the generative network, wherein the first information is determined by empirical information matching the intent; Send the first information to the first network element.

18. The method according to claim 16, characterized in that, The method further includes: Based on empirical information corresponding to the category of the generative network, rule information is determined, which describes the rules for processing the task under the category of the generative network. The rule information is sent to the first network element.

19. The method according to claim 16, characterized in that, The generative network corresponds to multiple categories, and the method further includes: Receive first rule information from the first network element, the first rule information being used to describe the rules corresponding to processing the task under at least one of the multiple categories of the generative network; Based on the category of the generative network, second rule information is generated. The second rule information is used to describe the rules for processing the task in other categories besides the at least one category among the multiple categories of the generative network. Based on the first rule information and the second rule information, generate rule information; The rule information is sent to the first network element.

20. The method according to any one of claims 12 to 15, characterized in that, The method further includes: Receive information from the first subtask of the first network element, the information of the first subtask being used to indicate the multiple subtasks after the task is decomposed.

21. The method according to claim 20, characterized in that, The method further includes: Based on empirical information corresponding to the category of the generative network, a first toolset is determined, which is used to provide services for the subtask.

22. The method according to claim 20, characterized in that, The method further includes: Receive rule information from the first network element.

23. The method according to claim 22, characterized in that, The method further includes: Based on the category of the generative network, query the storage unit for the third rule information corresponding to the category of the generative network; Based on the rule information and the third rule information, the fourth rule information is determined.

24. The method according to claim 23, characterized in that, The method further includes: Based on the fourth rule information and the empirical information corresponding to the category of the generative network, a second toolset is determined. The second toolset is used to provide services for the subtask under the fourth rule information.

25. A communication device, characterized in that, The apparatus includes a module for performing the method as described in any one of claims 1 to 11; or, a module for performing the method as described in any one of claims 12 to 24.

26. A communication system, characterized in that, The system includes: a first network element, a second network element, and a third network element; The first network element is used to implement the method as described in any one of claims 1 to 11, and the second network element or the third network element is used to implement the method as described in any one of claims 12 to 24.

27. The system according to claim 26, characterized in that, The system further includes a terminal device, which is used to send a service request message to the first network element. The service request message is used to indicate the task requested by the user and to determine the type of generative network, which is used to provide services for the task.

28. A chip, characterized in that, The chip includes at least one processor and a communication interface, the communication interface being coupled to the at least one processor, the at least one processor being configured to run computer programs or instructions to implement the method as described in any one of claims 1 to 11, or the method as described in any one of claims 12 to 24, the communication interface being configured to communicate with other modules outside the chip.

29. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, implement the method of any one of claims 1 to 11, or the method of any one of claims 12 to 24.

30. A computer program product, characterized in that, The computer program product includes computer instructions that instruct a computing device to perform the steps of the method according to any one of claims 1 to 11, or instruct a computing device to perform the steps of the method according to any one of claims 12 to 24.

Images