Communication methods and related apparatus

Abstract

The embodiments of the present application provide communication methods and a related apparatus. A communication method comprises: an agent sending a request message, wherein the request message comprises task information described in natural language, the task information is used for determining a first function, the first function is determined on the basis of function information, and the function information is obtained by means of a first component analyzing the task information on the basis of a network large model (NLM); and in response to the request message, the agent receiving first configuration information, wherein the first configuration information is configuration information of the first function. Thus, an agent can perform function queries using task information described in the form of natural language, without the need to use structured signaling for querying, thereby implementing a more intelligent and convenient function query mechanism.

Description

A communication method and related apparatus

[0001] This application claims priority to Chinese Patent Application No. 202411998496.6, filed with the State Intellectual Property Office of China on December 31, 2024, entitled “A Communication Method and Related Device”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of communication technology, and in particular to a communication method and related apparatus. Background Technology

[0003] In the 5G system architecture defined by the 3rd Generation Partnership Project (3GPP), when a network function (NF) or application needs to use a specific service, the NF can send a discovery request to the network repository function (NRF) through a service registration and discovery mechanism, specifying the required service type, capabilities, and other conditions. The NRF then searches its database to find a matching NF service and returns detailed information about that NF service, such as service endpoints and capabilities. In this way, the NF or application can discover and connect to the required service.

[0004] In future mobile communication systems, in addition to providing communication connectivity services, new services such as artificial intelligence (AI), sensing, and computing will also be required. This necessitates stronger on-demand customization capabilities to orchestrate and configure various service functions (NFs) and resources in a more flexible and dynamic manner. However, in existing service registration and discovery mechanisms, NF service invocation requires a standard signaling structure, necessitating the specification of the required service type or conditions in the request to obtain the corresponding service. With the proliferation of new service functions in future mobile communication systems, and the interactions and influences between these functions, the standard signaling used to request NF services is difficult to define completely. Summary of the Invention

[0005] This application provides a communication method and related apparatus for an intelligent agent to send a request message. The request message includes task information described in natural language, which is used to determine a first function. The first function is determined based on function information obtained by a first component through NLM analysis of the task information. In response to the request message, the intelligent agent receives first configuration information, which is configuration information for the first function. The intelligent agent can use the task information described in natural language to perform function queries without needing to use structured signaling, thus achieving a more intelligent and convenient function query mechanism.

[0006] Firstly, this application provides a communication method in which an agent sends a request message including task information described in natural language, and in response to the request message, the agent receives first configuration information. The first configuration information is configuration information for a first function, the task information is used to determine the first function, the first function is determined based on function information, and the function information is obtained by a first component analyzing the task information based on a Network Model (NLM).

[0007] Based on the above scheme, the intelligent agent can perform function queries using task information described in natural language, without needing to use structured signaling. That is, the intelligent agent does not need to carry query parameters such as function name and category in the request message, providing a more flexible function query mechanism, simplifying the query operation, and improving the convenience of the function query mechanism. The first component uses its built-in NLM to analyze the task information, uncovering potential functional requirements within the task information and outputting the functional information needed to execute the task. The component library then queries based on this functional information to determine the configuration information that matches the functional information. This enables intelligent decision-making based on changes in task requirements, improving the intelligence of the function query mechanism.

[0008] In one possible implementation, the agent sends a subscription message, and in response to the subscription message, the agent receives a first notification message sent by the component library. The subscription message includes a function category and / or capability expectation described in natural language. The first notification message is determined by the component library based on at least one of the function category and capability expectation.

[0009] In this implementation, the agent sends a subscription message to the component library. The component library can manage subscriptions to functions related to the agent based on the function categories and capability expectations in the subscription message, and notify the agent in a timely manner. The component library can manage subscriptions to function categories and / or capability expectations described in natural language without using structured signaling, thus achieving a more intelligent and convenient function subscription mechanism.

[0010] In one possible implementation, the functional information includes functional categories and available network resources; the functional categories include application functions and network functions; the first configuration information includes at least one of the following: the identifier of the first function, the category of the first function, the model corresponding to the first function, the mandatory input information of the first function, the optional input information of the first function, and the output sample of the first function.

[0011] In this implementation, the functional information includes functional categories and available network resources. The component library can query based on the functional categories and available network resources to determine the first function and first configuration information that satisfy the functional category and available network resources.

[0012] Secondly, this application provides a communication method in which a first component receives a request message sent by an agent, the request message including task information described in natural language; the first component analyzes the task information in the request message based on a Network Model (NLM) to determine functional information; and based on the functional information, the first component sends a query message to a component library. The first component includes an NLM, and the query message is used to determine a first function and first configuration information, the first configuration information being the configuration information for the first function.

[0013] Based on the above scheme, the first component can use NLM to analyze the task information described in natural language form, mine the potential functional requirements in the task information through NLM, output the functional information required to execute the task, and then the component library can query the functional information to determine the first function and the first configuration information, thus realizing a more intelligent and convenient function query mechanism.

[0014] In one possible implementation, the functional information includes functional categories and available network resources. The query message is a structured message sent via the ABI, and includes a component library identifier, functional category, and available network resources. The component library identifier identifies component libraries with intelligent functional query capabilities. The ABI, acting as a declarative interface between the first component and the component library, can convert the functional information into a structured query message and forward it to the component library corresponding to the component library identifier for processing.

[0015] Thirdly, this application provides a communication method in which a component library receives a request message sent by an agent, the request message including task information described in natural language; the component library analyzes the task information in the request message based on a Network Model (NLM) to determine first configuration information; and the component library sends the first configuration information. The first configuration information is configuration information for a first function, the first function matches the task information, and the first function is registered in the component library.

[0016] Based on the above scheme, the component library includes a first component with built-in NLM. The component library can use NLM to analyze task information in natural language form, uncover potential functional requirements within the task information, and output the functional information required to execute the task. Furthermore, the component library includes various network functions and application functions, along with corresponding configuration information. It can directly query the database based on the functional information to determine the first function and corresponding first configuration information matching that function, without requiring the first component to send a query message to the component library. This gives the component library not only simple search functionality but also intelligent requirement analysis capabilities, allowing it to directly determine suitable functions and configuration information based on the agent's task information, thus achieving a more intelligent and convenient function query mechanism.

[0017] In one possible implementation, request messages are analyzed based on the network generative pre-trained model NetGPT and NLM to determine the first configuration information. NetGPT has a larger number of parameters than NLM, and its computational cost is also higher than that of NLM. This implementation, using NetGPT to assist NLM in analyzing task information, improves analysis efficiency and is applicable to multi-task scenarios.

[0018] In one possible implementation, the component library receives a query message sent by the first component; based on the query message, the component library determines a first function and first configuration information. The first configuration information is the configuration information for the first function, which is matched with the task information. The query message includes the function category and available network resources.

[0019] In one possible implementation, the component library receives subscription messages sent by the agent, the subscription messages including function categories and / or capability expectations described in natural language; when the component library detects that at least one of the function categories and capability expectations in the subscription messages triggers a change in the subscription list, the component library sends a first notification message.

[0020] In this implementation, when the component library detects the launch of a new function that meets the functional category, or when a registered function is updated to meet the capability expectations, it will trigger a change in the subscription list to promptly notify the agent of the changes in the relevant functions.

[0021] In one possible implementation, when a function associated with the agent is detected to be deregistered or malfunctioning, the component library sends a second notification message to the agent.

[0022] Fourthly, this application provides a communication device that can implement the methods in any possible implementation of any of the first to third aspects described above. The device includes corresponding units or modules for performing the aforementioned methods. The units or modules included in the device can be implemented by software and / or hardware. For example, the device can be a terminal or a base station, or it can be a component (e.g., a processor, chip, or chip system) in a terminal or base station, or it can also be a logic module or software capable of implementing all or part of the functions of a terminal or base station. The device includes a processing unit and a transceiver unit. The constituent modules of the communication device can also be used to perform the steps executed in any possible implementation of any of the first to third aspects and achieve the corresponding technical effects, as detailed in the foregoing description, which will not be repeated here.

[0023] Fifthly, this application provides a communication device including at least one processor for executing a program or instructions in a memory to enable the device to implement the methods executed in various possible implementations of any of the first to third aspects.

[0024] In a sixth aspect, this application provides a communication device including at least one logic circuit and an input / output interface; the logic circuit is used to perform the methods executed in various possible implementations of any of the first to third aspects described above.

[0025] In a seventh aspect, this application provides a computer-readable storage medium storing a computer program or instructions that, when executed by a communication device, cause the communication device to perform the methods executed in any of the possible implementations of the first to third aspects described above.

[0026] Eighthly, this application provides a computer program product, including a computer program or instructions, which, when executed by a communication device, cause the communication device to perform the methods executed in any of the possible implementations of the first to third aspects described above.

[0027] Ninthly, this application provides a chip system including at least one processor for supporting a communication device in implementing the methods executed in various possible implementations of any of the first to third aspects described above.

[0028] In one possible design, the chip system may further include a memory for storing program instructions and data necessary for the communication device. The chip system may be composed of chips or may include chips and other discrete devices. Optionally, the chip system may also include interface circuitry that provides program instructions and / or data to the at least one processor. Attached Figure Description

[0029] Figure 1 is a schematic diagram of the system architecture provided in this application;

[0030] Figure 2 is a schematic diagram of another system architecture provided in this application;

[0031] Figure 3 is a flowchart illustrating a communication method provided in this application;

[0032] Figure 4 is a structural diagram of a functional configuration information provided in this application;

[0033] Figure 5 is a flowchart illustrating another communication method provided in this application;

[0034] Figure 6 is a flowchart illustrating another communication method provided in this application;

[0035] Figure 7 is a flowchart illustrating another communication method provided in this application;

[0036] Figure 8 is a schematic diagram of the structure of a communication device provided in this application;

[0037] Figure 9 is a schematic diagram of another communication device provided in this application. Detailed Implementation

[0038] In the 5G system architecture defined by 3GPP, Network Functions (NFs) implement their functions through service and operation mechanisms. These mechanisms are based on a service-based architecture (SBA), which divides network functions into several reusable services. These services communicate using lightweight interfaces to provide more flexible, scalable, and manageable network services. NFs include authentication server functions (AUSF), access and mobility management functions (AMF), session management functions (SMF), network exposure functions (NEF), and policy control functions (PCF), among others.

[0039] The NF service registration and discovery mechanism allows NF instances to register the services they provide, and allows other NFs or applications to discover and use the services provided by registered NFs. Service provision is achieved through message exchange between producers and consumers. The process of the NF service registration and discovery mechanism can be summarized as follows: First, the NF service consumer sends a discovery request to the NRF, specifying the required service type, capabilities, or other conditions. This request can be Nnrf_NFDiscovery_Request(desired NF service name, desired NF instance NF type, NF consumer NF type). The parameters of this request may also include producer NF set identifier, NF service set identifier, subscription permanent identifier (SUPI), dataset identifier, user equipment (UE) routing indication, single network slice selection assistance information (S-NSSAI), network slice instance identifier (NSI ID), and other service-related parameters. Next, the NRF decides whether to allow the NF service consumer to discover the desired NF instance based on the desired NF service and the type of the NF service consumer. If permitted, the NRF determines the set of NF instances that match the Nnrf_NFDiscovery_Request and the NRF internal policy, and sends the NF profile of the determined NF instances. The NF profile includes information that the network element needs to register with the NRF, such as NF type, supported slices, priority, IP address, supported services, etc.

[0040] However, SBA-based services and operational mechanisms include a limited number of service categories, such as user plane services, control plane services, location management services, session management services, policy control services, authentication and authorization services, and network slicing management services. As new service functions emerge in future mobile communication systems, such as AI, sensing, and computing, new service categories need to be defined. Moreover, these services are implemented through standardized structured signaling. With the proliferation of new service functions and their interactions and influences, defining standard signaling for these new services becomes difficult. Furthermore, SBA-based NRFs are essentially simple databases, requiring precise service names for queries and matching. NFs or applications must specify the desired service type or conditions in their requests to obtain the corresponding service; NRFs themselves lack intelligent analysis capabilities.

[0041] In view of this, this application provides a communication method and related apparatus. In this method, an intelligent agent sends a request message, which includes task information described in natural language. In response to the request message, the intelligent agent receives configuration information of a first function that matches the task information. The configuration information of the first function is determined based on the functional information required to perform the task, and this functional information is obtained by a first component through NLM analysis of the task information. Therefore, by directly analyzing and processing the task information described in natural language and determining the configuration information of the functions required to perform the task, a more intelligent and convenient function query mechanism is achieved.

[0042] Please refer to Figure 1, which is a schematic diagram of the architecture of a communication system 1000 provided in an embodiment of this application. As shown in Figure 1, the communication system 1000 includes a radio access network (RAN) 100, wherein the RAN 100 includes at least one RAN node (110a and 110b in Figure 1, collectively referred to as 110), and may also include at least one terminal (120a-120j in Figure 1, collectively referred to as 120). The RAN 100 may also include other RAN nodes, such as wireless relay devices and / or wireless backhaul devices (not shown in Figure 1). The terminal 120 is wirelessly connected to the RAN node 110. Terminals and RAN nodes can be interconnected via wired or wireless means. The communication system 1000 may also include a core network 200. The RAN node 110 is connected to the core network 200 via wireless or wired means. The core network equipment in core network 200 and the RAN node 110 in RAN 100 can be independent and different physical devices, or they can be the same physical device that integrates the logical functions of the core network equipment and the logical functions of the RAN node. Communication system 1000 may also include Internet 300.

[0043] RAN100 can be an evolved universal terrestrial radio access (E-UTRA) system, a new radio (NR) system, a 6th generation (6G) radio access system, or a future radio access system as defined in the 3rd generation partnership project (3GPP), or it can be a WiFi system. RAN100 can also include two or more of the above-mentioned different radio access systems. RAN100 can also be an open RAN (O-RAN).

[0044] RAN nodes, also known as radio access network devices, RAN entities, or access nodes, are used to help terminals access communication systems wirelessly. In one application scenario, an RAN node can be a base station, an evolved NodeB (eNodeB), a transmission reception point (TRP), a next-generation NodeB (gNB) in a 5G mobile communication system, a base station in a future mobile communication system, or an access node in a WiFi system. RAN nodes can be macro base stations (as shown in Figure 1, 110a), micro base stations or indoor stations (as shown in Figure 1, 110b), relay nodes, or donor nodes.

[0045] In another application scenario, multiple RAN nodes can collaborate to help terminals achieve wireless access, with different RAN nodes implementing different functions of the base station. For example, a RAN node can be a central unit (CU), a distributed unit (DU), or a radio unit (RU). Here, the CU performs the functions of the base station's Radio Resource Control (RRC) and Packet Data Convergence Protocol (PDCP), and can also perform the functions of the Service Data Adaptation Protocol (SDAP). The DU performs the functions of the base station's Radio Link Control (RANC) and Medium Access Control (MAC) layers, and can also perform some or all of the physical layer functions. For specific descriptions of these protocol layers, refer to the relevant 3GPP technical specifications. The RU can be used to implement radio frequency signal transmission and reception. The CU and DU can be two independent RAN nodes or integrated into the same RAN node, such as within a baseband unit (BBU). The RU can be included in radio frequency equipment, such as in a remote radio unit (RRU) or an active antenna unit (AAU). The CU can be further divided into two types of RAN nodes: CU-control plane and CU-user plane.

[0046] In different systems, RAN nodes may have different names. For example, in an O-RAN system, a CU can be called an open CU (O-CU), a DU can be called an open DU (O-DU), and an RU can be called an open RU (O-RU). The RAN nodes in the embodiments of this application can be implemented through software modules, hardware modules, or a combination of software and hardware modules. For example, a RAN node can be a server loaded with the corresponding software modules. The embodiments of this application do not limit the specific technology or device form used in the RAN nodes. For ease of description, a base station is used as an example of a RAN node in the following description.

[0047] A terminal is a device with wireless transceiver capabilities, capable of sending signals to or receiving signals from a base station. Terminals can also be called terminal equipment, user equipment (UE), mobile station, mobile terminal, etc. Terminals can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IoT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, smart cities, etc. Terminals can be mobile phones, tablets, computers with wireless transceiver capabilities, wearable devices, vehicles, airplanes, ships, robots, robotic arms, smart home devices, etc. The embodiments of this application do not limit the specific technology or device form used in the terminal.

[0048] Base stations and terminals can be fixed or mobile. They can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can be deployed on aircraft, balloons, and satellites. The embodiments of this application do not limit the application scenarios of the base stations and terminals.

[0049] The roles of base stations and terminals can be relative. For example, the helicopter or drone 120i in Figure 1 can be configured as a mobile base station. For terminals 120j that access the wireless access network 100 through 120i, terminal 120i is a base station; however, for base station 110a, 120i is a terminal, meaning that 110a and 120i communicate via a wireless air interface protocol. Of course, 110a and 120i can also communicate via a base station-to-base station interface protocol. In this case, relative to 110a, 120i is also a base station. Therefore, both base stations and terminals can be collectively referred to as communication devices. 110a and 110b in Figure 1 can be called communication devices with base station functions, and 120a-120j in Figure 1 can be called communication devices with terminal functions.

[0050] Communication between base stations and terminals, between base stations, and between terminals can be conducted using licensed spectrum, unlicensed spectrum, or both simultaneously. Communication can be conducted using spectrum below 6 GHz, spectrum above 6 GHz, or both simultaneously. The embodiments of this application do not limit the spectrum resources used for wireless communication.

[0051] In this embodiment of the application, the terminal or RAN node can implement the functions of the intelligent agent, the first component, and the component library.

[0052] Please refer to Figure 2, which is a schematic diagram of an A-Core system architecture provided in this application. A-Core is a next-generation core network based on intelligent agents. This system is not simply an integration or addition of intelligent features to an existing core network, but rather a closed-loop generation system formed through the cooperation of intelligent agents and various network functions. It can flexibly orchestrate, deploy, execute, and reclaim end-to-end network slices for new services, and improve performance through feedback from the network environment. A-Core can generate end-to-end (E2E) network slices according to user intent without providing detailed network configuration information. The lifecycle management of E2E network slices is completed by a combination of various agents.

[0053] In this context, an Agent is defined as an intelligent system within the network that can interact with its environment, collect data, invoke tools, and learn from past experiences to improve its decision-making capabilities and execute specific network tasks. Agents include planning agents, assembling agents, connection agents, and execution agents. Network slicing is a novel network architecture that allows operators to build multiple dedicated, virtualized, and isolated logical networks on top of a common physical network to meet the diverse network capability requirements of different customers.

[0054] As shown in Figure 2, this communication system includes multiple agents and various common components (toolboxes). Agents and common components can be deployed in the same network element or in different network elements; for example, agents and common components can be deployed in the RAN node. Each network element plays a different role (Actor), including the RAN node, UE, and A-GW. The RAN node, UE, and A-GW can be interconnected. For a description of the RAN node and UE, please refer to the preceding sections; it will not be repeated here. The A-GW is a data proxy between the RAN node and the E2E slice instance, supporting data routing based on E2E slice tunnels. As the executor of the intelligent agent, the Actor can execute relevant instructions based on the intelligent decisions generated by the intelligent agent, including E2E network slice management and control, task control for function execution in the component library, and dynamic network resource scheduling.

[0055] The following section introduces the various network elements involved in this system architecture.

[0056] The planning agent is the entity that receives customized network slicing requests, is responsible for understanding user intent, further decomposing the user intent into multiple executable tasks, and outputting a list of decomposed tasks. Each item in the task list includes a task description, task dependency metrics, quality of service (QoS) requirements, etc.

[0057] The assembly agent is responsible for receiving output from the planning agent, selecting appropriate network functions and / or application functions for each task, determining the functional topology based on task dependencies, and outputting a two-part function list. The first part is a function list for configuring the UE and / or RAN, and the second part is a function list for deployment in the core network. Each entry in the function list includes a function identifier, input parameters, type (configuration or deployment), and function dependency indicator.

[0058] The connecting agent is responsible for receiving the first part of the output from the assembling agent. Based on the function list used to configure the UE and / or RAN, it determines which UE(s) and / or RAN(s) will disrupt the E2E slice and determines their configuration parameters and resources. It manages and controls the connection topology of the E2E slice, and supports UE access management and slice selection. The connecting agent outputs instructions to the Actor to configure the UE and / or RAN, and connects the RAN to the function instance.

[0059] The executing agent is responsible for receiving the second part of the output from the assembling agent, managing and controlling the lifecycle of network slices, including deploying and linking function instances of the network and / or application, monitoring the running status of network slices, dynamically updating and recycling network slices, and outputting instructions for the Actor to deploy, update, and recycle function instances and chained function instances.

[0060] Toolbox is a marketplace containing a vast amount of network and application functionalities (such as functions, components, models, APIs, etc.) and configuration files. In other words, Toolbox can be understood as a component library. This functional information and configuration files include function name, description, type, required parameters, optional parameters, output examples, and usage methods. A component refers to an independent, reusable functional unit that can be called and integrated by other systems or applications. In the embodiments of this application, components, functions, or models are not specifically distinguished and can be substituted for each other as long as there is no logical conflict. Developers can register their developed functions, components, or models in Toolbox, where intelligent agents can orchestrate and call them. Functions can be non-standardized. Toolbox can receive queries or subscription requests for functions from other components and output responses or notifications regarding relevant functional information or configuration files.

[0061] The shared memory is responsible for collecting, vectorizing, and storing network data and knowledge not present in the network generative pre-trained transformer (NetGPT), as well as indexing this data or knowledge. The shared memory can receive data query requests from other components and output the requested data to consumers. It can also support retrieval-augmented generation (RAG) to insert relevant knowledge as hints to improve NetGPT's inference results. As a shared component in A-Core, NetGPT provides advanced network inference capabilities and can be invoked by any Agent. Based on input hints, it assists AgentGPT in each Agent to generate inference results related to network slices (such as task decomposition, function selection, etc.).

[0062] An agent-based interface (ABI) represents a declarative interface between an agent and a public component. It uses NetGPT to understand the purpose of a message and determines how to forward it to the appropriate agent or public component. Unlike imperative interfaces that specify service operations, ABIs specify goals. When a consumer invokes an agent or public component, it only needs to provide a goal description. The ABI understands the message's purpose and forwards it to the appropriate agent or public component for processing based on that goal.

[0063] The foregoing section describes the system architecture involved in the embodiments of this application. Based on the A-Core system architecture, this application designes an intelligent function query (IFQ) component. The IFQ component can understand task information described by the agent in natural language, analyze the network functions and / or application functions required to execute the task information, and output detailed function configuration information. Thus, a more intelligent function query mechanism for network functions and application functions is implemented through the IFQ component. The IFQ component can be used as an internal function of the ABI or built into the Toolbox. In this application embodiment, the first component can be the IFQ component, the intelligent agent can be the aforementioned planning intelligent agent, assembly intelligent agent, connection intelligent agent, and execution intelligent agent, etc., and the component library can be the Toolbox.

[0064] Example 1: The IFQ component is an internal function of the ABI.

[0065] Please refer to Figure 3, which is a flowchart illustrating a communication method provided in this application.

[0066] 301. The agent sends a request message.

[0067] The request message includes task information described in natural language. This task information can describe tasks related to user needs, such as "identify all vehicles in the current community" or "find the nearest restaurant for me," or it can describe tasks undertaken by the intelligent agent based on network slicing business requirements. For example, in network slicing for smart cities, this task information could be described as "continuously monitoring the network performance of devices such as traffic cameras, sensors, and data centers." This task information is used to determine the first function and the first configuration information. The first configuration information is the configuration information for the first function, which is the network function and / or application function required to perform the task.

[0068] The agent sends a request message to the first component via the ABI to determine the configuration information of the network functions and / or application functions required to perform the task. Correspondingly, the first component receives the request message via the ABI. When the first component acts as an internal function of the ABI, the ABI can be understood as an intermediate layer responsible for the interaction of requests or services between the first component and the agent and component library. Upon receiving the request message, the first component can retrieve the request message through the ABI.

[0069] 302. The first component analyzes the task information in the request message based on the Network Model (NLM) to determine the functional information.

[0070] The first component includes a network large model (NLM), which can be used to analyze or understand semantic content described in natural language. The NLM can be a model with a relatively small number of parameters, such as Mixtral-7B or Llama3, where 7B indicates that the Mixtral model has 7 billion parameters. The Mixtral-7B model is suitable for tasks such as text generation, dialogue systems, machine translation, and text understanding. The NLM built into the first component has a small number of parameters, allowing for efficient deployment in resource-constrained environments, such as edge computing devices and small cloud servers. Through the NLM, the first component can analyze or understand the task information described in natural language in request messages, determine the functional information required to execute the task, and query network load to obtain available network resources for the current task.

[0071] The functional information includes the names of the functions required to perform the task, the required function categories, and available network resources. Function categories include network functions and application functions. Optionally, the component library can be further categorized into multiple levels for registered network and application functions. For example, application functions can be subdivided into natural language processing, image analysis, and search recommendation, while image analysis functions can be subdivided into object recognition, image enhancement, and image classification. As another example, network functions can include basic core network functions such as session establishment and policy control, and can also include functions for new services in future mobile communication systems, such as base station awareness, network edge computing, model training, and computation offloading. Available network resources include available central processing unit (CPU) resources, available graphics processing unit (GPU) resources, available network bandwidth, and available memory space. Available CPU / GPU resources include the model, quantity, utilization rate, and remaining available computing power of the CPU / GPU. Available network bandwidth includes current bandwidth usage and remaining bandwidth.

[0072] The following describes the process by which the first component uses NLM to analyze task information in request messages to determine functional information. First, NLM performs semantic analysis on the task information described in natural language, understands the task intent, and extracts key information. This process can include entity recognition, intent understanding, and context understanding. Entity recognition identifies important entities in the task information, such as time, people, locations, and events. Intent understanding identifies the task's goal or the specific operations required to perform the task. Context understanding determines which information in the task information is relevant and which is irrelevant based on the contextual relationships. For example, in the task information "Help me find the nearest restaurant," NLM can identify "me" and "restaurant" as entities, "find restaurant" as the task goal, and clearly define "nearest" as a query constraint relevant to the task.

[0073] Secondly, the first component can determine the functional information required to execute the task based on the extracted key information. The first component can decompose the task into smaller, manageable steps using a chain-of-thought (CoT) approach, and obtain network data and knowledge stored in common memory via the ABI, such as the capabilities and locations of the UE and RAN, and the execution status of network slices. This network data and knowledge is then used as background information for function matching, and the internal NLM is invoked for analysis to determine the functional information corresponding to the task. For example, the NLM might analyze the task information "Find me the nearest restaurant" and determine that the corresponding function category is "Application Function," the function name is "Search Recommendation," and the function description could be "Obtain the user device's geographical location; query restaurants within 50 meters of this geographical location," with available network resources described as "Computing performance of 19.5 trillion floating-point operations per second; uplink bandwidth of 500 Mbps." In this way, the functional information corresponding to the task information is determined.

[0074] Optionally, the first component can combine multiple network models to analyze task information. For example, NLM can be used for semantic understanding of the task information. Traditional machine learning models, such as decision trees and support vector machines (SVM), can be combined to classify intent and determine the functional category of the task information. Models such as Transformer can also be combined to predict the network resources required to perform the task, such as predicting dynamic changes in network bandwidth and fluctuations in network load.

[0075] In this step, the first component decomposes the task through a thought chain, uses ABI to obtain network-related information as background information, and combines it with NLM for analysis to determine the functional information required to execute the task, thereby enabling the first component to have a more intelligent task processing capability.

[0076] 303. The first component sends a query message to the component library based on the functional information.

[0077] The component library can be used to query network functions or application functions that meet functional information. The first component sends a query message to the component library, which includes functional information. The component library receives this functional information and queries the database to determine functions that match the functional information. In this embodiment, the first component is an internal function of the ABI. The first component can send a query message to the component library through the ABI. The query message includes a component library identifier and functional information required to perform the task (such as function category, available network resources, etc.). The component library identifier can be used to identify component libraries with intelligent function query capabilities. The ABI can convert the functional information described in natural language sent by the first component into structured information. This query message is a structured message sent by the ABI; for example, the query message can be represented as IFQ_functionretrive_request(Tool-box ID, function type, useful resources, etc.).

[0078] 304. The component library determines the first function and the first configuration information based on the query message.

[0079] The component library, based on query parameters in the query message, such as function category and available network resources, queries the database for registered application functions and network functions to determine the first function matching the function information and its corresponding function configuration information (first configuration information). The first configuration information includes at least one of the following: the identifier of the first function, the category of the first function, the model corresponding to the first function, the required input information of the first function, the optional input information of the first function, and the output example of the first function. The identifier of the first function is used to distinguish different functions in the component library. The model corresponding to the first function can be a specific algorithm or program that implements the function, such as a YOLOv7 model. The required input information of the first function can be the input data necessary for the normal operation of the function. The optional input information of the first function can be additional input data that is selectively provided; these data are not necessary conditions for the execution of the function, but may affect the behavior or output of the function. The output example can be an example of the possible output results after the function is executed. The structure of the first configuration information can be seen in Figure 4.

[0080] For example, if the task information is "identify all vehicles in the current cell," the IFQ component can analyze this task information to determine the required function description, such as "it could be an object recognition function that can quickly and accurately detect objects in an image, etc." The IFQ component can also determine that the function category is "application function." The IFQ component sends a query message to the component library, which includes the above function information. The component library can retrieve the application function category and match application functions that meet the above function description. In this way, the application function and its configuration information can be determined.

[0081] 305. The component library sends the first configuration information.

[0082] The component library can first send the first configuration information to the first component, which then converts the first configuration information into natural language and sends the first configuration information described in natural language to the agent. Alternatively, the component library can directly send the first configuration information to the agent; this is not limited here.

[0083] 306. The first component sends first configuration information to the agent, and the agent receives the first configuration information accordingly.

[0084] In response to the request message sent by the agent in step 301, the agent receives the first configuration information. The first configuration information is the configuration information for the first function.

[0085] In summary, the agent sends a request message, which includes task information described in natural language. The first component receives the request message and uses NLM to analyze the task information to determine functional information, including the function category and available network resources. Based on this functional information, the first component sends a query message to the component library, which includes the functional information. The component library receives the query message, performs a query based on the functional information included in the query message, determines a first function and its corresponding first configuration information, and sends this first configuration information to the first component. The first component then forwards this first configuration information to the agent.

[0086] In this embodiment, the intelligent agent can perform function queries using task information described in natural language, without needing to use structured signaling. That is, the intelligent agent does not need to carry query parameters such as function name and category in the request message, providing a more flexible function query mechanism, simplifying the query operation, and improving the convenience of the function query mechanism. The task information is analyzed using the NLM built into the first component. The NLM uncovers potential functional requirements in the task information and outputs the functional information required to execute the task. The component library then queries based on this functional information to determine the configuration information that matches the functional information. This enables intelligent decision-making based on changes in task requirements, improving the intelligence of the function query mechanism.

[0087] Example 2: The IFQ component is an internal component in the component library, meaning the component library includes the IFQ component. Please refer to Figure 5, which is a flowchart illustrating a communication method provided in this application.

[0088] 501. The agent sends a request message to the component library.

[0089] The request message includes task information described in natural language. This task information can describe tasks related to user needs, or tasks undertaken by the agent based on network slice business requirements. This task information is used to determine a first function and first configuration information. The first configuration information is the configuration information for the first function, which is the network function and / or application function required to achieve the task. The agent can send the request message to the component library via the ABI. Correspondingly, the component library receives the request message.

[0090] 502. The component library determines the first configuration information based on the task information in the NLM analysis request message.

[0091] The component library has a built-in NLM (Natural Language Manager). NLM can analyze and understand the task information described in natural language within request messages, determine the functional information required to execute the task, query network load, and output the available network resources for the current task. Functional information includes the name of the function required to execute the task, the required function category, and available network resources. The process of using NLM to analyze task information in request messages can be found in step 302, and will not be repeated here. The component library can use the function category and available network resources from the functional information as query parameters to search and filter the database, determining the first function and first configuration information that matches the given function category and available network resources.

[0092] 503. The component library sends the first configuration information to the agent.

[0093] The component library can send initial configuration information to the agent via the ABI. Correspondingly, the agent receives this initial configuration information through the ABI interface. Since the agent receives structured messages, the ABI interface can act as a message forwarding middleware between the component library and the agent, converting the message format; for example, converting the initial configuration information, described in natural language form output by the component library, into a structured format.

[0094] In this embodiment, the intelligent agent sends a request message to a component library, which includes a first component with a built-in Natural Language Analyzer (NLM). The NLM analyzes the task information described in natural language in the request message and outputs the functional information required to execute the task. The component library includes various network functions and application functions, as well as corresponding configuration information. The component library can directly query the database based on the functional information to determine the first function and corresponding first configuration information that match the functional information. This provides a more flexible functional query mechanism, simplifies the query operation, and improves the convenience and intelligence of the functional query mechanism. Furthermore, as an internal function of the component library, the first component enables the component library to not only have simple search functions but also intelligent demand analysis capabilities. It can directly determine the appropriate functions and configuration information based on the intelligent agent's task information without the first component sending a query message to the component library.

[0095] Example 3: The component library includes the IFQ component and uses NetGPT for auxiliary analysis. Please refer to Figure 6, which is a flowchart illustrating a communication method provided in this application.

[0096] 601. The agent sends a request message to the component library.

[0097] For details, please refer to step 501 above; it will not be repeated here.

[0098] 602. The component library analyzes the request message based on NLM and NetGPT to determine the first configuration information.

[0099] Specifically, the component library can use the task information described in natural language within the NLM analysis request message to determine the functional information required to perform the task. It then uses this functional information to query the database to determine the first function that satisfies the required functional information. When the first function cannot be determined, the component library can use NetGPT for assisted analysis. NetGPT has a larger number of parameters and a higher computational cost than NLM. That is, when the NLM analysis request message cannot determine the first configuration information, NetGPT, with its larger number of parameters and higher computational cost, can be used. NetGPT offers more efficient performance and faster inference speed and is suitable for various task scenarios.

[0100] NetGPT is an extended model based on the generative pre-trained transformer (GPT) architecture. It combines the natural language understanding and generation capabilities of large language models with internet search functionality and external knowledge base query capabilities to better handle tasks requiring real-time data, dynamic information, or external context. NetGPT includes Llama3, neural network-based GPT models (such as GPT-3 and GPT-4), and the T5 model, among others.

[0101] GPT is a Transformer-based autoregressive language model that models the text generation process through multi-layer self-attention and feedforward neural networks. By pre-training on a large amount of text data, it learns the generation relationships between words to predict the probability of the next word in the sequence. The model generates text in an autoregressive manner.

[0102] The T5 model is a unified text-to-text model that can transform all NLP tasks (such as classification, translation, summarization, question answering, etc.) into text generation tasks. Based on an Encoder-Decoder architecture, T5 uses a Transformer encoder to encode the input text and a decoder to generate the target text. By transforming all tasks into a unified text generation problem, a single model can solve multiple types of NLP tasks and possesses strong transfer learning capabilities.

[0103] The following describes the process by which NetGPT assists NLM in analyzing request messages to determine the first configuration information. First, the component library can use the functional information obtained from NLM analysis as prompts for NetGPT, inputting these prompts into NetGPT for analysis assistance. For example, the prompt could be described as: "The agent's task is to check if there are any available vehicles in a certain area; available computing resources include two NVIDIA A100 GPU servers and two Intel Xeon Platinum 8280 CPUs @ 2.70GHz; the IFQ component found the function category as application function - image analysis; please help match functions in the component library to complete the task." Second, NetGPT can output network function information or application function information that can perform the task based on this prompt. NetGPT sends this functional information to the component library, which receives it. The component library can use the functional information output by NetGPT to query the database to obtain the first configuration information, or it can directly convert the functional information output by NetGPT into a semi-structured form and use this semi-structured functional information as the first configuration information; the specifics are not limited here.

[0104] 603. The component library sends the first configuration information to the agent, and the agent receives the first configuration information accordingly.

[0105] For details, please refer to step 503 above; it will not be repeated here.

[0106] In this embodiment, the component library includes a first component, which incorporates NLM and NetGPT. NLM analyzes the task information described in natural language within the request message and outputs the functional information required to execute the task. The component library then queries the database based on this functional information. When the component library cannot find first configuration information matching the functional information, it can use NetGPT to assist in the analysis. The component library can use the functional information obtained from NLM analysis as prompts for NetGPT. NetGPT can then output functional information that enables the task to be executed based on these prompts. The component library can then query the database based on this functional information to determine the first configuration information matching the functional information, or it can directly use the functional information as the first configuration information. NetGPT has a larger number of parameters than NLM, and its computational complexity is greater than that of NLM. Using NetGPT to assist NLM in analyzing task information improves analysis efficiency and is applicable to multi-task scenarios.

[0107] Examples one through three describe the function query mechanism based on the A-Core architecture. Example four below describes the function subscription mechanism based on the A-Core architecture. Please refer to Figure 7, which is a flowchart illustrating a communication method provided in this application.

[0108] 701. The agent sends a subscription message to the component library, which includes functional categories and / or capability expectations described in natural language.

[0109] For functional categories, please refer to the aforementioned descriptions. Capability expectations can be the functional or performance requirements that an intelligent agent expects a certain component or service to possess, which can be specific requirements in terms of performance, response speed, reliability, scalability, etc. For example, for base station perception functions, capability expectations can be described as "expecting to support point cloud data in PLY format; requiring bandwidth to be limited to below 10Mbps; and requiring a perception and recognition accuracy of over 90%", etc.

[0110] The subscription message may also include the agent's identifier. The component library receives the subscription message and can determine whether the agent has been added to the subscription list based on the agent's identifier. If it is determined that the agent has not been added to the subscription list, the component library can add the agent's identifier, function category, and expected capabilities to the subscription list.

[0111] 702. When the component library detects that at least one of the functional categories and capability expectations in the subscription message triggers a change in the subscription list, it sends a first notification message.

[0112] The component library includes a first component. The component library possesses intelligent analysis capabilities, allowing it to use Natural Language Modeling (NLM) to analyze functional categories and capability expectations described in natural language. When the component library detects the launch of a new function satisfying a functional category, or when an already registered function is updated to meet capability expectations, it triggers a change in the subscription list. The component library can proactively send a first notification message to agents subscribed to that functional category or capability expectation to inform them of the new function's capabilities and configuration information. For example, when the base station sensing function in the component library is updated, the component library can proactively send an update message to agents subscribed to the base station sensing function. The first component can periodically query various local functions, models, APIs, etc., based on the capability expectations described in the subscription message for subsequent tasks, and analyze whether each functional characteristic meets the capability expectations. If a function or tool that meets the capability expectations exists, the component library can send a notification message to the agent.

[0113] 703. When the component library detects that a function associated with the agent has been deregistered or has malfunctioned, it sends a second notification message.

[0114] The second notification message is used to inform the agent that the function associated with that agent has been deactivated or has malfunctioned. For example, when the component library detects that the image recognition function has been deactivated, it can send a deactivation message to agents that have subscribed to the image recognition function.

[0115] When the conditions of steps 702 and 703 are met, the agent can receive the first notification message and / or the second notification message.

[0116] In this embodiment, the agent sends a subscription message to the component library. The component library can manage the agent's subscriptions. When the component library detects at least one of the function categories and capability expectations triggering a change in the subscription list, it sends a first notification message. When the component library detects that a function associated with the agent has been deregistered or has malfunctioned, it sends a second notification message. Thus, the component library can manage subscriptions to function categories and / or capability expectations described in natural language without using structured signaling, thereby achieving a more intelligent and convenient function subscription mechanism. The agent does not need to frequently query the component library for function information; when the component library detects a function related to the agent, it automatically sends a notification message. Based on the function categories and capability expectations required by the agent, the component library can provide available function information in a timely manner through the subscription mechanism.

[0117] It is understood that, in order to achieve the functions in the above embodiments, the intelligent agent, the first component, and the component library include hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and method steps of the various examples described in conjunction with the embodiments disclosed in this application, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in a hardware-driven or software-driven manner depends on the specific application scenario and design constraints of the technical solution.

[0118] Figures 8 and 9 are schematic diagrams of possible communication devices provided in embodiments of this application. These communication devices can be used to implement the functions of the intelligent agent, the first component, and the component library in the above method embodiments, and thus can also achieve the beneficial effects of the above method embodiments. In the embodiments of this application, the communication device can be the terminal 120 shown in Figure 1, the base station 110 shown in Figure 1, or a module (such as a chip) applied to the terminal or base station.

[0119] As shown in Figure 8, the communication device 800 includes a processing unit 810 and a transceiver unit 820. The communication device 800 is used to implement the functions of the intelligent agent, the first component, or the component library in the method embodiments shown in Figures 3, 5, 6, or 7.

[0120] When the communication device 800 is used to implement the function of the intelligent agent in the method embodiment shown in FIG3: the transceiver unit 820 is used to send a request message; the transceiver unit 820 is also used to receive first configuration information in response to the request message.

[0121] When the communication device 800 is used to implement the function of the first component in the method embodiment shown in FIG3: the processing unit 810 is used to determine the function information based on the task information in the request message analyzed by NLM; the transceiver unit 820 is used to send a query message to the component library based on the function information; the transceiver unit 820 is also used to receive the first configuration information.

[0122] When the communication device 800 is used to implement the function of the component library in the method embodiment shown in FIG3: the transceiver unit 820 is used to receive a query message; the processing unit 810 is used to determine the first function and the first configuration information based on the query message; the transceiver unit 820 is also used to send the first configuration information.

[0123] When the communication device 800 is used to implement the function of the intelligent agent in the method embodiment shown in FIG5: the transceiver unit 820 is used to send a request message; the transceiver unit 820 is also used to receive first configuration information in response to the request message.

[0124] When the communication device 800 is used to implement the function of the component library in the method embodiment shown in FIG5: the transceiver unit 820 is used to receive the request message; the processing unit 810 is used to analyze the task information in the request message based on NLM to determine the first configuration information; and the transceiver unit 820 sends the first configuration information.

[0125] When the communication device 800 is used to implement the function of the intelligent agent in the method embodiment shown in FIG6: the transceiver unit 820 is used to send a request message; the transceiver unit 820 is also used to receive first configuration information in response to the request message.

[0126] When the communication device 800 is used to implement the function of the component library in the method embodiment shown in FIG6: the transceiver unit 820 is used to receive the request message; the processing unit 810 is used to analyze the task information in the request message based on NLM and NetGPT to determine the first configuration information; the transceiver unit 820 sends the first configuration information.

[0127] When the communication device 800 is used to implement the function of the intelligent agent in the method embodiment shown in FIG7: the transceiver unit 820 is used to send subscription messages; the transceiver unit 820 is also used to receive a first notification message, and / or a second notification message.

[0128] When the communication device 800 is used to implement the function of the component library in the method embodiment shown in FIG7: the transceiver unit 820 is used to receive subscription messages; the transceiver unit 820 is also used to send a first notification message when it detects that at least one of the function category and capability expectation in the subscription message triggers a change in the subscription list; the transceiver unit 820 is also used to send a second notification message when it detects that the function associated with the agent has been deregistered or has malfunctioned.

[0129] For a more detailed description of the processing unit 810 and the transceiver unit 820, please refer to the relevant descriptions in the method embodiments shown in Figures 3, 5, 6, or 7.

[0130] As shown in Figure 9, the communication device 900 includes a processor 910 and an interface circuit 920. The processor 910 and the interface circuit 920 are coupled together. It is understood that the interface circuit 920 can be a transceiver or an input / output interface. Optionally, the communication device 900 may also include a memory 930 for storing instructions executed by the processor 910, input data required for executing instructions by the processor 910, or data generated after the processor 910 executes instructions. Sometimes, the interface circuit 920 can also be understood as part of the processor 910, in which case the communication device 900 includes the processor 910.

[0131] When the communication device 900 is used to implement the method shown in FIG3, FIG5, FIG6 or FIG7, the processor 910 is used to implement the function of the processing unit 810, and the interface circuit 920 is used to implement the function of the transceiver unit 820.

[0132] When the aforementioned communication device is a chip applied to a terminal, the terminal chip implements the functions of the terminal in the above method embodiments. The terminal chip receives information from the base station, which can be understood as the information being first received by other modules in the terminal (such as an RF module or antenna), and then sent to the terminal chip by these modules. The terminal chip sends information to the base station, which can be understood as the information being first sent to other modules in the terminal (such as an RF module or antenna), and then sent to the base station by these modules.

[0133] When the aforementioned communication device is a chip applied to a base station, the base station chip implements the functions of the base station in the above method embodiments. The base station chip receives information from the terminal, which can be understood as the information being first received by other modules in the base station (such as an RF module or antenna), and then sent to the base station chip by these modules. The base station chip sends information to the terminal, which can be understood as the information being sent down to other modules in the base station (such as an RF module or antenna), and then sent to the terminal by these modules.

[0134] In this application, entity A sends information to entity B, either directly or indirectly through other entities. Similarly, entity B receives information from entity A, either directly or indirectly through other entities. Entities A and B can be RAN nodes or terminals, or modules within RAN nodes or terminals. Information transmission and reception can be between RAN nodes and terminals, such as between a base station and a terminal; between two RAN nodes, such as between a CU and a DU; or between different modules within a single device, such as between a terminal chip and other modules of the terminal, or between a base station chip and other modules of the base station.

[0135] It is understood that the processor in the embodiments of this application can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays, or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0136] The method steps in the embodiments of this application can be implemented in hardware or in software instructions executable by a processor. The software instructions can consist of corresponding software modules, which can be stored in random access memory, flash memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, registers, hard disks, portable hard disks, optical discs, or any other form of storage medium well known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. The storage medium can also be a component of the processor. The processor and the storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the ASIC can reside in a base station or terminal. The processor and the storage medium can also exist as discrete components in the base station or terminal.

[0137] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video optical disc; or it can be a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or non-volatile storage medium, or may include both types of storage media.

[0138] In the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions of different embodiments are consistent and can be referenced by each other. The technical features of different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0139] In this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. In the textual description of this application, the character " / " generally indicates an "or" relationship between the preceding and following related objects; in the formulas of this application, the character " / " indicates a "division" relationship between the preceding and following related objects. "Including at least one of A, B, and C" can mean: including A; including B; including C; including A and B; including A and C; including B and C; including A, B, and C.

[0140] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application. The order of the process numbers described above does not imply the order of execution; the execution order of each process should be determined by its function and internal logic.

Claims

1. A communication method, characterized in that, include: The agent sends a request message, which includes task information described in natural language. The task information is used to determine a first function, which is determined based on function information. The function information is obtained by the first component based on the analysis of the task information using a large network model (NLM). In response to the request message, the intelligent agent receives first configuration information, which is the configuration information for the first function.

2. The method according to claim 1, characterized in that, The method further includes: The agent sends a subscription message, which includes functional categories and / or capability expectations described in natural language; In response to the subscription message, the agent receives a first notification message sent by the component library, the first notification message being determined by the component library based on at least one of the function category and the capability expectation.

3. The method according to claim 1 or 2, characterized in that, The functional information includes functional categories and available network resources; the functional categories include application functions and network functions; the first configuration information includes at least one of the following: the identifier of the first function, the category of the first function, the model corresponding to the first function, the required input information of the first function, the optional input information of the first function, and the output sample of the first function.

4. A communication method, characterized in that, include: Receive a request message sent by an intelligent agent, the request message including task information described in natural language; The first component analyzes the task information in the request message based on the Network Model (NLM) to determine the functional information. The first component includes the NLM. Based on the functional information, the first component sends a query message to the component library. The query message is used to determine the first function and the first configuration information, where the first configuration information is the configuration information of the first function.

5. The method according to claim 4, characterized in that, The functional information includes functional categories and available network resources. The query message is a structured message sent through the agent-based interface (ABI), and the query message includes the component library identifier, the functional category, and the available network resources.

6. A communication method, characterized in that, include: Receive a request message sent by an intelligent agent, the request message including task information described in natural language; The component library analyzes the task information in the request message based on the Network Model (NLM) to determine the first configuration information, which is the configuration information of the first function. The first function matches the task information and is registered in the component library. Send the first configuration information.

7. The method according to claim 6, characterized in that, The method further includes: The request message is analyzed based on the network generative pre-trained model NetGPT and the NLM to determine the first configuration information. The number of parameters of NetGPT is greater than the number of parameters of the NLM, and the computational cost of NetGPT is greater than that of the NLM.

8. The method according to claim 6 or 7, characterized in that, The method further includes: The component library receives a query message sent by the first component; Based on the query message, a first function and first configuration information are determined. The first configuration information is the configuration information of the first function. The first function matches the task information. The query message includes function category and available network resources.

9. The method according to any one of claims 6 to 8, characterized in that, The method further includes: Receive subscription messages sent by the intelligent agent, the subscription messages including function categories and / or capability expectations described in natural language; When at least one of the function category and the capability expectation in the subscription message is detected to trigger a change in the subscription list, a first notification message is sent.

10. The method according to any one of claims 6 to 9, characterized in that, The method further includes: When a function associated with the agent is detected to be deregistered or malfunctioning, a second notification message is sent to the agent.

11. A communication device, characterized in that, The device includes a processor and an interface circuit, wherein the interface circuit is used to receive signals from other communication devices and transmit them to the processor or to send signals from the processor to other communication devices, and the processor is used to implement the method as described in any one of claims 1 to 10 through logic circuits or executing code instructions.

12. A computer-readable storage medium, characterized in that, The storage medium stores a computer program or instructions, which, when executed by a communication device, implement the method as described in any one of claims 1 to 10.

13. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by the communication device, the method as described in any one of claims 1 to 10 is implemented.

Images