Cooperative communication method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a cooperative communication method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: receiving a user communication intention sent by a side agent; the user communication intention is obtained by the end-side agent through recognition according to terminal acquisition data of the user terminal and scene description data of the current communication scene; mapping the user communication intention into communication sub-tasks corresponding to the function modules according to the function description information corresponding to the function modules, and sending the communication sub-tasks to the corresponding function modules; the communication subtasks are used for the functional modules to perform reasoning and decision-making so as to generate respective task execution decision-making information; and under the condition that the task execution decision information of each communication sub-task is matched with the user communication intention, indicating each function module to execute the respective communication sub-task according to the respective analyzed task execution decision information. By adopting the method, the overall utilization efficiency of communication network resources can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a collaborative communication method and device, computer equipment, computer readable storage medium and computer program product. BACKGROUND

[0002] With the rapid development of artificial intelligence technology, a user can be provided with convenient auxiliary services by perceiving an environment, making decisions and performing actions through an artificial intelligence agent (AIAgent, also known as "agent").

[0003] A multi-agent system is composed of multiple interacting agents, each of which has certain sensing, computing and execution capabilities, and can communicate with other agents through a communication network. The goal of multi-agent collaborative communication and control is to make multiple agents cooperate by designing reasonable communication and control decisions, so as to complete tasks that a single agent cannot complete independently.

[0004] The multi-agent collaborative communication scheme in the related art mostly focuses on the automation operation and single-point optimization of the network, and it is difficult to perform cross-layer network resource evaluation and collaborative optimization, often resulting in waste or insufficient allocation of communication network resources such as frequency and bandwidth, which restricts the network performance.

[0005] Therefore, the multi-agent collaborative communication scheme in the related art has the problem of low utilization rate of communication network resources. SUMMARY

[0006] Therefore, it is necessary to provide a collaborative communication method, device, computer equipment, computer readable storage medium and computer program product capable of improving network resource utilization.

[0007] In a first aspect, the present application provides a collaborative communication method applied to a network-side agent coordinator, comprising:

[0008] receiving a user communication intention sent by an end-side agent; the user communication intention being obtained by the end-side agent according to terminal acquisition data of a user terminal and scene description data of a current communication scene;

[0009] According to the function description information corresponding to each function module, the user communication intention is mapped to a communication subtask corresponding to each function module, and each communication subtask is sent to the corresponding function module; the communication subtask is used for the function module to make inductive decision to generate its own task execution decision information; the task execution decision information is used to indicate the configuration of the communication network resource associated with the function module; each function module includes an application layer function module in the cloud side intelligent agent and a network layer function module and a physical layer function module in the network side intelligent agent;

[0010] In the case that the task execution decision information of each communication subtask matches the user communication intention, each function module is instructed to execute its own communication subtask according to the task execution decision information parsed by itself.

[0011] In one embodiment, in the case that the task execution decision information of each communication subtask does not match the user communication intention, or in the case that there is a resource scheduling conflict in the task execution decision information of each communication subtask, the method further comprises:

[0012] According to the user communication intention, the weight information corresponding to each function module is adjusted to obtain the adjusted weight information corresponding to each function module; the weight information is used to represent the importance of the corresponding function module to the user communication intention;

[0013] The adjusted weight information corresponding to each function module is issued to the corresponding function module to instruct each function module to re-induce and decide its own communication subtask according to the adjusted weight information to generate new task execution decision information;

[0014] It is judged whether each new task execution decision information matches the user communication intention.

[0015] In one embodiment, the weight information includes module weight, and the adjustment of the weight information corresponding to each function module according to the user communication intention comprises:

[0016] According to the user communication intention, a target function module is determined in each function module; the matching degree of the function description information corresponding to the target function module and the user communication intention is higher than the matching degree of the function description information corresponding to other function modules and the user communication intention;

[0017] The module weight corresponding to the target function module is increased, and the module weight corresponding to the other function modules is decreased.

[0018] In one of the embodiments, the user communication intention is mapped to the communication sub-tasks corresponding to the respective function modules according to the function description information corresponding to the respective function modules, including:

[0019] The function description information currently corresponding to the respective function modules stored in the agent function library is obtained through an information query interface provided by the agent function library.

[0020] The communication service type and the communication network resource type matching the user communication intention are determined.

[0021] The communication service type and the communication network resource type are matched with the function description information currently corresponding to the respective function modules to determine the communication sub-tasks corresponding to the respective function modules.

[0022] In a second aspect, the application provides a collaborative communication method applied to an end-side agent, including:

[0023] Terminal collection data of a user terminal and scene description data of a current communication scene are obtained.

[0024] The communication demand information corresponding to the current communication scene is obtained according to the terminal collection data and the scene description data.

[0025] The communication demand information is quantized, and the quantized communication demand information is taken as the user communication intention.

[0026] The user communication intention is sent to a network-side agent coordinator; the network-side agent coordinator is used to execute the collaborative communication method applied to the network-side agent coordinator as described above.

[0027] In one of the embodiments, the communication demand information corresponding to the current communication scene is obtained according to the terminal collection data and the scene description data, including:

[0028] The terminal collection data and the scene description data are input into a pre-trained local AI inference model to output the communication demand information.

[0029] The communication demand information is a key user experience quality index of the current communication scene identified by the pre-trained local AI inference model according to the terminal collection data and the scene description data, and the communication demand information is determined according to the key user experience quality index and combined with application layer semantic information; the application layer semantic information is used to indicate the network performance demand required by the current communication scene.

[0030] In a third aspect, the application further provides a collaborative communication device applied to a network-side agent coordinator, including:

[0031] receive a user communication intention sent by an end-side agent; the user communication intention is obtained by the end-side agent according to terminal collection data of a user terminal and scene description data of a current communication scene;

[0032] map the user communication intention to a communication subtask corresponding to each function module according to function description information corresponding to each function module, and send each communication subtask to a corresponding function module; the communication subtask is used for reasoning and decision making by the function module to generate task execution decision information of each function module; the task execution decision information is used for indicating a communication network resource associated with the function module; each function module includes an application layer function module in a cloud-side agent and a network layer function module and a physical layer function module in a network-side agent;

[0033] match the task execution decision information of each communication subtask with the user communication intention, and instruct each function module to execute a corresponding communication subtask according to task execution decision information analyzed by each function module.

[0034] In a fourth aspect, the present application further provides a collaborative communication device applied to an end-side agent, comprising:

[0035] a data acquisition unit configured to acquire terminal collection data of a user terminal and scene description data of a current communication scene;

[0036] a demand acquisition unit configured to acquire communication demand information corresponding to the current communication scene according to the terminal collection data and the scene description data;

[0037] a quantification unit configured to quantize the communication demand information, and use the quantized communication demand information as the user communication intention;

[0038] a sending unit configured to send the user communication intention to a network-side agent coordinator; the network-side agent coordinator is configured to execute the collaborative communication method applied to the network-side agent coordinator as described above.

[0039] In a fifth aspect, the present application further provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to realize the steps of the above method.

[0040] In a sixth aspect, the present application further provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to realize the steps of the above method.

[0041] In a seventh aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when executed by a processor, implements the steps of the method described above.

[0042] The above-mentioned cooperative communication method, device, computer device, computer readable storage medium and computer program product are applied to a network side intelligent agent coordinator, a user communication intention sent by an end side intelligent agent is received, the user communication intention is obtained by the end side intelligent agent according to terminal acquisition data of a user terminal and scene description data of a current communication scene, the user communication intention is mapped to communication sub-tasks corresponding to respective function modules according to function description information corresponding to the respective function modules, and the communication sub-tasks are sent to the respective function modules, the communication sub-tasks are used for reasoning and decision-making by the function modules to generate respective task execution decision information, the task execution decision information is used for instructing configuration of communication network resources associated with the function modules, the function modules include an application layer function module in a cloud side intelligent agent and a network layer function module and a physical layer function module in the network side intelligent agent, and in the case that the task execution decision information of the communication sub-tasks matches the user communication intention, the function modules are instructed to execute the respective communication sub-tasks according to the respective task execution decision information analyzed by the function modules.

[0043] Thus, the user communication intention can be more accurately recognized by the terminal-side agent according to the terminal collection data of the user terminal and the scene description data of the current communication scene. After receiving the user communication intention, the network-side agent coordinator can map the user communication intention to the communication sub-tasks corresponding to the respective function modules according to the function description information of the respective function modules, and send the communication sub-tasks to the corresponding function modules, so as to convert the abstract user personalized intention into specific communication sub-tasks that can be understood and executed by the network, for the application layer function module in the cloud-side agent and the network layer function module and the physical layer function module in the network-side agent to make inference decisions to generate respective task execution decision information, to provide more accurate decision basis for communication network resource scheduling, so that in the case that the task execution decision information of each communication sub-task matches the user communication intention, each function module can execute the respective communication sub-task according to the respective task execution decision information. This method of cross-physical layer, network layer and application layer cooperation among the terminal-side agent, the network-side agent and the cloud-side agent can realize terminal-side user communication intention perception, and drive the network-side agent and the cloud-side agent to carry out distributed intelligent decision and communication network resource scheduling, which can consider the user communication intention, the state and constraint of multiple levels (physical layer, network layer and application layer) in a global perspective, jointly optimize the communication network resource arrangement, so that the communication network resources can be dynamically allocated to the most needed place as needed and accurately, avoiding suboptimal solution or resource conflict caused by local optimization, and improving the overall utilization efficiency of spectrum, bandwidth, computing power and other communication network resources. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the description of the embodiments of the present application or the related art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other related drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 An application environment diagram of a collaborative communication method in an embodiment;

[0046] Figure 2 A flowchart of a collaborative communication method in an embodiment;

[0047] Figure 3 A flowchart of another collaborative communication method in an embodiment;

[0048] Figure 4 An interaction diagram of a multi-agent collaborative communication in an embodiment;

[0049] Figure 5 An architecture diagram of a multi-agent collaborative communication for an embodiment;

[0050] Figure 6 A flowchart of a collaborative communication method for another embodiment;

[0051] Figure 7 A structural block diagram of a collaborative communication device for an embodiment;

[0052] Figure 8 A structural block diagram of another collaborative communication device for an embodiment;

[0053] Figure 9 An internal structural diagram of a computer device for an embodiment. DETAILED DESCRIPTION

[0054] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not intended to limit the present application.

[0055] It should be noted that the terms "first", "second", and the like used in the present application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "include" and "have" and any variations thereof used in the present application are intended to cover non-exclusive inclusion. The term "multiple" used in the present application refers to two or more. The term "and / or" used in the present application refers to one of the options or any combination of multiple options.

[0056] Before the embodiments of the present application are further described in detail, the terms and phrases involved in the embodiments of the present application are explained, and the terms and phrases involved in the embodiments of the present application are applicable to the following explanations.

[0057] AI agent (AIAgent): an intelligent entity (software or hardware) with the ability of environment perception, autonomous decision-making, and task execution. The AI agent dynamically perceives and understands the environment state by processing multi-modal data, autonomously generates action strategies based on preset goals or real-time instructions, calls tools to perform corresponding operations, and continuously learns and optimizes its strategies and behaviors according to internal and external feedback, forming a task closed loop. In a multi-agent collaborative system, AI agents can collaborate to complete complex tasks through knowledge sharing and resource coordination.

[0058] Intelligent agent communication network: a new network architecture for the large-scale collaboration needs of AI intelligent agents, which can realize safe and efficient communication, autonomous task collaboration and complementary capability enhancement among large-scale and multi-type AI intelligent agents through standardized protocols, dynamic networking mechanisms, cross-layer resource scheduling, etc.

[0059] Intention-driven network: a new network paradigm that can dynamically perceive and understand the communication intentions of users and applications, and based on these intentions, automatically and intelligently invoke and orchestrate cross-layer resources on the network side to autonomously meet user / application requirements, thereby significantly reducing the operational complexity of manual configuration and intervention, and realizing the fundamental transformation of network roles from "passive connection pipeline" to "active intention understanding and executor".

[0060] The collaborative communication method provided by the embodiments of the present application can be applied to the application environment as shown in Figure 1 The method is applied to a network side intelligent agent coordinator, which is deployed in a network side intelligent agent. The network side intelligent agent coordinator receives a user communication intention sent by an end side intelligent agent; the user communication intention is obtained by the end side intelligent agent according to terminal acquisition data of a user terminal and scene description data of a current communication scene; the network side intelligent agent coordinator maps the user communication intention to communication sub-tasks corresponding to respective function modules according to function description information corresponding to the respective function modules, and sends the communication sub-tasks to the corresponding function modules; the communication sub-tasks are used for reasoning and decision making by the function modules to generate respective task execution decision information; the task execution decision information is used to indicate configuration of communication network resources associated with the function modules; the function modules include an application layer function module in a cloud side intelligent agent and a network layer function module and a physical layer function module in the network side intelligent agent; the network side intelligent agent coordinator instructs the function modules to execute the communication sub-tasks according to the task execution decision information parsed by the respective function modules, in a case where the task execution decision information of the communication sub-tasks matches the user communication intention.

[0061] In an exemplary embodiment, as shown in Figure 2 A collaborative communication method is provided, and the method is applied to a network side intelligent agent coordinator in Figure 1 The following steps S210 to S230 are described by way of example. Wherein:

[0062] Step S210, receiving a user communication intention sent by an end side intelligent agent.

[0063] The user communication intention is obtained by the end side intelligent agent according to terminal acquisition data of a user terminal and scene description data of a current communication scene.

[0064] The terminal collection data can include interactive data generated by a user at a user terminal, such as voice data, image data, video data, text data, and other types of user instruction data input by the user at the user terminal in a current communication scenario. The terminal collection data can also include device state data of the user terminal. It can be understood that the terminal collection data can also include interactive data generated by the user terminal in a historical communication scenario.

[0065] The user terminal includes, but is not limited to, a personal computer, a notebook computer, a smart phone, a tablet computer, a drone, a low-altitude aircraft, an Internet of Things device, and a portable wearable device. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle device, a projection device, and the like. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, and the like. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, and the like.

[0066] The scene description data refers to data for describing a current communication scenario, including but not limited to communication scenario identification data, network state data (network type, transmission rate, and the like), and the like.

[0067] The user communication intent is used to indicate a communication intent of the user in a current communication scenario.

[0068] The current communication scenario includes, but is not limited to, a live communication scenario, a game scenario, a VR (Virtual Reality) conference scenario, and the like.

[0069] In a specific implementation, the end-side agent can identify a user communication intent according to terminal collection data of a user terminal and scene description data of a current communication scenario, and send the user communication intent to a network-side agent coordinator. The network-side agent coordinator can receive the user communication intent sent by the end-side agent.

[0070] In actual application, the end-side agent can obtain terminal collection data of a user terminal and scene description data of a current communication scenario by using a multi-modal sensor, analyze user semantics in combination with a local intent recognition large model, and output a user personalized communication intent.

[0071] Further, the end-side agent can first determine communication demand information of the user in the current communication scenario according to the terminal collection data and the scene description data in combination with the local intent recognition large model to analyze user semantics, and the communication demand information can be used to indicate at least one of a communication service type, a priority, and a key demand of the current communication scenario. Then, the end-side agent can convert the communication demand information into the user communication intent and send the user communication intent to the network-side agent coordinator.

[0072] In step S220, the user communication intention is mapped into the communication sub-tasks corresponding to the respective function modules according to the function description information corresponding to the respective function modules, and the communication sub-tasks are sent to the corresponding function modules.

[0073] The function modules include the application layer function module in the cloud-side intelligent agent and the network layer function module and the physical layer function module in the network-side intelligent agent.

[0074] The function description information refers to information for describing the functions of the corresponding function modules, including but not limited to skill description, performance index, and other data.

[0075] The communication sub-tasks are used for the function modules to make inference decisions to generate respective task execution decision information.

[0076] The task execution decision information is used to indicate the configuration of the communication network resources associated with the function modules.

[0077] The communication network resources can refer to various schedulable resources supporting network communication, including at least one of physical resources (such as antennas, base stations, wireless spectrum, etc.), logical scheduling resources (bandwidth, time slots, frequency blocks, routing, slices, transmission queues, isolation channels, cache spaces, etc.), and function resources (such as QoS (Quality of Service) resources, synchronization resources).

[0078] In a specific implementation, the network-side intelligent agent coordinator can map the user communication intention into the communication sub-tasks to be executed corresponding to the respective function modules according to the function description information corresponding to the respective function modules, and send the communication sub-tasks to be executed to the corresponding function modules.

[0079] In this embodiment, the user communication intention is obtained by quantizing the communication demand information, including specific numerical configurations corresponding to each communication index. For example, in a VR immersive conference scenario, the communication demand includes zero stall and audio-visual synchronization, which is quantized into the user communication intention {resolution 4K@120fps, upper limit of time delay 80ms…}. For another example, in an 8K ultra-high-definition live broadcast scenario, the communication demand includes extreme picture quality and audio-visual synchronization, which is quantized into the user communication intention {resolution 8K@90fps, audio-visual synchronization deviation <10ms…}.

[0080] Thus, in combination with the skill description, performance indicators and other function description information corresponding to each function module, the user communication intention can be mapped to the communication sub-tasks corresponding to each function module. For example, in a live broadcast scene, for the user communication intention {resolution 4K@120fps, minimum code rate 25Mbps, upper limit of latency 80ms, packet tolerance <0.01%, etc.}, the user communication intention is mapped to the communication sub-tasks corresponding to each function module: application layer enhanced video stream, guarantee key frame priority scheduling; network layer creates a live broadcast dedicated slice to guarantee low latency routing; physical layer estimates channel quality and switches to the best channel.

[0081] In some embodiments, the user communication intention can be mapped to the communication sub-tasks corresponding to each function module according to the communication service type and communication network resource type required to meet the user communication intention.

[0082] Thus, after each function module receives the respective communication sub-tasks, it can make inferences and decisions on the respective communication sub-tasks to generate respective task execution decision information.

[0083] Specifically, after the application layer function module receives the communication sub-tasks, during the inference and analysis process of the communication sub-tasks, it can make business KQI (Key Quality Indicators) decision and business feature adjustment decision to generate task execution decision information. Accordingly, the task execution decision information of the application layer function module can include business KQI decision results and business feature adjustment results.

[0084] After the network layer function module receives the communication sub-tasks, during the inference and analysis process of the communication sub-tasks, it can make network layer KPI (Key Performance Indication) decision and communication network resource scheduling decision to generate task execution decision information. Accordingly, the task execution decision information of the network layer function module can include network layer KPI decision results and communication network resource scheduling decision results.

[0085] The network layer KPI can include network routing, slice, and other communication bandwidth-related performance indicators corresponding to the network layer.

[0086] After the physical layer function module receives the communication sub-tasks, during the inference and analysis process of the communication sub-tasks, it can make physical layer KPI decision and communication network resource scheduling decision to generate task execution decision information. Accordingly, the task execution decision information of the physical layer function module can include physical layer KPI decision results and communication network resource scheduling decision results.

[0087] The physical layer KPIs can include performance indicators related to air interface quality characteristics such as subcarrier switching corresponding to the physical layer.

[0088] In this way, each functional module can send its own task execution decision information to the network side intelligent agent coordinator.

[0089] At step S230, if the task execution decision information of each communication subtask matches the user communication intent, each functional module is instructed to execute its own communication subtask according to the task execution decision information it has parsed.

[0090] In a specific implementation, after receiving the task execution decision information of each communication subtask, the network side intelligent agent coordinator can perform demand matching degree detection on the task execution decision information of each communication subtask and the user communication intent. If the task execution decision information of each communication subtask matches the demand of the user communication intent (i.e., demand alignment), each functional module is instructed to execute its own communication subtask according to the task execution decision information it has parsed, and the end side intelligent agent is notified that the user communication intent has been matched.

[0091] If the task execution decision information of each communication subtask does not match the user communication intent, or if there is a resource scheduling conflict in the task execution decision information of each communication subtask, the network side intelligent agent coordinator can resolve the conflict between the decisions of each layer functional module through a dynamic weighted conflict resolution mechanism, feed back the conflict resolution result to each layer functional module, and notify each layer functional module to perform communication subtask parsing again. That is, each layer functional module can be instructed to perform communication subtask parsing again by adjusting the weight information corresponding to each layer functional module.

[0092] In the above collaborative communication method, the network side intelligent agent coordinator receives the user communication intent sent by the end side intelligent agent. The user communication intent is identified by the end side intelligent agent based on terminal acquisition data of the user terminal and scene description data of the current communication scene. The user communication intent is mapped to communication subtasks corresponding to each functional module according to the functional description information of each functional module, and each communication subtask is sent to the corresponding functional module. The communication subtask is used for reasoning and decision making by the functional module to generate its own task execution decision information. The task execution decision information is used to instruct the configuration of the communication network resources associated with the functional module. The functional modules include application layer functional modules in the cloud side intelligent agent and network layer functional modules and physical layer functional modules in the network side intelligent agent. If the task execution decision information of each communication subtask matches the user communication intent, each functional module is instructed to execute its own communication subtask according to the task execution decision information it has parsed.

[0093] Thus, by the terminal-side agent, the user communication intention can be more accurately identified according to the terminal collection data of the user terminal and the scene description data of the current communication scene. After receiving the user communication intention, the network-side agent coordinator can map the user communication intention to the communication sub-tasks corresponding to the respective function modules according to the function description information of the respective function modules, and send the communication sub-tasks to the corresponding function modules, so as to convert the abstract user personalized intention into specific communication sub-tasks that can be understood and executed by the network, for the application layer function module in the cloud-side agent and the network layer function module and the physical layer function module in the network-side agent to make inference decisions to generate respective task execution decision information, to provide more accurate decision basis for communication network resource scheduling, so that in the case that the task execution decision information of each communication sub-task matches the user communication intention, each function module can execute the respective communication sub-task according to the task execution decision information analyzed by itself. This method of cross-physical layer, network layer and application layer cooperation among the terminal-side agent, the network-side agent and the cloud-side agent can consider the user communication intention, the state and constraint of multiple levels (physical layer, network layer and application layer) in a global perspective, jointly optimize the communication network resource arrangement, so that the communication network resources can be dynamically allocated to the places that need them most on demand and accurately, avoiding suboptimal solution or resource conflict caused by local optimization, and improving the overall utilization efficiency of communication network resources such as spectrum, bandwidth and computing power.

[0094] In one embodiment, in the case that the task execution decision information of each communication sub-task does not match the user communication intention, or in the case that there is a resource scheduling conflict in the task execution decision information of each communication sub-task, the method further comprises: adjusting the weight information corresponding to each function module according to the user communication intention to obtain the adjusted weight information corresponding to each function module; the weight information is used to represent the importance of the corresponding function module to the user communication intention; and the adjusted weight information corresponding to each function module is issued to the corresponding function module to instruct each function module to re-infer and decide the respective communication sub-task according to the adjusted weight information of itself to generate new task execution decision information; and it is judged whether the new task execution decision information matches the user communication intention.

[0095] In a specific implementation, each function module has corresponding weight information, and the weight information is used to represent the importance of the corresponding function module to the user communication intention. Specifically, the weight information includes module weight, and in the initialization stage, the module weight corresponding to the three-layer function module is 1 / 3.

[0096] In a case where the task execution decision information of each communication subtask does not match the user communication intention, or in a case where the task execution decision information of each communication subtask exists resource scheduling conflict, the network side intelligent agent coordinator can adjust the weight information corresponding to each functional module according to the user communication intention, obtain the adjusted weight information corresponding to each functional module, and issue the adjusted weight information corresponding to each functional module to the corresponding functional module, so as to instruct each functional module to reason and decide the corresponding communication subtask according to the adjusted weight information, generate new task execution decision information, and send the new task execution decision information to the network side intelligent agent coordinator. The network side intelligent agent coordinator performs the step of demand matching degree detection of the task execution decision information of each communication subtask and the user communication intention, to determine whether the new task execution decision information matches the user communication intention, until the task execution decision information of each communication subtask matches the user communication intention.

[0097] The technical scheme of the embodiment can, in a case where the task execution decision information of each communication subtask does not match the user communication intention, or in a case where the task execution decision information of each communication subtask exists resource scheduling conflict, adjust the weight information corresponding to each functional module according to the user communication intention, obtain the adjusted weight information corresponding to each functional module, issue the adjusted weight information corresponding to each functional module to the corresponding functional module, so as to instruct each functional module to reason and decide the corresponding communication subtask according to the adjusted weight information, generate new task execution decision information, and determine whether the new task execution decision information matches the user communication intention.

[0098] In this way, in a case where the task execution decision information of each communication subtask does not match the user communication intention, or in a case where the task execution decision information of each communication subtask exists resource scheduling conflict, the weight information corresponding to each functional module can be adjusted, and the adjusted weight information corresponding to each functional module can be issued to the corresponding functional module, so as to instruct each functional module to reason and decide the corresponding communication subtask, generate new task execution decision information, and determine whether the new task execution decision information matches the user communication intention, so as to avoid suboptimal solution or resource conflict caused by local optimization. This cyclic manner enables the finally generated task execution decision information to meet the user communication intention.

[0099] In an embodiment, the weight information comprises module weights, and the adjusting the weight information corresponding to each function module according to the user communication intention comprises: determining a target function module in each function module according to the user communication intention; the matching degree of the function description information corresponding to the target function module and the user communication intention is higher than the matching degree of the function description information corresponding to other function modules and the user communication intention; and the module weight corresponding to the target function module is increased, and the module weight corresponding to other function modules is decreased.

[0100] In a specific implementation, in the process of adjusting the weight information corresponding to each function module according to the user communication intention, the network side intelligent agent coordinator can determine the target function module according to the matching degree of the function description information corresponding to each function module and the user communication intention; wherein the matching degree of the function description information corresponding to the target function module and the user communication intention is higher than the matching degree of the function description information corresponding to other function modules and the user communication intention. The other function modules refer to the function modules other than the target function module in each function module.

[0101] In this way, the network side intelligent agent coordinator can increase the module weight corresponding to the target function module and decrease the module weight corresponding to other function modules.

[0102] For example, if the physical layer function module needs to save power in the current task execution decision information, and cannot meet the transmission bandwidth of the network layer function module. That is, there is a resource scheduling conflict between the task execution decision information of the physical layer function module and the task execution decision information of the network layer function module, if it is judged that the matching degree of the function description information corresponding to the network layer function module and the user communication intention is higher, the network side intelligent agent coordinator can increase the module weight corresponding to the network layer function module and decrease the module weight corresponding to the physical layer function module.

[0103] In some embodiments, in the process of each function module re-reasoning and deciding the corresponding communication subtask according to the adjusted weight information of each function module to generate new task execution decision information, each function module can re-reason and decide the corresponding KQI or KPI configuration according to whether the module weight is increased or decreased, to generate new task execution decision information.

[0104] For example, if the module weight corresponding to the application layer function module is decreased, the application layer function module can decrease at least part of the business KQI index data corresponding to the application layer function module, so that the business of the application layer function module is more smooth under the limited communication network resource.

[0105] For another example, if the module weight corresponding to the network layer function module is increased, the network layer function module can increase at least part of the network layer KPI index data (such as bandwidth) corresponding to the network layer function module based on the increased module weight, to improve the communication quality.

[0106] The technical scheme of the embodiment, the weight information includes module weights, a target function module is determined in each function module according to the user communication intention; the matching degree of the function description information corresponding to the target function module and the user communication intention is higher than the matching degree of the function description information corresponding to other function modules and the user communication intention; the module weight corresponding to the target function module is increased, and the module weight corresponding to other function modules is decreased. In this way, the weight information is used to represent the importance of the corresponding function module to the user communication intention, the module weight corresponding to the target function module with a higher matching degree of the corresponding function description information and the user communication intention is increased, and the module weight corresponding to other function modules is decreased, so as to instruct each function module to re-reason and decide the communication subtask of each function module according to the adjusted weight information of each function module, so that the communication network resources can be dynamically allocated to the most needed place as needed and accurately.

[0107] In one embodiment, according to the function description information corresponding to each function module respectively, the user communication intention is mapped into the communication subtask corresponding to each function module respectively, including: acquiring the function description information currently corresponding to each function module stored in the agent function library through an information query interface provided by the agent function library; determining the communication service type and the communication network resource type matched with the user communication intention; matching the communication service type and the communication network resource type with the function description information currently corresponding to each function module to determine the communication subtask corresponding to each function module respectively.

[0108] Among them, the network side agent coordinator includes an agent function library, the agent function library is used for dynamically maintaining the network layer function module and the physical layer function module of the registered network side agent, and the latest function of the application layer function module in the cloud side agent. The function description information of each layer function module is stored, and the latest function description information of different function modules is dynamically synchronized, a query interface is provided for other modules of the agent coordinator, and the matching of task requirements and registered function modules in different types and environments is realized.

[0109] In the specific implementation, in the process of mapping the user communication intention into the communication subtask corresponding to each function module respectively according to the function description information corresponding to each function module respectively, the network side agent coordinator can acquire the function description information currently corresponding to each function module stored in the agent function library through an information query interface provided by the agent function library, determine the communication service type and the communication network resource type matched with the user communication intention, that is, the communication service type and the communication network resource type required to meet the user communication intention, match the communication service type and the communication network resource type with the function description information currently corresponding to each function module, and determine the communication subtask corresponding to each function module respectively.

[0110] For example, if the communication service type matching the user communication intention is real-time live event broadcast, the communication network resource type is low-latency routing, dedicated bandwidth, and stable channel; each layer function module has clear function description information (such as the application layer function module "responsible for video stream encoding and data encapsulation", the network layer function module "responsible for routing planning and bandwidth allocation", and the physical layer function module "responsible for channel evaluation and frequency band switching"), the "real-time live event broadcast" communication service type is matched with the application layer function module "video stream encoding" function, and the "enhanced video stream" subtask is determined; the "low-latency routing, dedicated bandwidth" communication network resource type is matched with the network layer function module "routing / bandwidth management" function, and the "create live broadcast dedicated slice" subtask is determined; and the "stable channel" communication network resource type is matched with the physical layer function module "channel evaluation" function, and the "channel quality estimation and switching" subtask is determined.

[0111] The technical scheme of the embodiment acquires the function description information currently respectively corresponding to each function module stored in the agent function library through the information query interface provided by the agent function library; determines the communication service type and the communication network resource type matching the user communication intention; and matches the communication service type and the communication network resource type with the function description information currently respectively corresponding to each function module, to determine the communication subtasks respectively corresponding to each function module. In this way, by matching the communication service type and the communication network resource type required to meet the user communication intention with the function description information currently respectively corresponding to each function module, the communication subtasks respectively corresponding to each function module can be more accurately allocated, the subtasks undertaken by each function module are more adapted to the capability of the function module, execution inefficiency or failure caused by mismatching of subtasks and module functions is avoided, and directional allocation and efficient utilization of communication network resources are realized.

[0112] In an exemplary embodiment, as shown in Figure 3 , a collaborative communication method is provided, which is applied to an end-side agent in Figure 1 for example, and includes the following steps S310 to S340. Wherein:

[0113] Step S310, terminal acquisition data of a user terminal and scene description data of a current communication scene are acquired.

[0114] The user terminal can be provided with various embedded sensors (such as visual, tactile, and motion sensors), software and hardware interfaces, and application programs, which collect multi-modal data such as text, voice, images, and videos in real time as terminal acquisition data.

[0115] Meanwhile, the end-side agent can also acquire scene description data of the current communication scene.

[0116] At step S320, the communication demand information corresponding to the current communication scenario is obtained according to the terminal collection data and the scene description data.

[0117] In specific implementation, the end-side agent can parse the user semantics in combination with the local intention recognition large model according to the terminal collection data and the scene description data, determine the communication demand information of the user in the current communication scenario.

[0118] At step S330, the communication demand information is quantified, and the quantified communication demand information is taken as the user communication intention.

[0119] In specific implementation, the end-side agent can quantify the communication demand information, and take the quantified communication demand information as the user communication intention. Therefore, the user communication intention can include specific numerical configurations corresponding to each communication index.

[0120] At step S340, the user communication intention is sent to the network-side agent coordinator.

[0121] The network-side agent coordinator is configured to execute the collaborative communication method applied to the network-side agent coordinator as described above.

[0122] In specific implementation, the end-side agent can send the user communication intention to the network-side agent coordinator, so that the network-side agent coordinator executes the collaborative communication method applied to the network-side agent coordinator as described above.

[0123] In the above collaborative communication method, the method is applied to the end-side agent, which obtains terminal collection data of a user terminal and scene description data of a current communication scenario; obtains communication demand information corresponding to the current communication scenario according to the terminal collection data and the scene description data; quantifies the communication demand information, takes the quantified communication demand information as a user communication intention; sends the user communication intention to a network-side agent coordinator; and the network-side agent coordinator is configured to execute the collaborative communication method as described above. In this way, the end-side agent can accurately identify the communication demand information according to the terminal collection data and the scene description data, quantify the communication demand information into the user communication intention, and send the user communication intention to the network-side agent coordinator, thereby providing accurate decision basis for communication network resource scheduling.

[0124] In an embodiment, the communication demand information corresponding to the current communication scenario is obtained according to the terminal collection data and the scene description data, including: inputting the terminal collection data and the scene description data into a pre-trained local AI inference model to output the communication demand information; the communication demand information is that the pre-trained local AI inference model identifies a key user experience quality index corresponding to the current communication scenario according to the terminal collection data and the scene description data, and determines the communication demand information according to the key user experience quality index in combination with application layer semantic information; the application layer semantic information is used to indicate a network performance demand required by the current communication scenario.

[0125] The pre-trained local AI inference model can be an intent recognition large model.

[0126] The key user experience quality index can be a high-priority user experience quality index.

[0127] In a specific implementation, in the process of obtaining the communication demand information corresponding to the current communication scenario according to the terminal collection data and the scene description data, the terminal-side intelligent agent can input the terminal collection data and the scene description data into the pre-trained local AI inference model to output the communication demand information. The pre-trained local AI inference model identifies a key user experience quality (QoE) index (such as frame rate stability in cloud gaming or AR special effect fluency in live streaming) corresponding to the current communication scenario according to the terminal collection data and the scene description data, and determines the communication demand information according to the key user experience quality index in combination with application layer semantic information. The application layer semantic information is used to indicate a network performance demand required by the current communication scenario, for example, the application layer semantic information can include “the user is in a fierce battle and needs low delay” or “the user triggers an AR special effect and needs high bandwidth”.

[0128] The technical scheme of the embodiment, by inputting the terminal collection data and the scene description data into the pre-trained local AI inference model to output the communication demand information, the communication demand information is that the pre-trained local AI inference model identifies a key user experience quality index corresponding to the current communication scenario according to the terminal collection data and the scene description data, and determines the communication demand information according to the key user experience quality index in combination with application layer semantic information; the application layer semantic information is used to indicate a network performance demand required by the current communication scenario. In this way, by obtaining the multi-dimensional data of the terminal collection data and the scene description data, the end-side local AI inference model is used for feature extraction, and complex user interaction content, application semantics and key user experience quality demand are converted into communication demand, to realize accurate user communication intent understanding.

[0129] For the convenience of those skilled in the art, Figure 4An interaction graph for multi-agent collaborative communication is provided. For example... Figure 4 As shown, the edge-side agent uses multimodal sensors to acquire terminal-collected data from the user terminal and scene description data of the current communication scenario. It then combines this with a local intent recognition model to parse user semantics and output personalized user communication intent. The cloud-side agent, based on the user communication intent perceived by the edge, makes business KQI decisions and business feature adjustment decisions for application-layer functional modules, generating task execution decision information for these modules. The network agent responds to the business KQI, making KPI decisions and communication network resource scheduling decisions for network-layer and physical-layer functional modules, generating task execution decision information for both network and physical-layer functional modules. When there is a mismatch between the task execution decision information and the user communication intent, or when there are resource scheduling conflicts among the task execution decision information, the network agent resolves conflicts in the decisions of each functional module through a dynamic weighted conflict resolution mechanism.

[0130] The collaboration among these three intelligent agents spans the physical, network, and application layers of communication, enabling proactive safeguarding of refined user interaction experiences through the communication network and enhancing the intelligence of human-computer interaction.

[0131] In yet other embodiments, Figure 4 On this basis, Figure 5 An architecture diagram for multi-agent collaborative communication is provided to illustrate the architecture of each agent. The task execution decision information in this application can also be named "parsed result". For example... Figure 5 As shown, the terminal-side intelligent agent (responsible for recognizing user communication intent), the network-side intelligent agent (responsible for multi-agent task coordination and sub-task parsing and execution of physical layer functional modules and network layer functional modules), and the cloud-side intelligent agent (responsible for sub-task parsing and execution of application layer functional modules of services).

[0132] Specifically, the network-side agent coordinator in the network-side agent includes the following modules:

[0133] Task decomposition module: Receives user communication intent from the edge agent, and based on the functional description information of the application layer functional modules, network layer functional modules, and physical layer functional modules provided by the agent function library (agent skill library), maps the user communication intent task into communication sub-tasks of multiple functional modules of the physical layer functional modules, network layer functional modules, and application layer functional modules. This achieves accurate matching between multi-dimensional resources and user communication intent, and provides the dependencies and execution order between each communication sub-task.

[0134] Task coordination module: send the communication sub-tasks generated by the task decomposition module to the application layer function module in the cloud-side agent and the network layer function module and the physical layer function module in the network-side agent, and receive the sub-task analysis results (task execution decision information) of the application layer function module, the network layer function module, and the physical layer function module. When the task execution decision information of each layer function module matches the user communication intent, notify each layer function module to execute the respective parsed communication sub-tasks. In the case where the task execution decision information of each layer function module does not match the user communication intent, or in the case where there is a resource scheduling conflict in the task execution decision information of each layer function module, implement conflict resolution of the decision of each layer function module based on a dynamic weighted conflict resolution mechanism technology, and instruct each layer function module to re-analyze the communication sub-tasks, so that multiple function modules across layers (physical, network, and application) share information (state, demand, and constraint) to complete multi-objective decision optimization.

[0135] Agent skill library maintenance: dynamically maintain the latest functions of the application layer function module in the cloud-side agent and the network layer function module and the physical layer function module in the network-side agent. Store the function description information of each layer function module and dynamically synchronize the latest function description information of different function modules to provide a query interface for other modules of the agent coordinator and realize the matching of task requirements and registered function modules in different types and environments.

[0136] Security and authentication: ensure the data and communication security between the agent coordinator and different function modules, as well as between the network-side agent and the cloud-side agent and the end-side agent. Verify the authenticity of the identity of each layer function module accessing the agent coordinator, control the scope of task requests, data access, and capability invocation of each layer function module, and protect sensitive data such as user communication intent, knowledge base of different agents, and operation log.

[0137] The end-side agent includes the following modules:

[0138] Multi-modal data perception module: through various embedded sensors (such as visual, tactile, and motion sensors), software and hardware interfaces, and application programs on the user terminal, real-time collection of multi-modal data such as text, speech, images, and videos, construction of a user interaction database, and provision of user individualized preference calculation basis for user communication intent analysis. Data content includes but is not limited to user instruction data, device state data, and environmental data.

[0139] A user communication intention analysis module: based on multi-modal perception data, combined with scene description data, using an end-side local lightweight AI inference model, real-time analysis of user interaction mode, identification of high priority QoE indicators (such as frame rate stability in cloud gaming, AR special effect fluency in live streaming), and combined with application layer semantic information (understanding "the user is in intense battle and needs low latency" or "the user triggers AR special effects and needs high bandwidth"), the complex user interaction content, application semantics and key user experience quality requirements are converted into communication requirements, and the quantized communication requirement elements (i.e. user communication intention) are transmitted to the network side. The communication requirement elements can include but are not limited to target objects, intention levels, service types, quality of service requirements, required resource types / levels, delay requirements, security levels, priorities, etc.

[0140] In another embodiment, a flowchart of another kind of collaborative communication method is provided, as shown in Figure 6 The steps include:

[0141] In step S610, the end-side intelligent agent obtains terminal acquisition data of the user terminal and scene description data, analyzes user semantics in combination with a local intention recognition large model, identifies user communication intention, and sends it to the network intelligent agent.

[0142] In step S620, after the network intelligent agent receives the user communication intention, the intelligent agent coordinator maps the user communication intention to the communication sub-tasks corresponding to the application layer function module, the network layer function module and the physical layer function module respectively, and sends each communication sub-task to the corresponding function module.

[0143] In step S630, after the application layer function module of the cloud-side intelligent agent receives the communication sub-task sent by the intelligent agent coordinator, it makes inference decision on the communication sub-task, generates task execution decision information of the application layer function module, and sends the task execution decision information to the intelligent agent coordinator.

[0144] In step S640, after the network layer function module of the network intelligent agent receives the communication sub-task sent by the intelligent agent coordinator, it makes inference decision on the communication sub-task, generates task execution decision information of the network layer function module, and sends the task execution decision information to the intelligent agent coordinator.

[0145] In step S650, after the physical layer function module of the network intelligent agent receives the communication sub-task sent by the intelligent agent coordinator, it makes inference decision on the communication sub-task, generates task execution decision information of the physical layer function module, and sends the task execution decision information to the intelligent agent coordinator.

[0146] In step S660, the intelligent agent coordinator performs demand matching degree detection on the task execution decision information of each communication sub-task and the user communication intention.

[0147] If matched, steps S670-S680 are performed:

[0148] In step S670, each function module executes its own communication subtask according to the task execution decision information parsed by itself.

[0149] In step S680, the end-side agent notifies the user that the communication intention has been matched.

[0150] If not matched, steps S690-S6100 are performed:

[0151] In step S690, the weight information corresponding to each function module is adjusted according to the user communication intention.

[0152] In step S610, the adjusted weight information corresponding to each function module is issued to the corresponding function module to instruct each function module to re-reason the communication subtask according to the adjusted weight information, and return to steps S630-S660 until the user communication intention is matched.

[0153] In this way, through the cooperation among the end-side agent, the network agent, and the cloud-side agent across the physical layer, the network layer, and the application layer, the end-side user communication intention is accurately perceived, and the network-side and cloud-side agents are driven to carry out distributed intelligent decision-making and resource arrangement to accurately guarantee the quality of communication services. At the same time, the agents share state, demand, and constraint information, and introduce a dynamic weighted conflict resolution mechanism and multi-objective decision optimization to form a cross-layer joint optimization strategy that meets the global optimum, so that communication network resources can be dynamically allocated to the places where they are needed, avoiding suboptimal solutions or resource conflicts caused by local optimization, improving the overall utilization efficiency of spectrum, bandwidth, computing power, and other communication network resources, and reducing the overall network operation cost while meeting high-demand services.

[0154] To facilitate understanding by those skilled in the art, the collaborative communication method of the present application is described in more detail in two application scenarios: a user live streaming a sports event using an AI phone and a user participating in an immersive conference using VR glasses.

[0155] Embodiment one: User live streams a sports event using an AI phone

[0156] 1. The user uses an AI phone to live stream a badminton event. At the game point of the third set of the women's singles final, the user says: The score is now 20:16, and Team A is leading. We may witness the birth of the women's singles champion at this event within one minute.

[0157] 2. AI phones collect user voice data via microphone, collect data on the number of viewers and the frequency of bullet screen interactions via live streaming software application interfaces, and collect current network status data (network type, transmission rate, etc.) via on-chip SoC (System on Chip).

[0158] 3. Based on the collected multimodal data, the AI ​​phone uses a local AI inference model to analyze the user's real-time communication needs {service type: live sports broadcast; priority: highest level; guarantee duration: 60s; key requirements: zero lag, ultra-high definition video}, and quantifies them into the user's communication intent {resolution 4K@120fps, minimum bitrate 25Mbps, latency limit 80ms, packet loss tolerance <0.01%, etc.}, and finally sends the user's communication intent to the network intelligent agent.

[0159] 4. The network agent receives the user's communication intent at the receiving end and generates a task request based on this intent. The agent coordinator, according to the functional descriptions of the application layer, network layer, and physical layer modules provided in the agent function library, and based on the communication service type and network resource type required to satisfy the user's communication intent, decomposes the user's communication intent into corresponding layer communication sub-tasks: the application layer module enhances the video stream, ensuring priority scheduling of key frames; the network layer module creates dedicated live-stream slices, ensuring low-latency routing; and the physical layer module estimates channel quality and switches to the best channel. The agent coordinator then assigns the communication sub-tasks to the respective layer modules.

[0160] 5. After receiving the application layer communication subtask from the agent coordinator, the application layer functional module of the cloud-side agent parses the communication subtask and obtains the parsing result (task execution decision information): adopts key frame hierarchical scheduling, establishes a three-level transmission queue, and creates a bullet screen traffic isolation channel.

[0161] 6. After receiving the network layer communication subtask from the agent coordinator, the network layer functional module of the network agent parses the communication subtask and obtains the parsing result: Create a hard-isolated live broadcast slice, configure 5QI=83, and reserve 30Mbps uplink bandwidth; Plan a low-latency deterministic route: AI mobile phone → edge UPF of the stadium (distance ≤300m) → SRv6 video dedicated trunk line.

[0162] 7、The physical layer function module of the network agent receives the physical layer communication subtask sent by the agent coordinator, analyzes the communication subtask, and obtains an analysis result: selecting a 28GHz frequency band, allocating 800MHz spectrum resources, configuring 256QAM (Quadrature Amplitude Modulation) and 4x4 UL-MIMO (Uplink Multiple-Input Multiple-Out-put), using 0.5ms TTI (Transmission Time Interval) to shorten the transmission interval, and pre-scheduling uplink authorization every 2ms to eliminate the request delay;

[0163] 8、The agent coordinator receives the analysis results fed back by the application layer function module, the network layer function module and the physical layer function module, detects the demand matching degree of each layer analysis result and the user communication intention, determines that each layer analysis result matches the user communication intention, and notifies each layer function module to execute the corresponding communication subtask according to the analysis result.

[0164] 9、After the cloud-side agent and the network agent each layer function module subtask execution is completed, the network agent notifies the end-side agent that the user communication intention has been matched, and ends this task.

[0165] Example Two: User participates in immersive conference using VR glasses

[0166] 1、The user participates in the immersive conference using the VR glasses, and during the conference, the user says that the picture is a little bit lagging, and there is an audio-visual asynchronization.

[0167] 2、The VR glasses collect the user's voice and collect network state data.

[0168] 3、Based on the collected multi-modal data, the VR glasses use the local AI inference model to analyze the user's real-time communication demand {business type: holographic interactive conference; priority: highest; key demand: zero lag, audio-visual synchronization}, and quantize it into communication intention {resolution 8K@90fps, minimum code rate 1.2Gbps, upper limit of delay 50ms, audio-visual synchronization deviation <10ms, etc.}, and finally send the user communication intention to the network agent.

[0169] 4、The network agent receives the end-side user communication intention, and generates a task request based on the intention. The agent coordinator decomposes the user communication intention into communication sub-tasks of corresponding layers according to the communication service type and communication network resource type required to meet the user communication intention, according to the function description information of the application layer function module, the network layer function module and the physical layer function module provided in the agent function library: the application layer optimizes the media stream synchronization mechanism; the network layer creates a URLLC (Ultra-reliable & Low-latency Communication) slice to guarantee deterministic low latency; the physical layer switches to a high-bandwidth millimeter wave channel to provide sufficient air interface resources. The agent coordinator assigns the communication sub-tasks to the function modules of each layer.

[0170] 5、The application layer function module of the cloud-side agent receives the application layer communication sub-task sent by the agent coordinator, parses the communication sub-task, and obtains the parsing result: deploy an accurate media synchronization mechanism to add high-precision timestamps to audio and video streams; request the network layer to provide the highest priority QoS marking for the media stream.

[0171] 6、The network layer function module of the network agent receives the network layer communication sub-task sent by the agent coordinator, parses the communication sub-task, and obtains the parsing result:

[0172] a) Create a dedicated URLLC slice, configure 5QI (5G QoS Indicator) = 80, reserve 1.5 Gbps downlink bandwidth, and the latency budget is ≤20 ms;

[0173] b) Deploy a deterministic forwarding path: VR glasses → local edge UPF (User Plane Function, deployed in the user's office building, distance <100m) → low-latency metropolitan SRv6 (Segment Routing IPv6, segment routing based on IPv6 forwarding plane) special line → cloud conference server.

[0174] 7、The physical layer function module of the network agent receives the physical layer communication sub-task sent by the agent coordinator, parses the sub-task, and obtains the parsing result: detect that the current connection is in the Sub-6GHz frequency band, switch to 39.5GHz; allocate 800MHz continuous frequency spectrum; use 256QAM modulation and 4x4 DL-MIMO (Downlink Multiple-Input Multiple-Output).

[0175] 8、The intelligent agent coordinator receives the analysis results fed back by the application layer function module, the network layer function module and the physical layer function module, and performs demand matching degree detection on the layer analysis results and the user communication intention. It is found that the analysis results of the network layer function module and the physical layer function module conflict. The conflict reason is that the channel switching scheme (millimeter wave) of the physical layer function module and the routing scheme (depending on the wired connection point of the local edge UPF) of the network layer function module exist physical access point mismatching contradiction, resulting in that the low latency path planned by the network layer cannot provide effective backhaul for the millimeter wave access selected by the physical layer.

[0176] 9、The intelligent agent coordinator realizes conflict resolution of decision of each layer function module based on dynamic weighted conflict resolution mechanism technology, and notifies each layer to redevelop communication subtask analysis.

[0177] 10、The application layer function module of the cloud side intelligent agent reanalyzes the communication subtask and obtains the analysis result: if the physical layer cannot provide >1Gbps stable bandwidth, temporarily reduce the resolution to 6K@90fps or 4K@120fps during the key speaking period, and preferentially guarantee the smoothness and synchronization; maintain the high-precision timestamp synchronization mechanism.

[0178] 11、The network layer function module of the network intelligent agent reanalyzes the communication subtask and obtains the analysis result: based on the effective bandwidth capability fed back by the physical layer, the bandwidth reservation is adjusted to 800Mbps; adjust the routing scheme: user VR glasses→current stable connection Sub-6GHz macro station→through metropolitan bearer network (enable low latency forwarding optimization)→the nearest available edge UPF node (distance about 2km)→SRv6 deterministic network trunk→conference server.

[0179] 12、The physical layer function module of the network intelligent agent reanalyzes the communication subtask and obtains the analysis result: keep the current Sub-6GHz frequency band, activate carrier aggregation; configure downlink 8x8 MU-MIMO (Multi-User Multiple-Input Multiple-Output, multi-user multiple-input multiple-output) or at least 4x4 SU-MIMO (Single-User Multiple-Input Multiple-Output, single-user multiple-input multiple-output); optimize beamforming.

[0180] 13、The intelligent agent coordinator receives the layer analysis results, and performs demand matching degree detection on the layer analysis results and the user communication intention. It is found that the layer analysis results are matched with the user communication intention, and the intelligent agent coordinator notifies each layer function module to execute the corresponding communication subtask according to the analysis result.

[0181] 14、Cloud-side intelligent agent and network intelligent agent each layer function module subtask execution ends, the network intelligent agent informs the end-side intelligent agent user communication intention has been matched, ending this task.

[0182] It should be understood that, although each step in the flowchart involved in each embodiment as described above is shown in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, there is no strict order limitation for the execution of these steps, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be alternately executed with at least part of other steps or steps or stages in other steps. It can be understood that the steps in different embodiments can be freely combined as needed, and various non-contradictory schemes formed by the combination are within the scope of protection of the present application.

[0183] Based on the same inventive concept, the embodiments of the present application also provide a collaborative communication device for implementing the above-mentioned collaborative communication method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more collaborative communication device embodiments provided below can refer to the limitations of the collaborative communication method in the above text, which will not be repeated here.

[0184] In one exemplary embodiment, as shown in Figure 7 A collaborative communication device is provided, applied to a network-side intelligent agent coordinator, comprising: a receiving unit 710, a mapping unit 720, and a matching unit 730, wherein:

[0185] The receiving unit 710 is configured to receive a user communication intention sent by an end-side intelligent agent; the user communication intention is obtained by the end-side intelligent agent according to terminal acquisition data of a user terminal and scene description data of a current communication scene.

[0186] The mapping unit 720 is configured to map the user communication intention into communication subtasks corresponding to respective function modules according to function description information corresponding to the respective function modules, and send the communication subtasks to the respective function modules; the communication subtasks are used for the function modules to make inference decisions to generate respective task execution decision information; the task execution decision information is used to instruct to configure communication network resources associated with the function modules; the function modules include an application layer function module in a cloud-side intelligent agent and a network layer function module and a physical layer function module in a network-side intelligent agent.

[0187] The matching unit 730 is configured to instruct each of the function modules to perform the respective communication subtask according to the task execution decision information parsed by each of the function modules, in a case where the task execution decision information of each of the communication subtasks matches the user communication intention.

[0188] In one of the embodiments, in a case where the task execution decision information of each of the communication subtasks does not match the user communication intention, or in a case where there is a resource scheduling conflict in the task execution decision information of each of the communication subtasks, the device further comprises a weight adjustment unit configured to adjust the weight information corresponding to each of the function modules according to the user communication intention to obtain adjusted weight information corresponding to each of the function modules, the weight information being used to represent the importance of the user communication intention to the corresponding function module, and the adjusted weight information corresponding to each of the function modules being delivered to the corresponding function module to instruct each of the function modules to infer and decide the respective communication subtask according to the adjusted weight information to generate new task execution decision information, and to determine whether the new task execution decision information matches the user communication intention.

[0189] In one of the embodiments, the weight information comprises module weight, and the weight adjustment unit is specifically configured to determine a target function module from the function modules according to the user communication intention, the matching degree of the function description information corresponding to the target function module and the user communication intention being higher than the matching degree of the function description information corresponding to other function modules and the user communication intention, the module weight corresponding to the target function module being increased, and the module weight corresponding to the other function modules being decreased.

[0190] In one of the embodiments, the mapping unit 720 is specifically configured to acquire the function description information currently corresponding to each of the function modules from the agent function library through an information query interface provided by the agent function library, determine the communication service type and the communication network resource type matching the user communication intention, and match the communication service type and the communication network resource type with the function description information currently corresponding to each of the function modules to determine the communication subtask corresponding to each of the function modules.

[0191] In one of the embodiments, as shown in Figure 8 A collaborative communication device applied to an end-side agent is provided, which comprises a data acquisition unit 810, a demand acquisition unit 820, a quantification unit 830 and a sending unit 840.

[0192] The data acquisition unit 810 is configured to acquire terminal collection data of a user terminal and scene description data of a current communication scene.

[0193] The requirement obtaining unit 820 is configured to obtain communication requirement information corresponding to the current communication scenario according to the terminal collection data and the scene description data.

[0194] The quantization unit 830 is configured to quantize the communication requirement information, and take the quantized communication requirement information as the user communication intention.

[0195] The sending unit 840 is configured to send the user communication intention to a network side intelligent agent coordinator, and the network side intelligent agent coordinator is configured to perform the collaborative communication method applied to the network side intelligent agent coordinator as described above.

[0196] In one of the embodiments, the requirement obtaining unit 820 is specifically configured to input the terminal collection data and the scene description data into a pre-trained local AI inference model, and output the communication requirement information; the communication requirement information is obtained by the pre-trained local AI inference model according to the terminal collection data and the scene description data, identifying a key user experience quality indicator corresponding to the current communication scenario, and determining the communication requirement information according to the key user experience quality indicator and application layer semantic information; the application layer semantic information is used to indicate network performance requirements required by the current communication scenario.

[0197] The above-mentioned modules in the collaborative communication device can be realized by software, hardware and combinations thereof in whole or in part. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to the above-mentioned modules.

[0198] In one exemplary embodiment, a computer device is provided, which can be a server, and the internal structure diagram thereof can be as shown in FIG. 8. Figure 9As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store function description information data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a collaborative communication method.

[0199] Those skilled in the art can understand that, Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0200] In one embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in each of the above method embodiments.

[0201] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in each of the above method embodiments.

[0202] In one embodiment, a computer program product is provided, including a computer program, and the computer program is executed by a processor to implement the steps in each of the above method embodiments.

[0203] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0204] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0205] The technical features of the above embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0206] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A cooperative communication method, characterized in that, The method, applied to a network-side agent coordinator, includes: The receiving end-side intelligent agent sends the user's communication intent; the user's communication intent is identified by the end-side intelligent agent based on the terminal-collected data of the user terminal and the scene description data of the current communication scenario; Based on the functional description information corresponding to each functional module, the user's communication intent is mapped to a communication subtask corresponding to each functional module, and each communication subtask is sent to the corresponding functional module. The communication subtask is used by the functional module to perform reasoning and decision-making to generate its own task execution decision information. The task execution decision information is used to instruct the configuration of the communication network resources associated with the functional module. Each functional module includes an application layer functional module in the cloud-side intelligent agent and a network layer functional module and a physical layer functional module in the network-side intelligent agent. When the task execution decision information of each of the communication subtasks matches the user's communication intent, each of the functional modules is instructed to execute its respective communication subtask according to the task execution decision information it has parsed.

2. The method according to claim 1, characterized in that, If the task execution decision information of each of the communication subtasks does not match the user's communication intent, or if there is a resource scheduling conflict in the task execution decision information of each of the communication subtasks, the method further includes: Based on the user's communication intent, the weight information corresponding to each of the functional modules is adjusted to obtain the adjusted weight information corresponding to each of the functional modules; the weight information is used to characterize the importance of the corresponding functional module to the user's communication intent. The adjusted weight information corresponding to each of the aforementioned functional modules is sent to their respective functional modules to instruct each of the aforementioned functional modules to re-reason and make decisions on their respective communication subtasks based on their adjusted weight information, and to generate new task execution decision information. Determine whether the new task execution decision information matches the user's communication intent.

3. The method according to claim 2, characterized in that, The weight information includes module weights, and adjusting the weight information corresponding to each functional module according to the user's communication intent includes: Based on the user's communication intent, a target functional module is determined among the functional modules; the degree of matching between the functional description information corresponding to the target functional module and the user's communication intent is higher than the degree of matching between the functional description information corresponding to other functional modules and the user's communication intent. Increase the module weight corresponding to the target functional module and decrease the module weight corresponding to the other functional modules.

4. The method according to claim 1, characterized in that, The step of mapping the user's communication intent to communication subtasks corresponding to each functional module based on the functional description information corresponding to each functional module includes: The function description information corresponding to each function module stored in the intelligent agent function library can be obtained through the information query interface provided by the intelligent agent function library. Determine the communication service type and communication network resource type that match the user's communication intent; The communication service type and the communication network resource type are matched with the functional description information corresponding to each functional module to determine the communication sub-task corresponding to each functional module.

5. A collaborative communication method, characterized in that, Applied to edge-side intelligent agents, the method includes: Acquire terminal-collected data from the user terminal and scene description data of the current communication scenario; Based on the data collected by the terminal and the scene description data, the communication requirement information corresponding to the current communication scene is obtained; The communication requirement information is quantified, and the quantified communication requirement information is used as the user's communication intent; The user's communication intent is sent to the network-side agent coordinator; the network-side agent coordinator is used to execute the cooperative communication method as described in any one of claims 1 to 4.

6. The method according to claim 5, characterized in that, The step of obtaining communication requirement information corresponding to the current communication scenario based on the data collected by the terminal and the scenario description data includes: The terminal-collected data and the scene description data are input into a pre-trained local AI inference model, which outputs the communication requirement information. The communication requirement information is determined by the pre-trained local AI inference model based on the terminal-collected data and the scene description data, identifying key user experience quality indicators corresponding to the current communication scenario, and determining the communication requirement information based on the key user experience quality indicators and application layer semantic information; the application layer semantic information is used to indicate the network performance requirements required for the current communication scenario.

7. A collaborative communication device, characterized in that, The device, applied to a network-side intelligent agent coordinator, includes: The receiving unit is used to receive the user communication intent sent by the end-side intelligent agent; the user communication intent is identified by the end-side intelligent agent based on the terminal-collected data of the user terminal and the scene description data of the current communication scenario; A mapping unit is used to map the user's communication intent into communication subtasks corresponding to each functional module based on the functional description information corresponding to each functional module, and to send each communication subtask to the corresponding functional module; the communication subtasks are used by the functional modules to perform reasoning and decision-making to generate their respective task execution decision information; the task execution decision information is used to instruct the configuration of the communication network resources associated with the functional module; each functional module includes an application layer functional module in the cloud-side intelligent agent and a network layer functional module and a physical layer functional module in the network-side intelligent agent; The matching unit is used to instruct each of the functional modules to execute their respective communication sub-tasks according to the parsed task execution decision information when the task execution decision information of each of the communication sub-tasks matches the user's communication intent.

8. A collaborative communication device, characterized in that, The device, applied to edge-side intelligent agents, includes: The data acquisition unit is used to acquire terminal-collected data from the user terminal and scene description data of the current communication scenario; The requirement acquisition unit is used to acquire communication requirement information corresponding to the current communication scenario based on the data collected by the terminal and the scenario description data. A quantization unit is used to quantize the communication requirement information and use the quantized communication requirement information as the user's communication intent. A sending unit is configured to send the user's communication intent to a network-side agent coordinator; the network-side agent coordinator is configured to execute the cooperative communication method as described in any one of claims 1 to 4.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4, and / or the steps of the method according to any one of claims 5 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4, and / or the steps of the method according to any one of claims 5 to 6.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4, and / or the steps of the method according to any one of claims 5 to 6.

Images