Intelligent communication system, method and equipment for next generation cloud wireless network

Abstract

The invention provides an intelligent communication system, an intelligent communication method and intelligent communication equipment for a next-generation clouded wireless network. The system comprises: a general control module configured to obtain global state information of a communication network, and determine a target control sub-module from a plurality of control sub-modules based on the global state information by using a pre-trained artificial intelligence large model; and the data communication module is configured to perform extraction processing on communication data based on a pre-trained semantic communication model to obtain a semantic feature vector in response to determining that the semantic control submodule exists in the target control submodule, and transmit the semantic feature vector through a wireless link.

Description

Technical Field

[0001] The present disclosure relates to the field of communication technologies, and in particular to an intelligent communication system, method, and device for next-generation cloud-based wireless networks. Background Art

[0002] Current communication networks are built based on Shannon's information theory, and the communication protocol stack adopts a layered modular design. Communication networks only focus on bit-level transmission performance and lack understanding and utilization of the semantic information of data. As a result, the communication network lacks the ability to process semantic information, resulting in low efficiency in effective information transmission.

[0003] In view of this, how to enable communication networks to process semantic information has become a technical problem that needs to be solved urgently. Summary of the Invention

[0004] In view of this, the purpose of the present disclosure is to propose an intelligent communication system, method and device for the next generation cloud-based wireless network to solve or partially solve the above technical problems.

[0005] Based on the above objectives, the first aspect of the present disclosure proposes an intelligent communication system for next-generation cloud-based wireless networks, the system comprising:

[0006] a universal control module configured to obtain global state information of the communication network and determine a target control submodule from a plurality of control submodules based on the global state information using a pre-trained artificial intelligence macromodel;

[0007] The data communication module is configured to extract and process the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector in response to determining that a semantic control submodule exists in the target control submodule, and transmit the semantic feature vector via a wireless link.

[0008] Based on the same inventive concept, a second aspect of the present disclosure proposes an intelligent communication method for next-generation cloud-based wireless networks, the method comprising:

[0009] The general control module obtains global state information of the communication network and uses a pre-trained artificial intelligence model to determine a target control submodule from multiple control submodules based on the global state information;

[0010] In response to determining that a semantic control submodule exists in the target control submodule, the data communication module extracts and processes the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link.

[0011] Based on the same inventive concept, the third aspect of the present disclosure proposes an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0012] Based on the same inventive concept, a fourth aspect of the present disclosure proposes a non-transitory computer-readable storage medium, which stores computer instructions for causing a computer to execute the method described above.

[0013] From the above, it can be seen that the present disclosure provides an intelligent communication system, method and device for the next generation cloud wireless network. The general control module obtains the global state information of the communication network, and uses a pre-trained artificial intelligence large model to determine the target control submodule from multiple control submodules based on the global state information. In this way, the artificial intelligence large model can accurately determine the target control submodule from multiple control submodules, so that the target submodule can be used to implement resource management and semantic communication, thereby realizing unified control of communication and resource calculation. When a semantic control submodule exists in the target control submodule, it means that the communication data meets the semantic transmission conditions. The data communication module extracts and processes the communication data based on the pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector through a wireless link. In this way, the semantic communication model in the data communication module can be used to process the semantic information, so that the communication system can parse the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0015] Figure 1 This is a schematic diagram of the structure of an intelligent communication system for next-generation cloud-based wireless networks according to an embodiment of the present disclosure;

[0016] Figure 2 This is a schematic structural diagram of a universal control module according to an embodiment of the present disclosure;

[0017] Figure 3 This is a schematic diagram of the structure of a universal intelligent plane platform architecture for 6G cloud networks according to an embodiment of the present disclosure;

[0018] Figure 4 This is a schematic diagram of the structure of the collaborative control submodule according to an embodiment of the present disclosure;

[0019] Figure 5 Schematic diagram of the structure of the semantic communication model of the embodiment of the present disclosure;

[0020] Figure 6 This is a flow chart of an intelligent communication method for next-generation cloud-based wireless networks according to an embodiment of the present disclosure;

[0021] Figure 7 Schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0023] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the described object changes, the relative position relationship may also change accordingly.

[0024] Based on the description of the background technology, the currently widely used fifth-generation mobile communication (5th Generation Mobile Communication Technology, referred to as 5G) system is built on Shannon's information theory. Its communication protocol stack adopts a layered modular design, which only focuses on the transmission performance at the bit level and lacks understanding and utilization of the semantic information of the data. With the continuous development of communication technology and application requirements, the business forms in the sixth-generation mobile communication (6th Generation Mobile Communication Technology, referred to as 6G) era are more diverse and complex. For example, emerging services such as augmented reality (AR), virtual reality (VR), holographic communication, and industrial Internet of Things not only require ultra-high-speed and ultra-low latency communication guarantees, but also put forward higher requirements on the intelligent processing capabilities of the network. The current 5G architecture faces deficiencies in the following aspects:

[0025] On the one hand, 5G networks lack the ability to process semantic information. Current communication systems only transmit bit sequences and do not understand the actual semantic content of the information, making it difficult to meet the semantic communication needs of the 6G era. Semantic communication aims to convey the meaning of information rather than just the bits, and can significantly improve the efficiency of effective information transmission even under bandwidth constraints. However, current network architectures are not designed for this purpose.

[0026] On the other hand, current 5G networks lack unified management and application support for heterogeneous artificial intelligence (AI) models. In future 6G scenarios, different services may require the deployment of different AI models (for example, image recognition models, natural language processing models, etc.) for content understanding and processing. In traditional networks, these models typically operate independently of the communication system and lack coordination with network protocols. Current systems are unable to dynamically load, update, and optimize multiple semantic communication models, making it difficult to flexibly introduce new intelligent functions based on business needs.

[0027] Furthermore, the resource scheduling mechanism of 5G networks primarily focuses on the allocation of communication resources (e.g., spectrum and time-frequency resource blocks), lacking the coordinated scheduling of computing resources and the management of semantic-level resources. In 6G cloud networks, the computing power required by edge computing nodes and large-scale AI models will be closely tied to communication resources. If these continue to operate independently and be scheduled separately, the utilization of communication and computing resources will be uncoordinated, preventing the full potential of computing-network convergence from being realized.

[0028] Furthermore, current networks lack autonomous learning and adaptive capabilities. 5G system parameter configuration and optimization largely rely on pre-set settings and manual adjustments. While some AI-assisted features have been introduced, an end-to-end intelligent self-optimization system has yet to be established. Faced with dynamically changing network environments and service demands, the current architecture struggles to timely update communication strategies or AI models to maintain optimal performance.

[0029] In summary, the current 5G communication architecture cannot meet the development needs of the 6G era, which demands "deep integration of communication and computing" and "intrinsic intelligence." 6G networks require a new architecture to achieve unified control of communication and computing resources, efficient transmission of semantic information, and continuous optimization of AI models to support the rich and diverse next-generation intelligent services.

[0030] Current 5G mobile communications have the following technical problems:

[0031] (1) The network architecture lacks the ability to recognize and process semantic information, making it difficult to achieve efficient communication at the semantic level;

[0032] (2) It is impossible to uniformly manage and schedule heterogeneous AI models, which limits the expansion of network intelligence capabilities;

[0033] (3) The separate scheduling of communication resources and computing resources leads to low efficiency of computing-network integration and cannot meet the 6G requirements for communication-computing collaboration;

[0034] (4) The network lacks self-learning and adaptive mechanisms, making it impossible to optimize communication protocols and model parameters in real time according to environmental changes and business needs.

[0035] To meet the agility, flexibility, and scalability requirements of diverse services, network transformation towards virtualization and cloud-native technologies is inevitable. Cloud-native radio access networks (RANs) are gaining widespread favor among operators worldwide. Cloud-native RANs help operators build cost-effective, operationally efficient, green, and fully automated wireless networks and deliver innovative services.

[0036] Similar to other cloud-native architectures, containerization, microservices architecture, and Kubernetes orchestration are core features of cloud-native RAN. Containerization and Kubernetes orchestration are more implementation methods, while microservices architecture represents a functional reconfiguration. Based on a microservices architecture, the RAN is constructed as a collection of loosely coupled services. Each service performs a specific function and can be independently developed, deployed, and scaled, and can be developed, deployed, and scaled on any infrastructure and container platform. However, microservices architecture is not the only architectural model that can be adopted during RAN transformation. Therefore, RANs that are functionally transformed based on cloud-native design principles will be referred to as service-oriented RANs.

[0037] The deep integration of 6G and artificial intelligence (AI) will drive the development of 6G networks toward intelligence. This integration not only improves operators' production and operational efficiency but also provides AI capabilities as a service to various industries and ordinary consumers, injecting intelligence into various industries and ushering in an intelligent society. To achieve this goal, 6G networks need to leverage technologies such as distributed AI, task-centric AI capabilities, intent-based networking, and AI-as-a-Service to meet their needs for AI functionality, performance, privacy, and personalization, enhance the network's own intelligence capabilities, and thus provide more flexible, efficient, and ubiquitous 6G intelligent solutions.

[0038] As mentioned above, how to realize the processing of semantic information by communication networks has become an important research issue.

[0039] Based on the above description, if Figure 1As shown, the intelligent communication system 100 for the next generation cloud-based wireless network proposed in this embodiment includes:

[0040] The general control module 110 is configured to obtain global status information of the communication network and determine a target control submodule from multiple control submodules based on the global status information using a pre-trained artificial intelligence model.

[0041] In specific implementations, the intelligent communication system 100 for next-generation cloud-based wireless networks is also called the Common Intelligent Plane (CIP) platform architecture for 6G cloud-based wireless networks. The Common Intelligent Plane platform architecture for 6G cloud-based wireless networks is deployed in the control plane and user plane of the 6G communication system. As a new intelligent control and communication plane, CIP connects to base stations (or wireless access network control planes), edge computing platforms, and core network management systems through standard interfaces.

[0042] The general control module 110 is also called the Common Control Function (CCF). The general control module 110 is used to uniformly control the computing resources of the artificial intelligence algorithm and the wireless communication resources. The general control module 110 obtains cross-layer global status information of the entire communication network, such as the current wireless link status of each cell, the computing load of each edge server, and the performance indicators of the registered semantic communication model. Based on this global status information, the general control module 110 generates a globally optimal resource scheduling strategy.

[0043] The universal control module 110 has multi-level control capabilities, from fine-grained to macro-global. At the millisecond level, the universal control module 110 can control physical layer parameters, such as modulation coding, orthogonal frequency division multiplexing (OFDM) symbol mapping, etc.; at the second level, the universal control module 110 can schedule computing tasks to appropriate edge servers, and at a higher level, it can also determine the allocation of network slice resources or service deployment. In this way, integrated real-time control of network communication and computing resources is achieved.

[0044] The data communication module 120 is configured to extract and process the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector in response to determining that a semantic control submodule exists in the target control submodule, and transmit the semantic feature vector via a wireless link.

[0045] In a specific implementation, the intelligent communication system 100 for the next generation cloud-based wireless network includes a data communication module 120 as the core carrier of semantic communication. The data communication module 120 is also called the Intelligent Data Function (IDF).

[0046] The data communication module 120 can receive data input in multiple modalities, for example, communication data including at least one of the following: text, audio, video, and image, and extract compact semantic feature vectors from the communication data using a pre-trained end-to-end semantic communication model.

[0047] The data communication module 120 acts as a pluggable semantic processing layer within the data transmission path. When the general control module 110 decides to enable semantic transmission for any service, communication data is semantically encoded by the data communication module 120 before being sent. Otherwise, communication data follows a traditional protocol stack path assisted by traditional AI-based user plane functional modules. Through this architectural design, the CIP platform is embedded within the 6G cloud network architecture, achieving the convergence and collaboration of communication and computing.

[0048] CIP includes a general control module 110 and a data communication module 120, and is internally coordinated by an intelligent collaboration mechanism. The general control module 110 is mainly used for decision-making and resource scheduling, while the data communication module 120 is used for semantic data processing. The two work together to complete business support.

[0049] Through the above embodiment, the general control module 110 obtains the global state information of the communication network, and uses the pre-trained artificial intelligence large model to determine the target control submodule from multiple control submodules based on the global state information. In this way, the artificial intelligence large model can accurately determine the target control submodule from multiple control submodules, so that the target submodule can be used to implement resource management and semantic communication, thereby realizing unified control of communication and resource calculation. When a semantic control submodule exists in the target control submodule, it means that the communication data meets the semantic transmission conditions. The data communication module 120 extracts and processes the communication data based on the pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link. In this way, the semantic communication model in the data communication module can realize the processing of semantic information, so that the communication system can parse and obtain the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission.

[0050] In some embodiments, the plurality of control submodules include: a computing power control submodule, a wireless control submodule, and a semantic control submodule;

[0051] The computing power control submodule is configured to determine the computing power scheduling of the edge computing nodes that process the communication data;

[0052] The wireless control submodule is configured to allocate spectrum resources and wireless links to communication data;

[0053] The semantic control submodule is configured to extract and process the communication data based on the computing power scheduling and the spectrum resources and based on a pre-trained semantic communication model to obtain a semantic feature vector, and transmit the semantic feature vector through the wireless link.

[0054] In specific implementations, the general control module 110 includes multiple control submodules, each of which manages a specific resource dimension. These include a computing power control submodule, a wireless control submodule, and a semantic control submodule. Each control submodule collaborates and manages resources in its corresponding domain.

[0055] The computing power control submodule is responsible for scheduling computing power for edge computing nodes, the wireless control submodule is responsible for allocating spectrum and wireless link resources, and the semantic control submodule coordinates the management, updating, and loading of the semantic communication module within the data communication module, thereby integrating the management and control of the semantic communication model into the overall network control system. Through this modular design, the universal control module 110 can adopt specialized control strategies for different resources.

[0056] Figure 2 FIG. 1 is a schematic diagram of the structure of the universal control module of the embodiment of the present disclosure. Figure 2 As shown, the computing power control submodule (i.e. Figure 2 The computing power control module in the system obtains computing power data from the computing power platform through the first interface (i.e., northbound interface), and sends computing power control decisions to the computing power platform through the second interface (i.e., southbound interface). Figure 2 The wireless resource control module in the ) obtains resource data from the traditional communication protocol stack through the first interface (i.e., the northbound interface), and sends the resource allocation decision to the traditional communication protocol stack through the second interface (i.e., the southbound interface). Figure 2 The IDF control module in the system obtains communication data from the IDF through the first interface (i.e., the northbound interface) and sends the semantic feature vector to the IDF through the second interface (i.e., the southbound interface).

[0057] Each control submodule operates on a different timescale within the general control module 110. The wireless control submodule typically makes scheduling decisions on a millisecond-level cycle to adapt to rapidly changing wireless links; the computing power control submodule may allocate edge computing tasks on a second-level cycle; and the semantic control submodule schedules the IDF on demand based on service occurrences or performance feedback.

[0058] Through the above scheme, the universal control module 110 unifies the management of various network resources. In addition to the scheduling of conventional wireless communication resources (for example, spectrum bandwidth, time slots, power), the universal control module 110 is also used to allocate and schedule computing resources (for example, the processing power of edge servers, cloud computing instances) and semantic resources (for example, AI semantic communication models and knowledge bases). This unified resource management logic enables the universal control module 110 to balance and coordinate between different resources. For example, when the network is congested, the universal control module 110 can reduce the semantic coding overhead of some services (reduce the amount of data) while reserving more wireless bandwidth for critical services; for example, when the edge computing load is too high, the universal control module 110 can temporarily adjust some semantic processing tasks to be executed locally by the terminal or delayed to balance the computing pressure.

[0059] In some embodiments, the universal control module 110 includes: a first interface, a collaborative control submodule, and a second interface;

[0060] The first interface is configured to obtain global status information of the communication network, wherein the global status information includes: usage of computing resources, performance status of the semantic communication model and real-time status of the communication link;

[0061] The collaborative control submodule is configured to determine a target control submodule from a plurality of control submodules based on the global state information using a pre-trained artificial intelligence large model;

[0062] The second interface is configured to send the target decision to the target control submodule and control the target control submodule to execute the target decision.

[0063] When implementing it specifically, Figure 3 This is a schematic diagram of the structure of the universal intelligent plane platform architecture for 6G cloud network according to the embodiment of the present disclosure. Figure 3 As shown, the universal control module 110 passes through the first interface (ie Figure 3 The general control module 110 obtains global status information from various resource platforms through the second interface (i.e., the northbound interface of the general control function in the general control module). The global status information includes the usage of computing resources from the graphics processing unit (GPU) scheduling platform, the performance status of each semantic communication model, and the real-time status of the communication link. Figure 3 The southbound interface of the general control function in the data communication module 120 sends the target decision to be executed. The general control module 110 determines the target control submodule from multiple control submodules according to the intelligent collaboration mechanism. Figure 3 The intelligent data function in the middle receives user data and outputs OFDM symbols.

[0064] To coordinate the operation of multiple control submodules, the intelligent collaboration mechanism within the collaborative control submodule is the scheduling decision-making core within the general control module 110. The intelligent collaboration mechanism is based on a pre-trained artificial intelligence model, with each control submodule acting as a task head for the model.

[0065] Through offline training and online fine-tuning of large amounts of network scenario data, large artificial intelligence models can learn strategies for collaborative optimization among communication, computing, and semantic resources. Figure 4 FIG. 1 is a schematic diagram of the structure of the collaborative control submodule of the embodiment of the present disclosure. Figure 4 As shown, at runtime, the artificial intelligence large model (i.e. Figure 4 The intelligent collaboration big model in the system receives global status information from each control submodule and the network environment as input, and outputs control decision suggestions for each control submodule at the same time through forward reasoning. For example, after inputting the current business context, the artificial intelligence big model can simultaneously make multiple decisions such as semantic communication model selection, computing task scheduling, and spectrum resource allocation. Each control submodule then performs specific operations based on these suggestions. Through such a unified decision-making agent, cross-module collaborative control is achieved, avoiding the suboptimal results that may be produced by the independent operation of each control submodule. The intelligent collaboration mechanism can also use multi-agent reinforcement learning to perform strategy modeling. Each control submodule acts as an agent, sharing global rewards and states under the framework of the artificial intelligence big model, to achieve multi-task collaborative reinforcement learning decision-making.

[0066] In order to avoid decision conflicts and inefficiencies between each control submodule, the general control module 110 is internally coordinated by an intelligent collaboration mechanism. Centralized intelligent control of multiple resource management is achieved through the general control module 110. For example, when a high-definition video service arrives, the general control module 110 can simultaneously decide to allocate a certain bandwidth for the service (decision made by the wireless control submodule) and reserve GPU computing power on the adjacent mobile edge computing (MEC) server for video semantic compression processing (decision made by the computing power control submodule), and select the semantic communication model that is most suitable for compressing this video (decision made by the semantic control submodule).

[0067] This series of decisions is synchronized and coordinated by the universal control module 110, enabling communication and computing resources to work together for the business. Furthermore, the universal control module 110 provides comprehensive control capabilities, from the OFDM symbol level to the network topology level. This allows the module to perform both real-time optimization of the physical layer (PHY) and media access control (MAC) layer, as well as global resource orchestration across multiple domains, meeting the flexibility and efficiency requirements of 6G networks.

[0068] The intelligent collaboration mechanism uses a large-scale artificial intelligence model as the core decision-making unit, combined with multiple control sub-modules (task heads) with specific functions to achieve multi-task collaborative strategy output. The artificial intelligence large model can be a pre-trained deep learning model or a knowledge-driven intelligent agent that has the ability to learn network operation rules and business models from a large amount of historical data. Through the intelligent collaboration mechanism, the artificial intelligence large model can receive global status information and business demand information gathered from the general control module 110. After internal analysis and processing, multiple task heads in the output layer simultaneously make decisions for different control areas.

[0069] Specifically, each task head of the artificial intelligence large model corresponds to a different control submodule, such as wireless communication resource allocation, computing resource scheduling, semantic communication model control, etc. Each task head generates control strategy parameters for the corresponding field. The wireless resource task head may output spectrum allocation and power control schemes, the computing resource task head outputs task allocation and migration decisions between edge clouds or terminals, and the semantic control task head outputs instructions on which semantic coding model to choose or whether to compress semantic information, etc. These decisions generated by the artificial intelligence large model will be sent to the corresponding control submodules respectively. The wireless control submodule in the general control module 110 executes spectrum and power allocation decisions, the computing power control submodule in the general control module 110 executes computing task scheduling decisions, and the semantic control submodule in the general control module 110 cooperates with the data communication module 120 to execute semantic communication model selection or parameter adjustment decisions. In this way, the originally scattered control tasks are unified at the policy level.

[0070] By utilizing a shared AI master model as the unified decision-making entity, each task head shares the global features and knowledge representation learned within the model, ensuring coordination and consistency among different control strategies. Compared to traditional approaches where independent decisions are made by different modules and then coordinated through manual rules, the intelligent collaborative mechanism directly outputs a multi-dimensional control decision combination optimized under the same objective function. For example, the AI master model can comprehensively consider the current multi-user channel conditions, server load, and the semantic importance of services to simultaneously determine how much bandwidth to allocate to each user, which computing tasks to offload to the edge, and which data to employ high-precision semantic encoding. The strategies output by each task head are mutually balanced. If the AI master model determines that a user's semantic data is not of significant benefit to the overall network, it may reduce the corresponding spectrum allocation to free up resources for other more critical tasks. Alternatively, when edge computing resources are limited, it may appropriately reduce the semantic resolution requirements for some services to reduce computational and data transmission. Through this mechanism, CIP achieves a deep integration of communication, computing, and semantic control, enabling adaptive optimization of overall network performance. The intelligent collaboration mechanism significantly improves the intelligence level of network control, enabling 6G networks to more effectively respond to diverse business needs and complex and changing environments.

[0071] Through the above solution, the general control module 110 adopts a bidirectional open interface, the first interface is used for information aggregation, and the second interface is used for control distribution. The first interface collects operating status and context information from each component of the communication network, including: channel quality indicators and interference levels at the wireless layer, traffic loads of each access point and terminal, and semantic transmission effect indicators (for example, semantic decoding success rate) and model status information from the IDF, etc. Through the cross-layer information aggregation of the first interface, the general control module 110 has a global perception of the operating status and semantic communication performance of the entire communication network. The second interface is used for the general control module 110 to issue control instructions, covering various control commands at the physical layer, network layer and application semantic layer. For example, the general control module 110 sends spectrum and power allocation plans to the base station through the second interface, sends task scheduling instructions to the edge computing node, or notifies the IDF to switch the semantic communication model version used, etc. The second interface ensures that the decisions of the general control module 110 can be quickly transmitted to the execution unit, realizing direct control of network behavior. Through the cooperation of the first interface and the second interface, the universal control module 110 forms a closed-loop control system for the entire communication network, supporting cross-module status perception and instruction execution.

[0072] In some embodiments, the data communication module 120 includes: a first communication module on the transmitting end and a second communication module on the receiving end;

[0073] The first communication module is configured to encode the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and send the semantic feature vector to the second communication module via the wireless link;

[0074] The second communication module is configured to decode the semantic feature vector based on a pre-trained semantic communication model to obtain the communication data.

[0075] In specific implementations, the semantic encoder at the transmitting end encodes the communication data into semantic feature vectors representing the semantic content. The semantic feature vectors undergo necessary channel coding and modulation and are then transmitted over a wireless link. The semantic decoder at the receiving end then reconstructs the communication data or extracts the semantic information from the received semantic feature vectors for service use. For example, for a video stream, the data communication module 120 can encode the original video frames into semantic feature vectors representing the scene content, reconstructing the video image or identifying the target information at the receiving end. The data communication module 120 provides a universal semantic vector interface that represents the semantic feature vectors output by different semantic communication models in a standard data format, thereby decoupling the underlying communication process from the specific semantic communication model. This model-free design enables the system to be compatible with any heterogeneous semantic communication model. To meet different service needs, new semantic communication models can be connected to the data communication module 120 through the dynamic registration mechanism provided by the CIP. The model provider calls the registration interface to submit information such as the semantic communication model type, input and output dimensions, and initial parameters. After verification, the semantic communication model is attached to the data communication module 120 for subsequent communication calls. The data communication module 120 also supports joint source channel coding optimization of the registered model, that is, the semantic encoder is trained in combination with the wireless channel characteristics, so that the semantic encoder considers channel noise and bandwidth limitations in end-to-end optimization, thereby improving the robustness of semantic transmission.

[0076] In order to incorporate the management of semantic communication models into unified scheduling, the embodiment of the present disclosure designs a semantic control submodule within the general control module 110. The semantic control submodule is specifically used to interact with the data communication module 120 to coordinate various operations related to the semantic communication model, including the registration, loading and unloading, and version updates of the semantic communication model. In this way, the semantic communication model management function originally performed independently by the data communication module 120 is integrated into the control framework of the general control module 110, realizing centralized coordination. For example, when any semantic communication model needs to be updated, the data communication module 120 can choose the appropriate time to issue a model update instruction to the data communication module 120 based on the global status information and business needs mastered by the general control module 110, ensuring that the semantic communication model update process has the least impact on the business; when a new semantic communication model is introduced into the system, the data communication module 120 adds the information of the new semantic communication model to the list of global models, and notifies the relevant data communication modules 120 to load and use it on demand. In this way, the general control module 110 can globally track and control the usage and version evolution of the semantic communication model, avoid the disconnection between semantic communication model management and network resource scheduling, and improve the efficiency and coordination of semantic communication model management.

[0077] Figure 5 Schematic diagram of the structure of the semantic communication model of the embodiment of the present disclosure. Figure 5 As shown, data communication module 120 is used for semantic data processing and generation tasks, as well as providing certain AI capabilities for the traditional protocol stack user plane. A general semantic vector interface is designed within data communication module 120 to improve model versatility. Internal functionality is divided into two parts: a semantic communication model execution unit and a semantic communication model management interface, clarifying the responsibilities of data processing and model control.

[0078] The universal semantic vector interface of the data communication module 120 is used to encapsulate the semantic features output by different semantic communication models in a unified data format for transmission and exchange in the network. In this way, no matter which type of semantic communication model is used (for example, a language model for text, a visual model for images, or a semantic communication model of other modalities) to extract semantic information, it will be converted into a standardized vector representation and encapsulation format after processing by the universal semantic vector interface. The universal semantic vector interface can add necessary metadata to the semantic features, where the metadata includes model identification, feature dimensions, semantic confidence, etc., so that the receiving end can correctly interpret and use these semantic data. Through the universal semantic vector interface, semantic information can be transparently transmitted between different nodes without having to consider the specific output form of each semantic communication model, thereby realizing universal carrying of semantic data at the network level.

[0079] Within the data communication module 120, the semantic communication model execution unit is responsible for specific semantic encoding and decoding operations in the semantic communication link, and for tasks such as channel estimation, channel compensation, modulation and demodulation, and feedback retransmission based on the AI model in the traditional communication link. In the semantic communication link, at the transmitting end, the execution unit uses the selected semantic communication model to process the communication data (e.g., text, images) and extract the semantic feature vectors of the content expressed by the communication data; at the receiving end, the execution unit uses the corresponding semantic communication model to decode or infer the received semantic feature vectors and reconstruct the communication data or an equivalent semantic representation.

[0080] The semantic communication model management interface is used to manage and schedule the semantic communication model, and interact with the semantic control submodule of the general control module 110. Through the management interface, the data communication module 120 can receive control instructions from the general control module 110, including but not limited to loading any semantic communication model or unloading any semantic communication model, switching the currently used model version, or starting a model update training, etc., and feedback the status information of the semantic communication model to the general control module 110, including but not limited to the name and version of the currently used semantic communication model, the accuracy index of the semantic communication model operation, etc. This separation of the execution unit and the management interface allows the data communication module 120 to focus on high-speed processing of semantic data, and to flexibly respond to instructions from the control plane, thereby realizing the decoupling of data plane processing and control scheduling.

[0081] In the CIP architecture of the disclosed embodiment, the data communication module 120 supports an end-to-end semantic communication process. Take a semantic communication as an example: the data communication module 120 at the sending end obtains communication data (for example, a text or an image) from the upper-layer application, selects an appropriate semantic communication model through the semantic control submodule, and performs semantic encoding on the communication data to extract a high-dimensional semantic feature vector. The semantic feature vector is encapsulated into a semantic data packet in a standard format through the universal semantic vector interface and sent out via the communication resources of the network. After the data communication module 120 at the receiving end receives the semantic data packet, it parses the semantic feature vector therein through the universal semantic vector interface, and the semantic control submodule calls the corresponding semantic communication model to perform semantic decoding or reconstruction to restore the original communication data or semantically equivalent expression.

[0082] During this process, the data communication module 120 is responsible for ensuring the allocation and quality control of the underlying transmission resources, and coordinates the communicating parties to adopt a consistent semantic communication model through the semantic control submodule. For example, at the beginning of a communication session, the data communication module 120 can instruct the data communication modules 120 of the sending end and the receiving end to load the same version of the semantic communication model to ensure the consistency of encoding and decoding. When the semantic transmission effect is monitored to decrease during communication (for example, the accuracy of the information decoded by the receiving end decreases), the management interface of the data communication module 120 will feed back the performance changes to the general control module 110. After receiving the feedback, the general control module 110 can decide whether to trigger a model adjustment, for example, instructing both parties to switch to an alternative semantic communication model, or start a model update process to improve the quality of the semantic communication model. Through the close collaboration between the general control module 110 and the data communication module 120 in the semantic communication process, CIP can ensure the reliability and efficiency of semantic communication. The communication control layer can perceive the quality changes of semantic transmission and make timely adjustments to resources or models. The semantic processing layer (data communication module 120) also realizes consistent selection and dynamic optimization of models with the support of the general control module 110, thereby achieving efficient and reliable end-to-end semantic information transmission.

[0083] When a new semantic communication model is introduced into the system, the data communication module 120 registers the semantic communication model through the model management interface, adding the semantic communication model's identifier, version number, applicable service type, and other information to the local model library. It also notifies the semantic control submodule of the general control module 110 that the semantic communication model has been registered. Upon receiving the notification, the general control module 110 incorporates the new semantic communication model into the global semantic resource directory so that it can be factored into network scheduling decisions when needed. When selecting a semantic communication model for a particular application, the data communication module 120 preliminarily determines a list of candidate semantic communication models based on the application's semantic requirements and its own model library. It then communicates with the general control module 110 through the management interface to obtain network-side information (e.g., whether the current network bandwidth allows for the use of more complex models). After comprehensively considering the service semantic requirements and network resource conditions, the data communication module 120 ultimately determines which semantic communication model to load for the service's semantic encoding. The data communication module 120 approves this decision and, when necessary, provides policy recommendations (e.g., when network load is too high, recommending that the data communication module 120 select a lighter semantic communication model to reduce communication overhead).

[0084] When the existing semantic communication model needs to be updated and trained (for example, the model performance degrades over time and requires online optimization), the data communication module 120 is responsible for the specific update execution, including preparing training data, calling local computing power for training calculations, and coordinating federated learning with other nodes; the general control module 110 provides peripheral support, for example, by collecting relevant data indicators from the entire network to assist in determining when an update is needed, by allocating network resources to ensure data exchange and communication bandwidth during the model update process, and broadcasting information about the new semantic communication model to other nodes after the update is completed.

[0085] During the registration, selection, and update phases of the semantic communication model, the data communication module 120 leads the actual operations, while the general control module 110 provides global information and scheduling support, ensuring these processes are efficient and consistent with overall network operations. This collaborative control design ensures that the semantic communication model can be retrieved and updated promptly when needed, while also being integrated into the network's unified scheduling framework, rather than operating independently.

[0086] Through the above solution, the first communication module encodes the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and then transmits the semantic feature vector to the second communication module via a wireless link. The second communication module decodes the semantic feature vector based on the pre-trained semantic communication model to obtain the communication data. In this way, the semantic communication model in the data communication module can process semantic information, enabling the communication system to parse the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission.

[0087] In some embodiments,

[0088] The general control module 110 is configured to determine whether the current semantic communication model meets the preset update conditions;

[0089] The general control module 110 is configured to obtain semantic restoration accuracy data of each node in the communication network in response to determining that the current semantic communication model meets a preset update condition;

[0090] The data communication module 120 is configured to use each node to determine updated model parameters of the current semantic communication model through a federated learning algorithm according to the semantic restoration accuracy data, and send the updated model parameters to the central node;

[0091] The data communication module 120 is configured to aggregate the updated model parameters using the central node to obtain global model parameters, and send the global model parameters to each node;

[0092] The data communication module 120 is configured to utilize each node in the communication network to update the current semantic communication model according to the global model parameters to obtain the trained semantic communication model.

[0093] During specific implementation, the data communication module 120 takes the lead in executing the training and updating of the semantic communication model to ensure that the semantic communication model in the data communication module 120 can be dynamically optimized over time. Specifically, based on the trigger mechanism (for example, the semantic transmission performance is lower than the threshold or the periodic time is reached), the semantic communication model update is started, and multiple nodes use federated learning to train the semantic communication model in parallel on local data and aggregate the update results, or perform joint optimization adjustment of the semantic communication model parameters on a single node in combination with the current channel status. During the update process, the general control module 110 provides the necessary resource scheduling and signaling coordination support. After the update is completed, the model effect is evaluated through performance feedback to form a closed-loop mechanism for continuous optimization of the model. Through the above-mentioned model update method, the semantic communication model can continue to evolve with changes in the network environment and business needs, and always maintain optimal performance.

[0094] The initiation of the model update process depends on the preset trigger mechanism. The CIP system continuously monitors various performance indicators related to semantic communication, such as semantic decoding accuracy, semantic transmission delay, model response speed, and network resource consumption. When some indicators are detected to drop below the preset threshold (indicating that the current semantic communication model performance may no longer meet the requirements), a model update process will be triggered. In addition, the system can also trigger semantic communication model updates regularly according to policies (for example, by time period or number of uses) to prevent the performance of the semantic communication model from degrading due to long-term non-training. In addition to threshold triggering and periodic triggering, when a new business scenario is detected or a better external semantic communication model is available, the operation and maintenance personnel can also manually trigger the semantic communication model update. The above trigger mechanism ensures that the semantic communication model update has clear starting conditions and does not cause unnecessary interference to the system.

[0095] During the semantic communication model update process, the data communication module 120 employs a variety of intelligent optimization algorithms to improve the effectiveness of the semantic communication model. One of these algorithms is the federated learning (FL) mechanism: when data communication modules 120 distributed across different nodes in the network (e.g., multiple base stations or user terminals) each possess local training data, a federated learning process can be initiated under the coordination of the universal control module 110. The data communication module 120 at each node uses local data to iteratively train the semantic communication model (e.g., calculate gradients or update local model parameters). The local update results are then uploaded to a central node (which can be a server or cloud server hosting a data communication module 120) for aggregation without leaking the original data, generating improved global model parameters. After aggregation is complete, the universal control module 110 coordinates the distribution of the new global model parameters to the data communication modules 120 at each participating node, enabling the semantic communication model across the entire network to be synchronously updated. This federated learning approach can improve the quality of the semantic communication model by utilizing distributed data across the entire network while protecting the data privacy of all parties. The second is channel-model joint optimization: when training or updating the model, the data communication module 120 takes into account the communication channel status of the current communication network, and by designing a jointly optimized loss function or training process, the parameters of the semantic communication model are adaptively adjusted according to the current channel environment. For example, a penalty can be added to the semantic communication model training objective to make the semantic feature vector encoded by the semantic communication model more robust to typical channel noise, or the compression ratio of the semantic feature vector can be dynamically adjusted according to the channel bit error rate. By integrating the characteristics of the communication link into the model optimization, the new semantic communication model generated can better adapt to the transmission conditions of the communication network and improve the reliability of semantic transmission. The above two model optimization and update methods can be used separately or in combination according to the situation to cope with different model types and network scenarios. The entire semantic communication model update process is led by the data communication module 120 to complete the specific training calculations and parameter updates.

[0096] During the semantic communication model update process, the general control module 110 mainly provides collaboration and auxiliary support. First, under the federated learning solution, the general control module 110 uses its global vision to select the set of nodes participating in the update (for example, select those nodes with poor semantic transmission performance or rich training data), and issues instructions to these nodes to start local training through the second interface, while reserving necessary communication resources for uploading and issuing parameters during the federated learning process (for example, allocating dedicated time-frequency resources for transmitting model update information) to ensure the smooth progress of the federated learning process. During training, the general control module 110 collects feedback from each node in real time, such as training progress, local model performance, etc., and summarizes this information through the first interface and provides it to the data communication module 120 of the central node for reference, thereby assisting the aggregate optimization of global model parameters. For channel-model joint optimization, the general control module 110 is responsible for continuously providing the channel state information of the communication network (for example, the signal-to-noise ratio and bit error rate statistics of each link) to the data communication module 120 during the model training period, so that the semantic communication model optimization process can be based on the latest communication network status. The general control module 110 does not directly participate in the update calculation of model parameters, but ensures the node participation, resource allocation and information path are unobstructed, creating good conditions for the data communication module 120 to successfully complete the update of the semantic communication model.

[0097] After the semantic communication model is updated, the CIP enters the feedback verification phase. The updated semantic communication model is loaded by the data communication module 120 and applied to actual semantic encoding / decoding tasks. The general control module 110 and the data communication module 120 jointly monitor the performance indicators of the new semantic communication model during operation. If the new semantic communication model is found to have improved semantic transmission accuracy and latency compared to the old semantic communication model and meets the expected goals, the management interface confirms the successful closure of the semantic communication model update. The data communication module 120 marks the new semantic communication model as the currently used valid model version and notifies the general control module 110 that the network-wide update is complete. Conversely, if the new semantic communication model fails to achieve the expected performance improvement or introduces new issues (for example, a significant increase in computational overhead leading to increased latency), the system can trigger further measures through the feedback mechanism. For example, the data communication module 120 can initiate a new round of targeted training and optimization, or, if necessary, the general control module 110 can instruct the system to temporarily roll back to the old version of the semantic communication model to ensure that services are not seriously affected. At the same time, the system will record the update details to provide a reference for the next semantic communication model optimization. Through this closed-loop feedback mechanism, the semantic communication model update process forms a cycle of self-correction and continuous improvement, ensuring that the semantic communication model evolves toward optimized performance without deviating from its goals. In summary, the model update method of the disclosed embodiments enables the semantic communication model in CIP to continuously and adaptively evolve based on the network environment and service requirements, maintaining a high level of semantic communication service quality.

[0098] Through the above solution, the current semantic communication model meets the preset update conditions, and semantic restoration accuracy data is obtained for each node in the communication network. Each node uses the federated learning algorithm to determine the updated model parameters of the current semantic communication model based on the semantic restoration accuracy data, and then sends these updated model parameters to the central node. The central node aggregates the updated model parameters to obtain global model parameters, which are then sent to each node. Each node in the communication network then updates the current semantic communication model based on the global model parameters to obtain a trained semantic communication model. This allows the trained semantic communication model to better adapt to the transmission conditions of the communication network, improving the reliability of semantic transmission.

[0099] In some embodiments, the data communication module 120 is further configured to:

[0100] In response to determining that no semantic control submodule exists in the target control submodule, channel estimation processing and channel compensation processing are performed on the communication data based on a pre-trained data communication model to obtain transmission data, and the transmission data is transmitted via a wireless link.

[0101] In specific implementation, the data communication module 120 also includes an AI-based traditional user plane functional module, covering AI models for traditional user plane functions such as channel estimation, channel compensation, modulation and demodulation, feedback and retransmission. The AI model can use machine learning or deep learning methods to improve the performance of traditional communication functions. For example, through the neural network model, more accurate channel estimation and prediction, more effective channel compensation algorithms, more robust modulation and demodulation strategies, and more efficient feedback and retransmission mechanisms can be achieved. The design of the traditional user plane functional module based on the AI model enables the basic communication functions of the user plane to fully utilize the advantages of AI technology to further optimize the overall performance of the network.

[0102] By integrating computing control with communication control on the same platform, the joint scheduling of communication and computing resources is achieved. For example, during peak mission demand, CIP can intelligently coordinate the allocation of spectrum and edge computing power to ensure that heavy computing services and high-speed data transmission are optimized simultaneously, thereby maximizing overall system performance.

[0103] By leveraging semantic communication mechanisms, communication networks only transmit critical information relevant to the business, significantly reducing redundant data transmission. In bandwidth-limited or latency-sensitive scenarios (for example, real-time AR / VR interactions), CIP can deliver high-level semantic content with minimal overhead, improving the efficiency of effective information delivery.

[0104] CIP's intelligent collaborative scheduling flexibly adjusts the allocation ratio of communication and computing resources based on real-time business needs. Compared to traditional fixed allocations, CIP fully utilizes both spectrum and computing resources, allowing idle resources to be promptly reallocated to urgent tasks, significantly improving overall resource utilization.

[0105] Deep learning-based semantic communication and collaborative control are adaptive, automatically adjusting strategies based on environmental changes. For example, when wireless channels degrade, semantic coding adaptively extracts more robust features, and collaborative scheduling prioritizes resources for critical tasks. As a result, the system maintains stable and reliable service quality even in complex scenarios such as high-speed mobility and sudden interference.

[0106] The adoption of a universal interface without model binding and a dynamic model management mechanism enables the system to easily introduce new AI models or service types without major architectural modifications. This ensures the CIP platform's scalability and backward compatibility with emerging new services and models, protecting network investments and simplifying upgrade processes.

[0107] With the above solution, if the semantic control submodule does not exist in the target control submodule, it indicates that the communication data does not meet the semantic transmission conditions. Channel estimation and channel compensation are performed on the communication data based on the pre-trained data communication model to obtain transmission data, which is then transmitted via a wireless link. This pre-trained data communication model enables more accurate channel estimation and prediction, as well as more effective channel compensation.

[0108] Through the above embodiment, the general control module 110 obtains the global state information of the communication network, and uses the pre-trained artificial intelligence large model to determine the target control submodule from multiple control submodules based on the global state information. In this way, the artificial intelligence large model can accurately determine the target control submodule from multiple control submodules, so that the target submodule can be used to implement resource management and semantic communication, thereby realizing unified control of communication and resource calculation. When a semantic control submodule exists in the target control submodule, it means that the communication data meets the semantic transmission conditions. The data communication module 120 extracts and processes the communication data based on the pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link. In this way, the semantic communication model in the data communication module can realize the processing of semantic information, so that the communication system can parse and obtain the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission.

[0109] It should be noted that the method of the embodiments of the present disclosure can be performed by a single device, such as a computer or server. The method of the embodiments of the present disclosure can also be applied in a distributed scenario, where multiple devices cooperate to perform the method. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiments of the present disclosure, and the multiple devices will interact with each other to complete the method.

[0110] It should be noted that the above description is limited to some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] Based on the same inventive concept, corresponding to any of the above embodiments and methods, the present disclosure also provides an intelligent communication method for the next generation cloud wireless network. Figure 6 As shown, the method includes:

[0112] In step 201, the general control module obtains global status information of the communication network, and uses a pre-trained artificial intelligence model to determine a target control submodule from multiple control submodules based on the global status information.

[0113] In practice, the universal control module 110 obtains global, cross-layer state information for the entire communication network, such as the current wireless link status of each cell, the computing load of each edge server, and the performance indicators of registered semantic communication models. Based on this global state information, the universal control module 110 generates a globally optimal resource scheduling strategy.

[0114] In step 202 , in response to determining that a semantic control submodule exists in the target control submodule, the data communication module extracts and processes the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link.

[0115] In a specific implementation, the data communication module 120 can receive data input in multiple modalities, for example, communication data including at least one of the following: text, audio, video, and image, and extract compact semantic feature vectors from the communication data using a pre-trained end-to-end semantic communication model.

[0116] Through the above embodiment, the general control module 110 obtains the global state information of the communication network, and uses the pre-trained artificial intelligence large model to determine the target control submodule from multiple control submodules based on the global state information. In this way, the artificial intelligence large model can accurately determine the target control submodule from multiple control submodules, so that the target submodule can be used to implement resource management and semantic communication, thereby realizing unified control of communication and resource calculation. When a semantic control submodule exists in the target control submodule, it means that the communication data meets the semantic transmission conditions. The data communication module 120 extracts and processes the communication data based on the pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link. In this way, the semantic communication model in the data communication module can realize the processing of semantic information, so that the communication system can parse and obtain the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission.

[0117] In some embodiments, before step 202, the method further includes:

[0118] Step 202A: The general control module determines whether the current semantic communication model meets a preset update condition.

[0119] In step 202B, the general control module obtains semantic restoration accuracy data of each node in the communication network in response to determining that the current semantic communication model meets a preset update condition.

[0120] In step 202C, the data communication module uses each node to determine updated model parameters of the current semantic communication model through a federated learning algorithm according to the semantic restoration accuracy data, and sends the updated model parameters to the central node.

[0121] In step 202D, the data communication module aggregates the updated model parameters using the central node to obtain global model parameters, and sends the global model parameters to each node.

[0122] Step 202E: The data communication module uses each node in the communication network to update the current semantic communication model according to the global model parameters to obtain the trained semantic communication model.

[0123] In specific implementation, during the network initialization phase or when introducing new services, the corresponding semantic communication model is registered and deployed through the CIP data communication module 120, and the semantic communication model is preliminarily trained and adjusted using mechanisms such as federated learning to adapt the semantic communication model to the current network environment.

[0124] When a user service arrives (for example, the data flow of an AR application), the base station or edge server in the communication network calls the general control module 110 of the CIP to make a decision and schedule. The general control module 110 comprehensively considers the service requirements and network status, and the intelligent collaboration mechanism collaboratively determines the data transmission strategy of the service. For example, the general control module 110 determines whether the service requires semantic compression transmission. If so, it selects a suitable semantic communication model (provided by IDF) to encode the data and allocates the required communication resources and computing resources to the service (for example, edge computing nodes are used to assist in processing decoding).

[0125] Subsequently, the business data is processed by the CIP platform: for communication data that requires semantic transmission, the semantic feature vector is extracted by the semantic encoder of the data communication module 120 and sent through the wireless link; for communication data that does not require semantic processing, the traditional user-side functional module based on the AI model performs channel estimation, channel compensation, modulation and demodulation, feedback retransmission and other processing before sending it through the wireless link.

[0126] At the receiving end, the CIP data communication module 120 calls the corresponding semantic decoder to decode and reconstruct the received semantic feature vector to restore the original business content or usable semantic information, while the conventional data is demodulated and decoded by the traditional user plane function module based on AI.

[0127] During the entire process, the general control module 110 continuously monitors the communication channel status and service quality indicators, and issues new control commands through the second interface to make real-time adjustments when necessary. For example, it dynamically increases or decreases the spectrum resources allocated to the service or calls other edge computing power to assist in adapting to changes.

[0128] Finally, during the completion or execution of the business, CIP collects data transmission performance feedback for this business (for example, semantic restoration accuracy, latency jitter), and uses the performance feedback data to trigger the update and optimization process of the semantic communication model (for example, further federated learning iteration or adjustment of collaborative decision-making strategies) to gradually improve the system's support efficiency for subsequent similar businesses.

[0129] Through the above solution, the CIP platform realizes closed-loop control from business requests to resource scheduling, data transmission, result feedback and model self-optimization, greatly improving the intelligent response capability of the 6G network in the face of complex business scenarios.

[0130] Through the above embodiment, the general control module 110 obtains the global state information of the communication network, and uses the pre-trained artificial intelligence large model to determine the target control submodule from multiple control submodules based on the global state information. In this way, the artificial intelligence large model can accurately determine the target control submodule from multiple control submodules, so that the target submodule can be used to implement resource management and semantic communication, thereby realizing unified control of communication and resource calculation. When a semantic control submodule exists in the target control submodule, it means that the communication data meets the semantic transmission conditions. The data communication module 120 extracts and processes the communication data based on the pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link. In this way, the semantic communication model in the data communication module can realize the processing of semantic information, so that the communication system can parse and obtain the actual semantic content carried by the communication data, thereby significantly improving the efficiency of effective information transmission.

[0131] For the convenience of description, the above devices are described as being functionally divided into various modules. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0132] The apparatus of the above embodiment is used to implement the corresponding intelligent communication method for the next generation cloud wireless network in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0133] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements the intelligent communication method for the next-generation cloud-based wireless network described in any of the above embodiments.

[0134] Figure 7 10 is a schematic diagram showing a more specific hardware structure of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are communicatively connected to each other within the device via the bus 1050.

[0135] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0136] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 1020 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0137] The input / output interface 1030 is used to connect input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0138] The communication interface 1040 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (e.g., USB (Universal Serial Bus), network cable, etc.) or a wireless method (e.g., mobile network, WIFI (Wireless Fidelity), Bluetooth, etc.).

[0139] The bus 1050 comprises a path for transmitting information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).

[0140] It should be noted that although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040, and the bus 1050, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0141] The electronic device of the above embodiment is used to implement the corresponding intelligent communication method for the next generation cloud wireless network in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0142] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the intelligent communication method for the next-generation cloud-based wireless network as described in any of the above embodiments.

[0143] The computer-readable media of this embodiment include permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0144] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the intelligent communication method for the next generation cloud-based wireless network as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0145] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a computer program product, including computer program instructions. When the computer program instructions are run on a computer, the computer executes the intelligent communication method for the next-generation cloud-based wireless network as described in any of the above embodiments, which has the beneficial effects of the corresponding method embodiments and will not be repeated here.

[0146] It is understandable that before using the technical solutions of each embodiment of the present disclosure, the type, scope of use, usage scenarios, etc. of the personal information involved will be informed to the user in an appropriate manner, and the user's authorization will be obtained.

[0147] For example, in response to a user's active request, a prompt message is sent to the user to clearly inform the user that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operation of the disclosed technical solution based on the prompt message.

[0148] As an optional but non-limiting implementation, in response to a user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. Furthermore, the pop-up window may also contain a selection control for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0149] It is understandable that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0150] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure is limited to these examples. Based on the concept of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.

[0151] In addition, to simplify the description and discussion, and so as not to obscure the embodiments of the present disclosure, known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, devices may be shown in the form of block diagrams to avoid obscuring the embodiments of the present disclosure, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure are to be implemented (i.e., these details should be fully within the purview of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it will be apparent to those skilled in the art that the embodiments of the present disclosure may be implemented without these specific details or with variations in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0152] Although the present disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0153] The embodiments of the present disclosure are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the present disclosure. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.

Claims

1. An intelligent communication system for next-generation cloud-based wireless networks, characterized in that: The system comprises: a universal control module configured to obtain global state information of the communication network and determine a target control submodule from a plurality of control submodules based on the global state information using a pre-trained artificial intelligence macromodel; The data communication module is configured to extract and process the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector in response to determining that a semantic control submodule exists in the target control submodule, and transmit the semantic feature vector via a wireless link.

2. The system according to claim 1, wherein: The multiple control submodules include: a computing power control submodule, a wireless control submodule and a semantic control submodule; The computing power control submodule is configured to determine the computing power scheduling of the edge computing nodes that process the communication data; The wireless control submodule is configured to allocate spectrum resources and wireless links to communication data; The semantic control submodule is configured to extract and process the communication data based on the computing power scheduling and the spectrum resources and based on a pre-trained semantic communication model to obtain a semantic feature vector, and transmit the semantic feature vector through the wireless link.

3. The system according to claim 1, wherein: The universal control module includes: a first interface, a collaborative control submodule and a second interface; The first interface is configured to obtain global status information of the communication network, wherein the global status information includes: usage of computing resources, performance status of the semantic communication model and real-time status of the communication link; The collaborative control submodule is configured to determine a target control submodule from a plurality of control submodules based on the global state information using a pre-trained artificial intelligence large model; The second interface is configured to send the target decision to the target control submodule and control the target control submodule to execute the target decision.

4. The system according to claim 1, wherein: The data communication module includes: a first communication module at the sending end and a second communication module at the receiving end; The first communication module is configured to encode the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and send the semantic feature vector to the second communication module via the wireless link; The second communication module is configured to decode the semantic feature vector based on a pre-trained semantic communication model to obtain the communication data.

5. The system according to claim 1, wherein: The general control module is configured to determine whether the current semantic communication model meets the preset update conditions; The general control module is configured to obtain semantic restoration accuracy data of each node in the communication network in response to determining that the current semantic communication model meets a preset update condition; The data communication module is configured to use each node to determine updated model parameters of the current semantic communication model through a federated learning algorithm according to the semantic restoration accuracy data, and send the updated model parameters to a central node; The data communication module is configured to aggregate the updated model parameters using the central node to obtain global model parameters, and send the global model parameters to each node; The data communication module is configured to utilize each node in the communication network to update the current semantic communication model according to the global model parameters to obtain the trained semantic communication model.

6. The system according to claim 1, wherein: The data communication module is further configured to: In response to determining that no semantic control submodule exists in the target control submodule, channel estimation processing and channel compensation processing are performed on the communication data based on a pre-trained data communication model to obtain transmission data, and the transmission data is transmitted via a wireless link.

7. An intelligent communication method for next-generation cloud-based wireless networks, characterized in that: The method comprises: The general control module obtains global state information of the communication network and uses a pre-trained artificial intelligence model to determine a target control submodule from multiple control submodules based on the global state information; In response to determining that a semantic control submodule exists in the target control submodule, the data communication module extracts and processes the communication data based on a pre-trained semantic communication model to obtain a semantic feature vector, and transmits the semantic feature vector via a wireless link.

8. The method according to claim 7, characterized in that Before the data communication module extracts and processes the communication data based on the pre-trained semantic communication model to obtain the semantic feature vector, the method further includes: The general control module determines whether the current semantic communication model meets the preset update conditions; The general control module obtains semantic restoration accuracy data of each node in the communication network in response to determining that the current semantic communication model meets a preset update condition; The data communication module uses each node to determine the updated model parameters of the current semantic communication model through a federated learning algorithm according to the semantic restoration accuracy data, and sends the updated model parameters to the central node; The data communication module aggregates the updated model parameters using the central node to obtain global model parameters, and sends the global model parameters to each node; The data communication module uses each node in the communication network to update the current semantic communication model according to the global model parameters to obtain the trained semantic communication model.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the program, the method according to any one of claims 7 to 8 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a computer to execute the method according to any one of claims 7 to 8.

Images