BNC and 5G private network-based fixed mobile arithmetic intelligent fusion device and method

Abstract

The application discloses a kind of solid migration arithmetic intelligent fusion equipment and method based on BNC and 5G special network, solid migration arithmetic intelligent fusion equipment includes hardware layer, resource sharing layer, network function fusion layer and intelligent application layer, deployment in user side.Through constructing cloud intelligent agent application store;Solid migration network arithmetic intelligent fusion equipment is deployed in user side, accesses fixed network, mobile network terminal equipment;Realize that terminal equipment is forwarded through solid migration network arithmetic intelligent fusion equipment network function fusion layer between data cooperation, local unloading and safety isolation, intelligent agent application deployment, start power, storage, database and software resource scheduling, dynamically allocate shared resources.The application is characterized in that it solves the problem of high delay, complex operation and maintenance and resource waste caused by traditional solid migration network being separately arranged, controls the end-to-end delay within 10 milliseconds, and improves the resource utilization rate by more than 30%.It greatly shortens the industrial intelligent application landing period and provides an integrated solution for industrial internet scenarios.

Description

Technical Field

[0001] This invention relates to the field of 5G private network and fixed broadband private network technology, and is particularly applicable to fixed-mobile computing-data-intelligence converged equipment and methods based on BNC and 5G private networks. Background Technology

[0002] The Broadband Core Network (BNC) is a new type of network architecture for broadband networks. It is mainly responsible for access control, service management, and traffic forwarding of broadband services, and adopts a control-transfer-forwarding separation architecture. The BNC control plane (BNC-CP) and user plane (BNC-UP) are, in principle, centrally deployed to achieve fast forwarding and routing management of user traffic.

[0003] 5G private networks are dedicated communication networks built on the 5G technology architecture to meet the customized needs of vertical industries. They incorporate a UPF (User Plane Function) deployment architecture and combine key technologies such as network slicing, SDN, NFV, and MEC (Medium-and-Edge Computing) to achieve end-to-end communication services isolated from the public network. 5G private networks have been widely used in industrial control, port logistics, telemedicine, and smart cities. The UPF architecture has become a core deployment solution in scenarios such as energy mines, intelligent manufacturing, and smart ports, effectively solving the limitations of traditional public networks in terms of access reliability, bandwidth capacity, and data security. However, existing solutions still have some shortcomings, such as high deployment costs and energy consumption due to the separation of UPF and MEC networks, cumbersome network configuration in complex scenarios, difficulty in simultaneously meeting the collaborative needs of low latency and wide-area access in some architectures, and insufficient adaptability for seamless access for end users. These issues urgently require further optimization of performance and adaptability through technological innovation.

[0004] Meanwhile, in the traditional model of telecom operators providing network services to government and enterprise users, fixed-line and mobile networks are planned, deployed, and maintained independently. This makes it difficult to meet the "fixed-mobile integration" demands of government and enterprise users' production businesses. This results in separate hardware equipment (fixed-line switches and routers vs. mobile base stations and core network elements), network management systems, and maintenance teams for fixed and mobile networks. Government and enterprise users need to handle fixed-line leased line faults and mobile terminal access issues separately by connecting to multiple interfaces of the operator. Operators repeatedly invest in infrastructure resources such as data centers and transmission links, and perform independent troubleshooting, parameter configuration, upgrades, and maintenance for the two networks, leading to high network operation and maintenance costs and low management efficiency. Moreover, fixed-line bandwidth resources are relatively stable, but suffer from "insufficient peak bandwidth and wasted off-peak bandwidth," such as bandwidth shortages during enterprise office hours and idle resources during non-working hours. Mobile network bandwidth is greatly affected by the wireless channel environment, exhibiting fluctuations, and package bandwidth is usually ordered based on peak demand, resulting in low resource utilization during off-peak hours. Because the separation of fixed and mobile networks prevents the dynamic scheduling and complementarity of fixed and mobile resource pools, government and enterprise users need to purchase both fixed-line leased lines and mobile network traffic packages to ensure business continuity, resulting in double bandwidth costs. Meanwhile, operators, due to the inability to share resources, find it difficult to improve overall resource utilization through flexible scheduling, thus limiting the potential for cost reduction and efficiency improvement.

[0005] Furthermore, government and enterprise users have high requirements for data localization and isolation, especially in sensitive industries such as manufacturing and finance. In the traditional fixed-mobile separation model, the fixed network ensures data security through dedicated physical isolation or VPN tunnels, while the mobile network relies on core network authentication and encryption. The security policies of the two networks are configured independently, lacking a unified security management system. When business data is transmitted across fixed and mobile networks, security policy mismatches can easily occur. For example, when a mobile terminal accesses the fixed network intranet, the lack of unified identity authentication and access control poses a risk of data leakage or unauthorized intrusion. Independent security audit systems cannot achieve full-process data traceability across networks, making it difficult to meet compliance requirements. Moreover, the business needs of different government and enterprise sectors vary significantly. For example, the industrial internet requires low-latency, high-reliability, low-bandwidth transmission, while smart healthcare requires high-bandwidth, low-packet-loss image data transmission. In the traditional fixed-mobile separation model, network functions and parameter configurations are relatively fixed, making it difficult to flexibly customize them according to industry needs. For example, it is impossible to allocate differentiated fixed-mobile converged resources for different businesses of the same enterprise (such as production control and office internet access), and it is also difficult to quickly respond to the elastic needs of enterprises to expand or shrink capacity, resulting in low adaptability of network services to the actual business scenarios of government and enterprise users. Summary of the Invention

[0006] The purpose of this invention is to provide a fixed-mobile computing, data and intelligence convergence device and method based on BNC and 5G private network, which can solve the problems of poor service coordination, high operation and maintenance costs and low resource utilization caused by the separate deployment of traditional fixed network and 5G mobile private network.

[0007] To achieve the above objectives, the fixed-mobile computing-data-intelligence converged device based on BNC and 5G private network described in this invention includes a hardware layer, a resource sharing layer, a network function convergence layer, and an intelligent application layer, and is deployed on the user side. The hardware layer integrates a fixed network access module, a 5G wireless access module, and an edge computing module, and provides a unified deployment and installation interface; The resource sharing layer integrates a computing resource pool, a storage resource pool, a database cluster, and a software operating environment; The network function fusion layer integrates the broadband core network user plane functions with the 5G private network user plane functions, and integrates data packet routing and forwarding, local offloading, QoS guarantee, access authentication adaptation and cross-network data collaboration functions. It realizes unified control and management of fixed and mobile network user plane through SDN and NFV technologies. The intelligent application layer provides an intelligent agent application distribution interface, supporting the download, deployment, and operation of independent intelligent agent applications for the Industrial Internet processed by knowledge distillation technology.

[0008] Furthermore, the edge computing module in the hardware layer supports embedded Linux or HarmonyOS edge version operating systems, is equipped with a real-time kernel and containerized deployment environment, and is compatible with lightweight orchestration tools.

[0009] Furthermore, the network function fusion layer also integrates ATSSS access traffic routing function, supporting seamless switching, traffic splitting and collaborative scheduling between fixed network and 5G private network.

[0010] Furthermore, the intelligent agent application in the intelligent application layer generates a lightweight edge model from a large cloud model through knowledge distillation technology, supporting the operation of TensorFlow Lite for Microcontrollers or ONNX RuntimeEdge frameworks.

[0011] Furthermore, the resource sharing layer is also configured with a unified security management module to realize identity authentication, access control, encrypted data transmission, and cross-network security auditing functions.

[0012] Furthermore, the local offloading, in collaboration with the edge computing module, enables business data to be processed and forwarded locally on the fixed-mobile computing-data-intelligence converged device without needing to be transmitted back to the operator's core network, with end-to-end transmission latency controlled within 10 milliseconds.

[0013] Furthermore, the network function fusion layer supports network slicing, enabling the dynamic allocation of dedicated network slice resources based on business needs, and the isolated operation and differentiated QoS guarantees for different types of services.

[0014] Furthermore, the hardware layer also integrates a status monitoring module, which collects equipment operating parameters, resource utilization, and network transmission status data in real time, and uploads them to the management platform through a remote operation and maintenance interface, supporting fault early warning and remote diagnosis functions.

[0015] The fixed-mobile computing-data-intelligence fusion method based on BNC and 5G private network described in this invention includes the following steps: S1, builds a cloud-based intelligent agent application store for intelligent agents to be listed and retrieved; S2 deploys a fixed-mobile network computing, data and intelligence convergence device on the user side, and connects to fixed network and mobile network terminal devices, including IoT devices; S3 enables data collaborative forwarding, local offloading, and security isolation between terminal devices through the network function fusion layer of the fixed-mobile-network-computing-data-intelligence converged device. S4, terminal devices deploy and launch intelligent agent applications through the intelligent application layer of the fixed-mobile-network computing-data-intelligence fusion device; S5 enables terminal devices to schedule computing power, storage, databases, and software resources through the resource sharing layer of the fixed-mobile-network-computer-data-intelligence converged equipment, and dynamically allocate shared resources.

[0016] Furthermore, when the intelligent agent application is iterated and updated, the fixed-mobile-network-computer-data-intelligence fusion device completes the incremental upgrade of the intelligent agent through differential update technology based on the update notification pushed by the cloud-based intelligent agent application store.

[0017] The advantages of this invention lie in its unified design of the broadband core network user plane and the 5G private network user plane, sharing computing power, storage, database, and software resources. Through a layered architecture comprising a hardware layer, a resource sharing layer, a network function integration layer, and an intelligent application layer, it achieves collaborative forwarding, local offloading, and secure isolation of data between fixed and mobile networks. Simultaneously, based on knowledge distillation technology, industrial internet intelligent agents are transformed into lightweight edge applications, which users can download and deploy on demand through a portal, achieving instant access to intelligent applications. By integrating hardware and software resources and aggregating the ecosystem, it improves network deployment convenience and resource utilization, reduces operation and maintenance costs, and meets the low-latency, high-reliability business requirements of industrial scenarios. Attached Figure Description

[0018] Figure 1 This is a diagram of the architecture of the fixed-mobile computing-data-intelligence converged device based on BNC and 5G private network of the present invention.

[0019] Figure 2 This is a schematic diagram illustrating the deployment of the fixed-mobile computing-data-intelligence converged device based on BNC and 5G private network according to the present invention.

[0020] Figure 3 This is a flowchart illustrating the implementation of the fixed-mobile computing-data-intelligence fusion method based on BNC and 5G private network of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, the detailed architecture of the fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network described in this invention is as follows: The fixed-mobile-network-computer-data-intelligence integrated device is an integrated edge device that adopts a design concept of hardware integration, resource sharing, functional integration, and lightweight application. It is deployed at user-side edge locations such as parks and factories (e.g., edge data centers next to workshops). Figure 2 As shown, the core solution addresses pain points such as the separation of fixed and mobile networks, the fragmentation of computing power and network, and the cumbersome deployment of intelligent applications. The overall architecture is divided into four layers: hardware layer, resource sharing layer, network function integration layer, and intelligent application layer. Each layer works together to achieve the core goals of fixed-mobile convergence, computing-data synergy, and intelligent deployment.

[0023] The hardware layer serves as the carrier for multi-mode access and integrated deployment, forming the physical foundation for converged devices. Adopting an integrated design, its size is adapted to edge data center installation spaces, providing standardized deployment interfaces (such as rack mounts, power interfaces, and network interfaces). This eliminates the need for dispersed deployment of fixed-line, mobile, and computing power equipment, significantly reducing on-site deployment complexity. Core components of the hardware layer include: 1. Fixed network access module: integrates gigabit / 10-gigabit Ethernet interface and optical module, supports wired access methods such as all-optical network and leased line, and is compatible with fixed terminal access such as factory camera group, PLC, industrial control computer, etc., to ensure high bandwidth and stability of wired data transmission.

[0024] 2. 5G Wireless Access Module: Integrates 5G base station chip and antenna unit, supports SA / NSA dual-mode networking, adapts to wireless terminal access such as AGV and mobile quality inspection terminal, and reserves 4G / Wi-Fi compatibility capability to meet the heterogeneous terminal access needs of industrial scenarios.

[0025] 3. Edge computing hardware module: Equipped with a multi-core industrial-grade CPU, GPU accelerator card and embedded storage unit, it supports embedded Linux / HarmonyOS edge version operating system, and is equipped with a real-time kernel and K3s lightweight container orchestration tool to meet the real-time operation and reliability requirements of AI intelligent agents in industrial scenarios. At the same time, it integrates a status monitoring module to collect equipment operating parameters in real time (such as CPU utilization, network latency and storage capacity) to provide data support for remote operation and maintenance and fault early warning.

[0026] 5. Security Hardware Module: Integrates an encryption chip that supports national cryptographic algorithms (such as SM4) to provide hardware-level encryption protection for data transmission and storage, adapting to the high security requirements of industrial scenarios.

[0027] The resource sharing layer serves as a unified scheduling center for computing power, storage, and software. It is the core support layer for integrated fixed-mobile network computing, data, and intelligence equipment. It breaks down the traditional barriers between fixed and mobile network resources, constructing a unified resource pool to achieve the sharing and dynamic scheduling of computing power, storage, databases, and software resources, thereby improving resource utilization. Its specific structure and functions are as follows: 1. Unified Resource Pool: Includes a computing power resource pool (integrating CPU and GPU computing power, supporting elastic allocation according to task requirements), a storage resource pool (storing business data, intelligent agent models, and configuration files in blocks, supporting local data persistence), and a database cluster (using a lightweight edge database to store device operation logs, intelligent agent recognition results, and terminal access information). The resource pool supports dynamic scaling and can automatically adjust the resource allocation ratio according to network load and intelligent application needs.

[0028] 2. Resource Scheduling Engine: Based on SDN (Software Defined Networking) and NFV (Network Functions Virtualization) technologies, an intelligent scheduling algorithm is built to monitor network traffic, computing power utilization, and service priorities in real time, enabling dynamic allocation of resources among fixed network services, mobile network services, and intelligent applications, and ensuring resource supply for high-priority services (such as industrial quality inspection and production line control).

[0029] 3. Unified Security Management Module: Integrates identity authentication, access control, data encryption, and security auditing functions to achieve unified authentication for fixed and mobile terminal access, encrypted cross-network data transmission, and full-process log auditing. This solves the problem of fragmented security policies between traditional fixed and mobile networks and meets the requirements for localization and compliance of industrial data.

[0030] 4. Software runtime environment: Provides a standardized container runtime environment and model inference framework (such as TensorFlow Lite for Microcontrollers, ONNX Runtime Edge), adapts to lightweight intelligent agent applications after knowledge distillation, and supports parallel operation of multiple intelligent agents and resource conflict coordination.

[0031] The network function convergence layer is the network core of the fixed-mobile network computing, data, and intelligence convergence integrated equipment. It is the core carrier for the integration of fixed and mobile user plane functions. Its core innovation lies in the unified design of the user plane (UP) functions of the broadband core network (BNC) and the user plane (UPF) functions of the 5G private network. Through a unified user plane processing flow, it abandons the traditional independent user plane architecture of fixed and mobile networks, achieving collaborative forwarding, local offloading, and differentiated protection of data in fixed and mobile networks. Specifically, it includes: 1. UP / UPF integration function: integrates packet routing and forwarding, local traffic splitting, QoS (Quality of Service) guarantee, and access authentication adaptation functions. It is also compatible with fixed network IP protocol and 5G GTP-U protocol, realizing unified processing of fixed and mobile data. It eliminates the need for cross-user plane forwarding, shortens the transmission path, and controls the end-to-end latency to within 10 milliseconds, making it suitable for low-latency scenarios such as industrial quality inspection.

[0032] 2. ATSSS access traffic routing function: Supports seamless switching, traffic splitting and collaborative scheduling between fixed network and 5G private network. When mobile terminals (such as AGVs) move within the factory, the access link is automatically switched to ensure business continuity and avoid disconnection and latency jitter.

[0033] 3. Network slicing function: Supports dynamic allocation of dedicated network slices, which can allocate independent network resources according to business type (such as production control, office communication, intelligent quality inspection) to achieve isolated operation and differentiated QoS guarantee for different businesses. For example, low-latency slices can be allocated to quality inspection business, and high-bandwidth slices can be allocated to office business.

[0034] 4. Local offloading function: In collaboration with the edge computing module, industrial business data (such as quality inspection video data) can be forwarded and processed locally on the fixed-mobile-network-computer-data-intelligence integrated device without being transmitted back to the operator's core network, which reduces transmission bandwidth costs and ensures data security and privacy.

[0035] The intelligent application layer serves as the interface between fixed-mobile-grid integrated computing, data, and intelligence devices and the industrial internet ecosystem. It provides a lightweight entry point for deploying and operating intelligent agents, including downloading, deploying, running, and managing them, supporting a seamless application deployment experience. Specific functions include: 1. Intelligent Agent Application Distribution Interface: Provides a standardized API interface to connect with cloud-based industrial internet portal systems, supports rapid downloading, verification, and decryption of intelligent agent applications, and adapts to lightweight intelligent agent packaging formats from different service providers.

[0036] 2. Automated Deployment Engine: After receiving the intelligent agent application from the cloud, it automatically calls upon the computing power and storage resources of the resource sharing layer to complete the containerized deployment, environment configuration and startup of the intelligent agent without manual intervention, achieving deployment upon download.

[0037] 3. Agent lifecycle management: Supports the start, stop, uninstallation, and differential update functions of agents. Automatically triggers the uninstallation of idle agents based on device resource utilization. Reduces the amount of data transmitted during model upgrades and lowers bandwidth consumption through differential update technology.

[0038] 5. Parameter configuration and interaction interface: Provides local operation interface (such as touch screen, web management terminal) and remote interaction interface, supports users to configure intelligent application parameters (such as defect recognition accuracy, annotation form, alarm threshold), and synchronizes application running results (such as defect recognition annotation, equipment status) to user terminal.

[0039] Based on the aforementioned integrated fixed-mobile computing, data, and intelligence (BNC) and 5G private network-based fixed-mobile network computing, data, and intelligence fusion device, a complete solution for the deployment of intelligent industrial internet applications is provided. Taking the industrial component defect visual recognition scenario provided by service provider A and the deployment of lightweight intelligent agents as an example, the method of fixed-mobile network computing, data, and intelligence fusion is explained in detail, covering the entire process from intelligent agent packaging and racking, fusion device deployment, application acquisition and deployment to actual operation, encompassing the three-level collaborative logic of cloud-edge-field, such as... Figure 3 As shown, the specific steps are as follows: Step 1: Build a cloud-based intelligent agent application store, where service provider A's intelligent agents are packaged and uploaded to the cloud.

[0040] Service Provider A developed and streamlined AI-powered intelligent applications to address the defect detection needs of industrial components, such as... 1. Model Development: Based on industrial component defect datasets, a large cloud-based model is trained to achieve high-precision recognition of various defects such as scratches, dents, and cracks, with a core recognition accuracy of over 95%.

[0041] 2. Knowledge Distillation and Encapsulation: Using knowledge distillation technology, the large cloud model (teacher model) is compressed into a lightweight edge model (student model). While retaining more than 90% of the core recognition accuracy, the model size is compressed by more than 70%, adapting to the edge computing power and storage resources of fusion devices. At the same time, the lightweight model is containerized and encapsulated, and the model inference dependency library is integrated to generate a standardized intelligent agent application package.

[0042] 3. Cloud-based Upload: Build a cloud-based intelligent agent application store, upload the packaged intelligent agent application to the cloud-based intelligent agent application store, complete the core information entry, including functional summary (visual recognition of component defects, automatic annotation, real-time alarm), applicable scenarios (mechanical manufacturing, production line quality inspection), resource requirements (computing power ≥ 2 TOPS, storage ≥ 10GB), applicable industrial categories (metal components, plastic components), and add search keywords such as "component defects, visual recognition, surface inspection, industrial quality inspection, automated inspection", complete the cloud-based uploading and search configuration of the intelligent agent, and make it available for downstream factory users to choose from.

[0043] Step 2: On the factory side, for machinery manufacturing plants with industrial component defect detection needs, deploy and initialize the integrated fixed-mobile-grid-computer interface (PCI / AI) device. The specific steps are as follows: 1. Equipment Installation: Deploy the integrated fixed-mobile-network computing-intelligence device in the edge room next to the production workshop. Fix it through a standardized rack interface and connect it to an industrial power supply to complete the physical installation of the equipment. At the same time, build an all-optical network link to establish a wired connection between the integrated equipment and the camera group (multi-view deployment, covering the entire surface of the components) next to the production line conveyor belt. Reserve a 5G wireless access channel to adapt to the mobile quality inspection terminal in the workshop.

[0044] 2. System Initialization: Start the integrated fixed-mobile-network computing-data-intelligence device, complete self-tests of each hardware module (fixed network access, 5G access, computing power module, security module) to ensure normal hardware operation; initialize the resource sharing layer, start the computing power, storage, and database resource pools, and configure resource scheduling strategies and security control rules; activate the UP / UPF integration function of the network function fusion layer, configure QoS parameters and network slicing strategies to ensure low-latency transmission of quality inspection services; initialize the intelligent application layer, connect to the cloud-based industrial internet portal system, and complete device identity authentication and communication link establishment.

[0045] 3. Terminal Adaptation: Connect the camera array to the fixed network interface of the fusion device, complete device access authentication and parameter configuration, and set the camera acquisition frequency to 0.1 seconds / frame to ensure real-time acquisition of multi-view video data of components; simultaneously, establish communication between the factory management terminal, quality inspection terminal and the fusion device to achieve parameter configuration and result reception. Terminal devices also include user-side office computers, mobile phones and other IoT devices.

[0046] Step 3: On the factory side, the integrated fixed-mobile-network-computer-data-intelligence device retrieves and downloads the intelligent agent for detecting defects in industrial components. The specific operation is as follows: 1. Demand Retrieval: Access the cloud-based industrial internet portal system through the factory management terminal, and enter keywords such as "component defect visual recognition" and "industrial quality inspection" in the search entry. Based on the keyword matching algorithm and combined with the cloud-based annotation information of the intelligent agent, the cloud portal system quickly locates the defect visual recognition intelligent agent uploaded by service provider A, and displays the functional details, resource requirements, and suitable scenarios of the intelligent agent for user confirmation.

[0047] 2. Application Selection and Download: After quality inspectors confirm the target intelligent agent to be selected, they submit the selection instruction on the cloud portal. The cloud portal system generates an encrypted download link and pushes it to the local fixed-mobile-network-computer-data-intelligence integrated device in the factory. The fixed-mobile-network-computer-data-intelligence integrated device downloads the intelligent agent application package from the cloud through a dedicated interface of the intelligent application layer based on the encrypted link. At the same time, the application package is verified and decrypted to ensure the integrity and security of the application and prevent malicious tampering.

[0048] Step 4: The integrated fixed-mobile network computing, data, and intelligence device completes the automated deployment and personalized parameter configuration of the intelligent agent, achieving immediate activation. The specific operation is as follows: 1. Automated Deployment: The intelligent application layer of the fixed-mobile-network-computer-data-intelligence integrated device starts the automated deployment engine, reads the resource requirement information of the intelligent agent application package, and sends a resource allocation request to the resource sharing layer. The scheduling engine of the resource sharing layer allocates dedicated computing power (2TOPS), storage (10GB), and database space to the intelligent agent according to the current resource occupancy, and automatically completes container startup, model loading, and runtime environment configuration. The entire deployment process does not require intervention from professional technicians and takes less than 5 minutes.

[0049] 2. Personalized parameter configuration: Quality inspectors can access the intelligent agent parameter configuration page through the local operation interface of the fusion device or the factory management terminal to complete personalized settings, including defect recognition accuracy (such as high-precision mode / high-efficiency mode), defect labeling format (location coordinates + type + level), defect alarm threshold (such as triggering an alarm for level 3 defects and above), and result synchronization frequency (real-time synchronization / summarized synchronization every 10 seconds). After configuration, the intelligent agent will be started.

[0050] Step 5: The intelligent agent runs stably on the integrated fixed-mobile-grid computing-data-intelligence device, completing the real-time identification and labeling of component defects. The specific process is as follows: 1. Data Acquisition and Transmission: When the production line is running normally, the conveyor belt carries the components past the camera group at 0.1-second intervals. The camera group simultaneously collects multi-view video image data of the components and transmits the data to the hardware layer of the converged equipment with low latency through an all-optical network or 5G wireless link. The hardware layer forwards the data to the network function convergence layer, where the UP / UPF integrated module completes data parsing and priority marking, and prioritizes scheduling to the quality inspection business slice.

[0051] 2. AI Real-time Recognition and Labeling: The network function fusion layer forwards data to the intelligent application layer. The running defect visual recognition intelligent agent calls on the dedicated computing power allocated by the resource sharing layer to perform real-time AI inference and analysis on video image data, accurately detect defects such as scratches, dents, and cracks on the surface of components, and automatically labels the location coordinates, defect type (such as scratches, dents), and defect level (level 1 / 2 / 3) of each defect, generating detection result data.

[0052] 3. Result Output and Linkage: The test results are synchronized to the factory quality inspection terminal in real time for quality inspectors to view and verify; at the same time, when a level 3 or above serious defect is detected, the intelligent agent triggers a local alarm (audio and visual alarm) and synchronizes it to the production line control system. The conveyor belt can be suspended according to the preset strategy to prevent unqualified components from flowing into the next process; the test log and result data are stored in the database of the resource sharing layer for subsequent traceability and statistical analysis.

[0053] The entire data flow is as follows: Video data collected by the camera group is received by the hardware layer of the fixed-mobile network computing-data-intelligence integrated device through the all-optical network (fixed network) / wireless network (5G mobile network). The data is routed and forwarded by the network function fusion layer UP / UPF integration function, which realizes local traffic splitting, QoS guarantee, access authentication adaptation and cross-network data collaboration, etc. The dedicated computing power is allocated through the resource sharing layer. The defect identification intelligent agent in the intelligent application layer synchronizes the identification and labeling results from the fixed-mobile network computing-data-intelligence integrated device to the factory quality inspection terminal. The cloud portal and the fixed-mobile network computing-data-intelligence integrated device are connected for the download and distribution of intelligent agent applications. The factory management terminal and the cloud portal are connected for the interactive data of demand retrieval and intelligent agent selection.

[0054] Intelligent agent operation and maintenance and iteration The status monitoring module of the fixed-mobile network computing, data and intelligence integrated device collects the intelligent agent's operating status (such as recognition accuracy and computing power utilization) and device operating parameters in real time, and uploads them to the cloud management platform, supporting fault early warning and remote diagnosis. When service provider A releases an iterative version of the intelligent agent, the cloud portal system pushes an update notification. The fixed-mobile network computing, data and intelligence integrated device completes the incremental upgrade of the intelligent agent through differential update technology, without having to download the complete application package again, ensuring continuous business operation and realizing full lifecycle operation and maintenance and iterative optimization of intelligent application operation.

[0055] In addition, the fixed-mobile network computing, data and intelligence convergence method also provides data collaborative forwarding, local offloading and security isolation between terminal devices through the network function convergence layer of the fixed-mobile network computing, data and intelligence convergence integrated device, including data distribution between office computers, mobile phones and other IoT devices; and realizes the scheduling of computing power, storage, database and software resources and the dynamic allocation of shared resources through the resource sharing layer of the fixed-mobile network computing, data and intelligence convergence integrated device.

[0056] This invention achieves three core advantages through the collaborative four-layer architecture of the fixed-mobile network-computing-data-intelligence integrated device and the three-level linkage of cloud-edge-field: First, the unified UP / UPF and resource-sharing architecture solves the problems of high latency, complex operation and maintenance, and resource waste caused by the traditional separate fixed and mobile networks, controlling end-to-end latency to within 10 milliseconds and improving resource utilization by more than 30%; Second, lightweight intelligent agents and automated deployment significantly shorten the industrial intelligent application implementation cycle, requiring only less than 10 minutes from selection to operation, reducing the technical threshold for factory users; Third, the integration of fixed and mobile networks and the collaboration of computing, data, and intelligence adapt to the heterogeneous terminal access and high real-time business requirements of industrial scenarios, providing an integrated solution for industrial Internet scenarios.

Claims

1. A fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network, characterized in that: It includes a hardware layer, a resource sharing layer, a network function integration layer, and an intelligent application layer, and is deployed on the user side; The hardware layer integrates a fixed network access module, a 5G wireless access module, and an edge computing module, and provides a unified deployment and installation interface; The resource sharing layer integrates a computing resource pool, a storage resource pool, a database cluster, and a software operating environment; The network function fusion layer integrates the broadband core network user plane functions with the 5G private network user plane functions, and integrates data packet routing and forwarding, local offloading, QoS guarantee, access authentication adaptation and cross-network data collaboration functions. It realizes unified control and management of fixed and mobile network user plane through SDN and NFV technologies. The intelligent application layer provides an intelligent agent application distribution interface, supporting the download, deployment, and operation of independent intelligent agent applications for the Industrial Internet processed by knowledge distillation technology.

2. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The edge computing module in the hardware layer supports embedded Linux or HarmonyOS edge version operating systems, is equipped with a real-time kernel and containerized deployment environment, and is compatible with lightweight orchestration tools.

3. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The network function fusion layer also integrates ATSSS access traffic routing function, supporting seamless switching, traffic splitting and collaborative scheduling between fixed network and 5G private network.

4. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The intelligent application layer uses knowledge distillation technology to train a lightweight edge model from a large cloud model, supporting TensorFlow Lite for Microcontrollers or ONNX Runtime Edge framework.

5. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The resource sharing layer is also configured with a unified security management module to realize identity authentication, access control, encrypted data transmission and cross-network security auditing functions.

6. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The local offloading, in collaboration with the edge computing module, enables business data to be processed and forwarded locally on the fixed-mobile-computer-data-intelligence converged device without having to be transmitted back to the operator's core network, with end-to-end transmission latency controlled within 10 milliseconds.

7. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The network function fusion layer supports network slicing, enabling the dynamic allocation of dedicated network slice resources based on business needs, and providing isolated operation and differentiated QoS guarantees for different types of services.

8. The fixed-mobile computing-data-intelligence fusion device based on BNC and 5G private network according to claim 1, characterized in that: The hardware layer also integrates a status monitoring module, which collects equipment operating parameters, resource utilization, and network transmission status data in real time, and uploads them to the management platform through a remote operation and maintenance interface, supporting fault early warning and remote diagnosis functions.

9. A fixed-mobile computing-data-intelligence fusion method based on BNC and 5G private network using the device described in any one of claims 1-8, characterized in that, Includes the following steps: S1, builds a cloud-based intelligent agent application store for intelligent agents to be listed and retrieved; S2 deploys a fixed-mobile network computing, data and intelligence convergence device on the user side, and connects to fixed network and mobile network terminal devices, including IoT devices; S3 enables data collaborative forwarding, local offloading, and security isolation between terminal devices through the network function fusion layer of the fixed-mobile-network-computing-data-intelligence converged device. S4, terminal devices achieve intelligent agent application deployment and startup through the intelligent application layer of the integrated fixed-mobile-network computing-data-intelligence device; S5 enables terminal devices to schedule computing power, storage, databases, and software resources through the resource sharing layer of the fixed-mobile-network-computer-data-intelligence converged equipment, and dynamically allocate shared resources.

10. The fixed-mobile computing-data-intelligence fusion method based on BNC and 5G private network according to claim 9, characterized in that: When the intelligent agent application is iterated and updated, the fixed-mobile-network-computer-data-intelligence fusion device completes the incremental upgrade of the intelligent agent through differential update technology based on the update notification pushed by the cloud-based intelligent agent application store.

Images