5G edge gateway server, 5G intelligent edge gateway and system

Abstract

The invention discloses a 5G edge gateway server, a 5G intelligent edge gateway and a system, and the 5G edge gateway server carries out container deployment, operates an edge computing platform in the container deployment, and comprises a network communication module which carries out the communication with a user plane function through an N6 interface, and analyzes a network request when receiving the network request; the network request comprises a model data issuing request; the task receiving module obtains a task issued by the cloud, and the task is generated by the cloud according to the network request and determines an issuing mode according to a task issuing mechanism; the business processing module is used for locally deploying the model or executing the task according to the task and the resource issued by the cloud; the model is a model which is intensively trained by the cloud. According to the method, cloud edge collaboration is realized, and multi-scene edge computing task scheduling is supported.

Description

Technical Field

[0001] The present invention relates to the field of edge computing technology, and in particular to a 5G edge gateway server, a 5G intelligent edge gateway and a system. Background Art

[0002] With the rapid development of 5G technology, network transmission speed, bandwidth and latency performance have been significantly improved, promoting the rapid expansion of multiple fields such as the Internet of Things, smart cities, and smart manufacturing. However, the traditional cloud computing model can no longer meet the needs of low latency and high bandwidth. Edge computing has gradually become an important solution, which reduces data processing latency and improves response speed and service quality by sinking computing resources to the edge of the network.

[0003] Currently, most of the 5G edge gateway devices on the market are embedded devices for the Internet of Things. They have relatively simple functions, are difficult to achieve cloud-edge collaboration, and lack flexible local service support. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a 5G edge gateway server, a 5G intelligent edge gateway and a system to achieve cloud-edge collaboration and support edge computing task scheduling in multiple scenarios.

[0005] In order to solve the above technical problems, the present invention discloses a 5G edge gateway server, including: the 5G edge gateway server is deployed in a container, and an edge computing platform is run in the container deployment, including:

[0006] A network communication module communicates with the user plane function through the N6 interface and parses the network request upon receiving the network request; the network request includes a model data sending request;

[0007] A task receiving module, which obtains the task issued by the cloud, wherein the task is generated by the cloud according to the network request and the issuing method is determined according to the task issuing mechanism;

[0008] The business processing module locally deploys the model or executes the task according to the task and resources sent by the cloud; the model is a model that has been centrally trained on the cloud.

[0009] In some implementations, it also includes: a data reporting module, which reports designated fields to the cloud through the Device model, and the designated fields are used to report status information, device status, and execution status of the model.

[0010] In some implementations, the designated fields include model version, model md5 and / or classification information.

[0011] In some implementations, the data reporting module sends a text message via the MQTT protocol or uploads image information or stream information after being encrypted in a preset manner.

[0012] In some implementations, the sending method includes sending in full and sending in pieces, and sending in full means sending all the model data when the model data is not greater than a preset threshold;

[0013] The fragmented sending is that when the model data is larger than a preset threshold, the model data is fragmented into pieces according to a preset size and sent piece by piece.

[0014] In some embodiments, the 5G edge gateway server communicates with the cloud via an event bus, and the cloud communicates with the 5G edge gateway server via a service bus.

[0015] In some embodiments, it also includes: an auxiliary module, which is used to record operation logs in operation and read and parse configuration files; the operation logs include task distribution logs, model deployment logs and inference execution logs.

[0016] In some embodiments, an AI / ML machine learning module is also included, and the AI / ML machine learning module intelligently caches frequently accessed resources based on artificial intelligence / machine learning algorithms.

[0017] According to a second aspect of the present invention, a 5G intelligent edge gateway is disclosed, comprising at least one 5G edge gateway server and UPF user plane function as described above;

[0018] According to a third aspect of the present invention, a 5G cloud-edge collaborative intelligent edge network system is disclosed, comprising:

[0019] A 5G intelligent edge gateway as described above;

[0020] The 5G intelligent edge gateway is connected to the cloud;

[0021] The UPF user plane function is connected to the 5G wireless access network through the N3 interface.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] The present invention provides a 5G edge gateway server, a 5G smart edge gateway, and a system. The 5G edge gateway server provides high-performance data processing and transmission capabilities. Time-sensitive computing tasks can be placed on the edge server for computing, and non-real-time processing tasks are still kept on the cloud server for computing, effectively solving the problems of high latency and low reliability in the traditional cloud computing model, optimizing resource utilization, improving the overall performance of the system, solving the needs of 5G private networks for edge computing and cloud-edge collaborative computing, and providing efficient and low-latency edge services. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the structure of a 5G edge gateway server provided by the present invention;

[0025] Figure 2 A module schematic diagram of a 5G edge gateway server provided by the present invention;

[0026] Figure 3 A schematic diagram of a 5G network used by a 5G edge gateway server provided by the present invention. DETAILED DESCRIPTION

[0027] For better understanding and implementation, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0028] The terms "including" and "having" and any variations thereof in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products or apparatus.

[0029] An embodiment of the present invention discloses a 5G edge gateway server, which realizes cloud-edge collaboration and elastic expansion, and supports edge computing task scheduling in multiple scenarios.

[0030] A 5G edge gateway server provided in the present application includes a network communication module, a task receiving module and a business processing module. The network communication module communicates with the user plane function through the N6 interface, and parses the network request when receiving the network request; the network request includes a model data sending request; the task receiving module is used to obtain the task sent by the cloud, and the task is generated by the cloud according to the network request and the sending method is determined according to the task sending mechanism; the business processing module is used to locally deploy the model or execute the task according to the task and resources sent by the cloud; the model is a model trained centrally on the cloud.

[0031] like Figure 1 As shown in the figure, this server is based on general x86 hardware facilities and has powerful processing capabilities and scalability. The software architecture of this server adopts advanced virtualization deployment solutions, gradually building a flexible and efficient computing environment from the bottom to the top. Specifically, the server runs ESXi or PVE virtualization system at the bottom to provide virtualization capabilities and resource management. The virtualization system provides a basis for the creation and management of multiple virtual machines (VMs), so that resources can be flexibly allocated and efficiently utilized. On top of the virtualization system, virtual machines running the Linux operating system support commonly used Ubuntu Server or CentOS, providing a stable and secure operating environment and ensuring the reliability and compatibility of upper-layer applications.

[0032] Deploy a containerized environment, including Docker and Kubernetes, on top of the Linux operating system virtual machine. This containerization technology enables rapid deployment and efficient management of applications. Through containerization, applications are packaged into standardized units, and each container runs in an independent namespace without interfering with each other, thereby achieving better isolation and portability. Kubernetes further provides container orchestration capabilities, including container start, stop, expansion, load balancing, and fault recovery. Dynamically increase or decrease container instances based on application load and resource usage, support service discovery and load balancing, and provide a stable access entry for applications by defining Services. Support automated deployment, expansion, and management to improve the elasticity and scalability of the system. On this server, the containerized environment allows for rapid deployment and adjustment of applications to meet the dynamic needs of edge nodes.

[0033] In the container environment, this server runs the edge computing platform, providing computing resource scheduling and management functions. The edge computing platform can support a variety of edge computing tasks, including data processing, analysis, and storage, and provide a RESTful API interface to interact with other systems. Through this platform, the edge gateway server can effectively allocate computing resources, optimize task scheduling, and improve computing efficiency.

[0034] In AI edge applications, the cloud-edge collaboration and task scheduling of AI applications, that is, the edge nodes perform reasoning according to the AI ​​model to achieve distributed intelligence; the cloud conducts centralized training of AI models and sends the models to the edge, that is, the server or edge device. In ordinary k8s containerized application deployment, the application upgrade generally pulls the complete image, and the application in the same computer room can be used normally. However, when the cloud and edge are deployed in different computer rooms in edge applications and the network distance between them is relatively far and the network latency is large, if the upgrade adopts the complete image pull method, it will cause a lot of waste of network bandwidth and the upgrade time will increase exponentially.

[0035] Therefore, in this application, if Figure 2 As shown, a task receiving module and a business processing module are set up to package or publish images directly on the edge by receiving tasks sent from the cloud, so as to reduce network transmission time and image deployment time.

[0036] When the network communication module receives a network request, it parses the network request. The network request can be issued by the cloud, other edge terminals or user plane functions. Among them, the network request issued by the cloud usually includes model data sending requests, task scheduling instructions or synchronization configuration files, etc., the network request of other edge terminals can be reporting information, etc., and the network request of the user plane function (UPF) is a data request of a terminal device, such as an IoT device or a user terminal, or as a relay of a request, etc., which is not limited in the application. After parsing these network requests, the network communication module will generate tasks based on the content and pass them to the corresponding modules for processing to ensure efficient operation and smooth cloud-edge collaboration.

[0037] When a network request contains an inference task requirement, the cloud creates a task that includes model delivery, inference execution, etc. When the cloud receives a request to deliver model data, it centrally trains the model, generates tasks, and delivers them to the edge. The cloud determines the tasks to be delivered and the delivery method based on the task delivery mechanism.

[0038] The task receiving module is used to obtain the task issued by the cloud, which is generated by the cloud according to the network request and the issuing method is determined according to the task issuing mechanism. The task issuing mechanism is to determine how to allocate the task to the edge end according to factors such as the priority of the task, the type of task, and the load of the edge node. The issuing method includes full issuance and fragmented issuance. The full issuance is when the model data is not greater than the preset threshold, all the model data is issued; the fragmented issuance is when the model data is greater than the preset threshold, the model data is fragmented according to the preset size and issued piece by piece. In this application, the cloud obtains the size of the model data after completing the centralized model training. When the model data is not greater than 12M, the model data is packaged and sent to the edge end in full. If the model data is greater than 12M, the full issuance will occupy a lot of bandwidth, increase the network transmission time, and may even cause transmission interruption, so fragmented issuance is adopted. Fragmented issuance effectively solves the problems of bandwidth waste, transmission interruption and slow page response speed in the transmission process of large models, improves transmission efficiency, and is suitable for large model deployment requirements in cloud-edge collaboration and edge computing scenarios.

[0039] The task receiving module is responsible for receiving the tasks assigned from the cloud and preparing to pass these tasks to subsequent modules for processing. When full delivery is adopted, after the task receiving module receives the task, the business processing module deploys the local model according to the task sent from the cloud, that is, retains the model data and replaces the model data.

[0040] When sharding is used, the cloud will cut the trained model data into multiple shards according to a certain size, such as 1M or 2M per shard. At the same time, each shard is accompanied by verification information (such as MD5 value) to ensure data integrity. Each shard is transmitted separately, and the edge receives and verifies it and stores it locally. After receiving all the shards, the edge reassembles them into a complete model and verifies the integrity based on the verification information. If a shard is missing or the verification fails, you can request the cloud to resend the shard.

[0041] The business processing module locally deploys the model or executes the task according to the tasks and resources sent by the cloud; the model is a model that has been centrally trained on the cloud. Before executing a task, the business processing module will evaluate whether the task can be executed immediately or whether the model needs to be deployed first based on the current resources, such as CPU, memory, storage space, etc. and network conditions. If a new AI model is required, the business processing module will prioritize the local deployment of the model. If the model has been deployed in place, the task execution process is started directly.

[0042] Dynamically allocate resources through the business processing module to give priority to high-priority tasks while minimizing the impact of task execution on other system functions. According to the tasks and resource status issued by the cloud, you can flexibly choose to deploy the model locally or execute the task directly. This design not only ensures the efficient operation of the system, but also improves the adaptability and response speed of tasks.

[0043] Furthermore, it also includes a data reporting module, which reports the specified field to the cloud through the Device model, and the specified field is used to report the state information, device state, and execution status of the model. The specified field includes the model version, model md5 and / or classification information. The data reporting module feeds back the running status and task execution status of the edge to the cloud. By using the Device model, the specified field information is reported to the cloud regularly or under specific trigger conditions. Exemplarily, the key parameters of the device are defined, including device status, running information, task execution progress, etc., so as to facilitate remote monitoring, resource scheduling and model version management in the cloud. The model version represents the version number of the currently running AI model, which helps the cloud to confirm whether the device uses the latest version of the model, automatically generates tasks when the model version is not the latest version, and sends model data. The model MD5 is used to verify the integrity and consistency of the model to ensure that the transmitted or deployed model is not damaged or tampered with. The classification information may include model category, task type or model purpose, etc., to help the cloud identify the application of different models in different task scenarios. Through these fields, the cloud can accurately determine whether it is necessary to update the model, adjust the task scheduling strategy or perform troubleshooting.

[0044] The data reporting module sends text messages through the MQTT protocol or uploads image information or stream information after encryption in a preset manner. The MQTT protocol is used to transmit data to the cloud. MQTT (Message Queuing Telemetry Transport) is a lightweight message transmission protocol that is widely used in Internet of Things (IoT) applications and is particularly suitable for low-bandwidth, high-latency or unstable network environments. The MQTT protocol is used to send text messages, such as basic text information such as device status, model version, execution results, etc., and upload them in a timely manner for monitoring and decision-making in the cloud.

[0045] For larger or sensitive data, such as image information or stream information, the data reporting module encrypts it in a preset way before uploading. These encryption methods ensure that the data is not tampered with during transmission, thus ensuring the security of the data. The encrypted image or stream information can be uploaded to the cloud via MQTT or other protocols (such as HTTPs, WebSocket, etc.). For example, the image information or stream information is base64 encrypted into visible characters, and then decrypted into image or stream information using base64 after receipt. It ensures the integrity and security of the information, and can also efficiently support the real-time transmission of large amounts of data, which is suitable for scenarios that require rapid response and data security.

[0046] Furthermore, it also includes an auxiliary module, which is used to record the operation log in operation, read and parse the configuration file; the operation log includes the task distribution log, model deployment log and reasoning execution log. The detailed information of the task sent from the cloud is recorded by the operation log, including the task type, task reception time, the process of task distribution to the edge node, etc., which is used to track the processing flow and status of the task and ensure the correctness and timeliness of task allocation. The model deployment log records the deployment process of the model on the edge, including the model version, deployment time, whether the deployment is successful or not, etc., to monitor whether the model is successfully deployed and any abnormalities in the deployment process. The reasoning execution log records the execution process of the model reasoning task, including reasoning requests, input data, execution results or execution status, etc., to detect the execution of the reasoning task and perform performance analysis. By parsing these files, it is ensured that the edge device runs according to the preset parameters to ensure the flexibility and adaptability of this server.

[0047] Furthermore, it also includes an AI / ML machine learning module, which intelligently caches resources frequently accessed by the terminal based on artificial intelligence / machine learning algorithms. The AI / ML module uses artificial intelligence and machine learning algorithms to analyze the access patterns of terminal devices and applications, and intelligently identifies which resources are frequently accessed. By caching these resources on the edge, access requests to the cloud are reduced, thereby reducing latency and bandwidth consumption. The AI / ML machine learning module continuously learns and analyzes the access patterns of the device and dynamically adjusts the cache strategy. For example, based on the usage history and current load of the terminal device, it automatically decides when to cache resources, when to clear the cache, and the size of the cache, gradually adapting to changes in different environments and needs, optimizing the effectiveness and overall performance of the cache, and helping to reduce bandwidth consumption and cloud burden, with significant advantages in edge computing and cloud-edge collaborative systems.

[0048] The present application also provides a 5G intelligent edge gateway, including:

[0049] At least one 5G edge gateway server and UPF user plane function as described above.

[0050] like Figure 3 As shown, the present application also provides a 5G cloud-edge collaborative intelligent edge network system, including:

[0051] A 5G intelligent edge gateway as described above;

[0052] The 5G intelligent edge gateway is connected to the cloud;

[0053] The UPF user plane function is connected to the 5G wireless access network through the N3 interface.

[0054] Among them, UE is user equipment, such as user terminals, IoT devices, etc., to interact directly with the 5G network. 5GRAN is the core component of the 5G network, responsible for providing radio connections and data transmission, including base stations and related network nodes. User equipment is connected to the base station through the radio access network (RAN), and the base station communicates with the UE through radio channels. The radio access network provides high-speed data transmission and low-latency communication. The 5GC AMF (Access and Mobility Management Function) control plane is responsible for access management, mobility management and user authentication. The 5GC SMF (Session Management Function) control plane is responsible for session management, managing user data sessions and data paths. The UPF (User Plane Function) user plane function is responsible for data plane processing and transmitting user data. The core network manages and forwards data through different functional nodes, such as AMF, SMF, and UPF. The 5G intelligent edge gateway provides low-latency and high-bandwidth processing capabilities to further improve the performance and efficiency of the 5G network.

[0055] The edge gateway server provides high-performance data processing and transmission capabilities, and can flexibly respond to various business needs. Time-sensitive computing tasks can be placed on edge servers, and non-real-time processing tasks are still kept on cloud servers. It effectively solves the problems of high latency and low reliability in traditional cloud computing models, optimizes resource utilization, improves the overall performance of the system, solves the needs of 5G private networks for edge computing and cloud-edge collaborative computing, and provides efficient and low-latency edge services.

[0056] The processing method of the system can refer to the description of the above method and will not be repeated here.

[0057] The present invention also provides a device, which may include: a memory storing executable program code;

[0058] a processor coupled to the memory;

[0059] A transceiver used to communicate with other devices or communication networks and to receive or send network messages;

[0060] A bus used to connect memory, processors, and transceivers for internal communication.

[0061] The transceiver receives messages transmitted over the network and passes them to the processor via the bus. The processor calls the executable program code stored in the memory through the bus for processing, and passes the processing results to the transceiver via the bus for sending, thereby realizing the method provided in the embodiment of the present application.

[0062] An embodiment of the present application also provides a non-transitory machine-readable storage medium, on which an executable program is stored. When the executable program is executed by a processor, the processor executes the processing method provided in the above embodiment.

[0063] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program for electronic data exchange, wherein the computer program enables a computer to execute the described method.

[0064] An embodiment of the present invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the described method.

[0065] The embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without paying creative labor.

[0066] Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0067] Finally, it should be noted that the embodiments disclosed in the present invention are only preferred embodiments of the present invention, which are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A 5G edge gateway server, characterized in that: The 5G edge gateway server is deployed in a container, and an edge computing platform is run in the container deployment, including: A network communication module communicates with the user plane function through the N6 interface and parses the network request upon receiving the network request; the network request includes a model data sending request; A task receiving module, which obtains the task issued by the cloud, wherein the task is generated by the cloud according to the network request and the issuing method is determined according to the task issuing mechanism; The business processing module locally deploys the model or executes the task according to the task and resources sent by the cloud; the model is a model that has been centrally trained on the cloud.

2. A 5G edge gateway server according to claim 1, characterized in that: Also includes: The data reporting module reports the specified fields to the cloud through the Device model. The specified fields are used to report the status information, device status, and execution status of the model.

3. A 5G edge gateway server according to claim 2, characterized in that: The designated fields include model version, model md5 and / or classification information.

4. A 5G edge gateway server according to claim 2, characterized in that: The data reporting module sends text messages through the MQTT protocol or uploads image information or stream information after being encrypted in a preset manner.

5. A 5G edge gateway server according to any one of claims 1-4, characterized in that: The sending mode includes sending in full and sending in pieces. The sending in full means sending all the model data when the model data is not greater than a preset threshold value. The fragmented sending is that when the model data is larger than a preset threshold, the model data is fragmented into pieces according to a preset size and sent piece by piece.

6. A 5G edge gateway server according to claim 5, characterized in that: The 5G edge gateway server communicates with the cloud via an event bus, and the cloud communicates with the 5G edge gateway server via a service bus.

7. A 5G edge gateway server according to claim 6, characterized in that: Also includes: Auxiliary module, the auxiliary module is used to record the operation log in operation and read and parse the configuration file; the operation log includes task distribution log, model deployment log and reasoning execution log.

8. A 5G edge gateway server according to claim 1, characterized in that: It also includes an AI / ML machine learning module, which intelligently caches frequently accessed resources based on artificial intelligence / machine learning algorithms.

9. A 5G intelligent edge gateway, characterized in that: include: At least one 5G edge gateway server and UPF user plane function as described in any one of claims 1-8.

10. A 5G cloud-edge collaborative intelligent edge network system, characterized in that: include: The 5G intelligent edge gateway as claimed in claim 9; The 5G intelligent edge gateway is connected to the cloud; The UPF user plane function is connected to the 5G wireless access network through the N3 interface.

Images