Object storage-based cloud edge collaboration method and device and storage medium

Abstract

The present disclosure relates to a cloud-edge collaboration method and device based on object storage and a storage medium. A cloud-edge collaboration method comprises: storing, by a first device, data at an object storage server; receiving, by the first device, a shared link corresponding to the data from the object storage server; and sending, by the first device, the shared link to a second device, so that the second device can request the data from the object storage server based on the shared link, wherein the first device is one of a cloud device and an edge device, and the second device is the other of the cloud device and the edge device.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of edge computing, and more particularly, to an object storage-based cloud-edge collaboration method, device and storage medium. BACKGROUND

[0002] Edge computing is an open platform that integrates network, computing, storage and application core capabilities to provide computing services on the spot, which is a supplement and extension of cloud computing. The application program of edge computing is initiated on the edge side to produce faster network service response, and meets the basic needs of industry in real-time business, application intelligence, security and privacy protection, etc. Edge computing is between physical entities and industrial connections, or at the top of physical entities. While the cloud computing can still access the historical data of edge computing.

[0003] As an important technology of edge computing, cloud-edge collaboration realizes the complementary of cloud computing and edge computing, and collaborates with the cloud and edge to work together. However, the existing cloud-edge collaboration architecture has problems such as difficult expansion in space, poor metadata search and classification ability, and low efficiency of massive data access, which needs further improvement. SUMMARY

[0004] To solve the above problems, the present disclosure proposes an object storage-based cloud-edge collaboration technology to alleviate or solve the problems existing in the existing cloud-edge collaboration architecture.

[0005] One aspect of the present disclosure provides a cloud-edge collaboration method, comprising:

[0006] storing, by a first device, data at an object storage server;

[0007] receiving, by the first device, a shared link corresponding to the data from the object storage server; and

[0008] sending, by the first device, the shared link to a second device, so that the second device can request the data from the object storage server based on the shared link,

[0009] wherein the first device is one of a cloud device and an edge device, and the second device is the other of the cloud device and the edge device.

[0010] Another aspect of the present disclosure provides a cloud-edge collaboration method, comprising:

[0011] receiving, by a second device, a shared link from a first device, the shared link corresponding to data stored at an object storage server by the first device; and

[0012] requesting, by the second device, the data from the object storage server based on the shared link,

[0013] wherein the first device is one of a cloud device and an edge device, and the second device is the other of the cloud device and the edge device.

[0014] Another aspect of the present disclosure provides a cloud-edge collaboration apparatus, comprising:

[0015] a processor; and

[0016] a memory storing executable instructions that, when executed, cause the processor to perform the above cloud-edge collaboration method.

[0017] Another aspect of the present disclosure provides a computer-readable storage medium containing executable instructions that, when executed, cause performance of the above cloud-edge collaboration method. BRIEF DESCRIPTION OF DRAWINGS

[0018] The present disclosure can be better understood with reference to the following detailed description when considered in connection with the following drawings, in which like elements are numbered alike, and in which: FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0019] wherein:

[0020] Figure 1 FIG. 1 shows an existing cloud-edge collaboration architecture;

[0021] Figure 2 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0022] Figure 3 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0023] Figure 4 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0024] Figure 5 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0025] Figure 6 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0026] Figure 7 FIG. 1 shows a cloud-edge collaboration architecture according to an embodiment of the present disclosure;

[0027] Other features and advantages of the present disclosure will become apparent from the following description of the embodiments, taken in conjunction with the accompanying drawings. DETAILED DESCRIPTION

[0028] Various exemplary embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. For clarity and brevity, not all features of the embodiments are described in this specification. However, it should be noted that many implementation-specific settings can be made in implementing embodiments of this disclosure to achieve the developer's specific goals, such as complying with constraints related to system devices and services, and these constraints may vary depending on the implementation. Furthermore, it should be understood that while development work may be complex and time-consuming, such development work is merely a routine task for those skilled in the art who benefit from this disclosure.

[0029] Furthermore, it should be noted that, in order to avoid obscuring this disclosure with unnecessary details, only processing steps and / or device configurations closely related to at least the technical solutions according to this disclosure are shown in the accompanying drawings, while other details not closely related to this disclosure are omitted. The following description of exemplary embodiments is merely illustrative and is not intended to be a limitation of this disclosure or its application.

[0030] Edge computing involves the distributed deployment and unified management of infrastructure resources. Deployment points with relatively concentrated resources can be called "central clouds" (hereinafter referred to as "clouds"), while deployment points with fewer resources are called "edge clouds" (hereinafter referred to as "edges"). In an edge computing cloud platform, there are a few central clouds and a majority of edge clouds, so how to schedule platform resources becomes very important.

[0031] The collaboration between the cloud and the edge mainly includes the following three scenarios:

[0032] (1) Intelligent collaboration: The edge uses device data for analysis and reasoning, and the cloud performs more accurate model training and then sends it to the edge;

[0033] (2) Data collaboration: There is a reliable and efficient message channel between the edge and the cloud, which supports the resumption of text, image, video and other data transmissions after network interruption and incremental data transmission.

[0034] (3) Security collaboration: Edge nodes provide some security policies, including firewalls and certificates, while the cloud provides comprehensive security policies.

[0035] Figure 1 This illustrates an existing cloud-edge collaborative architecture. For example... Figure 1 As shown, the cloud uses a machine learning engine to train an artificial intelligence (AI) model, then stores the AI ​​model in a disk array and sends it to the edge via a data communication module for inference execution. The edge uses the data communication module to receive the AI ​​model, perform analysis and inference, and directly uploads the labeled images, videos, and other data from the inference process to the cloud for recalibration and training of the AI ​​model.

[0036] However, in actual application scenarios, efficient access of massive, multi-source heterogeneous data and efficient collaboration of AI model training and delivery have extremely high requirements on the storage architecture, and put forward higher requirements on data access efficiency, data security, scalability, cost, etc. For example, the existing cloud-edge collaboration architecture as shown in Figure 1 There may be the following problems with the existing cloud-edge collaboration architecture as shown in

[0037] 1. Difficult to expand storage space: massive data requires a large amount of storage resources, and AI model training requires a storage architecture that can be expanded indefinitely as data grows. However, existing technologies are generally based on file storage and block storage, and usually achieve vertical expansion by adding more computing resources to a single node, so the scalability will reach an upper limit when the data reaches hundreds of TB;

[0038] 2. Poor search and classification ability of metadata: detailed descriptive metadata helps the system to easily mark, search, locate and analyze data, while file and block systems cannot enable program or user-defined extended attributes;

[0039] 3. Low data channel transmission efficiency: data pipeline efficiency will directly affect model training efficiency, and in the face of large data sets and AI models, the data upload and download efficiency of existing architectures is low.

[0040] Therefore, the present disclosure proposes a cloud-edge collaboration architecture based on object storage to improve system scalability, metadata search and classification ability, data channel transmission efficiency, and ensure safe and efficient cloud-edge collaboration in AI application scenarios.

[0041] Figure 2 A schematic diagram of a cloud-edge collaboration architecture according to an embodiment of the present disclosure is shown, in which Figure 1 Compared with

[0042] As shown in Figure 2 The cloud-edge collaboration architecture of the present disclosure adopts object-based storage technology, deploys one or more object storage servers in the cloud for storing text, pictures, audio, video and other files. Object storage is a massive, secure, low-cost and highly reliable cloud storage service suitable for storing any type of file, and supports uploading and querying of data metadata. Examples of object storage technology include Amazon's Simple Storage Service (S3), Microsoft Azure's Blob storage or Google's cloud storage, but the object storage of the present disclosure can not be limited to these. In addition, according to an embodiment of the present disclosure, the object storage server supports generating shared links for various data.

[0043] According to embodiments of the present disclosure, corresponding function modules can be set up in the cloud and the edge, thereby realizing various cloud-edge collaboration.

[0044] In the example of intelligent collaboration, the cloud device utilizes a machine learning engine to perform model training to generate an AI model. To support intelligent collaboration from the cloud to the edge, a cloud model management module can be set up in the cloud, and an edge model management module can be set up in the edge. The cloud device can perform the cloud-edge collaboration method shown in Figure 3

[0045] In step 101, the cloud device (first device) can upload AI model file data to an object storage server for storage. Preferably, the first device can upload meta information about the AI model file to the object storage server together. The meta information can include various custom information facilitating data identification and classification, and can describe various extended attributes of the AI model;

[0046] In step 102, the cloud device can receive a shared link corresponding to the stored AI model file data from the object storage server. The shared link is generated by the object storage server, and can include storage information of the data, and preferably also includes an expiration time of the shared link;

[0047] In step 103, the cloud device can issue the shared link to a second device as an edge device (e.g., edge node 1 shown in Figure 2 In step 103, the cloud device can issue the shared link to a second device as an edge device (e.g., edge node 1 shown in Figure 2 In step 103, the cloud device can issue the shared link to a second device as an edge device (e.g., edge node 1 shown in

[0048] Correspondingly, the edge device can perform the cloud-edge collaboration method shown in Figure 4

[0049] In step 201, the second device as an edge device (e.g., edge node 1 shown in Figure 2 In step 201, the second device as an edge device (e.g., edge node 1 shown in

[0050] ​​In step 202, the edge device can request the object storage server to download the AI model file data based on the shared link. In the case where other target edge nodes are also specified in the task list, the edge device can send the downloaded AI model to these edge nodes (e.g. Figure 2 Edge node 2 shown in FIG. 1B).

[0051] In this way, efficient intelligent collaboration between the cloud and the edge is achieved by means of the efficient upload and download channel provided by the object storage server. However, although the process of collaboration from the cloud to the edge is described by taking intelligent collaboration as an example, it should be understood that the cloud-edge collaboration method of the present disclosure is not limited to transferring AI models, but can be applied to transferring any kind of data from the cloud to the edge, and the process is also as shown in Figure 3 and Figure 4 In the example of data collaboration, the edge device performs analysis and inference with the AI model, and generates source data such as labeled pictures and videos in the AI inference process. In order to support data collaboration from the edge to the cloud, a cloud data management module can be set up in the cloud, and an edge data management module can be set up in the edge. The edge device can perform the cloud-edge collaboration method shown in Figure 3 and Figure 4

[0052] In the example of data collaboration, the edge device performs analysis and inference with the AI model, and generates source data such as labeled pictures and videos in the AI inference process. In order to support data collaboration from the edge to the cloud, a cloud data management module can be set up in the cloud, and an edge data management module can be set up in the edge. The edge device can perform the cloud-edge collaboration method shown in Figure 3

[0053] In step 101, the edge device (first device) can upload source data to the object storage server for storage. Preferably, the first device can upload meta-information about the source data to the object storage server together, so as to facilitate the retrieval and classification of the source data;

[0054] In step 102, the edge device can receive a shared link corresponding to the stored source data from the object storage server;

[0055] In step 103, the edge device can report the shared link to the second device as the cloud device through the data communication module, so that the cloud device can request the object storage server to download the source data based on the shared link.

[0056] In the case where there are multiple edge nodes that need to upload source data, one edge node (e.g. Figure 2 Edge node 1 shown in FIG. 1B) can be selected as the master node, and other target edge nodes (e.g. Figure 2 Edge node 2 shown in FIG. 1B) within the same local area network, for example, can be collected to report the source data to be uploaded to the cloud through the above-mentioned cloud-edge collaboration method.

[0057] Correspondingly, the cloud device can perform the cloud-edge collaboration method shown in Figure 4 ​​The cloud-edge collaboration method shown in

[0058] In step 201, the second device as the cloud device can receive the sharing link from the edge device through the data communication module, so as to obtain the storage information of the source data at the object storage server;

[0059] In step 202, the cloud device can request the object storage server to download the stored source data based on the sharing link.

[0060] In this way, by means of the efficient upload and download channel provided by the object storage server, efficient data collaboration between the cloud and the edge is realized. Although the cloud-edge collaboration method is described by taking the data collaboration as an example, it should be understood that the cloud-edge collaboration method of the present disclosure is not limited to transmitting data for training an AI model, but can be applied to transmitting any kind of data of the edge to the cloud, and the process is also as shown in Figure 3 and Figure 4 The process of collaboration from the edge to the cloud is described by taking the data collaboration as an example, but it should be understood that the cloud-edge collaboration method of the present disclosure is not limited to transmitting data for training an AI model, but can be applied to transmitting any kind of data of the edge to the cloud, and the process is also as shown in Figure 3 and Figure 4

[0061] As shown in Figure 2 The cloud-edge collaboration architecture of the present disclosure can also include a node scheduling module in the cloud. Unlike the node management module in Figure 1 , the node scheduling module according to the present disclosure can select a master node from a group of edge nodes to perform the cloud-edge collaboration method described above with reference to Figure 3 or Figure 4 When the current edge master node fails, the node scheduling module can elect a new edge master node according to a predefined election algorithm.

[0062] According to an embodiment of the present disclosure, the node scheduling module can represent the comprehensive performance index of each edge node as a multi-parameter function H = f(memory, throughput, GPU), and then confirm the master node on the edge side according to the maximum value calculated by the target function H.

[0063] For example, the target function may be used, where MEM is the memory occupancy rate, MEM avg is the average memory occupancy rate of a group of edge nodes; TPS is the throughput, TPS avg is the average system throughput; CPU is the CPU occupancy rate, CPU avg is the average CPU occupancy rate.

[0064] Of course, the algorithm for selecting the master node can not be limited to this, and the node management module in the cloud can consider various factors such as computing resources, storage resources, network resources, etc., as long as a suitable node can be selected.

[0065] ​The cloud-edge collaboration architecture of the present disclosure can also include other module functions, such as the cloud end can include: a data communication module for communication routing with the edge end; a machine learning engine responsible for AI model training; an API Server used as an interactive interface for the front-end visual management system, etc. The edge end can include: a data communication module for communication routing with the cloud end or other edge nodes; an edge engine for carrying edge application execution environments such as edge container, function calculation, rule engine, etc.

[0066] According to an embodiment of the present disclosure, in order to implement a secure collaboration strategy, the communication interface adopts the HTTP / HTTPS standard protocol form, and as an example, the shared link generated by the object storage server can adopt the following format: https: / / [host:port] / [filepath]?AccessKeyId=[id]&Expires=[expires]&Signature=[signature]. Wherein:

[0067] host:port represents the address of the object storage server,

[0068] filepath represents the storage path of the data,

[0069] AccessKeyId represents the user id, i.e. the account registered with the object storage service, using this account can access the corresponding object storage resource pool. The cloud-edge collaboration system of the present disclosure can open a separate account, and this account will also be recorded in the cloud end and the edge end, so that both ends can access the object storage server using this account,

[0070] Expires represents the expiration time,

[0071] Signature represents the digital signature.

[0072] When it is necessary to download data from the object storage server, the cloud end device or the edge device can generate a download request based on the information in the shared link. The object storage server authenticates each request of the user (the cloud end device or the edge device initiating the download request) first, i.e. according to the server address, user id, file path, etc. Information in the download request, using the same signature generation algorithm as the digital signature in the shared link to calculate the digital signature, and compare it with the digital signature carried in the request. If the comparison is consistent, the authentication is passed, otherwise the request is rejected.

[0073] Only users with a signature authentication pass can download the file using the shared link, and when the request time exceeds the expiration time, the shared link is invalid, and the file cannot be downloaded using the shared link, thereby ensuring the security of data collaboration. It should be understood that the present disclosure can use any feasible data signature technology, including but not limited to that used in Amazon S3.

[0074] Figure 5 is an interaction flow diagram of intelligent collaboration using the cloud-edge collaboration technology of the present disclosure. As shown in Figure 5 The intelligent collaboration flow according to the present disclosure includes:

[0075] S11: After the machine learning engine in the cloud performs model training and generates a trained AI model, the cloud device initiates a model file data upload request to the object storage server (e.g., through a cloud model management module) to upload AI model file data (preferably, also corresponding meta information) to the object storage server;

[0076] S12: After the file is uploaded successfully, the object storage server returns a successful upload response;

[0077] S13: When the cloud needs to issue an AI model to the edge, the cloud device (e.g., through a cloud model management module) initiates a file data query request to the object storage server;

[0078] S14: The object storage server finds the corresponding AI model file and generates a shared link to return to the cloud. It should be understood that S13 and S14 are not necessary, and the object storage server can also send the shared link to the cloud device directly through the successful upload response in S12;

[0079] S15: The cloud device (through the cloud model management module) issues the obtained shared link to the edge device (edge node 1). The edge node 1 can be a master node in a group of edge nodes, and the node management module of the cloud device finds the master node according to a master node election algorithm. In addition, the cloud device simultaneously issues a task list, which specifies the model to be issued and the target edge node;

[0080] S16: After receiving the shared link, the model management module of the edge node 1 initiates a file download request to the object storage server;

[0081] S17: The edge node 1 downloads the model file data through the model management module;

[0082] S18: The edge node 1 sends the model file data to other edge nodes (e.g., edge node 2) listed in the task list;

[0083] S19: After the sending is successful, the edge node 2 informs the edge node 1 that the cloud model issuance is successful;

[0084] S110: The edge node 1 informs the cloud device of the success of the issuance.

[0085] Figure 6 is an interaction flow diagram of data collaboration using the cloud-edge collaboration technology of the present disclosure. As shown in Figure 6 The data collaboration flow according to the present disclosure includes:

[0086] S21: The edge device (for example, the edge node 2) analyzes and reasons using the AI model issued by the cloud, and sends the source data (such as pictures and videos) marked in the AI reasoning process to the edge host node (the edge node 1) in the same local area network through the edge-end data management module;

[0087] S22: The edge node 1 initiates a source data upload request to the object storage server to upload the source data (preferably, also the corresponding meta information) to the object storage server;

[0088] S23: After the upload is successful, the object storage server returns an upload success response to the edge node 1, which can include a sharing link corresponding to the source data. Alternatively, the upload success response can not include the sharing link, and the edge node 1 can subsequently initiate a query request to the object storage server, and the object storage server includes the sharing link in the response to the query request;

[0089] S24: The edge node 1 reports the received sharing link to the cloud device (the cloud data management module);

[0090] S25: The cloud device returns a sharing link reporting success response to the edge node 1;

[0091] S26: The edge node 1 returns a reporting success response to the edge node 2;

[0092] S27: The cloud data management module initiates a source data download request to the object storage server through the sharing link;

[0093] S28: After the source data is downloaded successfully, the object storage server returns a download success response to the cloud device;

[0094] S29: The cloud device can process and learn on the downloaded source data using the machine learning engine, and the cloud machine learning engine corrects the AI model using the processed data, and uploads the updated data to the object storage server by the cloud model management module. The upload request can carry custom meta information, which can identify and classify the data, and when a data query request is initiated, the meta information can be returned together with the data, so as to realize the search and positioning of the meta data;

[0095] S210: The object storage server returns an upload success response to the cloud device.

[0096] The cloud-edge collaboration technology according to the present disclosure has more advantages and effects. For example:

[0097] 1. Ultra-high elasticity scalability: using the cloud-edge collaboration technology according to the present disclosure, the storage space can be expanded by adding object storage server nodes anytime and anywhere, and the storage space can be elastically expanded to tens of PB or even more;

[0098] 2. Metadata search and classification ability: by adding custom metadata to the data, the data is labeled, improving the searchability of the stored data, and effectively managing large data sets;

[0099] 3. Improve data channel efficiency: data pipeline efficiency will directly affect model training efficiency. The present disclosure proposes that the cloud and edge share a link to upload and download data, and the edge node distributes data to other nodes in the same local area network, which can improve data transmission efficiency and speed up the training results on large data sets.

[0100] 4. Low-cost access: object storage can be delivered in batches, and its cost is only a small part of traditional proprietary enterprise storage. The data distribution strategy in the local area network can reduce the access bandwidth of the edge to the object storage and save traffic charges.

[0101] The following refers to Figure 7 An exemplary hardware configuration of a computing device 800 that can be used to implement the cloud-edge collaboration method according to the present disclosure is described. The computing device 800 can be any machine configured to perform processing and / or computation. The computing device 800 can include, but is not limited to, a workstation, a server.

[0102] As Figure 7As shown, computing device 800 may include one or more components that may be connected to or communicate with bus 820 via one or more interfaces. Bus 802 may include, but is not limited to, Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MCA) bus, Enhanced ISA (EISA) bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus. Computing device 800 may include, for example, one or more processors 804, one or more input devices 806, and one or more output devices 808. The one or more processors 804 may be any kind of processor and may include, but is not limited to, one or more general-purpose processors or special-purpose processors (such as dedicated processing chips). Input device 806 may be any type of input device capable of inputting information to the computing device and may include, but is not limited to, a mouse, keyboard, touchscreen, microphone, and / or remote controller. Output device 808 may be any type of device capable of presenting information and may include, but is not limited to, a monitor, speaker, video / audio output terminal, vibrator, and / or printer.

[0103] The computing device 800 may also include or be connected to a non-transitory storage device 814, which may be any non-transitory storage device capable of storing data, wherein executable computer instructions may be stored, which, when executed, cause the processor 804 to perform the above-mentioned... Figure 3 or Figure 4 The cloud-edge collaboration method is described. The non-transitory storage device 814 may include, but is not limited to, disk drives, optical storage devices, solid-state storage, floppy disks, flexible disks, hard disks, magnetic tapes or any other magnetic media, compressed disks or any other optical media, cache memory and / or any other storage chip or module, and / or any other medium from which a computer can read data, instructions, and / or code. The computing device 800 may also include random access memory (RAM) 810 and read-only memory (ROM) 812. ROM 812 may store executable programs, utilities, or processes in a non-volatile manner. RAM 810 provides volatile data storage and stores instructions related to the operation of the computing device 800. The computing device 800 may also include a network / bus interface 816 coupled to a data link 818. The network / bus interface 816 can be any kind of device or system capable of enabling communication with external devices and / or networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication devices and / or chipsets (such as Bluetooth™ devices, 1302.11 devices, WiFi devices, WiMax devices, cellular communication facilities, etc.).

[0104] It should be understood that each unit of the resource sequencing apparatus 100 described in the above embodiments is only a logical module according to the specific function implemented by it, and is not intended to limit the specific implementation. In actual implementation, each unit described above can be implemented as an independent physical entity, or can also be implemented by a single entity (for example, a processor (CPU or DSP, etc.), an integrated circuit, etc.).

[0105] Various aspects, embodiments, implementations or features of the foregoing embodiments can be used individually or in any combination. Various aspects of the foregoing embodiments can be implemented by software, hardware, or a combination of hardware and software.

[0106] For example, the foregoing embodiments can be embodied as computer readable code on a computer readable medium. The computer readable medium is any data storage device that can store data which can thereafter be read by a computer system. Examples of computer readable media include read-only memory, random-access memory, CD-ROMs, DVDs, magnetic tape, hard drives, solid state drives, and optical data storage devices. The computer readable medium can also be distributed among computer system(s) connected through a network such that the computer readable code is stored and executed as a distributed system.

[0107] For example, the foregoing embodiments can take the form of hardware circuitry. The hardware circuitry can include any combination of combinational logic circuitry, clocked storage devices (such as flip-flops, latches, etc.), finite state machines, memory such as static random access memory or embedded dynamic random access memory, custom designed circuitry, programmable logic arrays, etc.

[0108] In one embodiment, a hardware circuit according to the present disclosure can be implemented by encoding a circuit description in a hardware description language (HDL) such as Verilog or VHDL. The HDL description can be synthesized against a library of cells designed for a given integrated circuit fabrication technology, and can be modified for timing, power, and other reasons to obtain a final design database, which can be transmitted to a foundry for production of integrated circuits by a semiconductor fabrication system. The semiconductor fabrication system can produce integrated circuits by depositing semiconductor materials, removing materials, changing the shape of deposited materials, modifying materials (e.g., by doping materials or modifying the dielectric constant with ultraviolet treatment), etc., (e.g., on a wafer that can include masks). The integrated circuits can include transistors and can also include other circuit elements (e.g., passive elements such as capacitors, resistors, inductors, etc.) as well as interconnections between the transistors and circuit elements. Some embodiments can implement multiple integrated circuits coupled together to implement the hardware circuit, and / or can use discrete elements in some embodiments.

[0109] While some embodiments of the present disclosure have been shown in detail by way of example, it will be appreciated that the examples described above are intended to be illustrative only and are not limiting of the scope of the present disclosure. It will be appreciated that the embodiments described above can be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A cloud-edge collaboration method, comprising: The first device stores the data on the object storage server. The first device receives a shared link corresponding to the data from the object storage server. as well as The first device sends the shared link to the second device, enabling the second device to request the data from the object storage server based on the shared link. The first device is one of the cloud device and the edge device, and the second device is the other of the cloud device and the edge device. In the case that the first device is a cloud device, the data includes data about AI models generated by the cloud device using a machine learning engine. In the case where the first device is an edge device, the data includes tagged source data generated during the edge device's analysis and reasoning using an AI model.

2. The method as described in claim 1, wherein, The first device is a cloud device, and the second device is the master node in a group of edge nodes.

3. The method of claim 2, further comprising: The first device sends a task list to the second device, the task list specifying the AI ​​model to be sent and the target edge node.

4. The method of claim 2, further comprising: Based on the objective function H, a second device is selected as the master node from the set of edge nodes. Wherein, the objective function H = • Where MEM represents memory usage, MEM avg Average memory usage; TPS is the throughput. avg Average system throughput; CPU utilization; CPU avg This represents the average CPU utilization.

5. The method of claim 1, wherein, The first device is the master node in a set of edge nodes, and the second device is a cloud device.

6. The method of claim 5, wherein, Each of the set of edge nodes is configured to perform analytical reasoning using an AI model, and the data includes source data marked in the analytical reasoning.

7. The method of claim 6, wherein, The first device collects source data from other edge nodes in the set of edge nodes.

8. The method of claim 1, further comprising: The first device stores the data along with metadata about the data on an object storage server.

9. The method of claim 1, wherein, The shared link includes the object storage server address, the storage path of the data, the user ID, the expiration time, and the digital signature.

10. The method of claim 9, wherein, The object storage server calculates a digital signature based on the object storage server address, the storage path of the data, and the user ID. It then compares the calculated digital signature with the digital signature carried in the data download request and allows the data to be downloaded only if the comparison results match.

11. A cloud-edge collaboration method, comprising: The second device receives a shared link from the first device, and the shared link corresponds to data stored by the first device on an object storage server. as well as The second device requests the data from the object storage server based on the shared link. The first device is one of the cloud device and the edge device, and the second device is the other of the cloud device and the edge device. In the case that the first device is a cloud device, the data includes data about AI models generated by the cloud device using a machine learning engine. In the case where the first device is an edge device, the data includes tagged source data generated during the analysis and reasoning process performed by the edge device using an AI model.

12. The cloud-edge collaboration method as described in claim 11, wherein, The shared link includes the object storage server address, the storage path of the data, the user ID, the expiration time, and the digital signature.

13. A cloud-edge collaborative device, comprising: processor; A memory storing executable instructions that, when executed, cause the processor to perform the cloud-edge collaboration method according to any one of claims 1-12.

14. A computer-readable storage medium comprising executable instructions, which, when executed, cause the method according to any one of claims 1-12 to be performed.

Images