Federated Learning Client-Side Controller Based on IoT

Abstract

The present invention discloses a joint learning client-side controller based on the Internet of Things, including a client side, a cloud management side, and a joint learning side. The client side includes a client management console, a client-side controller, an Internet of Things layer, a hybrid cloud layer, and a hybrid cloud controller. The client-side controller in the client side serves as a management access node. The client-side controller controls the node connection to connect the client side to the cloud management side and the joint learning side. The client-side controller and the Internet of Things layer mutually perform tasks. The client-side controller provides a scheduling execution entry to connect the hybrid cloud management and device layer of the cloud management side to perform corresponding operations. The client-side controller and the resource node open up script delivery, script execution, and file delivery channels through mutual trust authentication. The present invention sets the client-side controller as a management access node, controls the node to connect the cloud management side and the client side, and accesses an environment isolated from the new intelligent cloud to ensure data docking between the cloud management side and the client side, and to ensure the progress of joint learning.

Description

Technical Field

[0001] The present invention relates to the technical field of joint learning, and in particular to a joint learning client-side controller based on the Internet of Things. Background Art

[0002] Federated Learning is an emerging basic technology for artificial intelligence. It was first proposed by Google in 2016. It was originally used to solve the problem of local model update for Android mobile phone terminal users. Its design goal is to carry out efficient machine learning among multiple participants or multiple computing nodes under the premise of ensuring information security during big data exchange, protecting terminal data and personal data privacy, and ensuring legality and compliance. Among them, the machine learning algorithms that can be used in federated learning are not limited to neural networks, but also include important algorithms such as random forests. Federated learning is expected to become the basis for the next generation of artificial intelligence collaborative algorithms and collaborative networks.

[0003] Federated learning is a machine learning setting in which many clients jointly train a model under the coordination of a central server while keeping the training data decentralized and distributed.

[0004] The long-term goal of federated learning is to analyze and learn data from multiple data owners without exposing the data.

[0005] The system architecture of federated learning is as follows: Taking a scenario involving two data owners (i.e., companies A and B) as an example, the system architecture of federated learning is introduced. This architecture can be extended to scenarios involving multiple data owners. Suppose companies A and B want to jointly train a machine learning model, and their business systems each have relevant data of their respective users. In addition, company B also has label data that the model needs to predict. For data privacy protection and security considerations, A and B cannot directly exchange data, and a federated learning system can be used to build a model. The system architecture of federated learning consists of three parts.

[0006] Federated learning can be classified into horizontal federated learning, vertical federated learning, and federated transfer learning (FmL).

[0007] Horizontal Federated Learning When there is a lot of overlap in user features of two data sets, but little overlap in users, we split the data sets horizontally (i.e., user dimension) and take out the data with the same user features but not exactly the same users for training. This method is called horizontal federated learning. For example, there are two banks in different regions, and their user groups come from their respective regions, with very little overlap. However, their businesses are very similar, so the recorded user features are the same. At this point, we can use horizontal federated learning to build a joint model. In 2016, Google proposed a data joint modeling solution for Android phone model updates: when a single user uses an Android phone, the model parameters are continuously updated locally and uploaded to the Android cloud, so that data owners with the same feature dimensions can establish a joint model.

[0008] In the case where there is a large overlap of users in two data sets but a small overlap of user features, the data sets are split vertically (i.e., feature dimension) and the data with the same users but different user features is taken out for training. This method is called vertical federated learning. For example, there are two different institutions, one is a bank in a certain place, and the other is an e-commerce company in the same place. Their user groups are likely to include most of the residents in the place, so the user intersection is large. However, since the bank records the user's income and expenditure behavior and credit rating, while the e-commerce company keeps the user's browsing and purchase history, the intersection of their user features is small. Vertical federated learning is to aggregate these different features in an encrypted state to enhance the model capability. At present, many machine learning models such as logistic regression models, tree structure models, and neural network models have gradually been proven to be able to be built on this federated system.

[0009] Federated transfer learning When there is little overlap between users and user features in the two datasets, we do not split the data, but use transfer learning to overcome the lack of data or labels. This method is called federated transfer learning. For example, there are two different institutions, one is a bank in China, and the other is an e-commerce company in the United States. Due to geographical restrictions, the user groups of these two institutions have little overlap. At the same time, due to the different types of institutions, the data features of the two only overlap to a small extent. In this case, in order to conduct effective federated learning, transfer learning must be introduced to solve the problem of small unilateral data size and few labeled samples, thereby improving the effect of the model.

[0010] Federated learning has the following advantages:

[0011] (1) Data isolation: data will not be leaked to the outside, meeting the needs of user privacy protection and data security;

[0012] (2) It can ensure that the model quality is intact and negative transfer does not occur, ensuring that the federated model is better than the fragmented independent model;

[0013] (3) The participants are equal in status and can achieve fair cooperation;

[0014] (4) It can ensure that all parties involved can exchange information and model parameters in an encrypted manner while maintaining their independence and achieving growth at the same time.

[0015] However, currently in the joint learning, there is a lack of a docking and scheduling controller between the client side 100, the cloud management side 200 and the joint learning side 300, making it difficult to maintain the joint learning in progress. Summary of the invention

[0016] The purpose of the present invention is to provide a joint learning client-side controller based on the Internet of Things. The client-side controller connects to the cloud management side and the client side, and fully schedules data so that it can maintain the normal progress of joint learning, which can solve the problems raised in the above background technology.

[0017] To achieve the above object, the present invention provides the following technical solutions:

[0018] The client-side controller of the federated learning based on the Internet of Things includes the client side, the cloud management side and the federated learning side. The client side includes the client management console, the client-side controller, the Internet of Things layer, the hybrid cloud layer and the hybrid cloud controller. The client-side controller in the client side serves as a management access node. The client-side controller controls the node connection to connect the client side to the cloud management side and the federated learning side. The client-side controller and the Internet of Things layer execute tasks with each other. The client-side controller provides a scheduling execution entry to connect to the hybrid cloud management and device layer on the cloud management side to perform corresponding operations. The script distribution, script execution and file distribution channels are opened between the client-side controller and the resource node through mutual trust authentication.

[0019] Furthermore, the client management console includes a resource management module, a learning management module and an operation and maintenance management module, wherein the learning management module is connected to the input interface of the client-side controller, and the learning management module inputs the data to be managed to the client-side controller.

[0020] Furthermore, the client-side controller includes an orchestration execution module, a configuration checking module, a configuration collection module, an information forwarding module, a performance collection module, and a device management module, wherein the orchestration execution module is controlled by the cloud management side and accepts instructions from the cloud management side to complete the orchestration execution task.

[0021] Furthermore, the IoT layer includes an edge module, which uploads results to the information forwarding module, and the information forwarding module sends instructions to the edge module to execute tasks. Mutual trust authentication is performed between the edge module and the device management module.

[0022] Furthermore, the hybrid cloud layer includes a hybrid cloud unmanaged module and a hybrid cloud management module. The hybrid cloud unmanaged module uploads results to the hybrid cloud controller, and the hybrid cloud controller sends instructions to the hybrid cloud unmanaged module to execute tasks.

[0023] Furthermore, the hybrid cloud controller includes an intelligent management module, a security protection module, a service management module and a basic function module, wherein:

[0024] The intelligent management module is used for intelligent management of the hybrid cloud controller to ensure the smooth progress of joint learning;

[0025] The security protection module is used for security protection on the client side, cloud management side, and federated learning side in federated learning;

[0026] The service management module is used to conduct service experience in federated learning;

[0027] The basic function module is used to provide a variety of basic functions in joint learning.

[0028] Furthermore, the intelligent management module includes a channel intelligent switching module, a power adjustment module, and a load balancing module;

[0029] The security protection module includes a device connection security protection module, a node access protection module, and an illegal device detection module;

[0030] The service management module includes a business service module, a location information service module, a rights management module, and a time application control module;

[0031] The basic functional modules include a data forwarding module, an alarm module, an automated operation and maintenance module, and a cloud account management module.

[0032] Furthermore, the hybrid cloud controller also includes a probe module and a management authentication module.

[0033] Furthermore, the cloud management side includes a cloud platform, which includes a usage statistics module, an orchestration and scheduling center, a physical data module, and a monitoring and management platform. The usage statistics module is connected to the monitoring and management platform, the monitoring and management platform is connected to the hybrid cloud controller, the physical data module transmits data to the orchestration and scheduling center, and the orchestration and scheduling center sends execution instructions to the orchestration execution module.

[0034] Furthermore, the joint learning side includes a joint learning server and a joint learning runtime module. The joint learning server receives work instructions from the learning management module and the orchestration execution module. The joint learning server and the joint learning runtime module transmit data to each other, and the joint learning runtime module then feeds back to the orchestration scheduling center.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The present invention sets a client-side controller as a management access node. The control node connects the cloud management side and the client side, and accesses an environment isolated from Xin Zhiyun. The client-side controller is a hardware and software device that can be connected to the outside and is provided by Xin Zhiyun or the client side. It ensures data connection between the cloud management side and the client side and ensures the correct progress of joint learning.

[0037] 2. The client-side controller of the present invention provides a scheduling execution portal to connect to the hybrid cloud management and device layer on the cloud to perform corresponding operations. It provides performance collection and configuration collection functions to collect performance data in the joint learning process, provides data forwarding channels to forward joint learning model results and intermediate results, provides configuration collection and inspection, and verifies the resource configuration participating in the joint learning.

[0038] 3. The hybrid cloud controller of the present invention performs data matching tasks at the hybrid cloud layer, optimizes data resources, and performs tasks with the monitoring management platform on the cloud management side, making the client side and the cloud management side more friendly. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is an architecture diagram of a joint learning client-side controller based on the Internet of Things according to the first embodiment of the present invention;

[0040] Figure 2 This is a diagram of the architecture of the intelligent management module of the present invention;

[0041] Figure 3 A schematic diagram of a load balancing algorithm of the present invention;

[0042] Figure 4 This is a structural diagram of the security protection module of the present invention;

[0043] Figure 5 This is an architecture diagram of the service management module of the present invention;

[0044] Figure 6 This is an architecture diagram of the basic functional modules of the present invention;

[0045] Figure 7 This is an architecture diagram of a joint learning client-side controller based on the Internet of Things according to Embodiment 2 of the present invention;

[0046] In the figure: 100, client side; 101, client management console; 1011, resource management module; 1012, learning management module; 1013, operation and maintenance management module; 102, client side controller; 1021, orchestration execution module; 1022, configuration check module; 1023, configuration collection module; 1024, information forwarding module; 1025, performance collection module; 1026, device management module; 103, IoT layer; 1031, edge module; 104, hybrid cloud layer; 1041, hybrid cloud unmanaged module; 1042, hybrid Hybrid cloud management module; 105, hybrid cloud controller; 1051, intelligent management module; 1052, security protection module; 1053, service management module; 1054, basic function module; 1055, probe module; 1056, management authentication module; 200, cloud management side; 201, cloud platform; 2011, usage statistics module; 2012, orchestration and scheduling center; 2013, entity data module; 2014, monitoring management platform; 300, joint learning side; 301, joint learning server; 302, joint learning runtime module. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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.

[0048] Example 1

[0049] Figure 1 The architecture diagram of the IoT-based federated learning client-side controller of this embodiment is shown. The IoT-based federated learning client-side controller includes a client side 100, a cloud management side 200, and a federated learning side 300. The client side 100 includes a client management console 101, a client-side controller 102, an IoT layer 103, a hybrid cloud layer 104, and a hybrid cloud controller 105. The client-side controller 102 in the client side 100 serves as a management access node. The client-side controller 102 controls the node connection to connect the client side 100 to the cloud management side 200 and the federated learning side 300, so that the client side 100, the cloud management side 200, and the federated learning side 300 can be connected to each other, thereby ensuring the normal progress of the federated learning. The client-side controller 102 and the IoT layer 103 execute tasks with each other. The client-side controller 102 provides a scheduling execution entry to connect to the hybrid cloud management and device layer of the cloud management side to perform corresponding operations. The client-side controller 102 and the resource node open up script distribution, script execution, and file distribution channels through mutual trust authentication.

[0050] The client management console 101 includes a resource management module 1011 , a learning management module 1012 and an operation and maintenance management module 1013 , wherein the learning management module 1012 is connected to the input interface of the client-side controller 102 , and the learning management module 1012 inputs the data to be managed to the client-side controller 102 .

[0051] The resource management module 1011 manages the customer resources of the client side 100, maintains good customers, and improves customer satisfaction.

[0052] The learning management module 1012 is used to manage the learning situation of the customers on the client side 100, such as the learning status, learning progress, learning effect, etc. of the customers, so as to arrange future learning plans.

[0053] The operation and maintenance management module 1013 performs operation and maintenance scheduling for the entire customer learning process of the customer side 100, and keeps abreast of the customer's learning problems.

[0054] The client-side controller 102 includes an orchestration execution module 1021, a configuration checking module 1022, a configuration collection module 1023, an information forwarding module 1024, a performance collection module 1025, and a device management module 1026, wherein the orchestration execution module 1021 is controlled by the cloud management side 200, and accepts instructions from the cloud management side 200 to complete the orchestration execution task.

[0055] The orchestration execution module 1021 can communicate with the orchestration scheduling center 2012 of the cloud management side 200, receive the orchestration execution instruction from the orchestration scheduling center 2012, and complete the orchestration operation.

[0056] The configuration check module 1022 provides a data forwarding channel to forward the joint learning model results and intermediate results, provides configuration collection and inspection, and verifies the resource configuration involved in the joint learning.

[0057] The configuration collection module 1023 provides performance collection and configuration collection functions, and collects performance data in the joint learning process.

[0058] The information forwarding module 1024 transmits data with the edge module 1031 of the IoT layer 103 to ensure task execution.

[0059] The performance collection module 1025 collects the performance of the device to facilitate further understanding of the operating parameters and functions of the device.

[0060] The IoT layer 103 includes an edge module 1031 , which uploads results to the information forwarding module 1024 , and the information forwarding module 1024 sends instructions to the edge module 1031 to execute tasks. The edge module 1031 and the device management module 1026 perform mutual trust authentication.

[0061] The hybrid cloud layer 104 includes a hybrid cloud unmanaged module 1041 and a hybrid cloud management module 1042 . The hybrid cloud unmanaged module 1041 uploads results to the hybrid cloud controller 105 , and the hybrid cloud controller 105 sends instructions to the hybrid cloud unmanaged module 1041 to execute tasks.

[0062] The hybrid cloud controller 105 includes an intelligent management module 1051, a security protection module 1052, a service management module 1053 and a basic function module 1054, wherein:

[0063] The intelligent management module 1051 is used for intelligent management of the hybrid cloud controller 105 to ensure the smooth progress of joint learning.

[0064] The security protection module 1052 is used for security protection of the client side 100, the cloud management side 200 and the joint learning side 300 in joint learning.

[0065] The service management module 1053 is used to perform service experience in joint learning.

[0066] The basic function module 1054 is used to have various basic functions in joint learning.

[0067] Figure 2 The architecture diagram of the intelligent management module of this embodiment is shown. The intelligent management module 1051 includes a channel intelligent switching module, a power adjustment module, and a load balancing module.

[0068] The channel intelligent switching module ensures that the channel allocated to each wireless access point is the most optimized through adaptive channel selection, and minimizes and avoids adjacent / co-channel interference as much as possible.

[0069] The power adjustment module can automatically adjust the power based on the device's RF environment, automatically amplify or reduce the device's RF power, realize automatic detection and compensation of the coverage area, and achieve the best coverage effect.

[0070] Figure 3The algorithm diagram of load balancing of this embodiment is shown. The load balancing module provides balancing based on the number of access users (when the number of users between wireless access points exceeds the set threshold, the uniform distribution of users between different wireless access points can be dynamically adjusted), provides load balancing based on traffic (when the traffic value between wireless access points exceeds the set threshold, the uniform distribution of users between different wireless access points can be dynamically adjusted), and provides load balancing based on frequency band.

[0071] Figure 4 The architecture diagram of the security protection module of this embodiment is shown, and the security protection module 1052 includes a device connection security protection module, a node access protection module, and an illegal device detection module.

[0072] The device connection security protection module is used for wireless access point devices, which are deployed in the open environment of the Internet. The access of fake wireless access point and the hijacking and theft of the wireless access point are two major device security issues. The TLS / DTLS network security protocol is used to reasonably solve the device identity authentication problem on the Internet of Things.

[0073] The node access protection module supports access control based on user blacklists and whitelists, supports SMS-based real-name Internet authentication, and intelligently identifies user identity features based on the way users access the network, allowing users to enjoy customized strategies and services.

[0074] The illegal device detection module provides wireless intrusion detection (WIPS) function, which continuously monitors the wireless space, searches for illegal access points and temporary networks, and provides warnings and attacks on illegal devices / users.

[0075] Figure 5 The architecture diagram of the service management module of this embodiment is shown. The service management module 1053 includes a business service module, a location information service module, a permission management module, and a time application control module.

[0076] Business service module can provide remote office services.

[0077] The location information service module has built-in wireless access point location information identification, which can set different network services based on physical location information. For example, the wireless access point at the front desk provides guests with Internet access function, while limiting access to the Internet and specified internal resources.

[0078] The permission management module can define the access / exit permissions of roles, define one or more roles for users, each role can have different access permissions, control the user's upstream and downstream bandwidth, and provide differentiated management services for users.

[0079] The time application control module can set time-based service control.

[0080] Figure 6 The architecture diagram of the service management module of this embodiment is shown, and the basic function module 1054 includes a data forwarding module, an alarm module, an automated operation and maintenance module, and a cloud account management module.

[0081] The data forwarding module is used for user traffic and can be forwarded locally by the wireless access point based on policy decisions. Under the same wireless access point, one data flow is forwarded locally and one or more data flows are forwarded centrally.

[0082] The alarm module provides interface status alarms, device offline alarms, illegal wireless access point / client alarms, and can customize the scope and services that trigger the warning line.

[0083] The automated operation and maintenance module ensures service reliability. When the network is idle, it can perform network upgrades or fault detection. All of this can be done automatically without the need for real-time human involvement, thus simplifying the difficulty of network operation and maintenance.

[0084] The cloud account management module is used for management and operation in the form of cloud services. The management and operation rights are independent of the ownership of the equipment, but are shared by operation entities at all levels and implemented remotely in the form of cloud accounts.

[0085] The cloud management side 200 includes a cloud platform 201, which includes a usage statistics module 2011, an orchestration and scheduling center 2012, an entity data module 2013, and a monitoring management platform 2014. The usage statistics module 2011 is connected to the monitoring management platform 2014, and the monitoring data of the monitoring management platform 2014 is counted by the usage statistics module 2011. The monitoring management platform 2014 is connected to the hybrid cloud controller 105. The entity data module 2013 transmits data to the orchestration and scheduling center 2012, and the orchestration and scheduling center 2012 sends an execution instruction to the orchestration execution module 1021.

[0086] The joint learning side 300 includes a joint learning server 301 and a joint learning runtime module 302. The joint learning server 301 receives work instructions from the learning management module 1012 and the orchestration execution module 1021. The joint learning server 301 and the joint learning runtime module 302 transmit data to each other, and the joint learning runtime module 302 then feeds back to the orchestration scheduling center 2012.

[0087] Example 2

[0088] Figure 7The architecture diagram of the IoT-based federated learning client-side controller of this embodiment is shown. The IoT-based federated learning client-side controller includes a client side 100, a cloud management side 200, and a federated learning side 300. The client side 100 includes a client management console 101, a client-side controller 102, an IoT layer 103, a hybrid cloud layer 104, and a hybrid cloud controller 105. The client-side controller 102 in the client side 100 serves as a management access node. The client-side controller 102 controls the node connection to connect the client side 100 to the cloud management side 200 and the federated learning side 300, so that the client side 100, the cloud management side 200, and the federated learning side 300 can be connected to each other, thereby ensuring the normal progress of the federated learning. The client-side controller 102 and the IoT layer 103 execute tasks with each other. The client-side controller 102 provides a scheduling execution entry to connect to the hybrid cloud management and device layer of the cloud management side to perform corresponding operations. The client-side controller 102 and the resource node open up script distribution, script execution, and file distribution channels through mutual trust authentication.

[0089] The client management console 101 includes a resource management module 1011 , a learning management module 1012 and an operation and maintenance management module 1013 , wherein the learning management module 1012 is connected to the input interface of the client-side controller 102 , and the learning management module 1012 inputs the data to be managed to the client-side controller 102 .

[0090] The resource management module 1011 manages the customer resources of the client side 100, maintains good customers, and improves customer satisfaction.

[0091] The learning management module 1012 is used to manage the learning situation of the customers on the client side 100, such as the learning status, learning progress, learning effect, etc. of the customers, so as to arrange future learning plans.

[0092] The operation and maintenance management module 1013 performs operation and maintenance scheduling for the entire customer learning process of the customer side 100, and keeps abreast of the customer's learning problems.

[0093] The client-side controller 102 includes an orchestration execution module 1021, a configuration checking module 1022, a configuration collection module 1023, an information forwarding module 1024, a performance collection module 1025, and a device management module 1026, wherein the orchestration execution module 1021 is controlled by the cloud management side 200, and accepts instructions from the cloud management side 200 to complete the orchestration execution task.

[0094] The orchestration execution module 1021 can communicate with the orchestration scheduling center 2012 of the cloud management side 200, receive the orchestration execution instruction from the orchestration scheduling center 2012, and complete the orchestration operation.

[0095] The configuration check module 1022 provides a data forwarding channel to forward the joint learning model results and intermediate results, provides configuration collection and inspection, and verifies the resource configuration involved in the joint learning.

[0096] The configuration collection module 1023 provides performance collection and configuration collection functions, and collects performance data in the joint learning process.

[0097] The information forwarding module 1024 transmits data with the edge module 1031 of the IoT layer 103 to ensure task execution.

[0098] The performance collection module 1025 collects the performance of the device to facilitate further understanding of the operating parameters and functions of the device.

[0099] The IoT layer 103 includes an edge module 1031 , which uploads results to the information forwarding module 1024 , and the information forwarding module 1024 sends instructions to the edge module 1031 to execute tasks. The edge module 1031 and the device management module 1026 perform mutual trust authentication.

[0100] The hybrid cloud layer 104 includes a hybrid cloud unmanaged module 1041 and a hybrid cloud management module 1042 . The hybrid cloud unmanaged module 1041 uploads results to the hybrid cloud controller 105 , and the hybrid cloud controller 105 sends instructions to the hybrid cloud unmanaged module 1041 to execute tasks.

[0101] The hybrid cloud controller 105 includes a probe module 1055 and a management authentication module 1056 .

[0102] The probe module 1055 supports the probe function, and can set the probe collection frequency according to the application scenario, and supports the wireless access point to switch to pure probe mode. As long as the WiFi client turns on the WiFi connection, the wireless access point can obtain the MAC of the mobile terminal, the associated third-party wireless access point, signal strength and other information. At the same time, the collected information can be submitted to the ABLOOMY offline data collection server (VDS), and combined with VDS, user positioning, historical trajectory and other services can be realized.

[0103] The management and authentication module can realize the integrated management and authentication of wired and wireless networks. Whether the user logs in to the wired network or the wireless network, the user's account is managed and authenticated.

[0104] The cloud management side 200 includes a cloud platform 201, which includes a usage statistics module 2011, an orchestration and scheduling center 2012, an entity data module 2013, and a monitoring management platform 2014. The usage statistics module 2011 is connected to the monitoring management platform 2014, and the monitoring data of the monitoring management platform 2014 is counted by the usage statistics module 2011. The monitoring management platform 2014 is connected to the hybrid cloud controller 105. The entity data module 2013 transmits data to the orchestration and scheduling center 2012, and the orchestration and scheduling center 2012 sends an execution instruction to the orchestration execution module 1021.

[0105] The joint learning side 300 includes a joint learning server 301 and a joint learning runtime module 302. The joint learning server 301 receives work instructions from the learning management module 1012 and the orchestration execution module 1021. The joint learning server 301 and the joint learning runtime module 302 transmit data to each other, and the joint learning runtime module 302 then feeds back to the orchestration scheduling center 2012.

[0106] The present invention uses a client-side controller to connect the cloud management side and the client side. The client side exchanges data with the joint learning side through the cloud management side, provides optimization services for joint learning, improves the efficiency of joint learning, and increases the coordination of joint learning.

[0107] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A joint learning client-side controller based on the Internet of Things, characterized in that: The system comprises a client side (100), a cloud management side (200), and a federated learning side (300). The client side (100) comprises a client management console (101), a client side controller (102), an IoT layer (103), a hybrid cloud layer (104), and a hybrid cloud controller (105). The client side controller (102) in the client side (100) serves as a management access node. The client side controller (102) controls the node connection to connect the client side (100) to the cloud management side (200) and the federated learning side (300). The client side controller (102) and the IoT layer (103) mutually execute tasks. The client side controller (102) provides a scheduling execution entry to connect to the hybrid cloud management and device layer of the cloud management side to perform corresponding operations. The client side controller (102) and the resource node open up script delivery, script execution, and file delivery channels through mutual trust authentication. The client management console (101) comprises a resource management module (1011), a learning management module (1012) and an operation and maintenance management module (1013), wherein the learning management module (1012) is connected to an input interface of the client-side controller (102), and the learning management module (1012) inputs data to be managed to the client-side controller (102); The client-side controller (102) comprises an orchestration execution module (1021), a configuration checking module (1022), a configuration collection module (1023), an information forwarding module (1024), a performance collection module (1025), and a device management module (1026), wherein the orchestration execution module (1021) is controlled by the cloud management side (200) and accepts instructions from the cloud management side (200) to complete the orchestration execution task; The cloud management side (200) comprises a cloud platform (201), the cloud platform (201) comprises a usage statistics module (2011), an orchestration and scheduling center (2012), an entity data module (2013), and a monitoring and management platform (2014), the usage statistics module (2011) is connected to the monitoring and management platform (2014), the monitoring and management platform (2014) is connected to the hybrid cloud controller (105), the entity data module (2013) transmits data to the orchestration and scheduling center (2012), and the orchestration and scheduling center (2012) sends an execution instruction to the orchestration execution module (1021); The joint learning side (300) includes a joint learning server (301) and a joint learning runtime module (302). The joint learning server (301) receives work instructions from a learning management module (1012) and an orchestration execution module (1021). The joint learning server (301) and the joint learning runtime module (302) transmit data to each other, and the joint learning runtime module (302) then feeds back to the orchestration scheduling center (2012).

2. The IoT-based joint learning client-side controller according to claim 1, characterized in that: The IoT layer (103) includes an edge module (1031), the edge module (1031) uploads results to the information forwarding module (1024), the information forwarding module (1024) sends instructions to the edge module (1031) to execute tasks, and mutual trust authentication is performed between the edge module (1031) and the device management module (1026).

3. The IoT-based joint learning client-side controller according to claim 2, characterized in that: The hybrid cloud layer (104) includes a hybrid cloud unmanaged module (1041) and a hybrid cloud management module (1042). The hybrid cloud unmanaged module (1041) uploads results to the hybrid cloud controller (105), and the hybrid cloud controller (105) sends instructions to the hybrid cloud unmanaged module (1041) to execute tasks.

4. The IoT-based joint learning client-side controller according to claim 3, characterized in that: The hybrid cloud controller (105) includes an intelligent management module (1051), a security protection module (1052), a service management module (1053) and a basic function module (1054), wherein: The intelligent management module (1051) is used for intelligent management of the hybrid cloud controller (105) to ensure smooth progress of joint learning; The security protection module (1052) is used for security protection of the client side (100), the cloud management side (200) and the joint learning side (300) in joint learning; The service management module (1053) is used to perform service experience in joint learning; The basic function module (1054) is used to have various basic functions in joint learning.

5. The IoT-based joint learning client-side controller according to claim 4, characterized in that: The intelligent management module (1051) includes a channel intelligent switching module, a power adjustment module, and a load balancing module; The security protection module (1052) includes a device connection security protection module, a node access protection module, and an illegal device detection module; The service management module (1053) includes a business service module, a location information service module, a rights management module, and a time application control module; The basic function module (1054) includes a data forwarding module, an alarm module, an automated operation and maintenance module, and a cloud account management module.

6. The IoT-based joint learning client-side controller according to claim 5, characterized in that: The hybrid cloud controller (105) also includes a probe module (1055) and a management authentication module (1056).

Images