Ad-hoc machine learning training with constraints, predictive traffic loading, and private end-to-end encryption

Abstract

The machine learning network includes a plurality of test and measurement devices, one or more of the test and measurement devices having one or more communication interfaces configured to enable the devices to receive and process physical layer signals, a memory, and one or more processors configured to execute a program that causes the one or more processors to receive physical layer data, perform one or more operations on the physical layer data in accordance with a machine learning model to generate modified physical layer data, and send the modified physical layer data to at least one other node in the machine learning neural network. The machine learning network may include a learning node.

Description

[Technical Field]

[0001] This disclosure claims the benefit of U.S. Provisional Patent Application No. 63 / 415,505, filed October 12, 2022, entitled "Episodic Machine Learning with Training Through Constraints, Predictive Traffic Load, and Private End-to-End Encryption," U.S. Provisional Patent Application No. 63 / 429,508, filed December 1, 2022, entitled "AI Machine Learning Cell Component for Communication Link Testing, Measurement, and Maintenance Networks," and U.S. Patent Application No. 18 / 482,801, filed October 6, 2023, entitled "Episodic Machine Learning with Training Through Constraints, Predictive Traffic Load, and Private End-to-End Encryption," the disclosures of which are incorporated herein by reference in their entireties.

[0002] The present disclosure relates to test and measurement systems, and more particularly to systems for testing, measuring, and sustainment (TMS) of physical layer signals (including, but not limited to, optical and electromagnetic signals) in a network. [Background technology]

[0003] The increasing popularity of faster communication standards (Wi-Fi, 6G (6th generation), Internet of Things (IoT)) not only enables new technologies but also creates opportunities for their practical application. Distributed computing, like machine learning, has become a very popular technology. On the other hand, interest in distributed machine learning is increasing due to the increase in communication speed and the stability of communication links.

[0004] Federated learning is one such architecture. In federated learning, the learning process is distributed to edge devices in the network, and edge devices use their own local data to train models. Typically, a generic model resides on a central server or data center, and a copy of the model is shared with edge devices. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-177487 [Patent Document 2] Special Publication No. 2023-548150 [Non-patent literature]

[0006] [Non-Patent Document 1] "Federated Learning," Glossary, Nomura Research Institute, Ltd., [online], [Retrieved June 14, 2025], Internet<https: / / www.nri.com / jp / knowledge / glossary / federated_learning.html> Summary of the Invention [Problem to be solved by the invention]

[0007] Edge devices train models on local data, but methods for implementing these systems are lacking.

[0008] One particular area where this type of system is lacking is in the area of ​​physical layer signal testing, measurement, and maintenance, as the communications network itself employs to test, measure, and verify the network infrastructure. This contrasts with the use of federated learning, which relies on the network infrastructure to perform other tasks. [Means for solving the problem]

[0009] Embodiments of the present application employ a network of nodes for test and measurement, including test and measurement of the network infrastructure itself. Nodes in the network function as nodes in a machine learning neural network, operating on physical layer (PHY) signals through the application of artificial intelligence (AI) / machine learning (ML) input vectors. As used herein, the term "physical layer data" includes both physical layer signals and the data that may be contained in those signals. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 shows a network diagram of a communication network having multiple nodes. [Figure 2] Figure 2 shows a diagram of a cell tuple in a communication network. [Figure 3] FIG. 3 shows a flowchart of an embodiment of a method for adding a new node to a communication network using machine learning. DETAILED DESCRIPTION OF THE INVENTION

[0011] FIG. 1 illustrates an example communication network 10 having multiple test and measurement devices as nodes (e.g., 12). Each node comprises a processing unit that can be trained to become a machine learning model—an artificial intelligence model. In this description, the terms "artificial intelligence" and "machine learning" are used to refer to algorithms and processes that can receive and act on information without human intervention. These algorithms and processes may be trained to converge the model, which means that the model's error rate / prediction rate does not increase with further training (learning) or input.

[0012] Once training is complete, the nodes operate to test and monitor the network infrastructure, communication links, and nodes. In many machine learning systems, the machine learning model may operate on information received from a variety of input sources to perform a specific task, such as image processing and recognition in a facial recognition system or collecting sustainment data in a telecommunications system.

[0013] Nodes represent hundreds of thousands, if not millions, of distributed sensors. These sensors can include test and measurement equipment, antennas, broadcast hubs, signal sensors such as spectrum sensors, and reconfigurable intelligent surfaces (RISs). Different nodes have different capabilities and can result in different trained models, such as non-independent and non-identically distributed (non-IID) models. Some nodes may consist of simple sensors with limited processing elements and memory. More powerful sensors may consist of computing devices (e.g., general-purpose computing devices, servers) or test and measurement equipment (e.g., spectrum analyzers, oscilloscopes, multimeters).

[0014] Furthermore, taken as a whole, the individual trained models comprise a corpus of fundamental component signatures, i.e., nodes (nuclei) that form cells with predetermined objectives, for efficient monitoring of systems of interest. In one embodiment, the systems of interest are 6G communication systems. For measurements of critical signatures necessary for basic operation, some embodiments focus on methods and apparatus for systematically testing, measuring, and maintaining these systems.

[0015] Returning to FIG. 1, each node (e.g., 12) is connected to other nodes (e.g., 14 and 16). The links between nodes shown here are merely representative of the connectivity of the nodes within the network. In some embodiments, a data center or central hub 18 may provide overall control of the network and provide general-purpose machine learning models or algorithms to the nodes within the network, as described in more detail below.

[0016] Each node or device in the network has an AI / ML convergence engine that can receive given information and act on it, ideally through the execution of an ML model. Nodes can request parameter changes for given data based on the state of the communication link over which they reside and return the task to the node that provided it. Nodes can also establish bidirectional links, allowing them to send training data (vectors) over the network and send physical (PHY) layer data up (back to the network) for cause-and-effect correlation and recording the state of the system over time.

[0017] In some embodiments, nodes can acquire or receive data processed by more capable nodes (sensors). In some embodiments, the system uses predictive assessments of traffic loading, along with knowledge of sensor capabilities, routing delays, and security needs, to determine optimal distribution of training vectors across a network, such as the TekCloud® network, allowing correlation of measured physical parameters with distributed AI / ML training data.

[0018] Various nodes may utilize User Datagram Protocol (UDP) delivery methods rather than IP delivery because UDP offers the highest speeds (i.e., multicast) and employs zero-proof knowledge of the link before transmission. In some instances, the link may also employ blockchain verification as an additional layer of security, both to protect the integrity of the data and to allow owner-controlled distribution of the data through fungible tokens, proprietary keys, or a combination of both.

[0019] This network can systematically distribute AI / ML training data to target nodes and use corresponding physical measurements to form a "living" history of the system, which, among other benefits, enables versioning of changes. Systems using the disclosed technology are detectable by the system owner's ability to reconfigure the system to a previous state, eliminating the need to notify end users of outages.

[0020] Some embodiments include the ability to add nodes to the network. As noted above, nodes can consist of any of a wide variety of types of devices. "Adding" a node involves retraining a node whose model has "collapsed." This "collapse" occurs when the model no longer operates within a certain error / prediction accuracy range.

[0021] A node, referred to herein as a "learning node," operates on principles similar to federated learning in machine learning. In federated learning, edge devices (devices at the edge of the network) receive a basic model that may or may not have undergone some training. The edge device then trains a model running on the device using a data set derived from the device's local data. This can result in various devices having slightly different trained models, such as different weights assigned to outcomes. In large networks, this has the advantage of allowing multiple devices across the network to operate in different environments, such as taking advantage of differences in transmission media or having various devices form neighboring nodes.

[0022] The present embodiment provides a basis for messaging between nodes of such a network to perform machine learning related operations. The present embodiment defines a pod or tuple of nodes that are willing to participate in determining a desired ML model resolution for a learning node. Figure 2 illustrates an embodiment of a tuple. The tuple includes a learner node, a neighbor node, and at least one validator node. A neighbor node may consist of a node selected by the learning node from among many other nodes, as described in more detail below.

[0023] Embodiments generally use ML-enabled cells. FIG. 2 illustrates an embodiment of a learning node 20, with the understanding that any other node within the tuple or throughout the network may have the same structure. In FIG. 2, the learning node has some processing element, including a general-purpose processor, a graphics processing unit (GPU), a digital signal processor, a microcontroller, a field-programmable gate array (FPGA), etc. For purposes of this description, a "processor" 30 refers to any device capable of executing programs, including programs as part of a machine learning model. The node also has at least one (and typically multiple) communication link interfaces, such as 34, 36, and 38. These may include wired, wireless (including Wi-Fi, other wireless protocols, near-field communication), optical interfaces, etc. Each node will have at least some form of memory 32. As described in more detail below, each node's memory may maintain model versions and history.

[0024] The following discussion uses the tuples shown in FIG. 2 and the flowchart in FIG. 3. Initially, a learning node 20 is added to a network of nodes at 40 in FIG. 3. As mentioned above, the learning node may represent an existing node that needs to be retrained, essentially starting over. The learning node receives a generic model at 42. In the case of an existing node being retrained, this may include accessing its memory to retrieve the initial model it was trained on. Using the local data as training data, the node is trained to produce a learned model at 44. The node then needs to discover neighboring nodes.

[0025] All nodes can discover each other in one of two ways. A new node may receive information through beacons or other communications over one of the communication interfaces and store it in memory. For example, once a learning node has completed training a generic model using local data, it will discover one or more neighboring nodes by accessing previously received information, sending a request to one or possibly many of these nodes, and then waiting to receive a response. In a second method, the learning node can broadly send a message to all neighboring nodes and then receive the responses. The learning node may select a neighboring node based on either the response or previously received information about the node and the link between the two nodes.

[0026] This information may include an analysis of the available links between the two nodes. Some communication links may be fast enough but not accurate, or slow but accurate. Some communication links may be unavailable, for example, because one of the devices lacks an optical communication channel. Nodes may also analyze the time other nodes estimate it will take to associate with the learning node, possibly taking into account the overall time given to the learning node. All participating nodes (learning node, neighboring nodes, and validator nodes) may share their contribution to the overall time.

[0027] Based on this information, the node selects a node, shown as 22 in Figure 2, as a neighboring node. The two nodes then perform a comparison between their models, including weights and error levels, at 46. If there are no differences or differences that exceed an acceptable threshold, the process ends, the network identifies the node as trained, and the node begins operation at 52. If there are differences that require reconciliation, a third node in the tuple, a validator node, shown as either node 24, 26, or 28 in Figure 2, enters the process at 48. The process of discovering available validator nodes may take the form of the discovery process described above, or information gathered during this process may enable the identification and selection of validator nodes.

[0028] The learning node or neighboring node sends the results of the comparison (differences) to the validator node. The validator node analyzes the differences, adjusts weights, etc. of the trained model on the learning node, and returns these changes to the learning node. The learning node then adjusts the trained model at 50 and initiates operation at 52. In a derivative process, the validator node may also propagate the changes to neighboring nodes. Once operation is initiated, the new node may become available as a neighboring or validator node for new nodes or nodes being retrained.

[0029] In this way, all participating nodes in the process know the ML task times for each node, and the means by which those tasks are completed are governed by the global control parameters given to each node. These may include power consumption, maximum wake time, etc., to name just a few. This is translated into policy parameters for the participating products / devices / nodes to perform the required ML tasks.

[0030] Embodiments of the disclosed technology include methods, attributes, and component interactions that form the minimum architecture necessary to support the integration of ML information in a distributed sensing, wired, or wireless communication environment. Embodiments of the disclosed technology further include using this architecture in conjunction with physical (PHY) layer measurements to establish a metric against which another (third-party) ML engine can calibrate itself.

[0031] Some embodiments include, but are not limited to, use in sounding measurements of calibrated RF channels, a priori determination of optimal beam alignment for 6G applications (resulting in reduced codebook dimensionality), and determining the time-invariant channel (TIV) between 6G users involved in point-to-point communication links. Some embodiments utilize advance forwarding of data to enable better and faster data transmission. Embodiments are generally implemented through calibration of adjacent links to mainline communication protocols, either during or prior to the launch of the 6G link. ML techniques determine this channel and provide results using the architecture described herein.

[0032] Aspects of the disclosure may operate on specially created hardware, firmware, digital signal processors, or specially programmed general-purpose computers, including processors that operate according to programmed instructions. As used herein, the terms controller or processor are intended to encompass microprocessors, microcomputers, application-specific integrated circuits (ASICs), and dedicated hardware controllers. One or more aspects of the disclosure may be embodied in computer-usable data and computer-executable instructions, such as one or more program modules, executed by one or more computers (including monitoring modules) or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types when executed by a processor of a computer or other device. Computer-executable instructions may be stored on non-transitory computer-readable media, such as hard disks, optical disks, removable storage media, solid-state memory, or random access memory (RAM). As will be appreciated by one of ordinary skill in the art, the functionality of the program modules may be combined or distributed in various ways as desired. Furthermore, functionality may be embodied in whole or in part in firmware or hardware equivalents such as integrated circuits, FPGAs, etc. Certain data structures may be used to more effectively implement one or more aspects of the disclosure, and such data structures are contemplated within the scope of computer-executable instructions and computer-usable data described herein.

[0033] Aspects of the disclosed technology may operate on specially created hardware, firmware, digital signal processors, or specially programmed general-purpose computers, including processors that operate according to programmed instructions. The terms "controller" or "processor" herein contemplate microprocessors, microcomputers, ASICs, and dedicated hardware controllers, among others. Aspects of the disclosed technology may be implemented with computer-usable data and computer-executable instructions, such as one or more program modules, executed by one or more computers (including a monitoring module) or other devices. Generally, program modules include routines, programs, objects, components, data structures, etc., which, when executed by a processor in a computer or other device, perform particular tasks or implement particular abstract data types. Computer-executable instructions may be stored in computer-readable storage media, such as hard disks, optical disks, removable storage media, solid-state memory, RAM, etc. Those skilled in the art will appreciate that the functionality of the program modules may be combined or distributed as desired in various embodiments. Furthermore, such functionality may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, field programmable gate arrays (FPGAs), etc. Certain data structures may be used to more effectively implement one or more aspects of the disclosed technology, and such data structures are considered within the scope of the computer-executable instructions and computer-usable data described herein.

[0034] "Computer storage media" means any medium that can be used to store computer-readable information. By way of example and not limitation, computer storage media may include random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory and other memory technologies, compact disc read-only memory (CD-ROM), digital video disc (DVD) and other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage and other magnetic storage devices, and any other volatile or nonvolatile, removable or non-removable medium implemented in any technology. "Computer storage media" excludes signals themselves and transitory forms of signal transmission.

[0035] A communication medium means any medium usable for communicating computer-readable information. By way of example, and not limitation, communication media may include coaxial cable, fiber optic cable, air, or any other medium suitable for communicating electrical, optical, radio frequency (RF), infrared, acoustic, or other types of signals. Example

[0036] The following examples are provided to aid in understanding the technology disclosed in this application. Embodiments of the technology may include one or more of the examples described below, and any combination thereof.

[0037] Example 1 is a machine learning network comprising a plurality of test and measurement devices, one or more of which include one or more communication interfaces configured to enable the test and measurement devices to receive and process physical layer signals, a memory, and one or more processors configured to execute a program that causes the one or more processors to receive physical layer data, perform one or more operations on the physical layer data in accordance with a machine learning model to generate modified physical layer data, and transmit the modified physical layer data to at least one other node in the machine learning neural network.

[0038] Example 2 is the machine learning network of claim 1, wherein the plurality of test and measurement devices include one or more of a test and measurement device, a sensor, an antenna, a reconfigurable intelligent surface, a general-purpose computing device, and a server.

[0039] Example 3 is the machine learning network of either Example 1 or Example 2, wherein the physical layer signal includes a transmission rate, encoding, a transmission medium, and an interface.

[0040] Example 4 is a machine learning network described in any of Examples 1 to 3, wherein the program that causes the one or more processors to perform a process of executing an operation includes a program that causes the one or more processors to perform at least one of a process of determining change parameters for returning data to a node that transmitted the physical layer data, a process of determining beam alignment, a process of canceling interference, and a process of estimating a channel.

[0041] Example 5 is a machine learning network according to any one of Examples 1 to 4, wherein signals within the machine learning network use UDP (User Datagram Protocol) signaling for signals transmitted between nodes.

[0042] Example 6 is a learning node comprising one or more communication interfaces, a memory, and one or more processors, each of which is configured to execute a program that causes the processor to perform the following processes: receive a generic machine learning model via one of the one or more communication interfaces; train the generic machine learning model using data local to the learning node; detect one or more neighboring nodes; communicate with the one or more neighboring nodes to compare the learned (trained) model with those of the neighboring nodes; determine the difference between the learned model and the neighboring models; detect one or more verification nodes; send the difference to the one or more verification nodes; receive input from the one or more verification nodes; and adjust the learned model as needed based on the input to complete the learned model.

[0043] Example 7 is a learning node of Example 6, wherein the one or more processors are further configured to execute a program for receiving information about the one or more neighboring nodes and information about each of the connections to each of the neighboring nodes via one of the one or more communication interfaces before executing a program that causes the one or more processors to perform a process of detecting the one or more neighboring nodes.

[0044] Example 8 is a learning node of Example 7, wherein the program executed by the one or more processors to cause the one or more processors to perform a process of detecting the one or more neighboring nodes includes a program that causes the one or more processors to access the memory and perform a process of obtaining data regarding the one or more neighboring nodes and connections between the learning node and the one or more neighboring nodes.

[0045] Example 9 is a learning node of any of Examples 6 to 8, wherein the program executed by the one or more processors to cause the one or more processors to perform a process of detecting the one or more neighboring nodes includes a program to cause the one or more processors to perform a process of sending a request and receiving at least one response from at least one of the one or more neighboring nodes, wherein the response includes information regarding at least one communication link between the learning node and at least one of the one or more neighboring nodes and information regarding a job completion time for at least one of the one or more neighboring nodes.

[0046] Example 10 is a learning node of any of Examples 6 to 9, wherein the program that causes the one or more processors to execute a program for performing processing to communicate with the one or more neighboring nodes causes the one or more processors to perform processing to communicate with the neighboring nodes based on information regarding the communication link and the job completion time.

[0047] Example 11 is a learning node of Example 10, wherein the information about the communication link includes at least one of a response time, a selected one of the one or more communication interfaces, a required accuracy, a power consumption of the neighboring node, and a maximum startup time of the neighboring node.

[0048] Example 12 is a learning node of any of Examples 6 to 11, wherein the program executed by the one or more processors for the process of transmitting the difference to a verification node includes a program that causes the one or more processors to perform the process of transmitting the difference to one verification node among the one or more verification nodes based on information regarding a communication link between the learning node and the one verification node.

[0049] Example 13 is a learning node of any of Examples 6 to 12, wherein the program that causes the one or more processors to perform a process of adjusting the trained model includes a program that causes the one or more processors to perform a process of adjusting weights in the trained model based on the input.

[0050] Example 14 is a learning node of any of Examples 6 to 13, wherein the one or more processors are further configured to execute a program for performing processing that incurs a maximum time required to complete the trained model.

[0051] Example 15 is the learning node of any of Examples 6 to 14, wherein when the trained model is completed, the learning node becomes at least one of a neighboring node or a validation node.

[0052] Example 16 is a learning node of any of Examples 6 to 15, wherein the one or more processors are further configured to execute a program to store in the memory a history and version of the trained model, a completion time of the one or more neighboring nodes, and the one or more validation nodes.

[0053] Example 17 is a learning node of any of Examples 6 to 16, wherein the one or more processors are further configured to, upon completion of the trained model, join the communication network as a sensor node on which the trained model is running.

[0054] Additionally, the description of this application refers to specific features. It should be understood that the disclosure herein includes all possible combinations of these specific features. When a specific feature is disclosed in connection with a particular aspect or example, that feature can also be used in connection with other aspects and examples, to the extent possible.

[0055] Furthermore, when this application refers to a method having two or more defined steps or processes, these defined steps or processes may be performed in any order or simultaneously, unless the circumstances do not preclude this possibility.

[0056] All features disclosed in the specification, claims, abstract and drawings, and all steps in any disclosed method or process, may be combined in any combination, except where at least some of such features or steps are mutually exclusive combinations. Each feature disclosed in the specification, abstract, claims and drawings may be replaced by an alternative feature serving the same, equivalent or similar purpose, unless expressly stated otherwise.

[0057] Although specific embodiments of the invention have been illustrated and described for purposes of illustration, it will be appreciated that various modifications can be made therein without departing from the spirit and scope of the invention. Accordingly, the invention should not be limited except as by the appended claims.

Claims

1. A machine learning network, a plurality of test and measurement devices; One or more of the plurality of test and measurement devices: one or more communication interfaces configured to enable the test and measurement instrument to receive and process physical layer signals; Memory and one or more processors Equipped with the one or more processors receiving physical layer data; performing one or more operations on the physical layer data in accordance with the machine learning model to generate modified physical layer data; transmitting the modified physical layer data to at least one other node in the machine learning neural network; A machine learning network configured to execute a program that causes the one or more processors to perform the following.

2. 10. The machine learning network of claim 1, wherein the test and measurement device comprises one or more of a test and measurement device, a sensor, an antenna, a reconfigurable intelligent surface, a general-purpose computing device, and a server.

3. The machine learning network of claim 1 , wherein the physical layer signal includes a transmission rate, encoding, a transmission medium, and an interface.

4. The machine learning network of claim 1, wherein the program that causes the one or more processors to perform operations includes a program that causes the one or more processors to perform at least one of: a process for determining change parameters for returning data to a node that transmitted the physical layer data; a process for determining beam alignment; a process for canceling interference; and a process for estimating a channel.

5. The machine learning network of claim 1 , wherein signals within the machine learning network use UDP (User Datagram Protocol) signaling for signals transmitted between nodes.

6. A learning node, one or more communication interfaces; Memory and one or more processors Equipped with Each of the processors: receiving a generic machine learning model via one of the one or more communication interfaces; training the general-purpose machine learning model using data local to the learning node; detecting one or more neighboring nodes; communicating with the one or more neighboring nodes and comparing the trained model with the neighboring nodes; A process of calculating the difference between the trained model and a neighboring model; Discovering one or more validating nodes; sending the difference to the one or more validating nodes; receiving input from the one or more validating nodes; adjusting the trained model as needed based on the input, and completing the trained model; a learning node configured to execute a program that causes the processor to perform the above steps.

7. The learning node of claim 6, wherein the one or more processors are further configured to execute a program that performs a process of receiving information about the one or more neighboring nodes and information about each of the connections to each of the neighboring nodes via one of the one or more communication interfaces before executing a program that causes the one or more processors to perform a process of detecting the one or more neighboring nodes.

8. The learning node of claim 7, wherein the program executed by the one or more processors to cause the one or more processors to perform a process of detecting the one or more neighboring nodes includes a program that causes the one or more processors to perform a process of accessing the memory and obtaining data regarding the one or more neighboring nodes and connections between the learning node and the one or more neighboring nodes.

9. 7. The learning node of claim 6, wherein the program executed by the one or more processors to cause the one or more processors to perform a process of detecting the one or more neighboring nodes includes a program that causes the one or more processors to perform a process of sending a request and receiving at least one response from at least one of the one or more neighboring nodes, the response including information regarding at least one of a communication link between the learning node and at least one of the one or more neighboring nodes and information regarding a job completion time for at least one of the one or more neighboring nodes.

10. The learning node of claim 6, wherein the program for causing the one or more processors to execute a program for processing to communicate with the one or more neighboring nodes causes the one or more processors to perform processing to communicate with the neighboring nodes based on information about the communication link and the job completion time.

11. 11. The learning node of claim 10, wherein the information about the communication link includes at least one of a response time, a selected one of the one or more communication interfaces, a required accuracy, a power consumption of the neighboring node, and a maximum startup time of the neighboring node.

12. The learning node of claim 6, wherein the program executed by the one or more processors for the process of transmitting the difference to the validation node includes a program that causes the one or more processors to perform the process of transmitting the difference to one validation node among the one or more validation nodes based on information regarding a communication link between the learning node and the one validation node.

13. The learning node of claim 6, wherein the program that causes the one or more processors to perform a process of adjusting the trained model includes a program that causes the one or more processors to perform a process of adjusting weights in the trained model based on the input.

14. The learning node of claim 6 , wherein the one or more processors are further configured to execute a program for performing processing that incurs a maximum time required for completion of the trained model.

15. The learning node of claim 6, wherein, upon completion of the trained model, the learning node becomes at least one of a neighboring node or a validation node.

16. 7. The learning node of claim 6, wherein the one or more processors are further configured to execute a program to store in the memory a history and version of the trained model, and completion times of the one or more neighboring nodes and the one or more validation nodes.

17. The learning node of claim 6 , wherein the one or more processors are further configured to, once the trained model is completed, join the communication network as a sensor node on which the trained model is operating.

Images