Self-organizing machine learning training through constrained, predicted traffic load and private end-to-end encryption

Abstract

A machine learning network has a plurality of test and measurement devices, one or more of the test and measurement devices having one or more communication interfaces allowing the devices to receive and process physical layer signals, a memory, and one or more processors configured to execute code, and a processor to cause the one or more processors to receive the physical layer data, perform one or more operations on the physical layer data according to a machine learning model to produce changed physical layer data, and transmit the changed physical layer data to at least one other node in the machine learning neural network. The machine learning network may include learner nodes.

Description

[0001] Cross - Reference to Related Applications

[0002] This disclosure claims the benefit of U.S. Provisional Application 63 / 415,505, filed October 12, 2022, entitled "Self - Organizing Machine Learning Training via Constrained, Predicted Traffic Load, and Private End - to - End Encryption", U.S. Provisional Application 63 / 429,508, filed December 1, 2022, entitled "AI.ML Unit Components for Testing, Measuring, and Maintaining a Network in a Communication Link", and U.S. Non - Provisional Application 18 / 482,801, filed October 6, 2023, entitled "Self - Organizing Machine Learning Training via Constrained, Predicted Traffic Load, and Private End - to - End Encryption", the entire disclosures of which are incorporated herein by reference. Technical Field

[0003] The present invention relates to test and measurement systems, and more particularly to systems for testing, measuring, and maintaining physical layer signals (including but not limited to optical and electromagnetic signals) in a Test, Measurement, and Maintenance (TMS) network. Background Art

[0004] Faster communication standards, the increasing popularity of Wi - Fi, 6G (Sixth Generation), and the Internet of Things (IoT) have provided opportunities to make emerging technologies not only possible but also practical. Distributed computing has become a very popular technology, as has machine learning. However, the increased communication speed and stability of communication links have increased the interest in distributed machine learning.

[0005] Federated learning represents one such architecture. Federated learning involves distributing the learning process to edge devices in a network, where the edge devices train a model using their own local data. Typically, a global model resides on a central server or data center and shares copies of the model with the edge devices. The edge devices train the model on their local data. However, there is a lack of methods for implementing these types of systems.

[0006] A specific area lacking in this type of system is the testing, measuring, and maintenance of physical layer signals, such as the signals that the communication network itself uses to test, measure, and verify the network infrastructure. This is very different from the use of federated learning that relies on the network infrastructure to perform other tasks. Brief Description of the Drawings

[0007] Figure 1 A network diagram of a communication network having multiple nodes is shown.

[0008] Figure 2 A diagram of cell tuples in a communication network is shown.

[0009] Figure 3A flowchart showing an embodiment of a method for adding new nodes to a communication network using machine learning. Detailed Description

[0010] Embodiments herein use a network of nodes for testing and measurement, including testing and measurement of the network infrastructure itself. The nodes in the network operate as nodes of a machine learning neural network and operate on physical layer (PHY) signals by applying artificial intelligence (AI) / machine learning (ML) input vectors. The term "physical layer data" as used herein includes physical layer signals and data that may be contained in these signals.

[0011] Figure 1 An example of a communication network 10 with multiple test and measurement devices as nodes (e.g., 12) is shown. Each node includes a processing unit that can be trained to a machine learning artificial intelligence model. This discussion uses the terms "artificial intelligence" and "machine learning" to refer to algorithms and processes that can receive information and act on it without human intervention. These algorithms and processes can be trained to make the model converge, which means that no further training or input will increase the error rate / prediction percentage of the model.

[0012] Once trained, the nodes will test and monitor the network infrastructure, communication links, and nodes. In many machine learning systems, the machine learning model can operate on information received from various input sources to perform specific tasks, such as image processing and recognition in a facial recognition system, or collection of maintenance data in a telecommunications system.

[0013] These nodes represent hundreds of thousands or even millions of distributed sensors. These sensors may include test and measurement instruments, antennas, broadcast hubs, signal sensors such as spectral sensors, reconfigurable intelligent surfaces (RIS), etc. Different nodes may have different capabilities, resulting in different training models, such as non-independent and non-identically distributed models (non-iid). Some nodes may include simple sensors with processing elements and memory of limited capacity. More powerful sensors may include computing devices such as general-purpose computing devices, servers, and test and measurement instruments such as spectrum analyzers, oscilloscopes, multimeters, etc.

[0014] Furthermore, generally speaking, each individual training model forms a basic component feature, that is, a node (core) that forms a unit with a defined target and a predetermined purpose, which includes a corpus for effectively monitoring the target system. In one embodiment, the target system includes a 6G telecommunications system. In the case of operating on crucial flags necessary for basic operations, the embodiments are dedicated to methods and apparatuses for systematically testing, measuring, and maintaining these systems.

[0015] Back toFigure 1 Each node, such as 12, is connected to other nodes, such as 14 and 16. The links between the nodes shown here only include representations of the connections of the nodes in the network. In some embodiments, the data center or central hub 18 provides overall network control and can provide a general machine learning model or algorithm for the nodes in the network, as will be discussed in more detail below.

[0016] Each node or device in the network includes an AI / ML convergence engine that is capable of receiving prescribed information and acting on the information, ideally by executing an ML model. The node can determine change parameters for the prescribed data based on the state of the communication link on which the node resides, for passing back to the node that passed the task. The node can also establish a bi-directional link that allows the transmission of learning data (vectors) through the network and the transmission of PHY layer data up (back to the network) to enable causality and mark the system state over time.

[0017] In some embodiments, a node can acquire or receive data processed by a more capable node (sensor). In some embodiments, the system uses a predictive assessment of traffic load, as well as knowledge of sensor capabilities, routing latency, and security requirements, to determine the optimal distribution of learning vectors (training data) over the network (such as the network), thus allowing the association of measured physical parameters with the passed AI / ML learning (training) data.

[0018] Various nodes can employ UDP (User Datagram Protocol) routing methods, avoiding IP routing as it provides the best speed (i.e., multicast) and takes zero-proof knowledge of the link before transmission. In some cases, the link can also employ blockchain verification via replaceable tokens, proprietary keys, or a combination of both as an additional security layer for data integrity protection and distribution control by the data owner.

[0019] The network can use the organized passing of AI / ML learning (training) data to target nodes and their corresponding physical measurements to form a “living” history of the system, thus allowing advantages such as version control of changes. A system using the disclosed technology will be detectable by the system owner's ability to rebuild the system to a previous state without requiring communication interruption to the end user.

[0020] Some embodiments relate to adding nodes to the network. As described above, a node can include one of many different types of devices. “Adding” a node includes retraining a node in which the model has “collapsed,” where the model no longer operates within a given prediction error / accuracy range.

[0021] This node is referred to here as a "learning node" and operates in a manner similar to federated learning in machine learning. In federated learning, the edge devices, i.e., those at the edge of the network, receive a base model that may or may not have been trained to some extent. The edge devices then use a dataset derived from the device-local data to train the model operating on the device. This can result in slightly different trained models on different devices, such as weights assigned to different outcomes. In a large network, this has the advantage that the devices on the network operate in different environments, including differences in available transmission media, different devices making up nearby nodes, etc.

[0022] Embodiments relate to the basis for message passing between nodes of such a network to perform actions related to machine learning. Embodiments define a group of pods or tuples of nodes that are willing to participate in resolving the ML model desired by the learner node. Figure 2 An embodiment of the tuple is shown. The tuple will include a learner node, neighbor nodes, and at least one validator node. The neighbor nodes can include one of many nodes selected by the learner, as discussed in more detail below.

[0023] Embodiments generally use ML-enabled units. Figure 2 An embodiment of the learner node 20 is shown, it being understood that any other node in the tuple or on the network can have the same structure. In Figure 2 , the learner node has some processing element. For example, this may include a general-purpose processor, a graphics processing unit, a digital signal processor, a microcontroller, a field-programmable gate array (FPGA). This discussion will refer to any device capable of executing code (including code as part of a machine learning model) as a "processor" 30. The node also has at least one but typically several communication link interfaces, such as 34, 36, and 38. These may include wired, wireless (including wi-fi), other radio protocols, and near-field communication, as well as optical interfaces. Each node will have at least some form of memory 32. As discussed in more detail below, the memory of each node can maintain model versions and histories.

[0024] The following discussion uses Figure 2 the tuple shown in Figure 3 and the flowchart of Figure 3 . Initially, at 40 in , the learner node 20 is added to the node network. As described above, the learner node can represent an existing node that needs to be retrained, essentially starting over on its own. The learner node receives a general model at 42. For an existing node being retrained, this may involve accessing its memory to retrieve its initially trained model. Using local data as training data, the node trains at 44 to produce a trained model. Then, the node needs to discover its neighbors.

[0025] All nodes can discover each other in one of two ways. A new node can receive information either via a beacon or via other communication over one of the communication interfaces, and can then save it in its memory. For example, when a learner node finishes training a general model with its local data, the learner node will discover neighbor nodes by accessing the previously received information and sending requests to one or possibly many of those nodes, and then wait to receive a reply. In the second way, the learner node can send a broadcast message to all neighbor nodes and then receive replies. The learner node can select neighbors based on the replies or previously received information, based on the information received about the node and the link between two nodes.

[0026] This information can include an analysis of the available links between two nodes. Some communication links may have good speed but low accuracy, or low speed and high accuracy. Some communication links may not be available; for example, a device may not have an optical communication channel. The node can also analyze the amount of time that other nodes estimate is required for another node to work with the learner node, possibly based on the total time given to the learner node. All participants, the learner node, neighbor nodes, and validator nodes, can share their contributions to the total time.

[0027] The node selects a node as Figure 2 the neighbor node 22 as shown. Then, the two nodes perform a comparison between their models (such as weights, error levels, etc.) at 46. If there are no differences, or no differences beyond the difference threshold tolerance, the process can end, and the network will identify the node as trained, and the node enters operation at 52. If there are indeed differences that need to be coordinated, then the third node in the tuple, i.e., Figure 2 the validator node shown as one of nodes 24, 26, or 28 in enters the process at 48. The discovery process for available validator nodes can take the form of the previous discovery process, or the information collected during the process can allow for the identification and selection of validator nodes.

[0028] The learner node or neighbor node sends the comparison result, the differences, to the validator node. The validator node analyzes the differences and can adjust the weights, etc., of the trained model on the learner node, and returns these changes to the learner node. Then, the learner node adjusts its trained model at 50 and enters operation at 52. In a secondary process, the validator can also communicate the changes to neighbor nodes. Once in operation, the new node can become available as a neighbor node or as a validator for new nodes or those nodes undergoing retraining.

[0029] In this way, all participants in the process know the machine learning task times of each node, and the way they complete the tasks will be determined by the overall control parameters provided to each node. These may include power consumption, maximum wake-up time, etc., just to name a few. This will translate into policy parameters for the products / devices / nodes participating in the execution of the required machine learning tasks.

[0030] Embodiments of the present disclosure include the methods, attributes, and interactions between components that form the minimum architecture required to support ML joint intelligence in a distributed sensing, wired or wireless communication environment. Embodiments of the present disclosure also include using this architecture in combination with PHY layer measurements to set metrology standards that other third-party ML engines can adjust themselves to.

[0031] Some exemplary non-limiting embodiments include using calibrated RF channel sounding measurements or pre-determining the optimal beam alignment, thereby reducing the codebook size for 6G use cases and determining the time-invariant channel (TIV) between 6G users participating in a point-to-point communication link. Embodiments pre-use data transfer to allow for better and faster data transmission. Embodiments are generally achieved by calibrating the link adjacent to the main communication protocol during or before the start of the 6G link. ML techniques will determine this channel by using the architecture discussed herein to determine the results.

[0032] Aspects of the present disclosure may operate on specially created hardware, firmware, a digital signal processor, or a specially programmed general-purpose computer including a processor operating according to programming instructions. The terms controller or processor as used herein are intended to include a microprocessor, a microcomputer, an application-specific integrated circuit (ASIC), and a dedicated hardware controller. One or more aspects of the present disclosure may be embodied in computer-usable data and computer-executable instructions, such as embodied in 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. that perform specific tasks or implement specific abstract data types when executed by a processor in a computer or other device. The computer-executable instructions may be stored on a non-transitory computer-readable medium, such as a hard disk, an optical disk, a removable storage medium, a solid-state memory, a random access memory (RAM), etc. As will be understood by those skilled in the art, the functions of the program modules may be combined or distributed as needed in various aspects. In addition, the functions may be embodied in whole or in part in firmware or hardware equivalents, such as integrated circuits, FPGAs, etc. Specific data structures may be used to more effectively implement one or more aspects of the present disclosure, and these data structures are contemplated within the scope of the computer-executable instructions and computer-usable data described herein.

[0033] In some cases, the disclosed aspects may be implemented in hardware, firmware, software, or any combination thereof. The disclosed aspects may also be implemented as instructions carried or stored on one or more non-transitory computer-readable media and readable and executable by one or more processors. Such instructions may be referred to as a computer program product. As used herein, a computer-readable medium refers to any medium that can be accessed by a computing device. By way of example and not limitation, computer-readable media may include computer storage media and communication media.

[0034] Computer storage media refers to 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 or other storage technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, and any other volatile or non-volatile, removable or non-removable media implemented in any technology. Computer storage media does not include signals per se and transient forms of signal transmission.

[0035] Communication media refers to any medium that can be used 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.

[0036] Example

[0037] Illustrative examples of the disclosed technology are provided below. Embodiments of these technologies may include one or more of the examples described below and any combination of these examples.

[0038] Example 1 is a machine learning network, comprising: a plurality of test and measurement devices; one or more of the test and measurement devices comprising: one or more communication interfaces configured to allow the device to receive and process physical layer signals; a memory; and one or more processors configured to execute code to cause the one or more processors to receive physical layer data; perform one or more operations on the physical layer data according to a machine learning model to produce altered physical layer data; and transmit the altered physical layer data to at least one other node in the machine learning neural network.

[0039] Example 2 is the machine learning network according to claim 1, wherein the test and measurement device comprises one or more of a test and measurement instrument, a sensor, an antenna, a reconfigurable intelligent surface, a general computing device, and a server.

[0040] Example 3 is a machine learning network of any one of Examples 1 or 2, wherein the physical layer signal includes transmission rate, encoding, transmission medium, and interface.

[0041] Example 4 is a machine learning network of any one of Examples 1 to 3, wherein the code that causes one or more processors to perform operations includes code that causes one or more processors to perform at least one of determining a change parameter for returning data to a node that sends physical layer data, determining beam alignment, eliminating interference, and channel estimation.

[0042] Example 5 is a machine learning network of any one of Examples 1 to 4, wherein the signals in the network use User Datagram Protocol (UDP) signaling for signals sent between nodes.

[0043] Example 6 is a learner node, including: one or more communication interfaces; a memory; and one or more processors, each processor being configured to execute code to cause the processor to: receive a general machine learning model through one of the one or more communication interfaces; train the model using data local to the learning node; discover one or more neighbor nodes; communicate with the one or more neighbor nodes to compare the trained model with the neighbor nodes; determine the differences between the trained model and the neighbor models; discover one or more validator nodes; send the differences to the one or more validator nodes; receive inputs from the one or more validator nodes; and adjust the trained model as needed based on the inputs to complete the trained model.

[0044] Example 7 is the learner node of Example 6, wherein the one or more processors are further configured to, before executing code to cause the one or more processors to discover one or more neighbor nodes, execute code to receive information about one or more neighbor nodes and information about each connection to each neighbor node through one of the one or more communication interfaces.

[0045] Example 8 is the learner node of Example 7, wherein the code that causes one or more processors to discover one or more neighbor nodes includes code that causes one or more processors accessing the memory to retrieve data about the one or more neighbor nodes and the connections between the learning node and the one or more neighbor nodes.

[0046] Example 9 is a learner node of any one of Examples 6 to 8, wherein the code executed by one or more processors to cause the one or more processors to discover one or more neighbor nodes includes code to cause the one or more processors to issue a request and receive at least one response from at least one of the one or more neighbor nodes, the response including information about a communication link between the learner node and the at least one of the one or more neighbor nodes and at least one of the job completion times of the at least one of the one or more neighbor nodes.

[0047] Example 10 is a learner node of any one of Examples 6 to 9, wherein the code that causes one or more processors to execute code to communicate with one or more neighbor nodes causes the one or more processors to communicate with the neighbor nodes based on information about the communication link and job completion time.

[0048] Example 11 is a learner node of Example 10, wherein the information about the communication link includes at least one of a response time amount, a selected one of one or more communication interfaces, a required accuracy, a power consumption of the neighbor node, and a maximum wake-up time of the neighbor node.

[0049] Example 12 is a learner node of any one of Examples 6 to 11, wherein the code executed by one or more processors to send a difference to a validator node includes code to cause the one or more processors to send the difference to one of the one or more validator nodes based on information about a communication link between the learning node and a validator node.

[0050] Example 13 is a learner node of any one of Examples 6 to 12, wherein the code that causes one or more processors to adjust the trained model includes code to cause the one or more processors to adjust the weights in the trained model based on an input.

[0051] Example 14 is a learner node of any one of Examples 6 to 13, wherein one or more processors are further configured to execute code to receive a maximum time to complete the training model.

[0052] Example 15 is a learner node of any one of Examples 6 to 14, wherein after the trained model is completed, the learning node becomes at least one of a neighbor node or a validator node.

[0053] Example 16 is a learner node of any one of Examples 6 to 15, wherein one or more processors are further configured to execute code to store in a memory one or more training model histories and versions, completion times of one or more neighbor nodes, and one or more validator nodes.

[0054] Example 17 is a learner node of any one of Examples 6 to 16, wherein one or more processors are further configured to participate in a communication network as a sensor node that operates the trained model after the trained model is completed.

[0055] In addition, this written description also refers to specific features. It will be understood that the disclosure in this specification includes all possible combinations of these specific features. Where a specific feature is disclosed in the context of a particular aspect or example, that feature can also be used, to the extent possible, in the context of other aspects and examples.

[0056] In addition, when a method having two or more defined steps or operations is mentioned in this application, the defined steps or operations can be performed in any order or simultaneously, unless the context excludes these possibilities.

[0057] All features disclosed in the specification, including the claims, the abstract and the drawings, and all steps in any method or process disclosed, can be combined in any combination, except combinations where at least some of such features and / or steps are mutually exclusive. Unless otherwise expressly stated, each feature disclosed in the specification, including the claims, the abstract and the drawings, can be replaced by an alternative feature for the same, equivalent or similar purpose.

[0058] Although specific examples of the invention have been shown and described for purposes of illustration, it should be understood that various modifications can be made 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, comprising: a plurality of test and measurement devices; one or more of the test and measurement devices comprising: one or more communication interfaces configured to allow the device to receive and process physical layer signals; a memory; and one or more processors configured to execute code to cause the one or more processors to receive physical layer data; perform one or more operations on the physical layer data according to a machine learning model to produce altered physical layer data; and transmit the altered physical layer data to at least one other node in the machine learning neural network.

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

3. The machine learning network according to claim 1, wherein the physical layer signals include transmission rate, coding, transmission medium, and interface.

4. The machine learning network according to claim 1, wherein the code causing the one or more processors to perform operations includes code causing the one or more processors to perform at least one of determining altered parameters for returning data to the node that sent the physical layer data, determining beam alignment, eliminating interference, and channel estimation.

5. The machine learning network according to claim 1, wherein signals in the network use User Datagram Protocol (UDP) signaling for signals sent between nodes.

6. A learner node, comprising: one or more communication interfaces; a memory; and one or more processors, each processor configured to execute code to cause the processor: receive a general machine learning model through one of the one or more communication interfaces; train the model using data local to the learning node; discover one or more neighbor nodes; communicate with the one or more neighbor nodes to compare the trained model with the neighbor nodes; determine the differences between the trained model and the neighbor models; discover one or more validator nodes; send the differences to the one or more validator nodes; receive input from the one or more validator nodes; and adjust the trained model as needed based on the input to complete the trained model.

7. The learner node according to claim 6, wherein the one or more processors are further configured to, before executing code to cause the one or more processors to discover the one or more neighbor nodes, execute code to receive information about one or more neighbor nodes and information about each connection to each neighbor node through one of the one or more communication interfaces.

8. The learner node according to claim 7, wherein the code executed by the one or more processors to cause the one or more processors to discover the one or more neighbor nodes includes code causing the one or more processors accessing the memory to retrieve data about the one or more neighbor nodes and the connections between the learner node and the one or more neighbor nodes.

9. The learner node according to claim 6, wherein the code executed by the one or more processors to cause the one or more processors to discover the one or more neighbor nodes includes code to cause the one or more processors to issue a request and receive at least one response from at least one of the one or more neighbor nodes, the response including information about a communication link between the learner node and at least one of the one or more neighbor nodes, and at least one of the job completion times of at least one of the one or more adjacent nodes.

10. The learner node according to claim 6, wherein the code that causes the one or more processors to execute code to communicate with the one or more neighbor nodes causes the one or more processors to communicate with the neighbor nodes based on information about the communication link and the job completion time.

11. The learner node according to claim 10, wherein the information about the communication link includes at least one of an amount of response time, a selected one of the one or more communication interfaces, a required accuracy, a power consumption of the neighbor node, and a maximum wake-up time of the neighbor node.

12. The learner node according to claim 6, wherein the code executed by the one or more processors to send the difference to a validator node includes code to cause the one or more processors to send the difference to one of the one or more validator nodes based on information about a communication link between the learning node and one of the validator nodes.

13. The learner node according to claim 6, wherein the code that causes the one or more processors to adjust the trained model includes code to cause the one or more processors to adjust weights in the trained model based on the input.

14. The learner node according to claim 6, wherein the one or more processors are further configured to execute code to receive a maximum time to complete the trained model.

15. The learner node according to claim 6, wherein after the trained model is completed, the learning node becomes at least one of a neighbor node or a validator node.

16. The learner node according to claim 6, wherein the one or more processors are further configured to execute code to store in the memory one or more trained model histories and versions, and the completion times of the one or more neighbor nodes and the one or more validator nodes.

17. The learner node according to claim 6, wherein the one or more processors are further configured to participate in a communication network as a sensor node operating the trained model after the trained model is completed.

Images