Providing machine learning-based assistance to producer nodes

Abstract

According to one aspect, a method is disclosed, the method comprising receiving, from a producer node, producer node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity. According to another aspect, a method is disclosed, the method comprising identifying producer node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; and causing transmission of the producer node capabilities to a consumer node.

Description

Technical Field

[0001] The present application generally relates to the field of wireless communications. In particular, the present application relates to consumer nodes and producer nodes for wireless communications, and related methods and computer programs. Background Art

[0002] In a wireless communication network, machine learning algorithms may be executed by consumer nodes and / or producer nodes, for example, on the Radio Access Network (RAN) side (e.g., at a network node or base station), and on the user node side (e.g., at a user equipment (UE)). For example, a user equipment may have the functionality required to provide machine learning based assistance to the RAN. For example, these functions may include prediction (forecast) of certain events, such as handover (HO), cross reference symbol received power (RSRP) threshold, quality of service (QoS) change, mobility state change, etc. This forecast information then needs to be reported back to the serving gNB(s) to be used as input to the radio resource management (RRM) algorithms and RRM actions.

[0003] In general, a solution is needed to efficiently configure machine learning-based assistance while keeping the signaling overhead between consumer and producer nodes low. Summary of the invention

[0004] An example embodiment of a method includes receiving producer node capabilities including machine learning-based auxiliary capabilities from a producer node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity.

[0005] In an example embodiment, at least one machine learning mode associated with a machine learning entity includes at least one fallback operating mode.

[0006] In an example embodiment, the method further includes selecting at least one machine learning-based function among the machine learning-based functions; and causing transmission of an activation request to the producer node for activating the selected at least one machine learning-based function.

[0007] In an example embodiment, the method further includes receiving an activation request response from the producer node, the activation request response including an indication as to whether at least one machine learning-based functionality is activated.

[0008] In an example embodiment, the method further includes receiving, from a producer node, an activation request for activating at least one of the machine learning based functions.

[0009] In an example embodiment, the method further includes causing transmission of a machine learning based function activation response to the producer node.

[0010] In an example embodiment, the method further includes receiving, from a producer node, a change request to change at least one of the machine learning based functions.

[0011] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0012] In an example embodiment, the method further includes maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time when the change request includes a deactivation indication.

[0013] In one example embodiment, the method further includes causing transmission of a request to enable a deactivated reasoning report associated with the deactivated machine learning to the producer node.

[0014] In an example embodiment, the method further includes causing transmission of a response to the producer node indicating confirmation of the changed machine learning-based functionality.

[0015] In an example embodiment, the method further includes causing transmission of a change request to change at least one of the machine learning based functions to the producer node.

[0016] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0017] In an example embodiment, the method further comprises causing transmission of an inference report reconfiguration message to the producer node when the change request comprises a deactivation indication or a switch indication.

[0018] In an example embodiment, the method further includes maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time when the change request includes a deactivation indication.

[0019] In one example embodiment, the method further includes causing transmission of a request to enable a deactivated reasoning report associated with the deactivated machine learning to the producer node.

[0020] In an example embodiment, the method further includes receiving a response to the change request from the producer node.

[0021] An example embodiment of a method includes identifying producer node capabilities that include machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; and causing transmission of the producer node capabilities to a consumer node.

[0022] In an example embodiment, at least one machine learning mode associated with a machine learning entity includes at least one fallback operating mode.

[0023] In an example embodiment, the method further includes receiving, from a consumer node, an activation request for activating at least one of the machine learning based functions.

[0024] In an example embodiment, the method further includes causing transmission of an activation request response to the consumer node, the activation request response including an indication as to whether the at least one machine learning-based functionality is activated.

[0025] In an example embodiment, the method further includes causing transmission of an activation request to activate at least one of the machine learning based functions to a consumer node.

[0026] In an example embodiment, the method further includes receiving a machine learning based function activation response from the consumer node.

[0027] In an example embodiment, the method further includes causing transmission of a change request to change at least one of the machine learning based functions to a consumer node.

[0028] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0029] In an example embodiment, the method further includes maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time when the change request includes a deactivation indication.

[0030] In an example embodiment, the method further includes receiving a request from a consumer node to enable deactivated reasoning reporting associated with the deactivated machine learning.

[0031] In an example embodiment, the method further includes receiving a response from the consumer node indicating confirmation of the changed machine learning-based functionality.

[0032] In an example embodiment, the method further includes receiving, from a consumer node to a consumer node, a change request for changing at least one of the machine learning-based functions; and changing at least one of the machine learning-based functions based on the change request.

[0033] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0034] In an example embodiment, the method further includes receiving a reasoning report reconfiguration message from the consumer node when the change request includes a deactivation indication or a switch indication.

[0035] In an example embodiment, the method further includes maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time when the change request includes a deactivation indication.

[0036] In an example embodiment, the method further includes receiving a request from a consumer node to enable deactivated reasoning reporting associated with the deactivated machine learning.

[0037] In an example embodiment, the method further includes causing transmission of a reply to the change request to the consumer node.

[0038] An example embodiment of a consumer node includes at least one processor; and at least one memory including computer program code. The at least one memory and the computer program code are configured to, with the at least one processor, cause the consumer node to at least perform: receiving producer node capabilities including machine learning-based auxiliary capabilities from a producer node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity.

[0039] In an example embodiment, at least one machine learning mode associated with the machine learning entity includes at least one fallback operating mode.

[0040] In an example embodiment, at least one memory and computer program code are configured to, together with at least one processor, cause a producer node to at least perform: selecting at least one machine learning-based function among the machine learning-based functions; and causing transmission of an activation request to the producer node to activate the selected at least one machine learning-based function.

[0041] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving an activation request response from the producer node, the activation request response including an indication as to whether at least one machine learning-based functionality is activated.

[0042] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving an activation request from a producer node to activate at least one of the machine learning based functions.

[0043] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission of a machine learning based function activation response to the producer node.

[0044] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving a change request from a producer node to change at least one of the machine learning based functions.

[0045] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0046] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning-based functionality for a predetermined time period.

[0047] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission to the producer node of a request to enable a deactivation reasoning report associated with deactivated machine learning.

[0048] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission of a response to the producer node indicating confirmation of the changed machine learning-based functionality.

[0049] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission of a change request to the producer node for changing at least one of the machine learning-based functions.

[0050] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0051] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause the producer node to at least perform: when the change request includes a deactivation indication or a switch indication, causing transmission of an inference report reconfiguration message to the producer node.

[0052] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning-based functionality for a predetermined time period.

[0053] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission to the producer node of a request to enable a deactivation reasoning report associated with deactivated machine learning.

[0054] In an example embodiment, the at least one memory and the computer program code are configured to, with the at least one processor, cause the producer node to at least perform: receiving a reply to the change request from the producer node.

[0055] An example embodiment of a producer node includes at least one processor; and at least one memory including computer program code. The at least one memory and the computer program code are configured to, with the at least one processor, cause a consumer node to at least perform: identifying producer node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; and causing transmission of the producer node capabilities to a consumer node.

[0056] In an example embodiment, at least one machine learning mode associated with a machine learning entity includes at least one fallback operating mode.

[0057] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving an activation request from a consumer node to activate at least one of the machine learning based functions.

[0058] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: cause transmission of an activation request response to a consumer node, the activation request response including an indication as to whether at least one machine learning-based functionality is activated.

[0059] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission of an activation request to activate at least one of the machine learning-based functions to a consumer node.

[0060] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving a machine learning based function activation response from a consumer node.

[0061] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: causing transmission of a change request to change at least one of the machine learning-based functions to a consumer node.

[0062] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0063] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning-based functionality for a predetermined time period.

[0064] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving a request from a consumer node to enable deactivated reasoning reporting associated with deactivated machine learning.

[0065] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause the producer node to at least perform: receiving a response from the consumer node indicating a confirmation of the changed machine learning-based functionality.

[0066] In an example embodiment, at least one memory and computer program code are configured to, together with at least one processor, cause a producer node to at least perform: receiving a change request from a consumer node to the consumer node for changing at least one of the machine learning-based functions; and changing at least one of the machine learning-based functions based on the change request.

[0067] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0068] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: when the change request includes a deactivation indication or a switch indication, receive an inference report reconfiguration message from a consumer node.

[0069] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning-based functionality for a predetermined time period.

[0070] In an example embodiment, at least one memory and computer program code are configured to, with at least one processor, cause a producer node to at least perform: receiving a request from a consumer node to enable deactivated reasoning reporting associated with deactivated machine learning.

[0071] In an example embodiment, the at least one memory and the computer program code are configured to, with the at least one processor, cause the producer node to at least perform: causing transmission of a reply to the change request to the consumer node.

[0072] An example embodiment of a consumer node includes components for performing the following operations: receiving producer node capabilities including machine learning-based auxiliary capabilities from a producer node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity.

[0073] In an example embodiment, at least one machine learning mode associated with a machine learning entity includes at least one fallback operating mode.

[0074] In an example embodiment, the component is further configured to perform: selecting at least one machine learning-based function among the machine learning-based functions; and causing transmission of an activation request for activating the selected at least one machine learning-based function to the producer node.

[0075] In an example embodiment, the component is further configured to perform: receiving an activation request response from the producer node, the activation request response including an indication of whether at least one machine learning-based functionality is activated.

[0076] In an example embodiment, the component is further configured to perform: receiving an activation request from a producer node to activate at least one of the machine learning based functions.

[0077] In an example embodiment, the component is further configured to perform: causing transmission of a machine learning-based function activation response to the producer node.

[0078] In an example embodiment, the component is further configured to perform: receiving a change request from a producer node to change at least one of the machine learning based functions.

[0079] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0080] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

[0081] In an example embodiment, the component is further configured to perform: causing transmission of a request to enable deactivated reasoning reporting associated with the deactivated machine learning to the producer node.

[0082] In an example embodiment, the component is further configured to perform: causing transmission of a response to the producer node indicating confirmation of the changed machine learning-based functionality.

[0083] In an example embodiment, the component is further configured to perform: causing transmission of a change request for changing at least one of the machine learning based functions to the producer node.

[0084] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0085] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication or a switch indication, causing transmission of an inference report reconfiguration message to the producer node.

[0086] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

[0087] In an example embodiment, the component is further configured to perform: causing transmission of a request to enable deactivated reasoning reporting associated with the deactivated machine learning to the producer node.

[0088] In an example embodiment, the component is further configured to perform: receiving a reply to the change request from the producer node.

[0089] In an example embodiment, the component includes at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause execution of a consumer node.

[0090] An example embodiment of a producer node includes components for performing the following operations: identifying producer node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; and causing transmission of the producer node capabilities to a consumer node.

[0091] In an example embodiment, at least one machine learning mode associated with a machine learning entity includes at least one fallback operating mode.

[0092] In an example embodiment, the component is further configured to perform: receiving an activation request from a consumer node to activate at least one of the machine learning based functions.

[0093] In an example embodiment, the component is further configured to perform: causing transmission of an activation request response to the consumer node, the activation request response including an indication of whether at least one machine learning-based functionality is activated.

[0094] In an example embodiment, the component is further configured to perform: causing transmission of an activation request to activate at least one of the machine learning-based functions to a consumer node.

[0095] In an example embodiment, the component is further configured to perform: receiving a machine learning based function activation response from a consumer node.

[0096] In an example embodiment, the component is further configured to perform: causing transmission of a change request to change at least one of the machine learning-based functions to a consumer node.

[0097] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0098] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

[0099] In an example embodiment, the component is further configured to perform: receiving a request from a consumer node to enable deactivated reasoning reporting associated with the deactivated machine learning.

[0100] In an example embodiment, the component is further configured to perform: receiving a response from the consumer node indicating a confirmation of the changed machine learning-based functionality.

[0101] In an example embodiment, the component is further configured to perform: receiving a change request from a consumer node to a consumer node for changing at least one of the machine learning-based functions; and changing at least one of the machine learning-based functions based on the change request.

[0102] In an example embodiment, the change request includes an indication of a reason for the change, the reason indication including one of: a deactivation indication for deactivating the machine learning-based functionality; a switching indication for switching to another machine learning-based functionality; a pause indication for pausing the machine learning-based functionality for a set time period; and a reset indication for resetting or restarting the machine learning-based functionality.

[0103] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication or a switching indication, receiving an inference report reconfiguration message from the consumer node.

[0104] In an example embodiment, the component is further configured to perform: when the change request includes a deactivation indication, maintaining a state of a deactivation reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

[0105] In an example embodiment, the component is further configured to perform: receiving a request from a consumer node to enable deactivated reasoning reporting associated with the deactivated machine learning.

[0106] In an example embodiment, the component is further configured to perform: causing transmission of a reply to the change request to the consumer node.

[0107] In an example embodiment, the component includes at least one processor and at least one memory including computer program code, the at least one memory and the computer program code configured to, with the at least one processor, cause execution of a producer node.

[0108] An example embodiment of a computer program comprises instructions for causing an apparatus to perform the method of any of the above example embodiments.

[0109] An example embodiment of a computer readable medium includes program instructions for causing an apparatus to perform the method of any of the above-described example embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0110] The accompanying drawings are included to provide a further understanding of the embodiments and constitute a part of this specification, which illustrate the embodiments and together with the description help explain the principles of the embodiments. In the drawings:

[0111] Figure 1 An example embodiment of the subject matter described herein is shown, showing a consumer node.

[0112] Figure 2 An example embodiment of the subject matter described herein is shown, showing a producer node.

[0113] Figure 3A An example embodiment of the structure of a producer node's machine learning-based auxiliary capabilities is shown.

[0114] Figure 3B An example embodiment of a machine learning functional architecture is shown.

[0115] Figure 4A An example embodiment of signaling is shown in which a consumer node activates specific machine learning-based functionality in a producer node.

[0116] Figure 4B An example embodiment of signaling is shown in which a producer node requests activation of a machine learning based functionality.

[0117] Figure 4C An example embodiment of signaling is shown in which a producer node requests to deactivate a currently used machine learning based functionality.

[0118] Figure 4D An example embodiment of signaling is shown in which a consumer node requests to deactivate a currently used machine learning based functionality in a producer node.

[0119] Figure 4E An example embodiment of signaling is shown in which a consumer node uses a pause command to activate a currently used machine learning-based functionality in a producer node.

[0120] In the drawings, the same reference numerals are used to denote the same parts. DETAILED DESCRIPTION

[0121] Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. The detailed description provided below in conjunction with the accompanying drawings is intended as a description of this example and is not intended to represent the only form in which this example can be constructed or utilized. This description sets forth the functions of the example and the order of steps for constructing and operating the example. However, the same or equivalent functions and sequences may be accomplished by different examples.

[0122] As discussed herein, the terms "machine learning-based assistance for producer nodes", "machine learning-based assistance for user nodes", or "machine learning-based assistance" are broadly used to refer to functionality at a producer node (e.g., a user node or any other network node) for providing machine learning (ML)-based assistance associated with a network (e.g., a radio access network, cognitive radio, and software defined radio solutions). Even though the example embodiments discussed herein may relate to scenarios in which a user node has the functionality required to provide ML-based assistance to a radio access network, and in which these functionality may include, for example, prediction (forecast) of certain events (such as handover (HO), cross-reference symbol received power (RSRP) thresholds, quality of service changes, mobility state changes, etc.), these are only non-limiting examples. In addition, the employed ML algorithm for providing / generating machine learning-based assistance information for user nodes (e.g., RSRP values, CSI, prediction of HO events) may have been optimized (trained and tested) under a comprehensive set of operating modes (input data) and may have the ability to deliver at least one of typical ML / deep learning (DL) / artificial intelligence (AI) algorithm performance metrics.

[0123] Furthermore, as used herein, the labels “producer” and “consumer” may relate to machine learning prediction / inference.

[0124] Figure 1 Illustrative of an example embodiment of the subject matter described herein, a consumer node 100 is shown.

[0125] The consumer node 100 includes one or more processors 102 and one or more memories 104 with computer program code. The consumer node 100 may also include a transceiver 106 and other elements such as an input / output module ( Figure 1 not shown) and / or communication interface ( Figure 1 not shown).

[0126] Although consumer node 100 is depicted as including only one processor 102, consumer node 100 may include more than one processor. In an example embodiment, memory 104 is capable of storing instructions, such as an operating system and / or various applications.

[0127] In addition, the processor 102 is capable of executing the stored instructions. In one embodiment, the processor 102 may be embodied as a multi-core processor, a single-core processor, or a combination of one or more multi-core processors and one or more single-core processors. For example, the processor 102 may be embodied as one or more of various processing devices, such as a coprocessor, a microprocessor, a controller, a digital signal processor (DSP), a processing circuit system with or without an accompanying DSP, or various other processing devices, including integrated circuits, such as application-specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), microcontroller units (MCUs), hardware accelerators, special-purpose computer chips (e.g., machine learning special processors, such as tensor processing units (TPUs), etc.), etc. In one embodiment, the processor 102 may be configured to perform hard-coded functions. In an example embodiment, the processor 102 is embodied as an executor of software instructions, wherein the instructions may specifically configure the processor 102 to perform the algorithms and / or operations described herein when the instructions are executed.

[0128] The memory 104 may be embodied as one or more volatile memory devices, one or more non-volatile memory devices, and / or a combination of one or more volatile memory devices and non-volatile memory devices. For example, the memory 104 may be embodied as a semiconductor memory (such as a mask ROM, a PROM (programmable ROM), an EPROM (erasable PROM), a flash ROM, a RAM (random access memory), etc.).

[0129] For example, the consumer node 100 may be a network node, a base station, or any other network-side node or device that provides wireless communication (e.g., Long Term Evolution (LTE) or 5G wireless communication).

[0130] At least one memory 104 and the computer program code are configured to, together with at least one processor 102, cause the consumer node 100 to at least perform receiving user node capabilities including machine learning-based auxiliary capabilities from a user node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity.

[0131] also, Figure 1 Any combination of the illustrated components disclosed in the disclosure (e.g., at least one of the processor 102, the memory 104, and the transceiver 106) may constitute a component for receiving user node capabilities including machine learning-based auxiliary capabilities from a user node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity.

[0132] Figure 2 Illustrated is an example embodiment of the subject matter described herein, a producer node 200 is shown.

[0133] The producer node 200 includes one or more processors 202 and one or more memories 204 with computer program code. The producer node 200 may also include a transceiver 206 and other elements such as an input / output module ( Figure 2 not shown) and / or communication interface ( Figure 2 not shown).

[0134] Although producer node 200 is depicted as including only one processor 202, producer node 200 may include more than one processor. In an example embodiment, memory 204 is capable of storing instructions, such as an operating system and / or various applications.

[0135] In addition, the processor 202 is capable of executing the stored instructions. In one embodiment, the processor 202 may be embodied as a multi-core processor, a single-core processor, or a combination of one or more multi-core processors and one or more single-core processors. For example, the processor 202 may be embodied as one or more of various processing devices, such as a coprocessor, a microprocessor, a controller, a digital signal processor (DSP), a processing circuit system with or without an accompanying DSP, or various other processing devices, including integrated circuits, such as application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), microcontroller units (MCUs), hardware accelerators, special computer chips, etc. In one embodiment, the processor 102 may be configured to perform hard-coded functions. In an example embodiment, the processor 202 is embodied as an executor of software instructions, wherein the instructions may specifically configure the processor 202 to perform the algorithms and / or operations described herein when the instructions are executed.

[0136] The memory 204 may be embodied as one or more volatile memory devices, one or more non-volatile memory devices, and / or a combination of one or more volatile memory devices and non-volatile memory devices. For example, the memory 204 may be embodied as a semiconductor memory (such as a mask ROM, a PROM (programmable ROM), an EPROM (erasable PROM), a flash ROM, a RAM (random access memory), etc.).

[0137] For example, producer node 200 may be a user node or user device or any other type of device used by an end user and capable of communicating in a wireless network. Such devices include, but are not limited to, smartphones, tablet computers, smart watches, laptops, Internet of Things (IoT) devices, handheld or portable devices, etc.

[0138] At least one memory 204 and computer program code are configured to, together with at least one processor 202, cause the producer node 200 to at least perform identification of user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; and cause transmission of the producer node capabilities to the consumer node.

[0139] also, Figure 2 Any combination of the illustrated components disclosed in (e.g., at least one of the processor 202, the memory 204, and the transceiver 206) may constitute a component for identifying user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; and a component for causing transmission of producer node capabilities to a consumer node.

[0140] Figure 3A An example embodiment of the structure of the machine learning-based assistance capability 300 of the producer node is shown. In an example embodiment, the producer node may be configured to provide machine learning-based assistance to the consumer node, and expose its machine learning-based assistance capability to the consumer node by using the machine learning-based assistance capability 300 of the producer node. The machine learning-based assistance capability 300 of the producer node may be stored in the memory 204 of the producer node.

[0141] The machine learning-based auxiliary capability 300 of the producer node includes machine learning-based functions 302A, 302B, 302C. Each machine learning-based function 302A, 302B, 302C includes a machine learning entity 304A, 304B, 304C and at least one machine learning mode 306A, 306B, 306C, 306D, 306E associated with the machine learning entity 304A, 304B, 304C.

[0142] As an example, the machine learning entity 304A may be associated with a handover (HO) prediction. Two machine learning modes 306A, 306B have been associated with the machine learning entity 304A. For example, the first machine learning mode 306A may be associated with a HO prediction of 10 ms in advance with an accuracy of 95%. The second machine learning mode 306B may be a fallback operating mode that may be applied if the first machine learning mode 306A is not available. In this example, the second machine learning mode 306B (i.e., the fallback operating mode) is associated with a HO prediction of 200 ms in advance with an accuracy of 90%.

[0143] As an example, the machine learning entity 304B may be associated with a quality of service (QoS) prediction. Two machine learning modes 306C, 306D have been associated with the machine learning entity 304B. At least one machine learning mode 306C, 306D may be associated with the machine learning entity 304B. The machine learning mode 306D may be a fallback operating mode that may be applied in the event that the machine learning mode 306C is not available or cannot be used otherwise.

[0144] As another example, the machine learning-based auxiliary capability 300 of the producer node may also include a machine learning-based function 302A, which includes a machine learning mode 306E associated with only one machine learning entity 302C.

[0145] Generally related to the fallback operation mode, at least one fallback operation mode can be configured to be used when none of the configured machine learning based primary modes of the user node can be used. For example, this may occur when they are unavailable, no input is available / deactivated, there is a measurement or connection error, etc. The at least one fallback operation mode may include, for example, one of the following:

[0146] - A second producer node capability (based or not based on machine learning) that is different from the first auxiliary capability based on machine learning.

[0147] -Specific machine learning-based features that are part of the machine learning-based assistance capabilities.

[0148] -Specific rule-based capabilities that are part of the assistance capabilities based on machine learning.

[0149] - A combination of the above.

[0150] - Different fallback operating modes may be specified and configured for each of the different machine learning-based functions, for a set of machine learning modes, or for a set of machine learning entities. There may also be machine learning-based functions that do not require a fallback operating mode at all (or include a "do nothing" fallback operating mode).

[0151] Figure 3B An example embodiment of a machine learning functional architecture is shown. Figure 3BA simplified example is shown where both the radio access network (RAN) (e.g., gNB 100) and the user equipment (UE) 200 have machine learning based functions, namely, training 308, 312 and reasoning 310, 314. The training 308, 312 and reasoning 310, 314 functions are separated to illustrate that these functions can generally be decoupled and can be activated in parallel or at different times. In this case, training can mean that, for example, the (hyper) parameters of the machine learning algorithm are being learned, tested and validated. Reasoning means that data-driven predictions or decisions are taking place.

[0152] Figure 3B The control feedback message loop shown includes delivering configuration of (multiple) machine learning based functions and reporting of inference results. In an example embodiment, this may be a generic and minimal set of information that needs to be exchanged between the gNB 100 and the UE 100. Configuration of (multiple) machine learning based functions may include, for example, selection of machine learning models and layer architectures, possibly initializing weights, (hyper) parameters, selecting activation functions, etc. Both the gNB 100 and the UE 200 may be the source and target of such messages.

[0153] Figure 4A An example embodiment of signaling is shown in which a consumer node 100 activates a specific machine learning-based function in a producer node 200. The producer node 200 may be configured to provide machine learning-based assistance to the consumer node 100. In this example embodiment, the consumer node 100 is a network node, such as a gNB, and the producer node 200 is a user node, such as a user equipment. Obviously, in other environments, the actual nodes may be different.

[0154] At 400, the user node 200 exposes its machine learning-based auxiliary capabilities to the network node 100. The machine learning-based auxiliary capabilities may include machine learning-based functions, wherein each machine learning-based function includes a machine learning entity and at least one machine learning mode associated with the machine learning entity. The machine learning-based functions, machine learning entities, and machine learning modes have been described with respect to Figure 3A Discussed in more detail.

[0155] At 402, initial exploration and tuning of machine learning-based functions within the machine learning-based auxiliary capability may be performed. At this stage, some of the machine learning entities may be provided directly or indirectly by the network in the form of a generic description (such as a containerized implementation, etc.). In an example embodiment, the control signaling to perform this stage may be partially specific to the machine learning-based auxiliary capability and the exact definition of the machine learning-based functions within the auxiliary capability.

[0156] At 404 , the network node 100 may be configured to select an available machine learning based functionality. In the selection, the network node 100 may utilize knowledge about the machine learning based auxiliary capability information received from the user node 200 .

[0157] At 406, according to the needs of radio resource management (RRM), the network node 100 can send an activation request to the user node 200, and the activation request is used to activate one or more machine learning based functions (machine learning entities - machine learning modes) received from the user node 200.

[0158] At 408, the user node 200 is configured to determine whether the requested machine learning-based functionality is available.

[0159] At 410, the user node 200 is configured to transmit an activation response to the network node 100. For example, based on machine learning processing resource availability or radio measurement / condition detection, the user node 200 responds to the network node 100 to indicate whether the requested machine learning-based function can be utilized. If it is not possible to utilize the requested machine learning-based function, the user node 200 may indicate the reason for this in the activation response 410. For example, the user node 200 may indicate an alternative preferred machine learning-based function to be considered. As another example, the user node 200 may indicate a fallback operation mode included in the machine learning-based function to be used. As another example, the user node 200 may indicate that it does not have enough battery power to run the auxiliary algorithm. As another example, the user node 200 may indicate that a significant change in the channel radio condition has been detected, that is, the channel condition has changed significantly from the trained model and the accuracy of the auxiliary / inference based on machine learning is not within the target range.

[0160] If the user node 200 confirms the activation request in the activation response, the network node 100 enables the corresponding (specific) reasoning reporting scheme. The network node 100 may also provide additional configuration information to the user node 200 in the machine learning reasoning report configuration message 412, for example, one or more of the following:

[0161] -Configuration of reporting period. In an example embodiment, the default value may be part of the machine learning entity and / or machine learning model definition.

[0162] -Configuration of the actual content of the report. In an example embodiment, this can be part of the machine learning entity and / or machine learning model definition.

[0163] At 414, the user node 200 may send periodic user node reasoning reports to the network node 100 based on the requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0164] Figure 4B An example embodiment of signaling is shown in which a producer node 200 requests activation of a machine learning based functionality. The producer node 200 may be configured to provide machine learning based assistance to the consumer node 100. In this example embodiment, the consumer node 100 is a network node, such as a gNB, and the producer node 200 is a user node, such as a user equipment. Obviously, in other environments, the actual nodes may be different.

[0165] At 400, the user node 200 exposes its machine learning-based auxiliary capabilities to the network node 100. The machine learning-based auxiliary capabilities may include machine learning-based functions, wherein each machine learning-based function includes a machine learning entity and at least one machine learning mode associated with the machine learning entity. The machine learning-based functions, machine learning entities, and machine learning modes have been described with respect to Figure 3A Discussed in more detail.

[0166] At 402, initial exploration and tuning of machine learning-based functions within the machine learning-based auxiliary capability may be performed. At this stage, some of the machine learning entities may be provided directly or indirectly by the network in the form of a generic description (such as a containerized implementation, etc.). In an example embodiment, the control signaling to perform this stage may be partially specific to the machine learning-based auxiliary capability and the exact definition of the machine learning-based functions within the auxiliary capability.

[0167] Block 419 indicates that during certain time periods, the network node 100 does not need machine learning based functionality available.

[0168] At 420, the user node detects that machine learning based assistance is needed.

[0169] At 422, the user node 200 is configured to send a machine learning based assisted activation request to the network node 100. The request may request activation of one or more of the machine learning based functions exposed at 400. The machine learning based assisted activation request 422 may or may not include an explicit cause indication.

[0170] At 424, the network node 100 may be configured to send a machine learning based function activation response to the user node 200. The response serves as an acknowledgement to the activation request 422. It may also include the user node 200's reasoning report configuration information.

[0171] At 414, the user node 200 may send periodic user node reasoning reports to the network node 100 based on the requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0172] Figure 4C An example embodiment of signaling is shown, in which the producer node 200 requests to deactivate a currently used machine learning-based functionality. The producer node 200 may be configured to provide machine learning-based assistance to the consumer node 100. In this example embodiment, the consumer node 100 is a network node, such as a gNB, and the producer node 200 is a user node, such as a user equipment. Obviously, in other environments, the actual nodes may be different.

[0173] At 414, the user node 200 may send periodic user node reasoning reports to the network node 100 based on the requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0174] If the user node detects at 432 that the machine learning-based assistance should be changed, the user node 200 may be configured to send a change request 434 to the network node for deactivating or switching reasoning reporting. The change request 434 may also provide an indication of the reason for deactivation. The indicated reason may include, for example, one of the following:

[0175] - "Stop-bat": Due to low battery, the current machine learning-based assistance can no longer run.

[0176] - "Stop-comp": Due to low computing power, the current machine learning-based assistance can no longer run.

[0177] - "Stop-input": Due to a significant change in the input data (radio measurements) detected, the current machine learning-based assistance can no longer run.

[0178] - “Switch”: Machine learning-based assistance for switching to another configuration due to a detected change in input data (radio measurements) or a change in mobility status.

[0179] - "Pause-X": The current machine learning-based assistance is unable to run for the configured / indicated period of time X.

[0180] - "Reset-alg": Due to the particularity of algorithm implementation, it is necessary to reset / restart the machine learning based algorithm.

[0181] - "Reset-err": Too many errors have been detected, e.g. within a given time window or relative to a threshold. The ML-based algorithm needs to be reset / restarted.

[0182] The network node 100 may be configured to send a machine learning functionality deactivation response 436 to the user node 200 in response to the change request 434. The machine learning functionality deactivation response 436 provides confirmation of the deactivated machine learning-based assistance and reporting to the user node 200. In an example embodiment, the status of the deactivated reasoning reporting may be valid for a given time period / window, or until enabled again by the network node 100.

[0183] At 438, the network node deconfigures / deallocates user node inference reporting resources.

[0184] Figure 4D An example embodiment of signaling is shown in which a consumer node 100 requests to deactivate a currently used machine learning-based functionality in a producer node (UE). The producer node 200 may be configured to provide machine learning-based assistance to the consumer node 100. In this example embodiment, the consumer node 100 is a network node, such as a gNB, and the producer node 200 is a user node, such as a user equipment. Obviously, in other environments, the actual nodes may be different.

[0185] At 414, the user node 200 may send periodic user node reasoning reports to the network node 100 based on the requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0186] At 440, the network node 100 is configured to detect that machine learning-based assistance should be stopped. In an example embodiment, for example, machine learning assistance may not be needed (i.e., it may be stopped) in the following cases:

[0187] - No processing power available to exploit inference reports

[0188] - High (uplink) load detected and no traffic from user nodes for push

[0189] Capacity of management reports

[0190] - Too many errors detected (by any proprietary means)

[0191] - Due to periodic disabling / enabling as part of a larger machine learning / AI scheme.

[0192] At 442, the network node 100 is configured to send a machine learning assisted change request to the user node 200. The change request 442 requests the user node to deactivate or change one or more machine learning-based functions and corresponding reasoning reports of the current user. The change request 442 may also indicate a reason for deactivation, such as one of the following:

[0193] - "Stop": The current machine learning-based assistance is no longer needed.

[0194] - “Switch to X”: Switch to another configuration of machine learning-based auxiliary X (new machine learning entity and / or new machine learning mode).

[0195] - "Pause": The current machine learning-based assistance is not required for the configured / indicated period of time.

[0196] - "Reset": Too many errors were detected within a given time window or relative to a threshold.

[0197] Need to reset / restart machine learning based algorithms.

[0198] At 444, the user node 200 may be configured to send a machine learning assisted activation response 446 to the network node 100. The response may include, for example, an acknowledgment of the change request 442. In an example embodiment, the acknowledgment may be used by the network node 100 to reconfigure its own algorithms, disable reporting, etc. In the case of a negative acknowledgment (NACK) (e.g., the user node 200 cannot disable the specified machine learning assisted mode for some reason), the network node 100 may still decide to disable / deconfigure the associated reporting, but at the same time use this information in future signaling with the same user node 200 (e.g., when activation of the particular machine learning assisted mode is no longer required).

[0199] In the event that a "Switch to X" or "Reset" signal is indicated at 442, the network node 100 may be configured to send a machine learning inference report configuration message to the user node 200 at 446. The configuration message may instruct the user node 200 to reconfigure its inference reporting to the network node 100. In response to the configuration message at 446, the user node 200 is configured to again begin sending periodic user node inference reports to the network node 100 in accordance with the requested or set machine learning based functionality 416. At 418, the network node 100 may continuously monitor the quality of the inference reports from each activated machine learning based functionality.

[0200] Figure 4E An example embodiment of signaling is shown in which a consumer node 100 uses a pause command to deactivate a currently used machine learning-based function in a producer node. The producer node 200 may be configured to provide machine learning-based assistance to the consumer node 100. In this example embodiment, the consumer node 100 is a network node, such as a gNB, and the producer node 200 is a user node, such as a user equipment. Obviously, in other environments, the actual nodes may be different.

[0201] At 414, the user node 200 may send periodic user node reasoning reports to the network node 100 based on the requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0202] At 440, the network node 100 is configured to detect that machine learning based assistance should be stopped.

[0203] At 448, the network node 100 is configured to send a machine learning assistance change request including a pause indication to the user node 200. The pause indication may include a time period value 450 for which the current machine learning based assistance is not required.

[0204] When the time period 450 expires, the user node 200 may be configured to again begin sending periodic user node reasoning reports to the network node 100 based on the previously requested or set machine learning based functions 416. At 418, the network node 100 may continuously monitor the quality of the reasoning reports from each activated machine learning based function.

[0205] One or more of the illustrated example embodiments may implement a solution in which a combination of network node and user node reasoning is possible. In addition, one or more of the illustrated example embodiments may implement a solution in which a portion of the control based on machine learning may be delegated to the user node. This may make better use of the available user node radio information. In addition, user node quality monitoring of the reasoning results may be performed at both the user node and the network node, possibly at different time granularities and independent algorithms. In addition, one or more of the illustrated example embodiments may implement a solution in which the signaling load is reduced. The solution may make better use of the available user node machine learning capabilities without the need for explicit signaling for each machine learning-based functional configuration and / or change.

[0206] The exemplary embodiments described above may be performed by a user node (eg, user equipment) and / or a network node (eg, base station). In addition, a computer program including instructions for causing an apparatus to execute may perform the illustrated exemplary embodiments.

[0207] Although the subject matter has been described in language specific to structural features and / or acts, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Instead, the specific features and acts described above are disclosed as examples of implementing the claims, and other equivalent features and acts are intended to fall within the scope of the claims.

[0208] It should be understood that the above benefits and advantages may relate to one embodiment or may relate to several embodiments. The embodiments are not limited to embodiments that solve any or all of the problems described or embodiments that have any or all of the above benefits and advantages. It will also be further understood that reference to "an" item may refer to one or more of these items.

[0209] The steps of the methods described herein may be performed in any suitable order, or simultaneously where appropriate. In addition, individual blocks may be deleted from any method without departing from the spirit and scope of the subject matter described herein. Aspects of any of the above embodiments may be combined with aspects of any other embodiment described to form additional embodiments without losing the effects sought.

[0210] The term "comprising" is used herein to mean including the identified methods, blocks or elements, but such blocks or elements do not include an exclusive list and the method or apparatus may contain additional blocks or elements.

[0211] It should be understood that the above description is given as an example only, and various modifications may be made by those skilled in the art. The above description, examples and data provide a complete description of the structure and use of the exemplary embodiments. Although various embodiments have been described above with a certain degree of particularity or with reference to one or more individual embodiments, those skilled in the art may make many changes to the disclosed embodiments without departing from the spirit or scope of this specification.

Claims

1. A method of communication, comprising: receiving, at a radio access network (RAN) side at a network node, user node capabilities including machine learning-based assistance capabilities from a user node, the machine learning-based assistance capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; The method further comprises: Selecting at least one machine learning-based function among the machine learning-based functions; causing transmission of an activation request to activate the selected at least one machine learning-based functionality to the user node; receiving an activation request response from the user node, the activation request response comprising an indication of whether the at least one machine learning-based functionality is activated; causing transmission of a machine learning inference report configuration message to said user node; and A periodic user node reasoning report is received from the user node.

2. The method of claim 1 , wherein the at least one machine learning mode associated with the machine learning entity comprises at least one fallback operating mode.

3. The method according to any one of claims 1 to 2, further comprising: A change request for changing at least one activated machine learning-based function among the machine learning-based functions is received from the user node.

4. The method according to any one of claims 1 to 2, further comprising: Causing transmission of a change request to the user node for changing at least one activated one of the machine learning based functions.

5. A method of communication, comprising: receiving, at a radio access network (RAN) side at a network node, user node capabilities including machine learning-based auxiliary capabilities from a user node, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; The method further comprises: receiving, from the user node, an activation request for activating at least one of the machine learning based functions; causing transmission of a machine learning-based function activation response to the user node, wherein the activation response comprises an inference reporting configuration message for the user node; and A periodic user node reasoning report is received from the user node.

6. The method according to claim 5, further comprising: A change request for changing at least one activated machine learning-based function among the machine learning-based functions is received from the user node.

7. The method of claim 6, wherein the change request includes an indication of a reason for the change, the indication of a reason including one of: a deactivation indication for deactivating the machine learning based functionality; a switch indication for switching to another machine learning-based function; a pause indication for pausing the machine learning-based functionality for a set period of time; as well as A reset indication for resetting or restarting the machine learning based functionality.

8. The method according to claim 7, further comprising: When the change request includes the deactivation indication, a state of a deactivation reasoning report associated with the deactivated machine learning based functionality is maintained for a predetermined period of time.

9. The method according to claim 7, further comprising: Causing transmission of a request to enable deactivated reasoning reporting associated with deactivated machine learning to the user node.

10. The method according to claim 6, further comprising: causing transmission of a response to the user node indicating confirmation of the changed machine learning-based functionality.

11. The method according to claim 5, further comprising: Causing transmission of a change request to the user node for changing at least one activated one of the machine learning based functions.

12. The method of claim 11, wherein the change request includes an indication of a reason for the change, the indication of a reason including one of: a deactivation indication for deactivating the machine learning based functionality; a switch indication for switching to another machine learning-based function; a pause indication for pausing the machine learning-based functionality for a set period of time; as well as A reset indication for resetting or restarting the machine learning based functionality.

13. The method according to claim 12, further comprising: When the change request includes the deactivation indication or the handover indication, transmission of an inference report reconfiguration message to the user node is caused.

14. The method according to claim 12, further comprising: When the change request includes the deactivation indication, maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

15. The method according to claim 12, further comprising: Causing transmission of a request to enable deactivated reasoning reporting associated with deactivated machine learning to the user node.

16. The method according to claim 11, further comprising: A response to the change request is received from the user node.

17. A method of communication, comprising: identifying, at a user node, user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; as well as causing transmission of the user node capabilities to a network node at a radio access network (RAN) side; The method further comprises: receiving, from the network node, an activation request for activating at least one of the machine learning based functions; causing transmission of an activation request response to the network node, the activation request response comprising an indication of whether the at least one machine learning based functionality is activated; receiving a machine learning inference report configuration message from the network node; and The transmission of periodic user node inference reports to the network node is caused.

18. The method of claim 17, wherein the at least one machine learning mode associated with the machine learning entity comprises at least one fallback operating mode.

19. The method according to any one of claims 17-18, further comprising: Causing transmission of a change request to change at least one activated one of the machine learning based functions to the network node.

20. The method according to any one of claims 17-18, further comprising: receiving, from the network node, a change request for changing at least one activated one of the machine learning based functions; as well as The at least one machine learning-based function is changed based on the change request.

21. A method of communication, comprising: identifying, at a user node, user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; as well as causing transmission of the user node capabilities to a network node at a radio access network (RAN) side; The method further comprises: causing transmission of an activation request to activate at least one of the machine learning based functions to the network node; receiving a machine learning based function activation response from the network node, wherein the activation response comprises an inference reporting configuration message of the user node; and The transmission of periodic user node inference reports to the network node is caused.

22. The method according to claim 21, further comprising: Causing transmission of a change request to change at least one activated one of the machine learning based functions to the network node.

23. The method of claim 22, wherein the change request includes an indication of a reason for the change, the indication of a reason including one of: a deactivation indication for deactivating the machine learning based functionality; a switch indication for switching to another machine learning-based function; a pause indication for pausing the machine learning-based functionality for a set period of time; as well as A reset indication for resetting or restarting the machine learning based functionality.

24. The method according to claim 23, further comprising: When the change request includes the deactivation indication, maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

25. The method of claim 23, further comprising: A request is received from the network node to enable deactivated reasoning reporting associated with deactivated machine learning.

26. The method of claim 22, further comprising: A response is received from the network node indicating confirmation of the changed machine learning based functionality.

27. The method of claim 21, further comprising: receiving, from the network node, a change request for changing at least one activated one of the machine learning based functions; as well as The at least one machine learning-based function is changed based on the change request.

28. The method of claim 27, wherein the change request includes an indication of a reason for the change, the indication of a reason including one of: a deactivation indication for deactivating the machine learning based functionality; a switch indication for switching to another machine learning-based function; a pause indication for pausing the machine learning-based functionality for a set period of time; as well as A reset indication for resetting or restarting the machine learning based functionality.

29. The method according to claim 28, further comprising: When the change request includes the deactivation indication or the handover indication, an inference reporting reconfiguration message is received from the network node.

30. The method of claim 28, further comprising: When the change request includes the deactivation indication, maintaining a state of a deactivated reasoning report associated with the deactivated machine learning based functionality for a predetermined period of time.

31. The method of claim 28, further comprising: A request is received from the network node to enable deactivated reasoning reporting associated with deactivated machine learning.

32. The method of claim 27, further comprising: A transmission of a response to the change request is caused to the network node.

33. A network node (100), comprising: at least one processor (102); as well as at least one memory (104) comprising computer program code; The at least one memory (104) and the computer program code are configured to, together with the at least one processor (102), cause the network node (100) to at least execute: The network node is at a Radio Access Network (RAN) side; receiving, from a user node, user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; Selecting at least one machine learning-based function among the machine learning-based functions; causing transmission of an activation request to activate the selected at least one machine learning-based functionality to the user node; receiving an activation request response from the user node, the activation request response comprising an indication of whether the at least one machine learning-based functionality is activated; causing transmission of a machine learning inference report configuration message to the user node; as well as A periodic user node reasoning report is received from the user node.

34. The network node (100) according to claim 33, wherein the at least one memory (104) and the computer program code are configured to, together with the at least one processor (102), cause the network node (100) to perform at least the method according to any one of claims 2 to 4 and 6 to 16.

35. A user node (200), comprising: at least one processor (202); as well as at least one memory (204) comprising computer program code; The at least one memory (204) and the computer program code are configured to, together with the at least one processor (202), cause the user node (200) to at least execute: identifying user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; causing transmission of said user node capabilities to a network node at a radio access network side (RAN); as well as receiving, from the network node, an activation request for activating at least one of the machine learning based functions; causing transmission of an activation request response to the network node, the activation request response comprising an indication of whether the at least one machine learning based functionality is activated; receiving a machine learning inference report configuration message from the network node; and The transmission of periodic user node inference reports to the network node is caused.

36. The user node (200) of claim 35, wherein the at least one memory (104) and the computer program code are configured to, together with the at least one processor (102), cause the network node (100) to at least perform a method according to any one of claims 18 to 20 and 22 to 32.

37. A network node (100), comprising means for performing the following operations: The network node is at a radio access network (RAN) side, receiving, from a user node, user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning mode associated with the machine learning entity; receiving, from the user node, an activation request for activating at least one of the machine learning based functions; causing transmission of a machine learning-based function activation response to the user node, wherein the activation response comprises an inference reporting configuration message for the user node; as well as A periodic user node reasoning report is received from the user node.

38. The network node (100) according to claim 37, wherein the components are configured to perform the method according to any one of claims 2 to 4 and 6 to 16.

39. The network node (100) according to any one of claims 37 to 38, wherein the components include at least one processor and at least one memory including computer program code, and the at least one memory and the computer program code are configured to cause the execution of the network node (100) together with the at least one processor.

40. A user node (200), comprising components for performing the following operations: identifying user node capabilities including machine learning-based auxiliary capabilities, the machine learning-based auxiliary capabilities including machine learning-based functions, each machine learning-based function including a machine learning entity and at least one machine learning pattern associated with the machine learning entity; causing transmission of said user node capabilities to a network node at a radio access network side (RAN); causing transmission of an activation request to activate at least one of the machine learning based functions to the network node; receiving a machine learning based function activation response from the network node, wherein the activation response comprises an inference reporting configuration message of the user node; and The transmission of periodic user node inference reports to the network node is caused.

41. The user node (200) according to claim 40, wherein the components are configured to perform the method according to any one of claims 18 to 20 and 22 to 32.

42. A user node (200) according to any one of claims 40 to 41, wherein the components include at least one processor and at least one memory including computer program code, and the at least one memory and the computer program code are configured to cause the execution of the user node (200) together with the at least one processor.

43. A computer program product comprising instructions for causing an apparatus to perform a method according to any one of claims 1 to 16.

44. A computer program product comprising instructions for causing an apparatus to perform a method according to any one of claims 17 to 32.

45. A computer readable medium comprising a computer program, the computer program comprising instructions for causing an apparatus to perform a method according to any one of claims 1 to 16.

46. ​​A computer readable medium comprising a computer program, the computer program comprising instructions for causing an apparatus to perform a method according to any one of claims 17 to 32.

Images