Optimization and reporting of user terminal run-time capabilities for continual learning operation

Abstract

A method includes receiving, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; determining run-time capabilities limitations of the user device; determining that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmitting, to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receiving, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

Description

OPTIMIZATION AND REPORTING OF USER TERMINAL RUN-TIME CAPABILITIES FOR CONTINUAL LEARNING OPERATIONTECHNICAL FIELD

[0001] This description relates to wireless communications.BACKGROUND

[0002] A communication system may be a facility that enables communication between two or more nodes or devices, such as fixed or mobile communication devices. Signals can be carried on wired or wireless carriers.

[0003] An example of a cellular communication system is an architecture that is being standardized by the 3rd Generation Partnership Project (3GPP). A recent development in this field is often referred to as the long-term evolution (LTE) of the Universal Mobile Telecommunications System (UMTS) radio-access technology. E-UTRA (evolved UMTS Terrestrial Radio Access) is the air interface of 3GPP's Long Term Evolution (LTE) upgrade path for mobile networks. In LTE, base stations or access points (APs), which are referred to as enhanced Node AP (eNBs), provide wireless access within a coverage area or cell. In LTE, mobile devices, or mobile stations are referred to as user equipments (UE). LTE has included a number of improvements or developments. Aspects of LTE are also continuing to improve.

[0004] 5G New Radio (NR) development is part of a continued mobile broadband evolution process to meet the requirements of 5G, similar to earlier evolution of 3G and 4G wireless networks. In addition, 5G is also targeted at the new emerging use cases in addition to mobile broadband. A goal of 5G is to provide significant improvement in wireless performance, which may include new levels of data rate, latency, reliability, and security. 5G NR may also scale to efficiently connect the massive Internet of Things (loT) and may offer new types of mission-critical services. For example, ultra-reliable and low-latency communications (URLLC) devices may require high reliability and very low latency. 6G and other networks are also being developed.SUMMARY

[0005] A method may include receiving, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user deviceto carry out continual learning for a machine learning model; determining, by the user device, run-time capabilities limitations of the user device; determining, by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmitting, by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receiving, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

[0006] An apparatus may include means for receiving, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; means for determining, by the user device, run-time capabilities limitations of the user device; means for determining, by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; means for transmitting, by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and means for receiving, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

[0007] An apparatus may include 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 the apparatus at least to: receive, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; determine, by the user device, run-time capabilities limitations of the user device; determine, by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmit, by the user device to the network node, run-time capabilities limitation information related to aninability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receive, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

[0008] A method may include transmitting, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; receiving, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; determining, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and transmitting, by the network node to the user device, the continual learning configuration update.

[0009] An apparatus may include 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 the apparatus at least to: transmit, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; receive, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; determine, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and transmit, by the network node to the user device, the continual learning configuration update.

[0010] An apparatus may include means for transmitting, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; means for receiving, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; means for determining, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and means fortransmitting, by the network node to the user device, the continual learning configuration update.

[0011] Other example embodiments are provided or described for each of the example methods, including: means for performing any of the example methods; a non-transitory computer-readable storage medium comprising instructions stored thereon that, when executed by at least one processor, are configured to cause a computing system to perform any of the example methods; and an apparatus including 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 the apparatus at least to perform any of the example methods.

[0012] The details of one or more examples of embodiments are set forth in the accompanying drawings and the description below. Other features will be apparent from the description and drawings, and from the claims.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIG. l is a block diagram of a wireless network.

[0014] FIG. 2 is a flow chart illustrating operation of a user device (e.g., UE).

[0015] FIG. 3 is a flow chart illustrating operation of a network node (e.g., gNB).

[0016] FIG. 4 is a diagram illustrating operation of a system.

[0017] FIG. 5 is a block diagram of a wireless station or node (e.g., network node (such as gNB), user node or UE, relay node, or other node).DETAILED DESCRIPTION

[0018] FIG. 1 is a block diagram of a wireless network 130. In the wireless network 130 of FIG. 1, user devices 131, 132, 133 and 135, which may also be referred to as mobile stations (MSs) or user equipment (UEs), may be connected (and in communication) with a base station (BS) 134, which may also be referred to as an access point (AP), an enhanced Node B (eNB), a gNB or a network node. The terms user device and user equipment (UE) may be used interchangeably. A BS may also include or may be referred to as a RAN (radio access network) node, and may include a portion of a BS or a portion of a RAN node, such as (e.g., such as a centralized unit (CU) and / or a distributed unit (DU) in the case of a split BS or split gNB). At least part of the functionalities of a BS (e.g., access point (AP), base station (BS) or (e)Node B (eNB), gNB, RAN node) may also be carried out by any node, server or host which may be operably coupled to a transceiver, such as a remote radio head. BS (orAP) 134 provides wireless coverage within a cell 136, including to user devices (or UEs) 131, 132, 133 and 135. Although only four user devices (or UEs) are shown as being connected or attached to BS 134, any number of user devices may be provided. BS 134 is also connected to a core network 150 via a SI interface 151. This is merely one simple example of a wireless network, and others may be used.

[0019] A base station (e.g., such as BS 134) is an example of a radio access network (RAN) node within a wireless network. A BS (or a RAN node) may be or may include (or may alternatively be referred to as), e.g., an access point (AP), a gNB, an eNB, or portion thereof (such as a / centralized unit (CU) and / or a distributed unit (DU) in the case of a split BS or split gNB), or other network node.

[0020] According to an illustrative example, a BS node (e.g., BS, eNB, gNB, CU / DU, . . .) or a radio access network (RAN) may be part of a mobile telecommunication system. A RAN (radio access network) may include one or more BSs or RAN nodes that implement a radio access technology, e.g., to allow one or more UEs to have access to a network or core network. Thus, for example, the RAN (RAN nodes, such as BSs or gNBs) may reside between one or more user devices or UEs and a core network. According to an example embodiment, each RAN node (e.g., BS, eNB, gNB, CU / DU, ...) or BS may provide one or more wireless communication services for one or more UEs or user devices, e.g., to allow the UEs to have wireless access to a network, via the RAN node. Each RAN node or BS may perform or provide wireless communication services, e.g., such as allowing UEs or user devices to establish a wireless connection to the RAN node, and sending data to and / or receiving data from one or more of the UEs. For example, after establishing a connection to a UE, a RAN node or network node (e.g., BS, eNB, gNB, CU / DU, . . .) may forward data to the UE that is received from a network or the core network, and / or forward data received from the UE to the network or core network. RAN nodes or network nodes (e.g., BS, eNB, gNB, CU / DU, . . .) may perform a wide variety of other wireless functions or services, e.g., such as broadcasting control information (e.g., such as system information or on-demand system information) to UEs, paging UEs when there is data to be delivered to the UE, assisting in handover of a UE between cells, scheduling of resources for uplink data transmission from the UE(s) and downlink data transmission to UE(s), sending control information to configure one or more UEs, and the like. These are a few examples of one or more functions that a RAN node or BS may perform.

[0021] A user device or user node (user terminal, user equipment (UE), mobile terminal, handheld wireless device, etc.) may refer to a portable computing device that includeswireless mobile communication devices operating either with or without a subscriber identification module (SIM), including, but not limited to, the following types of devices: a mobile station (MS), a mobile phone, a cell phone, a smartphone, a personal digital assistant (PDA), a handset, a device using a wireless modem (alarm or measurement device, etc.), a laptop and / or touch screen computer, a tablet, a phablet, a game console, a notebook, a vehicle, a sensor, and a multimedia device, as examples, or any other wireless device. It should be appreciated that a user device may also be (or may include) a nearly exclusive uplink only device, of which an example is a camera or video camera loading images or video clips to a network. Also, a user node may include a user equipment (UE), a user device, a user terminal, a mobile terminal, a mobile station, a mobile node, a subscriber device, a subscriber node, a subscriber terminal, or other user node. For example, a user node may be used for wireless communications with one or more network nodes (e.g., gNB, eNB, BS, AP, CU, DU, CU / DU) and / or with one or more other user nodes, regardless of the technology or radio access technology (RAT). In LTE (as an illustrative example), core network 150 may be referred to as Evolved Packet Core (EPC), which may include a mobility management entity (MME) which may handle or assist with mobility / handover of user devices between BSs, one or more gateways that may forward data and control signals between the BSs and packet data networks or the Internet, and other control functions or blocks. Other types of wireless networks, such as 5G (which may be referred to as New Radio (NR)) may also include a core network.

[0022] In addition, the techniques described herein may be applied to various types of user devices or data service types, or may apply to user devices that may have multiple applications running thereon that may be of different data service types. New Radio (5G) development may support a number of different applications or a number of different data service types, such as for example: machine type communications (MTC), enhanced machine type communication (eMTC), Internet of Things (loT), and / or narrowband loT user devices, enhanced mobile broadband (eMBB), and ultra-reliable and low-latency communications (URLLC). Many of these new 5G (NR) - related applications may require generally higher performance than previous wireless networks.

[0023] loT may refer to an ever-growing group of objects that may have Internet or network connectivity, so that these objects may send information to and receive information from other network devices. For example, many sensor type applications or devices may monitor a physical condition or a status, and may send a report to a server or other network device, e.g., when an event occurs. Machine Type Communications (MTC, or Machine toMachine communications) may, for example, be characterized by fully automatic data generation, exchange, processing and actuation among intelligent machines, with or without intervention of humans. Enhanced mobile broadband (eMBB) may support much higher data rates than currently available in LTE.

[0024] Ultra-reliable and low-latency communications (URLLC) is a new data service type, or new usage scenario, which may be supported for New Radio (5G) systems. This enables emerging new applications and services, such as industrial automations, autonomous driving, vehicular safety, e-health services, and so on. 3 GPP targets in providing connectivity with reliability corresponding to block error rate (BLER) of 10-5 and up to 1 ms U-Plane (user / data plane) latency, by way of illustrative example. Thus, for example, URLLC user devices / UEs may require a significantly lower block error rate than other types of user devices / UEs as well as low latency (with or without requirement for simultaneous high reliability). Thus, for example, a URLLC UE (or URLLC application on a UE) may require much shorter latency, as compared to an eMBB UE (or an eMBB application running on a UE).

[0025] The techniques described herein may be applied to a wide variety of wireless technologies or wireless networks, such as 5G (New Radio (NR)), cmWave, and / or mmWave band networks, loT, MTC, eMTC, eMBB, URLLC, 6G, etc., or any other wireless network or wireless technology. These example networks, technologies or data service types are provided only as illustrative examples.

[0026] A machine learning (ML) model may be used within a wireless network to perform (or assist with performing) one or more tasks. In general, one or more nodes (e.g., BS, gNB, eNB, RAN node, user node, UE, user device, relay node, or other wireless node) within a wireless network may use or employ a ML model, e.g., such as, for example a neural network model (e.g., which may be referred to as a neural network, an artificial intelligence (Al) neural network, an Al neural network model, an Al model, a machine learning (ML) model or algorithm, a model, or other term) to perform, or assist in performing, one or more ML-enabled tasks. Other types of models may also be used. A ML-enabled task may include tasks that may be performed (or assisted in performing) by a ML model, or a task for which a ML model has been trained to perform or assist in performing).

[0027] ML-based algorithms or ML models may be used to perform and / or assist with performing a variety of wireless and / or radio resource management (RRM) and / or RAN- related functions or tasks to improve network performance, such as, e.g., in the UE for beam prediction (e.g., predicting a best beam or best beam pair based on measured reference signals), antenna panel or beam control, RRM (radio resource measurement) measurementsand feedback (channel state information (CSI) feedback), link monitoring, Transmit Power Control (TPC), etc. In some cases, ML models may be used to improve performance of a wireless network in one or more aspects or as measured by one or more performance indicators or performance criteria.

[0028] Models (e.g., neural networks or ML models) may be or may include, for example, computational models used in machine learning made up of nodes organized in layers. The nodes are also referred to as artificial neurons, or simply neurons, and perform a function on provided input to produce some output value. A neural network or ML model may typically require a training period to learn the parameters, i.e., weights, used to map the input to a desired output. The mapping may occur via the function that is learned from a given data for the problem in question. Thus, the weights are weights for the mapping function of the neural network. Each neural network model or ML model may be trained for a particular task.

[0029] To provide the output given the input, the ML functionality of a neural network model or ML model should be trained, which may involve learning the proper value for a large number of parameters (e.g., weights and / or biases) for the mapping function (or of the ML functionality of the ML model). For example, the parameters may be used to weight and / or adjust terms in the mapping function. This training may be an iterative process, with the values of the weights and / or biases being tweaked over many (e.g., tens, hundreds and / or thousands) of rounds of training episodes or training iterations until arriving at the optimal, or most accurate, values (or weights and / or biases). In the context of neural networks (neural network models) or ML models, the parameters may be initialized, often with random values, and a training optimizer iteratively updates the parameters (e.g., weights) of the neural network to minimize error in the mapping function. In other words, during each round, or step, of iterative training the network updates the values of the parameters so that the values of the parameters eventually converge to the optimal values.

[0030] ML models may be trained in either a supervised or unsupervised manner, as examples. In supervised learning, training examples are provided to the ML model or other machine learning algorithm. A training example includes the inputs and a desired or previously observed output. Training examples are also referred to as labeled data because the input is labeled with the desired or observed output. In the case of a neural network (which may be a specific case of ML model), the network (or ML model) learns the values for the weights used in the mapping function or ML functionality of the ML model that most often result in the desired output when given the training inputs. In unsupervised training,the ML model learns to identify a structure or pattern in the provided input. In other words, the model identifies implicit relationships in the data. Unsupervised learning is used in many machine learning problems and typically requires a large set of unlabeled data.

[0031] According to an example embodiment, a ML model may be classified into (or may include) two broad categories (supervised and unsupervised), depending on whether there is a learning “signal” or “feedback” available to a model. Thus, for example, within the field of machine learning, there may be two main types of learning or training of a model: supervised, and unsupervised. The main difference between the two types is that supervised learning is done using known or prior knowledge of what the output values for certain samples of data should be. Therefore, a goal of supervised learning may be to learn a function that, given a sample of data and desired outputs, best approximates the relationship between input and output observable in the data. Unsupervised learning, on the other hand, does not have labeled outputs, so its goal is to infer the natural structure present within a set of data points.

[0032] Supervised learning: The computer is presented with example inputs and their desired outputs, and the goal may be to learn a general rule that maps inputs to outputs. Supervised learning may, for example, be performed in the context of classification, where a computer or learning algorithm attempts to map input to output labels, or regression, where the computer or algorithm may map input(s) to a continuous output(s). Common algorithms in supervised learning may include, e.g., logistic regression, naive Bayes, support vector machines, artificial neural networks, and random forests. In both regression and classification, a goal may include finding specific relationships or structure in the input data that allow us to effectively produce correct output data. In some example cases, the input signal may be only partially available, or restricted to special feedback. Semi-supervised learning: the computer may be given only an incomplete training signal; a training set with some (often many) of the target outputs missing. Active learning: the computer can only obtain training labels for a limited set of instances (based on a budget), and also may optimize its choice of objects for which to acquire labels. When used interactively, these can be presented to the user for labeling.

[0033] Unsupervised learning: No labels are given to the learning algorithm, leaving it on its own to find structure in its input. Some example tasks within unsupervised learning may include clustering, representation learning, and density estimation. In these cases, the computer or learning algorithm is attempting to learn the inherent structure of the data without using explicitly-provided labels. Some common algorithms include k-means clustering,principal component analysis, and auto-encoders. Since no labels are provided, there may be no specific way to compare model performance in most unsupervised learning methods.

[0034] Continual Learning (CL) may refer to or may include a capability of the ML model to adapt to ever-changing (or continuously changing, or periodically changing) surrounding environment or data by learning or adapting the ML model continually based on incoming data (or new or updated data), e.g., without forgetting original or previous knowledge or ML model settings, and, e.g., which may be based on less than a full or complete set of data. For example, given a (e.g., potentially unlimited or continuous) stream of data (e.g., data reflecting changing or updated conditions or environment upon which the ML model should be updated), a continual learning (CL) algorithm may (or should) learn, e.g., by updating or adapting weights or other parameters of the ML model, based on a sequence of partial experiences or partial data (e.g., a most recent set of data) where all data may not be available at once, since new or updated data will be received later (thus, the new data potentially renders the weights or parameter settings of the ML model obsolete or inaccurate). Thus, a full or complete set of data may not be considered available at that time of ML model updating or adaptation, since the data or environment may be continuously or continually changing over time. Thus, at any given point or moment in time, data (upon which the ML model may be updated or adapted) may be considered incomplete because there may be a continuous stream of data. Thus, a CL algorithm may include or may refer to iteratively updating or adapting weights or other parameters of the ML model based on an updated set of data, and then repeating the learning or adaptation process for the ML model when a second (or later) set of updated data is received subsequently.

[0035] Thus, for example, where the ML model is trained (e.g., learning or adaptation is performed for the weights or other parameters of the ML model) for task A: in this case due to continuous context changes (e.g., changes in the environment or continuous stream of updated or new data that reflects the ever-changing current environment or context), the ML model trained for task A at time t would not fit at time t + 6t. Thus, a continual adaptation (or continual learning process) of the ML model may be very useful, and even needed in some cases, to ensure the required performance accuracy for the considered functionality that is performed or assisted by the ML model.

[0036] However, in some cases, limitations of run-time capabilities of the UE may limit or inhibit continual learning by the UE for a ML model. Run-time capabilities of the UE may include, e.g., any current capabilities or resources of the UE based on a current state of theUE due to its operation or running, and / or based on current resource usage of the UE or state of the UE.

[0037] FIG. 2 is a flow chart illustrating operation of a user device (or UE). Operation 210 includes receiving, by a user device (or UE) from a network node (e.g., a gNB or other network node), a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning (ML) model. Operation 220 includes determining, by the user device (or UE), run-time capabilities limitations of the user device. Operation 230 includes determining, by the user device (or UE), that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device. Operation 240 includes transmitting, by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device. And, operation 250 includes receiving, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

[0038] With respect to the method of FIG. 2, the continual learning configuration may include at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

[0039] With respect to the method of FIG. 2, the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, and may include at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

[0040] With respect to the method of FIG. 2, the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model may include at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machinelearning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

[0041] With respect to the method of FIG. 2, the received continual learning configuration update received from the network node may include at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

[0042] With respect to the method of FIG. 2, the run-time capabilities limitation information transmitted by the user device to the network node comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

[0043] With respect to the method of FIG. 2, the run-time capabilities limitation information is transmitted as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

[0044] FIG. 3 is a flow chart illustrating operation of a network node (e.g., gNB).Operation 310 includes transmitting, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model. Operation 320 includes receiving, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device. Operation 330 includes determining, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update. And, operation 340 includes transmitting, by the network node to the user device, the continual learning configuration update.

[0045] With respect to the method of FIG. 3, the continual learning configuration may include at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

[0046] With respect to the method of FIG. 3, the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, including at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

[0047] With respect to the method of FIG. 3, the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model may include at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continuallearning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

[0048] With respect to the method of FIG. 3, the continual learning configuration update may include at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

[0049] With respect to the method of FIG. 3, the run-time capabilities limitation information received by the network node from the user device may include one or more user device (UE) proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, including one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

[0050] With respect to the method of FIG. 3, the run-time capabilities limitation information is received as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

[0051] The description and figures provided herein, including the figures and description hereinbelow, provide further details, description and / or illustrative examples with respect to the methods of FIGs. 2 and 3.

[0052] As noted above, in some cases, limitations of run-time capabilities of the UE may limit or inhibit continual learning by the UE for a ML model. Run-time capabilities of the UE may include, e.g., any current capabilities of the UE based on a current state of the UE due to its operation or running, and / or based on current resource usage of the UE. Thus, forexample, UE run-time capabilities may include, e.g., any capabilities of the UE related to hardware, software, computational resources, memory resources, network resources, a status of the UE with respect to a power saving mode (e.g., whether the UE is in a power saving mode or not), battery or power resources (e.g., how much battery power is remaining, or whether the UE is in a power-saving mode or not), a current heat state of the UE (whether one or more components of the UE is overheated or not overheated, or a temperature of hardware, processor or other components as compared to an overheated temperature), or other UE capabilities.

[0053] In addition, UE run-time capabilities limitations may include any limitation of a UE run-time capability(ies). For example, run-time capabilities limitations of the UE, may be based on, may indicate or may reflect a current state of the UE with respect to one or more limitations or resources of the UE. Thus, for example, run-time capabilities limitations are not static capabilities of the UE that are typically reported in a capabilities exchange. Rather, runtime capabilities limitations are (or may include) current or updated capabilities limitations of the UE that may be dynamically (e.g., during operation of the UE) determined or updated by the UE based on a current UE status or current resource usage of the UE, while the UE is running or operating.

[0054] UE run-time capabilities limitations may include one or more of the following (as some illustrative examples):

[0055] 1) computational limitations or processing limitations of the UE, e.g., based on available processing resources or based on current computation level, such as, for example, processing load, or the available processing resources, such as of a central processing unit (CPU), a graphical processing unit (GPU), or other processor of the UE. Thus, for example, if there is less than a threshold percentage (e.g., less than 30% free processing / processor resources) or less than a threshold amount of processing resources available, this may inhibit or prevent further processing of data, such as for computational learning of a ML model, or may at least limit the carrying out of continual learning at least for some situations.

[0056] 2) Memory limitations of the UE, e.g., based on available memory resources and / or based on the amount or percentage of memory resources that are currently in use by the UE, such as what amount or what percentage of memory resources are in use, and / or what amount or percentage of memory resources are free or available. Thus, for example, if there is less than a threshold amount or percentage of free (available) memory resources, e.g., less than 25% free memory, or less than 5 GB free memory (as an example), or, e.g., more than 75% of memory resources are being used, or less than 25% of free / available memoryresources at the UE, this may inhibit or prevent processing of data by the UE, such as for computational learning, or may at least limit the carrying out of continual learning in at least some situations.

[0057] 3) Power (or battery)-related state or power or battery-related limitations of theUE, such as whether a power saving mode has been activated for the UE, or an amount of available battery resources of the UE. For example, if either the UE is in a power save mode or there is less than a threshold (e.g., less than 38%) amount of available battery power for the UE, this may prevent or inhibit the UE from performing continual learning, or at least limit the carrying out of continual learning in at least some situations.

[0058] 4) An overheating status or state of the UE, such as a temperature of some portion of the UE hardware such as a temperature of a processor or processor core, or other component, or a temperature measured by a temperature sensor on the UE, an indication of whether a UE temperature (e.g., temperature of a UE processor or UE component) is greater than a threshold, or an indication that the UE (or UE component) is either in an overheated state or within a threshold number of degrees of an overheated state. Any of these example temperature / heat-related limitations or overheating states may prevent or inhibit additional processing of data by that processor or processor core, such as for continual learning, or may at least limit the carrying out of continual learning in at least some situations. For example, if a processor (or other component), is either at an overheated temperature or within a threshold number of degrees within the overheated temperature, this may prevent the UE from performing continual learning or adaptation of the ML model, or may limit the carrying out of continual learning or adaptation of the ML model.

[0059] 5) Other resources of the UE, such as network resources (e.g., MAC / PHY resources), antenna resources, protocol entity resources, software resources, etc., may similarly have run-time capabilities limitations that may inhibit or limit the UE from carrying out the continual learning of the ML model. For example, if there are less than a threshold amount of network resources, antenna resources, protocol entity resources and / or software resources for the UE, this may inhibit the UE from carrying out the continual learning of the ML model.

[0060] Also, for example, the UE may compare the continual learning configuration to the run-time capabilities limitations of the UE, to determine if there are presently any UE runtime capabilities limitations that may prevent the UE from carrying out or performing continual learning for the ML model. Alternatively, or said another way, the UE may determine, based on a current state or status of its run-time capabilities and the continuallearning configuration, if there is sufficient UE run-time capabilities or resources available to carry out or perform continual learning of the ML model as indicated or instructed by (or in accordance with) the continual learning configuration. For example, if there is less than a first threshold (e.g., less than 35%) of battery power available, then continual learning may be prevented at the UE, as the UE does not have sufficient batter power to perform a continual learning of the ML model of the UE (or at least indicates the UE should not perform continual learning with less than the first threshold of battery power available). Or, if a processor of the UE is within a threshold number of degrees within an overheated state, or the amount of free memory or free processor resources is less than a threshold, this may indicate that the UE is unable to (or should not) carry out or perform continual learning (or an adaptation operation or iteration of the ML model based on current data) according to the received CL configuration, due to one or more run-time capabilities limitations of the UE.

[0061] Furthermore, different or various thresholds may be used by a UE to allow different outcomes or decisions for the determination by the UE as to whether continual learning may be performed by the UE. For example, a first threshold for a UE run-time capabilities limitation may be used to prevent the UE from carrying out the continual learning of the ML model, while a second threshold may be used to allow some limited continual learning to be performed by the UE, e.g., possibly with different CL configuration parameter(s). For example, if there is less than a first threshold (e.g., less than 35%) of battery power available at the UE, then continual learning may be completely prevented at the UE, as the UE does not have sufficient batter power to perform a continual learning of the ML model of the UE. However, if there is less than a second threshold (e.g., less than 50% battery power), but more than the first threshold (e.g., more than 35% battery power) at the UE, then the UE may be able to perform some limited continual learning of the ML model, e.g., based on an updated CL configuration that may allow the UE to perform a lesser amount of continual learning (or consume less battery power in performing continual learning).

[0062] For example, if the CL configuration received by the UE indicates, e.g., an adaptive ratio of 0.6 (indicating that 6 of the 10 layers of the ML model should be adapted as part of the continual learning operation), but the battery power level of 42% is not greater than 50% (the second threshold), but is greater than the first threshold (35%), the UE may be unable to perform the continual learning with adaptive ratio of 0.6. Thus, in this example, where the UE battery power level is 42% (greater than the first threshold but less than the second threshold), the UE may determine that the UE cannot perform continual learning according to the received CL configuration (with adaptive ratio of 0.6), but the UE canperform a lower level or lower amount of continual learning or adaptation than what is indicated by the CL configuration received by the UE, e.g., the UE can only perform continual learning or ML model adaptation for only 3 of the 10 layers (associated with an adaptive ratio of 0.3 for continual learning of the ML model, where 3 of the 10 layers of the ML model will be adapted, while 7 of the 10 layers will be held static or frozen / not adapted during this continual learning operation or iteration). Thus, in this example case, the UE may send or transmit to the gNB (or other network node) run-time capabilities limitation information for the UE that may include, e.g., information indicating that the UE is unable to perform the configured continual learning for the ML model in accordance with the received CL configuration (which indicates an adaptive ratio of 0.6) based on UE run-time capabilities limitation(s), information indicating the specific run-time capabilities limitation (battery power, in this example) that is preventing the UE from carrying out continual learning of the ML model, and / or one or more UE proposals (e.g., suggestions or recommendations) for one or more parameters of a CL configuration that the UE is able to perform or use for continual learning, based on its run-time capabilities limitations. In this example, the UE may transmit to the gNB a proposal for an adaptive ratio of 0.3 (e.g., thus, indicating that the UE can perform continual learning or adaptation of 3 of the 10 layers of the ML model), and may indicate that this proposal is based on (or due to) a run-time capabilities limitation related to battery power of the UE that prevents the UE from performing continual learning in accordance with the received continual learning configuration. The adaptive ratio is just one example, and the UE may provide a proposal (proposed value or parameter value) to the gNB for any of the UE run-time capabilities.

[0063] Based on the run-time capabilities limitation information received by the gNB from the UE (e.g., indicating that the UE is unable to perform continual learning based on the configured adaptive ratio of 0.6, and including the proposed adaptive ratio of 0.3 for which the UE can perform continual learning), the gNB may determine an updated CL configuration, and may transmit the updated CL configuration to the UE. The updated CL configuration may include, e.g., an updated parameter or parameter value, e.g., such as the adaptive ratio of 0.3 that is proposed by the UE or other value, or may include an indication to cancel (not perform) the continual learning, and / or may include an indication for the UE to use a current version (current set of weights) for the ML model.

[0064] FIG. 4 is a diagram illustrating operation of a system. A UE 410 may include a machine learning (ML) model 412. The UE 410 may perform or carry out continual learning for the ML model 412 based on a continual learning (CL) request and / or a continual learning(CL) configuration that may be received by UE 410 from a gNB (or other network node) 414. UE 410 may be in communication with gNB 414. In some cases, UE 410 may be connected to gNB 414.

[0065] At step 1 of FIG. 4, UE 410 receives a continual learning request, including a continual learning (CL) configuration. The CL configuration may include one or more parameters as part of the CL configuration. For example, the continual learning (CL) configuration may include a variety of parameters that may indicate parameters, conditions or requirements for the continuation learning (CL) for the ML model, including one or more of the following example parameters:

[0066] 1) An identification of a ML model (e.g., ML model ID), or a task or function(e.g., beam prediction) to be performed by the ML model.

[0067] 2) A request to perform or carry out continual learning on the ML model.

[0068] 3) An adaptive ratio, which may be or include a ratio of layers to be adapted vs. layers to be held static or frozen (not adapted), or an amount of the ML model or percentage or number of layers of the ML model on which to perform or carry out continual learning (to be adapted). An adaptive ratio may indicate a ratio of adaptive layers (layers to be adapted or trained / re-trained via continual learning) to frozen or static layers (layers of the ML model that are static or frozen or not to be adapted during this adaptation or continuation learning operation or iteration). For example, an adaptive ratio of 1 / 3 indicates that 1 / 3 (or one-third) of the ML model layers should have their weights adapted or re-trained during this continual learning operation or iteration, and the weights for the other 2 / 3 (two-thirds) of the layers of the ML model are not to be re-trained or adapted during this CL operation or iteration (thus, those 2 / 3 of the layers will be held static or frozen, and thus not changed or adapted during this continual learning operation or iteration). Thus, for example, if there are 20 layers of the ML model, an adaptive ratio of 0.5 indicates that weights of 10 of the 20 layers should be adapted or re-trained as part of this continual learning operation or iteration. Requiring more layers to be adapted (a higher adaptive ratio) requires more memory, processor, battery and / or time resources of the UE.

[0069] 4) An accuracy level for the machine learning model. A higher accuracy level(more accurate training or adapting of weights for the ML model) requirement will typically require, or may be associated with, more memory, processor, battery and time resources of the UE for ML model adaptation. For example, an accuracy level of the ML model may refer to or may include a ML model prediction accuracy, which is one example to indicate thereliability measure for the ML model. For example prediction accuracy may be determined or represented as: Prediction accuracy = (predicted value / true value) * 100%, which means that this example key performance indicator (KPI) is verified against the true label / data to determine an accuracy level or prediction accuracy. Also, or as another example, an accuracy level for a ML model may be determined based on one or more KPIs, such as based on an Fl score or other KPI.

[0070] 5) A training period for continual learning of the machine learning model, which may indicate a period (e.g., maximum period) of time during which training or adaptation of the ML model weights should be performed. A higher (larger) training (or adaptation) period will typically require, or may be associated with, more memory, processor, battery and / or time resources of the UE for ML model adaptation. Thus, the training period may refer to a maximum training period required by the UE. If this training period to 1 hour, then the ML model training / adaption (CL operation) for the ML model may not exceed this time limit. This parameter (training period) may also be referred to as the UE's (e.g., preferred or proposed) (re)training time limitation for continual learning.

[0071] 6) A periodicity for continual learning of the machine learning model. The periodicity may indicate how frequently (or a period) that retraining or adaptation should be performed for the ML model, e.g., every 100ms. A smaller periodicity will typically require, or may be associated with, more memory, processor, battery and / or time resources of the UE for ML model adaptation, since this smaller periodicity will require the UE to perform continual learning or model adaptation more frequently. These are some example parameters, and other parameters may be included.

[0072] At step 2 of FIG. 4, UE 410 may determine run-time capabilities limitations of the UE 410, and whether the UE 410 can perform or carry out continual learning in accordance with (or based on) the received continual learning configuration. UE run-time capabilities may include a current state of one or more capabilities or resources of the UE, such as a current state or status of UE processor resources, memory resources, other hardware resources, software resources, network resources, protocol entity resources, power or battery resources, or other resources or capabilities, and / or a state of the UE related to overheating or whether the UE is in power saving mode (since the ability of the UE to operate fully may be impacted by the UE being overheated and / or the UE being placed in a power saving mode).

[0073] The UE run-time capabilities limitations may reflect or indicate a current state of the UE with respect to one or more capabilities or resources of the UE, may include or may indicate whether the run-time capability is greater than or less than a threshold, for example.UE run-time capabilities limitations may include, for example, current a state or status of one or more of the following (based on a current state or status of the UE): processing / computational limitations of the UE based on available processing resources or based on current computation levels; memory limitations of the UE based on available memory resources and / or based on memory resources that are currently in use; whether a power saving mode has been activated for the UE; a battery or power state of the UE, or available battery resources of the UE; and / or an overheating status of the UE. In some cases, or for some capabilities or resources, the UE may determine its run-time capabilities limitations by determining an amount or percentage of a resource or capability that is free or available, such as based on a current state or current usage of the resource or capability (a current amount or percent of free or available processing resources or processing capability, memory resources, power or battery resources, network resources, other software or hardware resources, etc.)

[0074] Also, at step 2 of FIG. 4, the UE 410 determines whether its current run-time capabilities limitations will prevent the UE from performing continual learning in accordance with the received continual learning (CL) configuration. Thus, to determine whether the UE’s current run-time capabilities limitations may prevent the UE from performing the continual learning of the ML mode, the UE may compare its current run-time capabilities to one or more thresholds. For example, the UE may compare its current available battery power, current available memory resources, current available processing resources, current overheating status or processor temperature, etc., to one or more thresholds to determine if the UE is able or unable to perform continual learning based on the received continual learning configuration. For example, if the UE has less than 30% available processing or computational resources, or less than 5 GB of available or free memory, or less than 35% available battery power, then (any of these) may indicate that the UE is unable to perform continual learning for the ML model. Or the UE may also determine if the UE is in a power saving mode, or whether a temperature of a UE processor or UE component is within a threshold number (e.g., 10) degrees of an overheating temperature. If the UE is in a power saving mode or if the UE is in an overheated state or overheating temperature (e.g., a temperature of the UE or a component of the UE is greater than the overheated / overheating temperature) or the UE has a processor or other component temperature within a threshold number of degrees to the overheating temperature, then this indicates that the UE is unable to perform continual learning for the ML model, based on one or more of these UE run-time capabilities limitations.

[0075] Also, at step 2 of FIG. 4, the UE determining that the UE is able or unable to carry out continual learning in accordance with the CL configuration may be based on both the UE run-time capabilities limitations and the parameters or values of the continual learning configuration (CL configuration). For example, the various thresholds used by the UE for different UE run-time capabilities limitations may be based on (values for) one or more CL configuration parameters. Thus, for example, different CL configuration parameterdependent thresholds may be used to determine whether the UE is able to perform the continual learning in accordance with the

[0076] For example, the continual learning configuration received by the UE 410 from gNB 414 may include a number of values of one or more CL configuration parameters, such as values for an adaptive ratio, periodicity, training period, etc. In some cases, the values of one or more of the CL configuration parameters may indicate (or may be associated with) an amount or percentage of UE run-time capabilities required to carry out or perform continual learning in accordance with the received continual learning configuration. For example, an adaptive ratio greater than 0.5 (e.g., indicating that 50% of the ML model layers, e.g., 5 layers of the layers of the ML model should be adapted or trained during the continual learning operation) may require a UE battery power level of at least 50%, whereas an adaptive ratio of 0.3 or less may only require only a battery power level of at least 35%. Thus, for example, if the CL configuration received by the UE 410 indicates, e.g., an adaptive ratio of 0.6 (indicating that 6 of the 10 layers of the ML model should be adapted as part of the continual learning operation), the UE would then compare its current battery power level to determine if it is greater than or equal to 50% battery power level. If the current UE battery power level is less than 50% (e.g., 42%), this indicates (in this example) that the UE is unable to complete or carry out the continual learning for the ML model in accordance with the received CL configuration (which requires adaptive ratio of 0.6). On the other hand, if the UE battery power level is greater than or equal to 50%, this indicates that the UE is able to carry out the continual learning for the ML model in accordance with the received CL configuration.

[0077] More complicated scenarios may exist, and / or more complicated or more detailed analysis may be performed by the UE to determine if the UE run-time capabilities may limit or prevent the UE from performing continual learning in accordance with the received continual learning configuration. For example, if the CL configuration indicates a periodicity of continual learning (indicating a period of or time gap between each continual learning operation or iteration for the ML model) of less than 100ms, a battery level of at least 50% and available processor resources of at least 34% are required. On the other hand, aperiodicity that is greater than or equal to 100ms may require a battery level of at least 40% and at least 28% available processor resources. Thus, if the CL configuration indicates, for example, a periodicity of 150 ms (which is more than 100 ms), the UE would confirm or determine that the UE has a battery level of at least 40% and has at least 28% available processor resources, before transmitting a response to the gNB 414 confirming that the UE can perform or carry out the configured continual learning for the ML model. If either of these levels of at least 40% battery power level and 28% available processor resources are not available at the UE, then the UE may transmit to the gNB run-time capabilities limitation information that may inform the gNB that the UE is unable to perform the requested continual learning in accordance with the CL configuration, and may indicate the run-time capability that is the cause for CL rejection, and / or may provide a proposal (suggestion or recommendation) for one or more of these CL configuration parameters for which the UE is able to perform continual learning. Thus, depending on the parameters of the received CL configuration, the UE may compare the values of one or more UE run-time capabilities limitations to one or more, or to different, thresholds, to determine if the UE is able to carry out the continual learning in accordance with the received CL configuration.

[0078] Flow of FIG. 4 proceeds to operations 3-6 if the UE determines that it is unable to perform continual learning in accordance with the CL configuration due to UE run-time capabilities limitations. At step 3 of FIG. 4, the UE determines that one or more UE run-time capabilities limitations will prevent the UE from performing continual learning in accordance with the received CL configuration. See above for various examples. For example, for UE 410, in this example, an adaptive ratio greater than 0.5 may require a UE battery power level of at least 50%. Thus, if the continual learning configuration indicates an adaptive ratio greater than 0.5 (e.g., 0.6, 0.7, 0.8) and the current UE battery power level is less than 50%, then the UE is unable to perform continual learning for the ML model according to the received continual learning configuration.

[0079] At step 4 of FIG. 4, the UE 410 transmits to gNB 414 a UE assistance information message (which may be transmitted as a radio resource control (RRC) message). The UE assistance information may include, for example, run-time capabilities limitation information related to an inability of the UE to carry out continual learning for the ML model, e.g., including at least one of the following: information describing or relating to the inability of the UE to carry out continual learning for the ML model in accordance with the continual learning configuration; an indication that UE is unable to carry out continual learning of the ML model in accordance with the continual learning configuration based on the run-timecapabilities limitations of the UE; an indication of one or more of the run-time capabilities limitations of the UE that prevents the UE from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; and / or one or more UE proposals (e.g., UE suggestions, UE recommendations or UE preferences) for one or more parameters (values of parameters) of the continual learning configuration; or one or more UE proposals for one or more parameters (values of parameters) associated with the continual learning of the machine learning model. For example, the received CL configuration may indicate an adaptive ratio of 0.6 and a periodicity of 100ms. However, for a current UE battery power level of 42% (which is greater than 35% and less than 50%), the UE can only perform continual learning for a CL configuration with adaptive ratio of 0.3 or less, and a periodicity greater than 100ms. Thus, in this example, based on the current 42% battery power level of the UE measured by the UE, the UE may transmit a message to gNB 414 that includes UE assistance information that includes, for example, a proposal of 0.3 for an adaptive ratio and a 150 ms for periodicity. The run-time capabilities limitation information transmitted by the UE may include, or may be transmitted via: information or one or more parameters provided within the UE assistance message or UE assistance information; or, one or more power save parameters or information of one or more parameters provided within a power saving information element.

[0080] At step 5 of FIG. 4, the gNB 414 receives the UE assistance information message from the UE that includes the run-time capabilities limitation information related to the inability of the UE to carry out continual learning for the ML model, e.g., which may include a UE proposal (e.g., suggestion, recommendation, preference by the UE) for one or more continual learning configuration parameters or other parameters. For example, the UE assistance information message may include a UE proposal of an adaptive ratio of 0.6 and a periodicity of 150ms (based on the current run-time capabilities limitations of the UE). The gNB 414 may determine, based on the run-time capabilities limitation information received by gNB 414 from UE 410, a continual learning configuration update, e.g., which may be or may include an indication to cancel the continual learning request, to use the existing configuration (or set of weights) for the ML model (without requiring continual learning), and / or one or more updated / adjusted parameters (parameter values) for the continual learning configuration. The gNB 414 may determine whether the proposed parameter values would provide sufficient retraining or adaptation of the ML model to provide acceptable performance and / or accuracy for the function that the ML model is performing or assisting with. If the values of parameters proposed by the UE 410 would provide sufficientperformance, then the gNB 414 may send an updated CL request including an updated continual learning configuration that confirms the proposed parameters are acceptable, or indicates the parameters that are acceptable, for example. As an example, if the UE has proposed an adaptive ratio of 0.3 and a periodicity of 150ms, the gNB may accept the proposed values of these parameters, or may transmit an updated continual learning configuration with one or more different parameters than what the UE has proposed. Thus, the gNB 414 may transmit an updated continual learning configuration with the parameters of 0.3 for adaptive ratio and 150ms for periodicity, if those proposed values are acceptable to the gNB. Or, if those proposed values are not acceptable to the gNB, the gNB 414 may transmit the updated continual learning configuration indicating, for example, an adaptive ratio of 0.35 and a periodicity of 120 ms.

[0081] At step 6 of FIG. 4, the gNB 414 transmits to UE 410 the updated continual learning request, e.g., including the updated continual learning configuration.

[0082] After step 2 of FIG. 4, flow of FIG. 4 proceeds to steps 7-8, if the UE determines that the UE is able to perform continual learning in accordance with the CL configuration due to UE run-time capabilities limitations (no run-time capabilities limitations detected by UE 410 that would prevent the UE 410 from performing continual learning in accordance with the received continual learning configuration. After detecting that there are no UE run-time capabilities limitations that would prevent continual learning by the UE as configured by the gNB 414, operation proceeds to step 8. At step 8, the UE 410 performs a continual learning operation of the ML model (or layers of the ML model) in accordance with the received continual learning configuration.

[0083] An example UE assistance information may include preferences (or UE proposed values / proposals) for one or more parameters related to the continual learning configuration, such as proposed values for adaptive ratio, power limit, complexity limit, delay budge, and training time limit.

[0084] UEAssistanceInformation-rl9-IEs ::= SEQUENCE { cl -Preference- Adati veRati o-r 19 CL-Preference- Adati veRati o-r 19OPTIONAL, cl-Preference-PowerLimit-r 19 CL-Preference-PowerLimit-r 19OPTIONAL, cl-Preference-ComplexityLimit-rl9 CL-Preference-ComplexityLimit-rl9 OPTIONAL, cl-Preference-DelayBudget-r 19 CL-Preference-DelayBudget-r 19OPTIONAL, cl-Preference-T rainingTimeLimit-r 19 CL-Preference-T rainingTimeLimit-r 19 OPTIONAL.

[0085] The following are some example parameters or fields that may included within the UE assistance information. The UE may propose or suggest values for one or more of these example fields or parameters.

[0097] In another embodiment, some of the continual learning configuration parameters that are battery or power related may be provided within or embedded into power saving parameters, e.g., added to a RRC PowSav-Parameters IE (information element) as follows:

[0098] PowSav-Parameters-CL-rl9 ::= SEQUENCE ) cl-Preference-AdativeRatio-rl9 ENUMERATED {supported} OPTIONAL, cl-Preference-PowerLimit-rl9 ENUMERATED {supported} OPTIONAL, cl-Preference-ComplexityLimit-rl9 ENUMERATED {supported} OPTIONAL, } CL-Preference-AdativeRatio-rl9::= SEQUENCE { cl-Preference-AdativeRatio-rl9 ENUMERATED {oDot2, oDot4, oDot6, oDot8} OPTIONAL } CL-Preference-PowerLimit-rl9::= SEQUENCE { cl-Preference-PowerLimit-rl9 INTEGER (-30. 33) OPTIONAL} CL-Preference-ComplexityLimit-rl9::= SEQUENCE { cl-Preference-ComplexityLimit-rl9 INTEGER (0..200) OPTIONAL}

[0099] Some examples will now be described, based on the description and figures provided herein. A number of examples will now be described, based on the description and figures provided herein.

[0100] Example 1. An apparatus (e.g., 1300, FIG. 5; and / or UE 410 in FIG. 4) comprising: at least one processor (e.g., processor 1304, FIG. 5); and at least one memory (e.g., memory 1306, FIG. 5) storing instructions that, when executed by the at least one processor (1304), cause the apparatus at least to:

[0101] 1) Receive (step 1, FIG. 4), by a user device (e.g., UE 410, FIG. 4) from a network node (e.g., gNB 414, FIG. 4), a continual learning (CL) request comprising a continual learning configuration for the user device to carry out continual learning for amachine learning model. For example, at step 1 of FIG. 4, UE 410 receives a continual learning request, including a continuation learning (CL) configuration; the CL configuration may include one or more parameters as part of the CL configuration; for example, the continual learning (CL) configuration may include a variety of parameters that may indicate parameters, conditions or requirements for the continuation learning (CL) for the ML model (412, FIG. 4), including, e.g., a ML model ID, an adaptive ratio for CL, an accuracy level for the ML model, a training period for the CL, a periodicity for the CL, . . . ;

[0102] 2) Determine (step 2, FIG. 4), by the user device (UE 410), run-time capabilities limitations of the user device. For example, at step 2 of FIG. 4, UE 410 may determine runtime capabilities limitations of the UE 410, and whether the UE 410 can perform or carry out continual learning (CL) in accordance with (or based on) the received continual learning configuration. UE run-time capabilities may include, e.g., a current state of one or more capabilities or resources of the UE, such as a current state or status of UE processor resources, memory resources, other hardware resources, software resources, network resources, protocol entity resources, power or battery resources, or other resources or capabilities, and / or a state of the UE related to overheating or whether the UE is in power saving mode (since the ability of the UE to operate fully may be impacted by the UE being overheated and / or the UE being placed in a power saving mode).

[0103] 3) Determine, by the user device (UE 410, FIG. 4), that the user device is unable to carry out continual learning for the machine learning model (ML model 412, FIG. 4) according to the received continual learning configuration based on the run-time capabilities limitations of the user device. For example, at step 3 of FIG. 4, the UE may detect one or more run-time capabilities limitations that prevent (e.g., inhibit or limit) the UE 410 from carrying out or performing the continual learning in accordance with or based on the received continual learning configuration. To determine whether the UE’s current run-time capabilities limitations may prevent the UE from performing the continual learning of the ML mode, the UE 410 may compare its current run-time capabilities to one or more thresholds. For example, the UE may compare its current available battery power, current available memory resources, current available processing resources, current overheating status or processor temperature, etc., to one or more thresholds to determine if the UE is able or unable to perform continual learning based on the received continual learning configuration. For example, if the UE has less than 30% available processing or computational resources, or less than 5 GB of available or free memory, or less than 35% available battery power, then (any of these) may indicate that the UE is unable to perform continual learning for the ML model.Or the UE may also determine if the UE is in a power saving mode, or whether a temperature of a UE processor or UE component is within a threshold number (e.g., 10) degrees of an overheating temperature. If the UE is in a power saving mode or if the UE is in an overheated state or overheating temperature (e.g., a temperature of the UE or a component of the UE is greater than the overheated / overheating temperature) or the UE has a processor or other component temperature within a threshold number of degrees to the overheating temperature, then this indicates that the UE is unable to perform continual learning for the ML model, based on one or more of these UE run-time capabilities limitations. For example, for UE 410 to perform continual learning for the ML model based on an adaptive ratio greater than 0.5 requires a UE battery power level of at least 50%. Thus, if the continual learning configuration indicates an adaptive ratio greater than 0.5 (e.g., 0.6, 0.7, 0.8) and the current UE battery power level is less than 50%, then the UE is unable to perform continual learning for the ML model according to the received continual learning configuration. The UE 410 may compare various run-time capabilities of the UE 410 to different thresholds, to determine if any of the UE run-time capabilities have limitations that would prevent the UE from performing continual learning in accordance with received CL configuration.

[0104] 4) Transmit, by the user device (UE 410, FIG. 4) to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device. For example, at step 4 of FIG. 4, the UE 410 transmits to gNB 414 a UE assistance information message (which may be transmitted as a radio resource control (RRC) message). The UE assistance information may include, for example, run-time capabilities limitation information related to an inability of the UE to carry out continual learning for the ML model, e.g., including at least one of the following: information describing or relating to the inability of the UE to carry out continual learning for the ML model in accordance with the continual learning configuration; an indication that UE is unable to carry out continual learning of the ML model in accordance with the continual learning configuration based on the run-time capabilities limitations of the UE; an indication of one or more of the run-time capabilities limitations of the UE that prevents the UE from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; and / or one or more UE proposals (e.g., UE suggestions, UE recommendations or UE preferences) for one or more parameters (values of parameters) of the continual learning configuration; or one or more UE proposals for one or more parameters (values of parameters) associated with thecontinual learning of the machine learning model. For example, the received CL configuration may indicate an adaptive ratio of 0.6 and a periodicity of 100ms. However, for a current UE battery power level of 42% (which is greater than 35% and less than 50%), the UE can only perform continual learning for a CL configuration with adaptive ratio of 0.3 or less, and a periodicity greater than 100ms. Thus, in this example, based on the current 42% battery power level of the UE measured by the UE, the UE may transmit a message to gNB 414 that includes UE assistance information that includes, for example, a proposal of 0.3 for an adaptive ratio and a 150 ms for periodicity.

[0105] 5) Receive, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node. For example, at step 6 of FIG. 4, the UE 410 receives from gNB 414 an updated continual learning request, e.g., including an updated continual learning configuration. For example, the gNB 414 may determine, based on the run-time capabilities limitation information received by gNB 414 from UE 410, a continual learning configuration update, e.g., which may be or may include an indication to cancel the continual learning request, to use the existing configuration (or set of weights) for the ML model (without requiring continual learning), and / or one or more updated / adjusted parameters (parameter values) for the continual learning configuration. For example the gNB 414 may send an updated CL request including an updated continual learning configuration that confirms the proposed parameters (proposed by the UE 410) are acceptable, or indicates the parameters that are acceptable, for example. As an example, if the UE has proposed an adaptive ratio of 0.3 and a periodicity of 150ms, the gNB may accept the proposed values of these parameters, or may transmit to UE 410 (and UE 410 may receive) an updated continual learning configuration with one or more different parameters than what the UE has proposed.

[0106] Example 2. The apparatus of example 1, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

[0107] Example 3. The apparatus of any of examples 1-2, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memoryresources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

[0108] Example 4. The apparatus of any of examples 1-3, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

[0109] Example 5. The apparatus of any of examples 1-4, wherein the received continual learning configuration update received from the network node comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

[0110] Example 6. The apparatus of any of examples 1-5, wherein the run-time capabilities limitation information transmitted by the user device to the network node comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for atraining time limitation for continual learning of the machine learning model.

[0111] Example 7. The apparatus of any of examples 1-6, wherein the run-time capabilities limitation information is transmitted as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

[0112] Example 8. A method (see, e.g., flow chart of FIG. 2) comprising: receiving (210, FIG. 2, by a user device (UE 410, FIG. 4) from a network node (gNB 414, FIG. 4), a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; determining (220, FIG. 2), by the user device, run-time capabilities limitations of the user device; determining (230, FIG. 2), by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmitting (240, FIG. 2), by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receiving (250, FIG. 2), by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

[0113] Example 9. The method of example 8, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

[0114] Example 10. The method of any of examples 8-9, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

[0115] Example 11. The method of any of examples 8-10, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

[0116] Example 12. The method of any of examples 8-11, wherein the received continual learning configuration update received from the network node comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

[0117] Example 13. The method of any of examples 8-12, wherein the run-time capabilities limitation information transmitted by the user device to the network node comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

[0118] Example 14. The method of any of examples 8-13, wherein the run-time capabilities limitation information is transmitted as or via at least one of the following:information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

[0119] Example 15. An apparatus (e.g., 1300, FIG. 5, gNB 414, FIG. 4) comprising: at least one processor (e.g., processor 1304, FIG. 5); and at least one memory (e.g., memory 1306, FIG. 5) storing instructions that, when executed by the at least one processor (1304), cause the apparatus at least to:

[0120] 1) Transmit, by a network node (gNB 414, FIG. 4) to a user device (UE 410, FIG.4), a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model. For example, see step 1, FIG. 4, and related description.

[0121] 2) Receive, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device. For example, see step 4 of FIG. 4 and related description.

[0122] 3) Determine, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update. For example, see step 5 of FIG. 4, and related description.

[0123] 4) Transmit, by the network node to the user device, the continual learning configuration update. For example, see step 6 of FIG. 4 and related description.

[0124] Example 16. The apparatus of example 15, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

[0125] Example 17. The apparatus of any of examples 15-16, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, oravailable battery resources of the user device; or an overheating status of the user device.

[0126] Example 18. The apparatus of any of examples 15-17, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

[0127] Example 19. The apparatus of any of examples 15-18, wherein the continual learning configuration update comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

[0128] Example 20. The apparatus of any of examples 15-19, wherein the run-time capabilities limitation information received by the network node from the user device comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

[0129] Example 21. The apparatus of any of examples 15-20, wherein the run-timecapabilities limitation information is received as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

[0130] Example 22. A method (e.g., see flow chart of FIG. 3) comprising: transmitting (310, FIG. 3), by a network node (e.g., gNB 414, FIG. 4) to a user device (UE 410, FIG. 4), a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; receiving (320), by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; determining (330), by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and transmitting (340, FIG. 3), by the network node to the user device, the continual learning configuration update.

[0131] FIG. 5 is a block diagram of a wireless station or node (e.g., UE, user device, AP, BS, eNB, gNB, RAN node, network node, TRP, or other node) 1300 according to an example embodiment. The wireless station 1300 may include, for example, one or more (e.g., two as shown in FIG. 5) RF (radio frequency) or wireless transceivers 1302 A, 1302B, where each wireless transceiver includes a transmitter to transmit signals and a receiver to receive signals. The wireless station also includes a processor or control unit / entity (controller) 1304 to execute instructions or software and control transmission and receptions of signals, and a memory 1306 to store data and / or instructions.

[0132] Processor 1304 may also make decisions or determinations, generate frames, packets or messages for transmission, decode received frames or messages for further processing, and other tasks or functions described herein. Processor 1304, which may be a baseband processor, for example, may generate messages, packets, frames or other signals for transmission via wireless transceiver 1302 (1302A or 1302B). Processor 1304 may control transmission of signals or messages over a wireless network, and may control the reception of signals or messages, etc., via a wireless network (e.g., after being down-converted by wireless transceiver 1302, for example). Processor 1304 may be programmable and capable of executing software or other instructions stored in memory or on other computer media to perform the various tasks and functions described above, such as one or more of the tasks or methods described above. Processor 1304 may be (or may include), for example, hardware, programmable logic, a programmable processor that executes software or firmware, and / orany combination of these. Using other terminology, processor 1304 and transceiver 1302 together may be considered as a wireless transmitter / receiver system, for example.

[0133] In addition, referring to FIG. 5, a controller (or processor) 1308 may execute software and instructions, and may provide overall control for the station 1300, and may provide control for other systems not shown in FIG. 5, such as controlling input / output devices (e.g., display, keypad), and / or may execute software for one or more applications that may be provided on wireless station 1300, such as, for example, an email program, audio / video applications, a word processor, a Voice over IP application, or other application or software.

[0134] In addition, a storage medium may be provided that includes stored instructions, which when executed by a controller or processor may result in the processor 1304, or other controller or processor, performing one or more of the functions or tasks described above.

[0135] According to another example embodiment, RF or wireless transceiver(s) 1302A / 1302B may receive signals or data and / or transmit or send signals or data. Processor 1304 (and possibly transceivers 1302A / 1302B) may control the RF or wireless transceiver 1302 A or 1302B to receive, send, broadcast or transmit signals or data.

[0136] Example embodiments are provided or described for each of the example methods, including: An apparatus (e.g., 1300, FIG. 5) including means (e.g., processor 1304, RF transceivers 1302A and / or 1302B, and / or memory 1306, in FIG. 5) for carrying out any of the methods; a non-transitory computer-readable storage medium (e.g., memory 1306, FIG.5) comprising instructions stored thereon that, when executed by at least one processor (processor 1304, FIG. 5), are configured to cause a computing system (e.g., 1300, FIG. 5) to perform any of the example methods; and an apparatus (e.g., 1300, FIG. 5) including at least one processor (e.g., processor 1304, FIG. 5), and at least one memory (e.g., memory 1306, FIG. 5) including computer program code, the at least one memory (1306) and the computer program code configured to, with the at least one processor (1304), cause the apparatus (e.g., 1300) at least to perform any of the example methods.

[0137] Embodiments of the various techniques described herein may be implemented in digital electronic circuitry, or in computer hardware, firmware, software, or in combinations of them. Embodiments may be implemented as a computer program product, i.e., a computer program tangibly embodied in an information carrier, e.g., in a machine-readable storage device or in a propagated signal, for execution by, or to control the operation of, a data processing apparatus, e.g., a programmable processor, a computer, or multiple computers.Embodiments may also be provided on a computer readable medium or computer readable storage medium, which may be a non-transitory medium. Embodiments of the various techniques may also include embodiments provided via transitory signals or media, and / or programs and / or software embodiments that are downloadable via the Internet or other network(s), either wired networks and / or wireless networks. In addition, embodiments may be provided via machine type communications (MTC), and also via an Internet of Things (IOT).

[0138] As used in this application, the term ‘circuitry’ or “circuit” refers to all of the following: (a) hardware-only circuit implementations, such as implementations in only analog and / or digital circuitry, and (b) combinations of circuits and soft-ware (and / or firmware), such as (as applicable): (i) a combination of processor(s) or (ii) portions of processor(s) / software including digital signal processor(s), software, and memory(ies) that work together to cause an apparatus to perform various functions, and (c) circuits, such as a microprocessor s) or a portion of a microprocessor(s), that require software or firmware for operation, even if the software or firmware is not physically present. This definition of ‘circuitry’ applies to all uses of this term in this application. As a further example, as used in this application, the term ‘circuitry’ would also cover an implementation of merely a processor (or multiple processors) or a portion of a processor and its (or their) accompanying software and / or firmware. The term ‘circuitry’ would also cover, for example and if applicable to the particular element, a baseband integrated circuit or applications processor integrated circuit for a mobile phone or a similar integrated circuit in a server, a cellular network device, or another network device.

[0139] The computer program may be in source code form, object code form, or in some intermediate form, and it may be stored in some sort of carrier, distribution medium, or computer readable medium, which may be any entity or device capable of carrying the program. Such carriers include a record medium, computer memory, read-only memory, photoelectrical and / or electrical carrier signal, telecommunications signal, and software distribution package, for example. Depending on the processing power needed, the computer program may be executed in a single electronic digital computer, or it may be distributed amongst a number of computers.

[0140] Furthermore, embodiments of the various techniques described herein may use a cyber-physical system (CPS) (a system of collaborating computational elements controlling physical entities). CPS may enable the embodiment and exploitation of massive amounts of interconnected ICT devices (sensors, actuators, processors microcontrollers,...) embedded inphysical objects at different locations. Mobile cyber physical systems, in which the physical system in question has inherent mobility, are a subcategory of cyber-physical systems. Examples of mobile physical systems include mobile robotics and electronics transported by humans or animals. The rise in popularity of smartphones has increased interest in the area of mobile cyber-physical systems. Therefore, various embodiments of techniques described herein may be provided via one or more of these technologies.

[0141] A computer program, such as the computer program(s) described above, can be written in any form of programming language, including compiled or interpreted languages, and can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit or part of it suitable for use in a computing environment. A computer program can be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network.

[0142] Method steps may be performed by one or more programmable processors executing a computer program or computer program portions to perform functions by operating on input data and generating output. Method steps also may be performed by, and an apparatus may be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit).

[0143] Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer, chip or chipset. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. Elements of a computer may include at least one processor for executing instructions and one or more memory devices for storing instructions and data. Generally, a computer also may include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks. Information carriers suitable for embodying computer program instructions and data include all forms of non-volatile memory, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory may be supplemented by, or incorporated in, special purpose logic circuitry.

[0144] To provide for interaction with a user, embodiments may be implemented on a computer having a display device, e.g., a cathode ray tube (CRT) or liquid crystal display(LCD) monitor, for displaying information to the user and a user interface, such as a keyboard and a pointing device, e.g., a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0145] Embodiments may be implemented in a computing system that includes a back-end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an embodiment, or any combination of such back-end, middleware, or front-end components. Components may be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (LAN) and a wide area network (WAN), e.g., the Internet.

[0146] While certain features of the described embodiments have been illustrated as described herein, many modifications, substitutions, changes and equivalents will now occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such modifications and changes as fall within the true spirit of the various embodiments.

Claims

WHAT IS CLAIMED IS:

1. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: receive, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; determine, by the user device, run-time capabilities limitations of the user device; determine, by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmit, by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receive, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

2. The apparatus of claim 1, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

3. The apparatus of any of claims 1-2, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels;memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

4. The apparatus of any of claims 1-3, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

5. The apparatus of any of claims 1-4, wherein the received continual learning configuration update received from the network node comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

6. The apparatus of any of claims 1-5, wherein the run-time capabilities limitation information transmitted by the user device to the network node comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

7. The apparatus of any of claims 1-6, wherein the run-time capabilities limitation information is transmitted as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

8. A method comprising: receiving, by a user device from a network node, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; determining, by the user device, run-time capabilities limitations of the user device; determining, by the user device, that the user device is unable to carry out continual learning for the machine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; transmitting, by the user device to the network node, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for themachine learning model according to the received continual learning configuration based on the run-time capabilities limitations of the user device; and receiving, by the user device from the network node, a continual learning configuration update, based on the run-time capabilities limitation information transmitted to the network node.

9. The method of claim 8, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

10. The method of any of claims 8-9, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

11. The method of any of claims 8-10, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration;an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

12. The method of any of claims 8-11, wherein the received continual learning configuration update received from the network node comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node; an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

13. The method of any of claims 8-12, wherein the run-time capabilities limitation information transmitted by the user device to the network node comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; ora proposal for a training time limitation for continual learning of the machine learning model.

14. The method of any of claims 8-13, wherein the run-time capabilities limitation information is transmitted as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

15. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: transmit, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; receive, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; determine, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and transmit, by the network node to the user device, the continual learning configuration update.

16. The apparatus of claim 15, wherein the continual learning configuration comprises at least one of the following: an adaptive ratio of layers, or an amount or percentage of layers of the machine learning model to be adapted; an accuracy level for the machine learning model; a training period for continual learning of the machine learning model; or a periodicity for continual learning of the machine learning model.

17. The apparatus of any of claims 15-16, wherein the run-time capabilities limitations of the user device reflect a current state of the user device with respect to one or more limitations of the user device, comprising at least one of the following: processing limitations of the user device based on available processing resources or based on current computation levels; memory limitations of the user device based on available memory resources or based on memory resources that are currently in use; whether a power saving mode has been activated for the user device; a battery or power state of the user device, or available battery resources of the user device; or an overheating status of the user device.

18. The apparatus of any of claims 15-17, wherein the run-time capabilities limitation information related to an inability of the user device to carry out continual learning for the machine learning model comprises at least one of the following: information describing or relating to the inability of the user device to carry out continual learning for the machine learning model in accordance with the continual learning configuration; an indication that the user device is unable to carry out continual learning of the machine learning model in accordance with the continual learning configuration based on the run-time capabilities limitations of the user device; an indication of one or more of the run-time capabilities limitations of the user device that prevents the user device from carrying out continual learning of the machine learning model in accordance with the continual learning configuration; one or more user device proposals for one or more parameters of the continual learning configuration; or one or more user device proposals for one or more parameters associated with the continual learning of the machine learning model.

19. The apparatus of any of claims 15-18, wherein the continual learning configuration update comprises at least one of the following: an updated continual learning configuration that includes at least one parameter that has been changed or updated based on the run-time capabilities limitation information transmitted by the user device to the network node;an indication by the network node to cancel the continual learning request; and / or an indication by the network node for the user device to use a current version of the machine learning model.

20. The apparatus of any of claims 15-19, wherein the run-time capabilities limitation information received by the network node from the user device comprises one or more user device proposals for one or more parameters of the continual learning configuration or one or more parameters associated with the continual learning of the machine learning model, comprising one or more of the following: a proposal for an adaptive ratio of layers, or a preferred amount or percentage of layers of the machine learning model to be adapted; a proposal for an accuracy level for the machine learning model; a proposal for a power limitation for continual learning of the machine learning model; a proposal for a periodicity for continual learning of the machine learning model; a proposal for a complexity limitation for continual learning of the machine learning model; a proposal for a delay budget for continual learning of the machine learning model; or a proposal for a training time limitation for continual learning of the machine learning model.

21. The apparatus of any of claims 15-20, wherein the run-time capabilities limitation information is received as or via at least one of the following: information or one or more parameters provided within a user device assistance message; or one or more power save parameters or information or one or more parameters provided within a power saving information element.

22. A method comprising: transmitting, by a network node to a user device, a continual learning request comprising a continual learning configuration for the user device to carry out continual learning for a machine learning model; receiving, by the network node from the user device, run-time capabilities limitation information related to an inability of the user device to carry out continual learning for themachine learning model according to the continual learning configuration based on run-time capabilities limitations of the user device; determining, by the network node based on the received run-time capabilities limitation information, a continual learning configuration update; and transmitting, by the network node to the user device, the continual learning configuration update.

Images