Optimization and reporting of user terminal runtime capabilities for continuous learning operations

Abstract

A method includes: a user equipment receiving a continuous learning request from a network node, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning against a machine learning model; determining runtime capability limitations of the user equipment; determining, based on the runtime capability limitations of the user equipment, that the user equipment cannot perform continuous learning against the machine learning model according to the received continuous learning configuration; transmitting runtime capability limitation information to the network node, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning against the machine learning model according to the received continuous learning configuration based on the runtime capability limitation information transmitted to the network node; and the user equipment receiving a continuous learning configuration update from the network node based on the runtime capability limitation information transmitted to the network node.

Description

Technical Field

[0001] This manual relates to wireless communication. Background Technology

[0002] A communication system can 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 carrier waves.

[0003] An example of a cellular communication system is the architecture standardized by the 3rd Generation Partnership Project (3GPP). Recent developments in this field are often referred to as Long Term Evolution (LTE) of Universal Mobile Telecommunications System (UMTS) radio access technology. E-UTRA (Evolved UMTS Terrestrial Radio Access) is the air interface for the 3GPP LTE upgrade path for mobile networks. In LTE, base stations or access points (APs), referred to as enhanced node APs (eNBs), provide radio access within a coverage area or cell. In LTE, mobile devices or mobile stations are referred to as user equipment (UEs). LTE has incorporated numerous improvements and developments. All aspects of LTE continue to improve.

[0004] The development of 5G New Radio (NR) is part of an ongoing evolution of mobile broadband to meet the requirements of 5G, similar to the early evolution of 3G and 4G wireless networks. Furthermore, in addition to mobile broadband, 5G also targets emerging use cases. The goal of 5G is to deliver significant improvements in wireless performance, which can include new levels of data rates, latency, reliability, and security. 5G NR can also be extended to efficiently connect massive Internet of Things (IoT) networks and can provide new types of mission-critical services. For example, ultra-reliable and low-latency communication (URLLC) devices may require high reliability and very low latency. 6G and other networks are also under development. Summary of the Invention

[0005] A method may include: a user equipment receiving a continuous learning request from a network node, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning against a machine learning model; the user equipment determining runtime capability limitations of the user equipment; the user equipment determining, based on the runtime capability limitations, that the user equipment cannot perform continuous learning against the machine learning model according to the received continuous learning configuration; the user equipment transmitting runtime capability limitation information to the network node, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning against the machine learning model according to the received continuous learning configuration based on the user equipment's runtime capability limitations; and the user equipment receiving a continuous learning configuration update from the network node based on the runtime capability limitation information transmitted to the network node.

[0006] An apparatus may include: components for receiving a continuous learning request from a network node by a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; components for determining runtime capability limitations of the user equipment; components for determining, based on the user equipment's runtime capability limitations, that the user equipment cannot perform continuous learning for the machine learning model according to the received continuous learning configuration; components for transmitting runtime capability limitation information from the user equipment to the network node, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for the machine learning model according to the received continuous learning configuration based on the user equipment's runtime capability limitations; and components for receiving a continuous learning configuration update from the network node based on the runtime capability 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 are configured, together with the at least one processor, to cause the apparatus to at least: receive a continuous learning request from a network node by a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; determine runtime capability limitations of the user equipment; determine, based on the runtime capability limitations of the user equipment, that the user equipment cannot perform continuous learning for a machine learning model according to the received continuous learning configuration; transmit runtime capability limitation information to the network node, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for a machine learning model according to the received continuous learning configuration based on the runtime capability limitations of the user equipment; and receive a continuous learning configuration update from the network node based on the runtime capability limitation information transmitted to the network node.

[0008] One method may include: a network node transmitting a continuous learning request to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning against a machine learning model; the network node receiving runtime capability limitation information from the user equipment, the runtime capability limitation information relating to runtime capability limitations of the user equipment that prevent the user equipment from performing continuous learning against the machine learning model according to the continuous learning configuration; the network node determining a continuous learning configuration update based on the received runtime capability limitation information; and the network node transmitting the continuous learning configuration update to the user equipment.

[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 are configured, together with the at least one processor, to cause the apparatus to at least: transmit a continuous learning request from a network node to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning against a machine learning model; receive runtime capability limitation information from the user equipment by the network node, the runtime capability limitation information relating to runtime capability limitations of the user equipment that prevent the user equipment from performing continuous learning against the machine learning model according to the continuous learning configuration; determine a continuous learning configuration update by the network node based on the received runtime capability limitation information; and transmit the continuous learning configuration update to the user equipment by the network node.

[0010] An apparatus may include: components for transmitting a continuous learning request from a network node to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; components for receiving runtime capability limitation information from the user equipment by the network node, the runtime capability limitation information relating to runtime capability limitations of the user equipment that prevent the user equipment from performing continuous learning for the machine learning model according to the continuous learning configuration; components for determining a continuous learning configuration update by the network node based on the received runtime capability limitation information; and components for transmitting the continuous learning configuration update from the network node to the user equipment.

[0011] Other example implementations are provided or described for each example method, including: components for performing any example method; a non-transitory computer-readable storage medium including instructions stored thereon, which, when executed by at least one processor, are configured to cause a computing system to perform any example method; and means including at least one processor and at least one memory including computer program code, which, together with the at least one processor, causes the means to at least perform any example method.

[0012] Details of one or more examples of the embodiments are set forth in the accompanying drawings and the following description. Other features will be apparent from the specification, the drawings, and the claims. Attached Figure Description

[0013] Figure 1 This is a block diagram of a wireless network.

[0014] Figure 2 This is a flowchart illustrating the operation of a user equipment (e.g., a UE).

[0015] Figure 3 This is a flowchart illustrating the operation of a network node (e.g., a gNB).

[0016] Figure 4 This is a diagram illustrating the operation of the system.

[0017] Figure 5 It is a block diagram of a wireless station or node (e.g., a network node (such as a gNB), a user node or UE, a relay node or other node). Detailed Implementation

[0018] Figure 1 This is a block diagram of wireless network 130. Figure 1 In the wireless network 130, user equipment 131, 132, 133, and 135 (which may also be referred to as mobile stations (MS) or user equipment (UE)) can connect to (and communicate with) a base station (BS) 134 (which may also be referred to as an access point (AP), enhanced Node B (eNB), gNB, or network node). The terms user equipment and user equipment (UE) are used interchangeably. A BS may also include or 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., in the case of a split BS or a split gNB, such as a centralized unit (CU) and / or a distributed unit (DU)). At least a portion of the functionality of a BS (e.g., an access point (AP), base station (BS), or (e) Node B (eNB), gNB, RAN node) can also be performed by any node, server, or host operatively coupled to a transceiver (such as a remote wireless head). BS (or AP) 134 provides wireless coverage within cell 136, including coverage to user equipment (or UEs) 131, 132, 133, and 135. Although only four user equipment (or UEs) are shown connected to or attached to BS 134, any number of user equipment can be provided. BS 134 is also connected to core network 150 via S1 interface 151. This is just a simplified example of a wireless network, and other examples can 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 RAN node) can be or may include (or may alternatively be referred to as) such as an access point (AP), gNB, eNB, or a portion thereof (such as a centralized unit (CU) and / or distributed unit (DU) in the case of splitting a BS or gNB) or other network nodes.

[0020] According to the illustrative example, a BS node (e.g., BS, eNB, gNB, CU / DU…) or radio access network (RAN) can be part of a mobile telecommunications system. The RAN (radio access network) can include one or more BS or RAN nodes implementing radio access technologies, for example, to allow one or more UEs to access the network or core network. Therefore, for example, the RAN (RAN node, such as BS or gNB) can reside between one or more user equipment or UEs and the core network. According to the example implementation, each RAN node (e.g., BS, eNB, gNB, CU / DU…) or BS can provide one or more wireless communication services for one or more UEs or user equipments, for example, to allow the UE to wirelessly access the network via the RAN node. Each RAN node or BS can perform or provide wireless communication services, such as allowing the UE or user equipment to establish a wireless connection to the RAN node, and sending data to one or more UEs and / or receiving data from one or more UEs. For example, after establishing a connection to the UE, the RAN node or network node (e.g., BS, eNB, gNB, CU / DU…) can forward data received from the network or core network to the UE, 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, etc.) can perform a wide variety of other radio functions or services, such as broadcasting control information to UEs (e.g., system information or on-demand system information), paging UEs when data to be delivered to them is available, assisting UEs in handover between cells, scheduling resources for uplink data transmission from UEs and downlink data transmission to UEs, and sending control information to configure one or more UEs. These are just a few examples of one or more functions that a RAN node or BS can perform.

[0021] User equipment or user node (user terminal, user equipment (UE), mobile terminal, handheld wireless device, etc.) can refer to portable computing devices that operate with or without a subscriber identification module (SIM), including but not limited to the following types of devices: mobile station (MS), mobile phone, cellular phone, smartphone, personal digital assistant (PDA), cell phone, device using a wireless modem (alarm or measuring device, etc.), laptop and / or touchscreen computer, tablet computer, tablet phone, game console, laptop, vehicle, sensor and multimedia device, as an example, or any other wireless device. It should be understood that user equipment can also be (or may include) a virtually exclusive uplink-only device, an example of which is a camera or camcorder that uploads images or video clips to the network. Furthermore, user node can include user equipment (UE), user equipment, user terminal, mobile terminal, mobile station, mobile node, subscriber equipment, subscriber node, subscriber terminal, or other user node. For example, a user node can be used to wirelessly communicate 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), the core network 150 may be referred to as the Evolved Packet Core (EPC), which may include a Mobility Management Entity (MME) that can handle or assist user equipment mobility / handover between BSs, one or more gateways that can forward data and control signals between the BS and a packet data network 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] Furthermore, the technologies described in this paper can be applied to various types of user equipment or data service types, or to user equipment that can have multiple applications running on it, which can be different data service types. New 5G (NR) development can support a variety of different applications or data service types, such as: Machine-Type Communication (MTC), Enhanced Machine-Type Communication (eMTC), Internet of Things (IoT) and / or Narrowband IoT user equipment, Enhanced Mobile Broadband (eMBB), and Ultra-Reliable and Low-Latency Communication (URLLC). Many of these new 5G (NR) related applications often require higher performance than previous wireless networks.

[0023] The Internet of Things (IoT) can refer to a growing group of objects that possess internet or network connectivity, enabling them to send and receive information from other network devices. For example, many sensor-type applications or devices can monitor physical conditions or states and, for instance, send reports to servers or other network devices when events occur. Machine-type communication (MTC or machine-to-machine communication) can be characterized, for example, by the fully automated generation, exchange, processing, and actuation of data between intelligent machines with or without human intervention. Enhanced Mobile Broadband (eMBB) can support data rates significantly higher than those currently available in LTE.

[0024] Ultra-Reliable and Low-Latency Communication (URLLC) is a new type of data service or a new use case that can be supported for new radio (5G) systems. This enables emerging new applications and services such as industrial automation, autonomous driving, vehicle safety, and eHealth services. As an illustrative example, 3GPP aims to provide reliable connections with a block error rate (BLER) of 10⁻⁵ and U-plane (user / data plane) latency of up to 1 ms. Therefore, for example, URLLC user equipment / UEs may require significantly lower block error rates and low latency (with or without the need for high reliability) than other types of user equipment / UEs. Thus, for example, a URLLC UE (or URLLC applications on a UE) may require much shorter latency compared to an eMBB UE (or an eMBB application running on a UE).

[0025] The techniques described in this article can be applied to a wide variety of wireless technologies or wireless networks, such as 5G (New Radio (NR)), cmWave and / or millimeter-wave band networks, IoT, MTC, eMTC, eMBB, URLLC, 6G, and any other wireless network or wireless technology. These example networks, technologies, or data service types are provided as illustrative examples only.

[0026] Machine learning (ML) models can be used within a wireless network to perform (or assist in performing) one or more tasks. Typically, one or more nodes within a wireless network (e.g., BS, gNB, eNB, RAN node, user node, UE, user equipment, relay node, or other wireless node) can use or employ ML models, such as neural network models (e.g., which may be referred to as neural networks, artificial intelligence (AI) neural networks, AI neural network models, AI models, machine learning (ML) models or algorithms, models, or other terms) to perform or assist in performing one or more ML-enabled tasks. Other types of models may also be used. ML-enabled tasks can include tasks that can be performed (or assisted in performing) by an ML model, or tasks that an ML model has been trained to perform or assist in performing.

[0027] ML-based algorithms or models can be used to perform and / or assist in performing various radio and / or radio resource management (RRM) and / or RAN-related functions or tasks to improve network performance, such as beam prediction (e.g., predicting the optimal beam or optimal beam pair based on a measured reference signal), antenna panel or beam control, RRM (Radio Resource Measurement) measurements and feedback (Channel State Information (CSI) feedback), link monitoring, Transmit Power Control (TPC), etc., in UEs. In some cases, ML models can be used to improve the performance of wireless networks in one or more aspects, or as measured by one or more performance metrics or standards.

[0028] A model (e.g., a neural network or ML model) can be or can include a computational model consisting of nodes organized in layers, as used in machine learning, for example. Nodes, also called artificial neurons, or simply neurons, perform a function on a given input to produce some output value. Neural network or ML models typically require training periods to learn parameters, or weights, used to map inputs to desired outputs. This mapping can occur via a function learned from given data about the problem in question. Therefore, weights are the weights of the mapping function of the neural network. Each neural network model or ML model can be trained for a specific task.

[0029] To provide an output given an input, a neural network model or the ML function of an ML model should be trained. This may involve learning appropriate values ​​for a large number of parameters (e.g., weights and / or biases) of a mapping function (or the ML function of an ML model). For example, parameters can be used to weight and / or adjust terms in the mapping function. This training can be an iterative process in which the values ​​of the weights and / or biases are adjusted over many (e.g., tens, hundreds, and / or thousands) training rounds or iterations until optimal or most accurate values ​​(or weights and / or biases) are reached. In the context of a neural network (neural network model) or ML model, the parameters can typically be initialized with random values, and the trainer iteratively updates the parameters (e.g., weights) of the neural network to minimize the error in the mapping function. In other words, during each round or step of iterative training, the network updates the values ​​of its parameters such that the values ​​eventually converge to optimal values.

[0030] As an example, ML models can be trained in a supervised or unsupervised manner. In supervised learning, training examples are provided to the ML model or other machine learning algorithm. Training examples consist of inputs and the expected or previously observed output. Training examples are also called labeled data because the inputs are labeled with the expected or observed output. In the case of neural networks (which can be specific to ML models), the network (or ML model) learns the values ​​of the weights used in the ML model's mapping function or ML function that most frequently produces the expected output given the training inputs. In unsupervised training, the ML model learns to recognize structures or patterns in the provided inputs. In other words, the model recognizes implicit relationships in the data. Unsupervised learning is used for many machine learning problems and often requires large amounts of unlabeled data.

[0031] Based on the example implementation, ML models can be classified (or can be classified) into two broad categories (supervised and unsupervised), depending on the presence of learning "signals" or "feedback" available for the model. Thus, for example, within the field of machine learning, there may be two main types of model learning or training: supervised and unsupervised. The main difference between the two types is that supervised learning is accomplished using known or prior knowledge of what the output values ​​of certain samples of the data should be. Therefore, the goal of supervised learning can be to learn a function that best approximates the relationship between observable inputs and outputs in the data, given a sample of data and a desired output. On the other hand, unsupervised learning has no labeled outputs, so its goal is to infer the natural structure existing within a set of data points.

[0032] Supervised learning: A computer is presented with example inputs and their expected outputs, and the goal can be to learn general rules that map inputs to outputs. Supervised learning can be performed, for example, in the context of classification (where a computer or learning algorithm attempts to map inputs to output labels) or regression (where a computer or algorithm can map inputs to continuous outputs). Common algorithms in supervised learning can include, for example, logistic regression, Naive Bayes, support vector machines, artificial neural networks, and random forests. In regression and classification, the goal can include finding specific relationships or structures in the input data that allow us to efficiently produce correct output data. In some example cases, the input signal may be only partially available or limited to specific feedback. Semi-supervised learning: The computer may be given only incomplete training signals; some (often many) of the target outputs are missing from the training set. Active learning: The computer can only obtain training labels from a limited set of instances (based on a budget) and can also optimize its selection of objects for obtaining labels. These can be presented to the user for labeling when used interactively.

[0033] Unsupervised learning: No labels are given to the learning algorithm, allowing it to find structure on its own from its input. Some example tasks within unsupervised learning can include clustering, representation learning, and density estimation. In these cases, the computer or learning algorithm attempts to learn the inherent structure of the data without using explicitly provided labels. Some common algorithms include k-means clustering, principal component analysis, and autoencoders. Because no labels are provided, there may not be a specific way to compare model performance in most unsupervised learning methods.

[0034] Continuous learning (CL) can refer to or include the ability of an ML model to adapt to a constantly changing (or continuously changing or periodically changing) surrounding environment or data by continuously learning or adapting the ML model based on incoming data (or new or updated data), for example, without forgetting original or previous knowledge or ML model settings, and, for example, it can be based on less than a complete or full dataset. For example, given a (e.g., possibly infinite or continuous) stream of data (e.g., data reflecting changes or updated conditions or environments that should cause the ML model to be updated), a continuous learning (CL) algorithm can (or should) learn, for example, by updating or adapting the weights or other parameters of the ML model based on partial experience or a sequence of partial data (e.g., a recent dataset), where all data may not be available at once because new or updated data will be received later (therefore, the new data may make the weights or parameter settings of the ML model outdated or inaccurate). Therefore, a complete or full dataset may not be considered available when the ML model is updated or adapted, because the data or environment may change continuously or over time. Therefore, at any given point or moment, the data (on which the ML model can be updated or adapted) can be considered incomplete because there may be a continuous stream of data. Thus, a CL algorithm can include or may refer to iteratively updating or adapting the weights or other parameters of the ML model based on an updated dataset, and then repeating the learning or adaptation process for the ML model when a second (or later) updated dataset is subsequently received.

[0035] Therefore, for example, in the case of training an ML model for task A (e.g., performing learning or adaptation on the weights or other parameters of the ML model): In this case, due to continuous contextual changes (e.g., changes in the environment or a continuous stream of updates or new data reflecting the constantly changing current environment or context), in time... The ML model trained for task A in time This will not be suitable. Therefore, continuous adaptation (or continuous learning) of ML models can be very useful, and even necessary in some cases, to ensure the required performance accuracy of the considered function performed or assisted by the ML model.

[0036] However, in some cases, limitations in the runtime capabilities of the UE may restrict or inhibit the UE's continuous learning of the ML model. The runtime capabilities of the UE may include, for example, any current capabilities or resources of the UE based on its current state as a result of its operation or running and / or based on the UE's current resource usage or the UE's state.

[0037] Figure 2 This is a flowchart illustrating the operations of a user equipment (or UE). Operation 210 includes the user equipment (or UE) receiving a continuous learning request from a network node (e.g., a gNB or other network node), the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning (ML) model. Operation 220 includes the user equipment (or UE) determining runtime capability limitations of the user equipment. Operation 230 includes the user equipment (or UE) determining, based on the user equipment's runtime capability limitations, that the user equipment cannot perform continuous learning for the machine learning model according to the received continuous learning configuration. Operation 240 includes the user equipment transmitting runtime capability limitation information to the network node, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for the machine learning model according to the received continuous learning configuration based on the user equipment's runtime capability limitations. Furthermore, operation 250 includes the user equipment receiving a continuous learning configuration update from the network node based on the runtime capability limitation information transmitted to the network node.

[0038] about Figure 2 The method, the continuous learning configuration may include at least one of the following: the adaptive ratio of the layers to be adapted to the machine learning model, or the number or percentage of layers; the accuracy level for the machine learning model; the training period for the continuous learning of the machine learning model; or the cycle for the continuous learning of the machine learning model.

[0039] about Figure 2 The method wherein the runtime capability limitation of the user equipment reflects the current state of the user equipment relative to one or more limitations of the user equipment, and may include at least one of the following: processing limitations of the user equipment based on available processing resources or based on the current computing level; memory limitations of the user equipment based on available memory resources or based on currently used memory resources; whether a power saving mode has been activated for the user equipment; the battery or power state of the user equipment, or the available battery resources of the user equipment; or the overheating state of the user equipment.

[0040] about Figure 2The method may include runtime capability limitation information related to the user device's inability to perform continuous learning of a machine learning model, which may include at least one of the following: information describing the user device's inability to perform continuous learning of a machine learning model according to a continuous learning configuration or related to the user device's inability to perform continuous learning of a machine learning model according to a continuous learning configuration; an indication that the user device's runtime capability limitation prevents it from performing continuous learning of a machine learning model according to a continuous learning configuration; an indication of one or more runtime capability limitations of the user device that prevents the user device from performing continuous learning of a machine learning model according to a continuous learning configuration; one or more user device proposals for one or more parameters of the continuous learning configuration; and one or more user device proposals for one or more parameters associated with continuous learning of a machine learning model.

[0041] about Figure 2 The method may include, in which the received continuous learning configuration update received from the network node may include at least one of the following: an updated continuous learning configuration, which includes at least one parameter that has been changed or updated based on runtime capability limitation information transmitted from the user equipment to the network node; an indication from the network node to cancel the continuous learning request; and / or an indication from the network node to the user equipment to use the current version of the machine learning model.

[0042] about Figure 2 The method involves transmitting runtime capability constraint information from a user equipment to a network node, including one or more user equipment proposals for one or more parameters configured for continuous learning or associated with continuous learning of a machine learning model, including one or more of the following: a proposal for an adaptive ratio or preferred amount or percentage of layers to be adapted to the machine learning model; a proposal for an accuracy level of the machine learning model; a proposal for power constraints for continuous learning of the machine learning model; a proposal for the period used for continuous learning of the machine learning model; a proposal for complexity constraints for continuous learning of the machine learning model; a proposal for latency budgets for continuous learning of the machine learning model; or a proposal for training time constraints for continuous learning of the machine learning model.

[0043] about Figure 2 The method by which runtime capability limitation information is transmitted as or via at least one of the following: information or one or more parameters provided in a user equipment auxiliary message; or one or more power saving parameters or information or one or more parameters provided in a power saving information element.

[0044] Figure 3This is a flowchart illustrating the operation of a network node (e.g., a gNB). Operation 310 includes the network node transmitting a continuous learning request to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning against a machine learning model. Operation 320 includes the network node receiving runtime capability limitation information from the user equipment, the runtime capability limitation information relating to the user equipment's inability to perform continuous learning against the machine learning model according to the continuous learning configuration based on runtime capability limitations. Operation 330 includes the network node determining a continuous learning configuration update based on the received runtime capability limitation information. And, operation 340 includes the network node transmitting the continuous learning configuration update to the user equipment.

[0045] about Figure 3 The method, the continuous learning configuration may include at least one of the following: the adaptive ratio of the layers to be adapted to the machine learning model, or the number or percentage of layers; the accuracy level for the machine learning model; the training period for the continuous learning of the machine learning model; or the cycle for the continuous learning of the machine learning model.

[0046] about Figure 3 The method wherein the runtime capability limits of the user equipment reflect the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: processing limits of the user equipment based on available processing resources or based on the current computing level; memory limits of the user equipment based on available memory resources or based on currently used memory resources; whether a power saving mode has been activated for the user equipment; the battery or power state of the user equipment, or the available battery resources of the user equipment; or the overheating state of the user equipment.

[0047] about Figure 3 The method may include runtime capability limitation information related to the user equipment's inability to perform continuous learning of a machine learning model, which may include at least one of the following: information describing the user equipment's inability to perform continuous learning of a machine learning model according to a continuous learning configuration or information related to the user equipment's inability to perform continuous learning of a machine learning model according to a continuous learning configuration; an indication that the user equipment's runtime capability limitation prevents it from performing continuous learning of a machine learning model according to a continuous learning configuration; an indication of one or more runtime capability limitations of the user equipment that prevents the user equipment from performing continuous learning of a machine learning model according to a continuous learning configuration; one or more user equipment proposals for one or more parameters of the continuous learning configuration; or one or more user equipment proposals for one or more parameters associated with continuous learning of a machine learning model.

[0048] about Figure 3The method for updating the continuous learning configuration may include at least one of the following: an updated continuous learning configuration, which includes at least one parameter that has been changed or updated based on runtime capability limitation information transmitted from the user equipment to the network node; an indication from the network node to cancel the continuous learning request; and / or an indication from the network node to the user equipment to use the current version of the machine learning model.

[0049] about Figure 3 The method, wherein runtime capability constraint information received by the network node from the user equipment may include one or more user equipment (UE) proposals for one or more parameters configured for continuous learning or for one or more parameters associated with continuous learning of the machine learning model, including one or more of the following: a proposal for the adaptive ratio or preferred amount or percentage of layers to be adapted to the machine learning model; a proposal for the accuracy level of the machine learning model; a proposal for the power constraint of continuous learning of the machine learning model; a proposal for the period of continuous learning of the machine learning model; a proposal for the complexity constraint of continuous learning of the machine learning model; a proposal for the latency budget of continuous learning of the machine learning model; or a proposal for the training time constraint of continuous learning of the machine learning model.

[0050] about Figure 3 The method by which runtime capability limitation information is received as or via at least one of the following: information or one or more parameters provided in a user equipment auxiliary message; or one or more power saving parameters or information or one or more parameters provided in a power saving information element.

[0051] The description and figures provided herein (including the figures and description below) provide information about Figure 2 and Figure 3 Further details, descriptions, and / or illustrative examples of the method.

[0052] As described above, in some cases, limitations in the runtime capabilities of a UE may restrict or inhibit the UE's continuous learning of ML models. The runtime capabilities of a UE may include, for example, any current capabilities of the UE based on its current state as a result of its operation or run and / or based on the UE's current resource usage. Therefore, for example, UE runtime capabilities may include any capabilities of the UE related to hardware, software, computing resources, memory resources, network resources, the UE's state relative to a power-saving mode (e.g., whether the UE is in a power-saving mode), battery or power resources (e.g., how much battery power remains, or whether the UE is in a power-saving mode), the UE's current thermal state (whether one or more components of the UE are overheated, or the temperature of the hardware, processor, or other components compared to the overheating temperature), or other UE capabilities.

[0053] Furthermore, UE runtime capability limitations may include any restrictions on UE runtime capabilities(s). For example, UE runtime capability limitations may be based on, may indicate, or may reflect the current state of the UE relative to one or more limitations or resources. Therefore, for example, runtime capability limitations are not static capabilities of the UE typically reported in capability exchanges. Rather, runtime capability limitations are (or may include) current or updated capability limitations of the UE that can be dynamically determined (e.g., during UE operation) by the UE based on its current UE state or current resource usage as the UE runs or operates.

[0054] UE runtime capability limitations may include one or more of the following (as some illustrative examples): 1) Computational or processing limitations of the UE, for example, based on available processing resources or based on the current computing level, such as processing load, or available processing resources such as the central processing unit (CPU), graphics processing unit (GPU), or other processors of the UE. Therefore, for example, if there are less than a threshold percentage (e.g., less than 30% idle processing / processor resources) or less than a threshold amount of available processing resources, this may inhibit or prevent further processing of data, such as computational learning for ML models, or may at least limit continuous learning to be performed in some cases.

[0055] 2) UE memory limitations, for example, based on the amount or percentage of available memory resources and / or based on the amount or percentage of memory resources currently being used by the UE, such as the amount or percentage of memory resources being used, and / or the amount or percentage of free or available memory resources. Therefore, for example, if there is less than a threshold amount or percentage of free (available) memory resources, such as less than 25% free memory, or less than 5GB of free memory (as an example), or for example, if more than 75% of memory resources are used at the UE, or less than 25% of free / available memory resources, this may inhibit or prevent the UE from processing data, such as for computational learning, or may at least limit continuous learning to be performed in at least some cases.

[0056] 3) UE power (or battery) related status or power or battery related limitations, such as whether power saving mode has been activated for the UE, or the amount of available battery resources for the UE. For example, if the UE is in power saving mode or the UE's available battery power is less than a threshold amount (e.g., less than 38%), this can prevent or disable the UE from performing continuous learning, or at least limit continuous learning to at least some of the cases.

[0057] 4) Overheating conditions or states of the UE, such as the temperature of a part of the UE hardware (e.g., the temperature of the processor or processor core or other components), or the temperature measured by a temperature sensor on the UE, an indication that the UE temperature (e.g., the temperature of the UE processor or UE components) is greater than a threshold, or an indication that the UE (or UE components) is in an overheating state or within a threshold degree of an overheating state. Any of these example temperature / thermal-related limits or overheating states may prevent or inhibit additional processing of data by the processor or processor core, such as for continuous learning, or may at least limit continuous learning to be performed in at least some cases. For example, if the processor (or other component) is at an overheating temperature or within a threshold degree of an overheating temperature, this may prevent the UE from performing continuous learning or adaptation of ML models, or may limit the performance of continuous learning or adaptation of ML models.

[0058] 5) Other UE resources (such as network resources (e.g., MAC / PHY resources), antenna resources, protocol entity resources, software resources, etc.) may similarly have runtime capability limitations, which may inhibit or limit the UE's continuous learning of ML models. For example, if there are fewer than a threshold amount of network resources, antenna resources, protocol entity resources, and / or software resources for the UE, this may inhibit the UE's continuous learning of ML models.

[0059] Furthermore, for example, the UE can compare the continuous learning configuration with its runtime capability limitations to determine if any UE runtime capability limitations currently exist that might prevent the UE from performing or enforcing continuous learning for the ML model. Alternatively, or in other words, the UE can determine whether there are sufficient UE runtime capabilities or resources available to perform or enforcing continuous learning of the ML model, as indicated or instructed by the continuous learning configuration, based on the current state or condition of its runtime capabilities and the continuous learning configuration. For example, if the available battery power is less than a first threshold (e.g., less than 35%), continuous learning can be prevented at the UE because the UE does not have sufficient battery power to perform continuous learning of the UE's ML model (or at least indicate that the UE should not perform continuous learning with a battery power less than the first available battery power threshold). Alternatively, if the UE's processor is within a threshold degree of overheating, or the amount of free memory or free processor resources is less than a threshold, this can indicate that the UE cannot (or should not) perform or conduct continuous learning (or adaptation or iteration of the ML model based on current data) according to the received CL configuration due to one or more of the UE's runtime capability limitations.

[0060] Furthermore, the UE can use different or various thresholds to allow the UE to determine different outcomes or decisions regarding whether it can perform continuous learning. For example, a first threshold limiting the UE's runtime capabilities can be used to prevent the UE from performing continuous learning of the ML model, while a second threshold can be used to allow the UE to perform some limited continuous learning, for example, with different CL configuration parameters. For instance, if the battery power available at the UE is less than the first threshold (e.g., less than 35%), continuous learning can be completely blocked at the UE because the UE does not have enough battery power to perform continuous learning of the UE's ML model. However, if there is less than the second threshold (e.g., less than 50% battery power) but greater than the first threshold (e.g., greater than 35% battery power) at the UE, the UE may be able to perform some limited continuous learning of the ML model, for example, based on an updated CL configuration that allows the UE to perform less continuous learning (or consume less battery power while performing continuous learning).

[0061] For example, if the CL configuration received by the UE indicates, for instance, an adaptive ratio of 0.6 (indicating that 6 out of 10 layers of the ML model should be adapted as part of a continuous learning operation), but the battery power level of 42% is no greater than 50% (the second threshold) but greater than the first threshold (35%), the UE may not be able to perform continuous learning at an adaptive ratio of 0.6. Therefore, in this example, with the UE battery power level at 42% (greater than the first threshold but less than the second threshold), the UE can determine that it cannot perform continuous learning according to the received CL configuration (with an adaptive ratio of 0.6). However, the UE can perform continuous learning or adaptation at a lower level or with a lower amount than indicated by the received CL configuration. For example, the UE may only perform continuous learning or ML model adaptation on only 3 out of 10 layers (associated with an adaptive ratio of 0.3 for continuous learning of the ML model, where 3 out of 10 layers of the ML model will be adapted, while 7 out of 10 layers will remain static or frozen / unadapted during the continuous learning operation or iteration). Therefore, in this example scenario, the UE can send or transmit runtime capability limitation information to the gNB (or other network node). This runtime capability limitation information may include, for example, information indicating that the UE cannot perform continuous learning of the ML model configuration based on the received CL configuration (which indicates an adaptive ratio of 0.6) due to UE runtime capability limitations, information indicating specific runtime capability limitations (in this example, battery power) that prevent the UE from performing continuous learning of the ML model, and / or one or more UE suggestions (e.g., recommendations or suggestions) for one or more parameters of the CL configuration that the UE can perform or use for continuous learning based on its runtime capability limitations. In this example, the UE may transmit a proposal to the gNB for an adaptive ratio of 0.3 (e.g., thus indicating that the UE can perform continuous learning or adaptation of 3 out of 10 layers of the ML model), and may indicate that the proposal is based on (or due to) runtime capability limitations related to the UE's battery power that prevent the UE from performing continuous learning according to the received continuous learning configuration. The adaptive ratio is merely an example, and the UE may provide the gNB with proposals (proposed values ​​or parameter values) for any UE runtime capability.

[0062] Based on runtime capability limitation information received by the gNB from the UE (e.g., an indication that the UE cannot perform continuous learning based on a configured adaptive ratio of 0.6, and including a suggested adaptive ratio of 0.3 that the UE can perform continuous learning under), the gNB can determine an updated CL configuration and can transmit the updated CL configuration to the UE. The updated CL configuration may include, for example, updated parameters or parameter values, such as an adaptive ratio of 0.3 or other values ​​proposed by the UE, or may include an indication to cancel (not perform) continuous learning, and / or may include an indication that the UE will use the current version (current weight set) for the ML model.

[0063] Figure 4 This is a diagram illustrating the operation of the system. UE 410 may include a machine learning (ML) model 412. UE 410 may perform or implement continuous learning for the ML model 412 based on continuous learning (CL) requests and / or continuous learning (CL) configurations that can be received by UE 410 from gNB (or other network nodes) 414. UE 410 may communicate with gNB 414. In some cases, UE 410 may be connected to gNB 414.

[0064] exist Figure 4 Step 1: UE 410 receives a continuous learning request, including a continuous learning (CL) configuration. The CL configuration may include one or more parameters as part of the CL configuration. For example, the continuous learning (CL) configuration may include various parameters that can indicate parameters, conditions, or requirements for continuous learning (CL) of the ML model, including one or more of the following example parameters: 1) The identifier of the ML model (e.g., ML model ID), or the task or function to be performed by the ML model (e.g., beam prediction).

[0065] 2) Requests to perform or continuously learn ML models.

[0066] 3) Adaptive ratio, which can be or includes the ratio of layers to be adapted to layers to remain static or frozen (unadapted), or the amount of the ML model or the percentage or number of layers of the ML model on which continuous learning (adaptation) is performed or implemented. The adaptive ratio can indicate the ratio of adaptive layers (layers to be adapted or trained / retrained via continuous learning) to frozen or static layers (layers of the ML model that are static or frozen or not adapted during the adaptation or continuous 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 retrained during the continuous learning operation or iteration, and the weights of the other 2 / 3 (two-thirds) of the ML model layers are not retrained or adapted during the CL operation or iteration (therefore, those 2 / 3 of these layers will remain static or frozen, and thus not change or adapt during the continuous learning operation or iteration). Therefore, for example, if there are 20 layers in an ML model, an adaptation ratio of 0.5 indicates that the weights of 10 of the 20 layers should be adapted or retrained as part of the continuous learning operation or iteration. Adapting more layers (higher adaptation ratios) requires more memory, processor, battery, and / or time resources from the UE.

[0067] 4) Accuracy Level of the Machine Learning Model. Higher accuracy levels (more accurate training or adaptation of the ML model weights) typically require, or may be associated with, more memory, processor, battery, and time resources for the UE (User Equipment) for ML model adaptation. For example, the accuracy level of an ML model can refer to or include the ML model's prediction accuracy, which is an example of a metric indicating the reliability of the ML model. For instance, prediction accuracy can be determined or represented as: Prediction Accuracy = (Predicted Value / True Value) * 100%, meaning that this example of a Key Performance Indicator (KPI) is validated against real labels / data to determine the accuracy level or prediction accuracy. Furthermore, or as another example, the accuracy level of the ML model can be determined based on one or more KPIs (such as based on the F1 score or other KPIs).

[0068] 5) The training period for continuous learning of a machine learning model can indicate the time period (e.g., the maximum period) during which training or adaptation of the ML model weights should be performed. A longer (larger) training (or adaptation) period will typically require, or may be associated with, more memory, processor, battery, and / or time resources from the UE for ML model adaptation. Therefore, the training period can refer to the maximum training period required by the UE. If the training period is 1 hour, the ML model training / adaptation (CL operation) of the ML model may not exceed this time limit. This parameter (training period) can also be referred to as the (e.g., preferred or proposed) (re)training time limit for the UE for continuous learning.

[0069] 6) Cycles for continuous learning of the machine learning model. Cycles can indicate the frequency (or number of cycles) at which retraining or adaptation should be performed on the ML model, for example, every 100ms. Smaller cycles will typically require more memory, processor, battery, and / or time resources from the UE, or may be associated with them for ML model adaptation, as the smaller cycle will require the UE to perform continuous learning or model adaptation more frequently. These are some example parameters, and others may be included.

[0070] exist Figure 4 In step 2, UE 410 can determine its runtime capability limitations and whether UE 410 can perform or implement continuous learning based on (or according to) the received continuous learning configuration. UE runtime capabilities may include the current state of one or more of the UE's capabilities or resources, such as the current state or condition of the UE's 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 the UE's state related to overheating or whether the UE is in power-saving mode (because the UE's ability to operate fully may be affected by UE overheating and / or the UE being placed in power-saving mode).

[0071] UE runtime capability limitations can reflect or indicate the current state of the UE relative to one or more capabilities or resources, and may include or indicate whether runtime capabilities are greater than or less than a threshold. UE runtime capability limitations may include, for example, the current state or condition of one or more of the following (based on the UE's current state or condition): processing / computing limitations of the UE based on available processing resources or based on the current computing level; memory limitations of the UE based on available memory resources and / or based on currently used memory resources; whether a power-saving mode has been activated for the UE; the UE's battery or power state, or the UE's available battery resources; and / or the UE's overheating state. In some cases, or for some capabilities or resources, the UE may determine its runtime capability limitations by determining the amount or percentage of idle or available resources or capabilities, such as based on the current state or current use of resources or capabilities (current amount or percentage of idle or available processing resources or processing capabilities, memory resources, power or battery resources, network resources, other software or hardware resources, etc.). In addition, Figure 4In step 2, UE 410 determines whether its current runtime capability limitations will prevent the UE from performing continuous learning according to the received continuous learning (CL) configuration. Therefore, to determine whether the UE's current runtime capability limitations might prevent the UE from performing continuous learning of the ML model, the UE can compare its current runtime capabilities with one or more thresholds. For example, the UE can compare its currently available battery power, currently available memory resources, currently available processing resources, current overheating state, or processor temperature with one or more thresholds to determine whether the UE can perform continuous learning based on the received continuous learning configuration. For example, if the UE has less than 30% available processing or computing resources, or less than 5GB of available or free memory, or less than 35% available battery power, then (any of these) could indicate that the UE cannot perform continuous learning of the ML model. Alternatively, the UE can also determine whether the UE is in a power-saving mode, or whether the temperature of the UE processor or UE components is within a threshold number (e.g., 10) degrees of overheating temperature. If the UE is in power-saving mode or if the UE is in an overheated state or at an overheated temperature (e.g., the temperature of the UE or its components is greater than the overheated / overheated temperature), or if the UE has a processor or other component temperature that is within a threshold degree difference from the overheated temperature, this indicates that the UE cannot perform continuous learning for the ML model based on one or more of these UE runtime capability limitations.

[0072] In addition, Figure 4 Step 2, the UE determines whether it can or cannot perform continuous learning according to the CL configuration, which can be based on parameters or values ​​of the UE runtime capability limitations and the continuous learning configuration (CL configuration). For example, various thresholds used by the UE for different UE runtime capability limitations can be based on the values ​​of one or more CL configuration parameters. Therefore, for example, different thresholds related to different CL configuration parameters can be used to determine whether the UE can perform continuous learning.

[0073] For example, the continuous learning configuration received by UE 410 from gNB 414 may include multiple values ​​of one or more CL configuration parameters, such as the adaptive ratio, period, training period, etc. In some cases, the value of one or more CL configuration parameters may indicate (or be associated with) the amount or percentage of UE runtime capability required to perform or implement continuous learning according to the received continuous learning configuration. For example, an adaptive ratio greater than 0.5 (e.g., indicating that 50% of the ML model layers should be adapted or trained during continuous learning operations, such as 5 layers of the ML model) may require at least 50% of the UE battery power level, while an adaptive ratio of 0.3 or less may require only at least 35% of the battery power level. Therefore, for example, if the CL configuration received by UE 410 indicates, for example, an adaptive ratio of 0.6 (indicating that 6 out of 10 layers of the ML model should be adapted as part of the continuous learning operation), the UE will compare its current battery power level to determine whether it is greater than or equal to 50% of the 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 cannot complete or perform continuous learning of the ML model based on the received CL configuration (which requires an 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 perform continuous learning of the ML model based on the received CL configuration.

[0074] More complex scenarios may exist, and / or the UE may perform more complex or detailed analyses to determine whether the UE's runtime capabilities might limit or prevent the UE from performing continuous learning according to the received continuous learning configuration. For example, if the CL configuration indicates that the continuous learning period (indicating the time interval or time gap between each continuous learning operation or iteration of the ML model) is less than 100ms, at least 50% battery level and at least 34% available processor resources are required. On the other hand, a period greater than or equal to 100ms may require at least 40% battery level and at least 28% available processor resources. Therefore, if the CL configuration indicates, for example, a period of 150ms (which is greater than 100ms), the UE will confirm or determine that it has at least 40% battery level and at least 28% available processor resources before transmitting a response to the gNB 414 confirming that the UE can perform or implement continuous learning for the ML model configuration. If either of these levels—at least 40% battery power and 28% available processor resources—is unavailable at the UE, the UE can transmit runtime capability limitation information to the gNB. This information can notify the gNB that the UE cannot perform the requested continuous learning according to the CL configuration, and can indicate the runtime capabilities that serve as the reason for CL rejection, and / or can provide a proposal (suggestion or recommendation) for one or more of these CL configuration parameters that would enable the UE to perform continuous learning. Therefore, depending on the parameters of the received CL configuration, the UE can compare one or more UE runtime capability limitation values ​​with one or more different thresholds to determine whether the UE can perform continuous learning according to the received CL configuration.

[0075] If the UE determines that it cannot perform continuous learning according to the CL configuration due to limitations in its runtime capabilities, then Figure 4 The process proceeds to steps 3 through 6. Figure 4 Step 3: The UE determines that one or more UE runtime capability limitations will prevent the UE from performing continuous learning according to the received CL configuration. See the various examples above. For example, for UE 410, in this example, an adaptive ratio greater than 0.5 may require at least 50% of the UE's battery power level. Therefore, if the continuous 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%, the UE cannot perform continuous learning on the ML model according to the received continuous learning configuration.

[0076] exist Figure 4At step 4, UE 410 transmits a UE assistance information message (which may be transmitted as a Radio Resource Control (RRC) message) to gNB 414. The UE assistance information may include, for example, runtime capability limitation information related to the UE's inability to perform continuous learning for the ML model, such as including at least one of the following: information describing or related to the UE's inability to perform continuous learning for the ML model according to the continuous learning configuration; an indication that the UE cannot perform continuous learning of the ML model according to the continuous learning configuration based on the UE's runtime capability limitations; an indication of one or more runtime capability limitations of the UE that prevent the UE from performing continuous learning of the machine learning model according to the continuous learning configuration; and / or one or more UE suggestions (e.g., UE suggestions, UE recommendations, or UE preferences) for one or more parameters (parameter values) of the continuous learning configuration; or one or more UE suggestions for one or more parameters (parameter values) associated with continuous learning of the machine learning model. For example, the received CL configuration may indicate an adaptive ratio of 0.6 and a period of 100ms. However, for a current UE battery power level of 42% (greater than 35% and less than 50%), the UE can only perform continuous learning for a CL configuration with an adaptive ratio of 0.3 or less and a period greater than 100ms. Therefore, in this example, based on the UE's current 42% battery power level as measured by the UE, the UE can transmit a message to gNB 414 including UE assistance information, such as a proposal for an adaptive ratio of 0.3 and a period of 150ms. The runtime capability limitation information transmitted by the UE may include or may be transmitted via: information or one or more parameters provided within a UE assistance message or UE assistance information; or information on one or more power saving parameters or one or more parameters provided within a power saving information element.

[0077] exist Figure 4At step 5, gNB 414 receives a UE assistance information message from the UE. This UE assistance information message includes runtime capability limitation information related to the UE's inability to perform continuous learning for the ML model. For example, it may include UE proposals (e.g., UE suggestions, recommendations, preferences) for one or more continuous learning configuration parameters or other parameters. For example, the UE assistance information message may include a UE proposal with an adaptation ratio of 0.6 and a period of 150ms (based on the UE's current runtime capability limitations). gNB 414 may determine continuous learning configuration updates based on the runtime capability limitation information received by gNB 414 from UE 410. For example, this may be or may include an indication to cancel the continuous learning request to use the existing configuration (or weight set) for the ML model (which does not require continuous learning), and / or one or more updated / adjusted parameters (parameter values) for the continuous learning configuration. gNB 414 may determine whether the proposed parameter values ​​will provide sufficient retraining or adaptation for the ML model to provide acceptable performance and / or accuracy for the functions being performed or assisted by the ML model. For example, if the parameter values ​​proposed by UE 410 would provide sufficient performance, gNB 414 can send an updated CL request including an updated continuous learning configuration that confirms or indicates that the proposed parameters are acceptable. As an example, if the UE has proposed an adaptive ratio of 0.3 and a period of 150ms, the gNB can accept the proposed values ​​of these parameters, or it can transmit an updated continuous learning configuration with one or more parameters different from those already proposed by the UE. Therefore, if these proposed values ​​are acceptable to the gNB, gNB 414 can transmit an updated continuous learning configuration with parameters of 0.3 for the adaptive ratio and 150ms for the period. Alternatively, if these proposed values ​​are unacceptable to the gNB, gNB 414 can transmit an updated continuous learning configuration indicating, for example, an adaptive ratio of 0.35 and a period of 120ms.

[0078] exist Figure 4 In step 6, gNB 414 transmits an updated continuous learning request to UE 410, including, for example, an updated continuous learning configuration.

[0079] exist Figure 4 After step 2, Figure 4The process proceeds to steps 7-8. If the UE determines that it can perform continuous learning according to the CL configuration due to UE runtime capability limitations (UE 410 has not detected any runtime capability limitations that would prevent UE 410 from performing continuous learning according to the received continuous learning configuration), and after detecting that there are no UE runtime capability limitations that would prevent the UE from performing continuous learning as configured by gNB 414, the operation proceeds to step 8. In step 8, UE 410 performs continuous learning operations on the ML model (or multiple layers of the ML model) according to the received continuous learning configuration.

[0080] Example UE assistance information may include preferences (or UE-proposed values / proposals) for one or more parameters related to the continuous learning configuration, such as proposed values ​​for adaptive ratio, power limit, complexity limit, latency budget, and training time limit.

[0081] UEAssistanceInformation-r19-IEs ::= SEQUENCE { cl-Preference-AdativeRatio-r19CL-Preference-AdativeRatio-r19OPTIONAL, cl-Preference-PowerLimit-r19CL-Preference-PowerLimit-r19OPTIONAL, cl-Preference-ComplexityLimit-r19CL-Preference-ComplexityLimit-r19OPTIONAL, cl-Preference-DelayBudget-r19CL-Preference-DelayBudget-r19OPTIONAL, cl-Preference-TrainingTimeLimit-r19CL-Preference-TrainingTimeLimit-r19OPTIONAL. The following are some example parameters or fields that can be included in the UE assistance information. The UE may suggest or recommend values ​​for one or more of these example fields or parameters.

[0082]

[0083] In another implementation, some continuously learned configuration parameters related to the battery or power can be provided within or embedded in the power-saving parameters, for example, added to the RRC PowSav-Parameters IE (information element), as shown below: PowSav-Parameters-CL-r19 ::=SEQUENCE { cl-Preference-AdativeRatio-r19ENUMERATED {supported}OPTIONAL, cl-Preference-PowerLimit-r19ENUMERATED {supported}OPTIONAL, cl-Preference-ComplexityLimit-r19ENUMERATED {supported}OPTIONAL, … } CL-Preference-AdativeRatio-r19::=SEQUENCE { cl-Preference-AdativeRatio-r19ENUMERATED {oDot2, oDot4, oDot6, oDot8}OPTIONAL } CL-Preference-PowerLimit-r19::=SEQUENCE { cl-Preference-PowerLimit-r19INTEGER (-30..33)OPTIONAL } CL-Preference-ComplexityLimit-r19::= SEQUENCE { cl-Preference-ComplexityLimit-r19INTEGER (0..200)OPTIONAL } Several examples will now be described based on the descriptions and figures provided herein.

[0084] Example 1. A device (e.g., Figure 5 1300; and / or Figure 4 UE 410 in the example includes: at least one processor (e.g., Figure 5 The processor 1304); and at least one memory for storing instructions (e.g., Figure 5 The memory 1306), when executed by at least one processor (1304), causes the device to at least: 1) By user equipment (e.g., UE 410, Figure 4 From network nodes (e.g., gNB 414, Figure 4 ) Receive (Step 1, Figure 4 A continuous learning (CL) request, which includes a continuous learning configuration for a user device to implement continuous learning against a machine learning model. For example, in Figure 4 At step 1, UE 410 receives a continuous learning request, including a continuous learning (CL) configuration; the CL configuration may include one or more parameters as part of the CL configuration; for example, the continuous learning (CL) configuration may include parameters that can indicate the parameters used for the ML model (412, Figure 4 The parameters, conditions, or requirements of continuous learning (CL) of ML model include, for example, the ML model ID, the adaptive ratio for CL, the accuracy level for the ML model, the training period for CL, the cycle for CL, ... 2) Determined by the user equipment (UE 410) (Step 2, Figure 4 The runtime capabilities of the user equipment. For example, in Figure 4 At step 2, UE 410 can determine the runtime capability limitations of UE 410 and whether UE 410 can perform or implement continuous learning (CL) according to (or based on) the received continuous learning configuration. UE runtime capabilities may include, for example, the current state of one or more capabilities or resources of the UE, such as the current state or condition 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 the UE's state related to overheating or whether the UE is in power-saving mode (because the UE's ability to operate fully may be affected by UE overheating and / or the UE being placed in power-saving mode).

[0085] 3) By user equipment (UE 410, Figure 4 Based on the runtime capability limitations of the user equipment, it is determined that the user equipment cannot execute a machine learning model (ML model 412) according to the received continuous learning configuration. Figure 4 Continuous learning. For example, in Figure 4At step 3, the UE may detect one or more runtime capability limitations that prevent (e.g., disable or restrict) the UE 410 from performing or implementing continuous learning according to or based on the received continuous learning configuration. To determine whether the UE's current runtime capability limitations might prevent the UE from performing continuous learning of the ML model, the UE 410 may compare its current runtime capabilities with one or more thresholds. For example, the UE may compare its currently available battery power, currently available memory resources, currently available processing resources, current overheating state, or processor temperature with one or more thresholds to determine whether the UE can perform continuous learning based on the received continuous learning configuration. For example, if the UE has less than 30% available processing or computing resources, or less than 5GB of available or free memory, or less than 35% available battery power, then (any of these) may indicate that the UE cannot perform continuous learning of the ML model. Alternatively, the UE may also determine whether the UE is in a power-saving mode, or whether the temperature of the UE processor or UE components is within a threshold number (e.g., 10) degrees of overheating temperature. If the UE is in power-saving mode, or if the UE is in an overheated state or at an overheated temperature (e.g., the temperature of the UE or its components exceeds the overheated / overheated temperature), or if the UE has a processor or other component temperature that differs from the overheated temperature by a threshold degree, this indicates that the UE cannot perform continuous learning for the ML model based on one or more of these UE runtime capability limitations. For example, for UE 410 to perform continuous learning for the ML model based on an adaptive ratio greater than 0.5, at least 50% of the UE battery power level is required. Therefore, if the continuous 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%, the UE cannot perform continuous learning for the ML model according to the received continuous learning configuration. UE 410 can compare its various runtime capabilities with different thresholds to determine whether any UE runtime capability has a limitation that would prevent the UE from performing continuous learning according to the received CL configuration.

[0086] 4) By user equipment (UE 410, Figure 4 This involves transmitting runtime capability limitation information to network nodes. This runtime capability limitation information relates to user equipment-based runtime capability limitations preventing the user equipment from implementing continuous learning for a machine learning model based on the received continuous learning configuration. For example, in... Figure 4At step 4, UE 410 transmits a UE assistance information message (which may be transmitted as a Radio Resource Control (RRC) message) to gNB 414. The UE assistance information may include, for example, runtime capability limitation information related to the UE's inability to perform continuous learning for the ML model, such as including at least one of the following: information describing or related to the UE's inability to perform continuous learning for the ML model according to the continuous learning configuration; an indication that the UE cannot perform continuous learning of the ML model according to the continuous learning configuration based on the UE's runtime capability limitations; an indication of one or more runtime capability limitations of the UE that prevent the UE from performing continuous learning of the machine learning model according to the continuous learning configuration; and / or one or more UE suggestions (e.g., UE suggestions, UE recommendations, or UE preferences) for one or more parameters (parameter values) of the continuous learning configuration; or one or more UE suggestions for one or more parameters (parameter values) associated with continuous learning of the machine learning model. For example, the received CL configuration may indicate an adaptive ratio of 0.6 and a period of 100ms. However, for a current UE battery power level of 42% (greater than 35% and less than 50%), the UE can only perform continuous learning for CL configurations with an adaptive ratio of 0.3 or less and a period greater than 100ms. Therefore, in this example, based on the UE's current 42% battery power level as measured by the UE, the UE can transmit a message to gNB 414 including UE assistance information, such as a proposal for an adaptive ratio of 0.3 and a period of 150ms.

[0087] 5) The user equipment receives continuously learned configuration updates from the network node based on runtime capability limitation information transmitted to the network node. For example, in Figure 4 In step 6, UE 410 receives an updated continuous learning request from gNB 414, which may include an updated continuous learning configuration. For example, gNB 414 may determine the continuous learning configuration update based on runtime capability constraint information received by gNB 414 from UE 410. This update may include, for example, an indication to cancel the continuous learning request to use the existing configuration (or weight set) for the ML model (which does not require continuous learning), and / or one or more updated / adjusted parameters (parameter values) for the continuous learning configuration. For example, gNB 414 may send an updated CL request including an updated continuous learning configuration that confirms the proposed parameters (proposed by UE 410) are acceptable, or indicates, for example, acceptable parameters. As an example, if the UE has proposed an adaptive ratio of 0.3 and a period of 150ms, gNB may accept the proposed values ​​of these parameters, or it may transmit (and UE 410 may receive) an updated continuous learning configuration with one or more parameters different from those already proposed by the UE.

[0088] Example 2. The apparatus according to Example 1, wherein the continuous learning configuration includes at least one of the following: an adaptive ratio or amount or percentage of layers to be adapted to the machine learning model; an accuracy level for the machine learning model; a training period for continuous learning of the machine learning model; or a cycle for continuous learning of the machine learning model.

[0089] Example 3. An apparatus according to any one of Examples 1-2, wherein the runtime capability limitation of the user equipment reflects the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: processing limitations of the user equipment based on available processing resources or based on the current computing level; memory limitations of the user equipment based on available memory resources or based on currently used memory resources; whether a power saving mode has been activated for the user equipment; the battery or power state of the user equipment, or the available battery resources of the user equipment; or the overheating state of the user equipment.

[0090] Example 4. The apparatus according to any one of Examples 1-3, wherein the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for a machine learning model includes at least one of the following: information describing that the user equipment cannot perform continuous learning for a machine learning model according to a continuous learning configuration or related to the user equipment's inability to perform continuous learning for a machine learning model according to a continuous learning configuration; an indication that the user equipment cannot perform continuous learning for a machine learning model according to a continuous learning configuration based on the user equipment's runtime capability limitations; an indication of one or more runtime capability limitations of the user equipment's runtime capability limitations, the one or more runtime capability limitations preventing the user equipment from performing continuous learning for a machine learning model according to a continuous learning configuration; one or more user equipment proposals for one or more parameters of the continuous learning configuration; or one or more user equipment proposals for one or more parameters associated with continuous learning for a machine learning model.

[0091] Example 5. The apparatus according to any one of Examples 1-4, wherein the received continuous learning configuration update received from the network node includes at least one of the following: an updated continuous learning configuration, the updated continuous learning configuration including at least one parameter that has been changed or updated based on runtime capability limitation information transmitted from the user equipment to the network node; an indication from the network node to cancel the continuous learning request; and / or an indication from the network node to the user equipment to use the current version of the machine learning model.

[0092] Example 6. An apparatus according to any one of Examples 1-5, wherein the runtime capability limiting information transmitted from the user equipment to the network node includes one or more user equipment proposals for one or more parameters configured for continuous learning or for one or more parameters associated with continuous learning of the machine learning model, including one or more of the following: a proposal for an adaptive ratio or preferred amount or percentage of layers to be adapted to the machine learning model; a proposal for an accuracy level of the machine learning model; a proposal for a power limit for continuous learning of the machine learning model; a proposal for a period for continuous learning of the machine learning model; a proposal for a complexity limit for continuous learning of the machine learning model; a proposal for a latency budget for continuous learning of the machine learning model; or a proposal for a training time limit for continuous learning of the machine learning model.

[0093] Example 7. An apparatus according to any one of Examples 1-6, wherein runtime capability limitation information is transmitted as or via at least one of: information or one or more parameters provided in a user equipment auxiliary message; or one or more power saving parameters or information or one or more parameters provided in a power saving information element.

[0094] Example 8. One method (see example) Figure 2 The flowchart includes: by user equipment (UE 410, Figure 4 From network node (gNB 414, Figure 4 ) Receive (210, Figure 2 ) Continuous learning request, which includes a continuous learning configuration for the user equipment to implement continuous learning for the machine learning model; determined by the user equipment (220, Figure 2 The runtime capability limitations of the user equipment; the user equipment determines (230, based on its runtime capability limitations) Figure 2 The user equipment cannot implement continuous learning for the machine learning model based on the received continuous learning configuration; the user equipment transmits (240, Figure 2 Runtime capability limitation information, which relates to the user equipment's inability to perform continuous learning for the machine learning model based on the received continuous learning configuration; and the user equipment receiving (250, ...) runtime capability limitation information from the network node based on the runtime capability limitation information transmitted to the network node. Figure 2 Continuous learning configuration updates.

[0095] Example 9. The method according to Example 8, wherein the continuous learning configuration includes at least one of the following: an adaptive ratio or number or percentage of layers to be adapted to the machine learning model; an accuracy level for the machine learning model; a training period for continuous learning of the machine learning model; or a cycle for continuous learning of the machine learning model.

[0096] Example 10. The method according to any one of Examples 8-9, wherein the runtime capability limitations of the user equipment reflect the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: processing limitations of the user equipment based on available processing resources or based on the current computing level; memory limitations of the user equipment based on available memory resources or based on currently used memory resources; whether a power saving mode has been activated for the user equipment; the battery or power state of the user equipment, or the available battery resources of the user equipment; or the overheating state of the user equipment.

[0097] Example 11. The method according to any one of Examples 8-10, wherein the runtime capability limitation information related to the user equipment's inability to perform continuous learning for the machine learning model includes at least one of the following: information describing that the user equipment cannot perform continuous learning for the machine learning model according to the continuous learning configuration or related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration; an indication that the user equipment cannot perform continuous learning for the machine learning model according to the continuous learning configuration based on the user equipment's runtime capability limitations; an indication of one or more runtime capability limitations of the user equipment's runtime capability limitations, the one or more runtime capability limitations preventing the user equipment from performing continuous learning for the machine learning model according to the continuous learning configuration; one or more user equipment proposals for one or more parameters of the continuous learning configuration; or one or more user equipment proposals for one or more parameters associated with the continuous learning of the machine learning model.

[0098] Example 12. The method according to any one of Examples 8-11, wherein the received continuous learning configuration update received from the network node includes at least one of the following: an updated continuous learning configuration, the updated continuous learning configuration including at least one parameter that has been changed or updated based on runtime capability limitation information transmitted from the user equipment to the network node; an indication from the network node to cancel the continuous learning request; and / or an indication from the network node to the user equipment to use the current version of the machine learning model.

[0099] Example 13. The method according to any one of Examples 8-12, wherein the runtime capability constraint information transmitted from the user equipment to the network node includes one or more user equipment proposals for one or more parameters configured for continuous learning or for one or more parameters associated with continuous learning of the machine learning model, including one or more of the following: a proposal for an adaptive ratio or preferred amount or percentage of layers to be adapted for the machine learning model; a proposal for an accuracy level for the machine learning model; a proposal for a power constraint for continuous learning of the machine learning model; a proposal for a period for continuous learning of the machine learning model; a proposal for a complexity constraint for continuous learning of the machine learning model; a proposal for a latency budget for continuous learning of the machine learning model; or a proposal for a training time constraint for continuous learning of the machine learning model.

[0100] Example 14. A method as in any of Examples 8-13, wherein runtime capability limitation information is transmitted as or via at least one of the following: information or one or more parameters provided within a user equipment auxiliary message; or one or more power saving parameters or information or one or more parameters provided within a power saving information element.

[0101] Example 15. A device (e.g., Figure 5 1300 Figure 4 gNB 414), including: at least one processor (e.g., Figure 5 The processor 1304); and at least one memory for storing instructions (e.g., Figure 5 The memory 1306), when executed by at least one processor (1304), causes the device to at least: 1) By network node (gNB 414, Figure 4 ) to user equipment (UE 410, Figure 4 The system transmits a continuous learning request, which includes a continuous learning configuration for the user device to implement continuous learning against a machine learning model. For example, see step 1. Figure 4 And related descriptions.

[0102] 2) The network node receives runtime capability limitation information from the user equipment. This runtime capability limitation information relates to the user equipment's inability to perform continuous learning for the machine learning model based on its runtime capability limitations. For example, see... Figure 4 Step 4 and related descriptions.

[0103] 3) Network nodes determine continuous learning configuration updates based on received runtime capability limitation information. For example, see... Figure 4 Step 5 and related descriptions.

[0104] 4) Continuously learned configuration updates are transmitted from network nodes to user equipment. For example, see... Figure 4 Step 6 and related descriptions.

[0105] Example 16. The apparatus according to Example 15, wherein the continuous learning configuration includes at least one of the following: an adaptive ratio or amount or percentage of layers to be adapted to the machine learning model; an accuracy level for the machine learning model; a training period for continuous learning of the machine learning model; or a cycle for continuous learning of the machine learning model.

[0106] Example 17. An apparatus according to any one of Examples 15-16, wherein the runtime capability limitations of the user equipment reflect the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: processing limitations of the user equipment based on available processing resources or based on the current computing level; memory limitations of the user equipment based on available memory resources or based on currently used memory resources; whether a power saving mode has been activated for the user equipment; the battery or power state of the user equipment, or the available battery resources of the user equipment; or the overheating state of the user equipment.

[0107] Example 18. An apparatus according to any one of Examples 15-17, wherein runtime capability limitation information relating to the user equipment's inability to perform continuous learning for a machine learning model includes at least one of the following: information describing that the user equipment cannot perform continuous learning for a machine learning model according to a continuous learning configuration or information relating to the user equipment's inability to perform continuous learning for a machine learning model according to a continuous learning configuration; an indication that the user equipment cannot perform continuous learning for a machine learning model according to a continuous learning configuration based on the user equipment's runtime capability limitations; an indication of one or more runtime capability limitations of the user equipment's runtime capability limitations, the one or more runtime capability limitations preventing the user equipment from performing continuous learning for a machine learning model according to a continuous learning configuration; one or more user equipment proposals for one or more parameters of the continuous learning configuration; or one or more user equipment proposals for one or more parameters associated with continuous learning for a machine learning model.

[0108] Example 19. An apparatus according to any one of Examples 15-18, wherein a continuous learning configuration update includes at least one of the following: an updated continuous learning configuration, the updated continuous learning configuration including at least one parameter that has been changed or updated based on runtime capability limitation information transmitted from the user equipment to the network node; an indication from the network node to cancel a continuous learning request; and / or an indication from the network node to the user equipment to use the current version of a machine learning model.

[0109] Example 20. An apparatus according to any one of Examples 15-19, wherein the runtime capability constraint information received by the network node from the user equipment includes one or more user equipment proposals for one or more parameters configured for continuous learning or for one or more parameters associated with continuous learning of the machine learning model, including one or more of the following: a proposal for an adaptive ratio or preferred amount or percentage of layers to be adapted to the machine learning model; a proposal for an accuracy level of the machine learning model; a proposal for a power constraint for continuous learning of the machine learning model; a proposal for a period of continuous learning of the machine learning model; a proposal for a complexity constraint for continuous learning of the machine learning model; a proposal for a latency budget for continuous learning of the machine learning model; or a proposal for a training time constraint for continuous learning of the machine learning model.

[0110] Example 21. The apparatus according to any one of Examples 15-20, wherein runtime capability limitation information is received as or via at least one of: information or one or more parameters provided in a user equipment auxiliary message; or one or more power saving parameters or information or one or more parameters provided in a power saving information element.

[0111] Example 22. A method (for example, see...) Figure 3 The flowchart includes: consisting of network nodes (e.g., Figure 4 gNB414) to user equipment ( Figure 4 UE 410) transmission ( Figure 3 (310) A continuous learning request, which includes a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; (320) runtime capability limitation information received by the network node from the user equipment, which relates to the user equipment's inability to perform continuous learning for a machine learning model according to the continuous learning configuration based on the user equipment's runtime capability limitations; (330) a continuous learning configuration update determined by the network node based on the received runtime capability limitation information; and (310) a transmission by the network node to the user equipment of (320) a continuous learning configuration update. Figure 3 (340) Continuous learning configuration updates.

[0112] Figure 5 This is a block diagram of a wireless station or node (e.g., UE, user equipment, AP, BS, eNB, gNB, RAN node, network node, TRP, or other node) 1300 according to an example implementation. The wireless station 1300 may include, for example, one or more (e.g., such as...) Figure 5The two RF (radio frequency) or wireless transceivers shown are 1302 A and 1302 B, each of which includes a transmitter for transmitting signals and a receiver for receiving signals. The wireless station also includes a processor or control unit / entity (controller) 1304 for executing instructions or software and controlling the transmission and reception of signals, and a memory 1306 for storing data and / or instructions.

[0113] Processor 1304 may also make decisions or determinations, generate frames, packets, or messages for transmission, decode received frames or messages for further processing, and perform other tasks or functions described herein. For example, processor 1304, which may be a baseband processor, may generate messages, packets, frames, or other signals for transmission via wireless transceiver 1302 (1302A or 1302B). Processor 1304 may control the transmission of signals or messages on a wireless network and may control the reception of signals or messages via a wireless network (e.g., after down-conversion by wireless transceiver 1302). Processor 1304 may be programmable and capable of executing software or other instructions stored in memory or 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) hardware, programmable logic, a programmable processor executing software or firmware, and / or any combination of these. Using other terms, processor 1304 and transceiver 1302 together may be considered, for example, a wireless transmitter / receiver system.

[0114] Additionally, refer to Figure 5 The controller (or processor) 1308 can execute software and instructions, and can provide overall control for station 1300, and can provide... Figure 5 Other systems, not shown, provide control, such as controlling input / output devices (e.g., a display, a keypad), and / or can execute software that can perform one or more applications available on the wireless station 1300, such as, for example, an email program, an audio / video application, a word processor, a VoIP application, or other applications or software.

[0115] Alternatively, a storage medium containing stored instructions may be provided, which, when executed by a controller or processor, may cause processor 1304 or other controllers or processors to perform one or more of the functions or tasks described above.

[0116] According to another example implementation, the RF or wireless transceiver 1302 A / 1302 B can receive signals or data and / or transmit or send signals or data. The processor 1304 (and possibly the transceiver 1302 A / 1302 B) can control the RF or wireless transceiver 1302 A or 1302 B to receive, transmit, broadcast, or transmit signals or data.

[0117] For each example method, an example implementation is provided or described, including: an apparatus (e.g., Figure 5 1300), including means for performing any method (e.g., Figure 5 The processor 1304, RF transceiver 1302 A and / or 1302 B and / or memory 1306; a non-transitory computer-readable storage medium (e.g., Figure 5 The memory 1306 includes instructions stored thereon, which are processed by at least one processor. Figure 5 The processor 1304 is configured to enable the computing system (e.g., Figure 5 (1300) executes any example method; and a device (e.g., Figure 5 The 1300), including at least one processor (e.g., Figure 5 The processor 1304) and at least one memory including computer program code (e.g., Figure 5 The memory (1306) and computer program code are configured together with at least one processor (1304) to enable the device (e.g., 1300) to perform at least any of the example methods.

[0118] Implementations of the various technologies described herein can be implemented in digital electronic circuits, or in computer hardware, firmware, software, or combinations thereof. Implementations can be implemented as computer program products, i.e., computer programs tangibly embodied in an information carrier (e.g., in a machine-readable storage device or in a propagating signal) for execution by or control of a data processing device (e.g., a programmable processor, a computer, or multiple computers). Implementations can also be provided on computer-readable media or computer-readable storage media (which may be non-transitory media). Implementations of the various technologies can also include implementations provided via transient signals or media, and / or program and / or software implementations downloadable via the Internet or other networks (wired and / or wireless networks). Additionally, implementations can be provided via machine-type communication (MTC) and also via the Internet of Things (IoT).

[0119] As used herein, the term “circuit system” or “circuit” means all of the following: (a) a hardware circuit implementation only, such as an implementation in analog and / or digital circuits only; and (b) a combination of circuits and software (and / or firmware), such as (if applicable): (i) a combination of (multiple) processors or (ii) a portion of (multiple) processors / software, including (multiple) digital signal processors, software, and (multiple) memories, which work together to enable a device to perform various functions; and (c) circuits, such as (multiple) microprocessors or a portion of (multiple) microprocessors, which require software or firmware to operate, even if the software or firmware is not physically present. This definition of “circuit system” applies to all uses of the term in this application. As another example, as used herein, the term “circuit system” will also cover an implementation of a processor (or multiple processors) or a portion of a processor and its accompanying software and / or firmware only. For example, and if applicable to a particular element, the term “circuit system” will also cover a baseband integrated circuit or application processor integrated circuit for a mobile phone or a similar integrated circuit in a server, cellular network device, or another network device.

[0120] Computer programs can be in the form of source code, object code, or some intermediate form, and they can be stored on some carrier, distribution medium, or computer-readable medium, which can be any entity or device capable of carrying the program. Such carriers include, for example, recording media, computer memory, read-only memory, photoelectric and / or electrical carrier signals, telecommunication signals, and software distribution packages. Depending on the required processing power, a computer program can be executed in a single electronic digital computer, or it can be distributed across multiple computers.

[0121] Furthermore, implementations of the various technologies described herein can utilize network-physical systems (CPS) (systems that control collaborative computing elements of physical entities). CPS enables the implementation and utilization of a large number of interconnected ICT devices (sensors, actuators, processors, microcontrollers, etc.) embedded in physical objects at different locations. Mobile network-physical systems, which are inherently mobile physical systems, are a subcategory of network-physical systems. Examples of mobile physical systems include mobile robots and electronic devices transported by humans or animals. The increasing prevalence of smartphones has increased interest in the field of mobile network-physical systems. Therefore, various implementations of the technologies described herein can be provided via one or more of these technologies.

[0122] Computer programs such as those described above can be written in any programming language, including compiled or interpreted languages, and can be deployed in any form, including as a standalone program or as a module, component, subroutine, or other unit or part thereof suitable for use in a computing environment. Computer programs can be deployed to execute on a single computer, at a single site, or on multiple computers distributed across multiple sites and interconnected via a communication network.

[0123] The method steps can be executed by one or more programmable processors that execute a computer program or a portion thereof to perform a function by manipulating input data and generating output. The method steps can also be executed by special-purpose logic circuitry, and the apparatus can be implemented as special-purpose logic circuitry, such as an FPGA (Field-Programmable Gate Array) or an ASIC (Application-Specific Integrated Circuit).

[0124] Processors suitable for executing computer programs include, for example, general-purpose and special-purpose microprocessors, and any one or more processors of any kind of digital computer, chip, or chipset. Typically, a processor receives instructions and data from read-only memory or random access memory, or both. The elements of a computer may include at least one processor for executing instructions and one or more memory devices for storing instructions and data. Typically, a computer may also include one or more mass storage devices for storing data, such as magnetic disks, magneto-optical disks, or optical disks, or operatively coupled thereto to receive data from or transfer data to or both. Information carriers suitable for embodying computer program instructions and data include all forms of non-volatile memory, including, for example, semiconductor memory devices such as EPROM, EEPROM, and flash memory devices; magnetic disks, such as internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and memory may be supplemented by or incorporated into special-purpose logic circuitry.

[0125] To provide interaction with the user, the implementation can 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 pointing device, such as a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback, such as visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including sound, speech, or tactile input.

[0126] The implementation scheme can be implemented in a computing system that includes backend components, such as a data server, or middleware components, such as an application server, or frontend components, such as a client computer with a graphical user interface or web browser through which a user can interact with the implementation scheme, or any combination of such backend, middleware, or frontend components. Components can be interconnected via digital data communication of any form or medium, such as a communication network. Examples of communication networks include local area networks (LANs) and wide area networks (WANs), such as the Internet.

[0127] While certain features of the described embodiments have been shown as described herein, many modifications, substitutions, alterations, and equivalents will now occur to those skilled in the art. Therefore, it should be understood that the appended claims are intended to cover all such modifications and alterations falling within the true spirit of the various embodiments.

Claims

1. An apparatus comprising: At least one processor; as well as At least one memory storing instructions that, when executed by the at least one processor, cause the device to at least: The user equipment receives a continuous learning request from a network node, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; The user equipment determines the runtime capability limitations of the user equipment; The user equipment determines, based on its runtime capability limitations, that it cannot perform continuous learning for the machine learning model according to the received continuous learning configuration; The user equipment transmits runtime capability limitation information to the network node, the runtime capability limitation information being related to the user equipment's inability to perform continuous learning for the machine learning model based on the received continuous learning configuration; as well as The user equipment receives continuous learning configuration updates from the network node based on the runtime capability limitation information transmitted to the network node.

2. The apparatus of claim 1, wherein the continuous learning configuration comprises at least one of the following: The adaptive ratio, or the number or percentage of layers to be adapted to the machine learning model; Regarding the accuracy level of the machine learning model; Training periods used for continuous learning of the machine learning model; or The cycle used for continuous learning of the machine learning model.

3. The apparatus according to any one of claims 1 to 2, wherein the runtime capability limitation of the user equipment reflects the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: Based on available processing resources or based on the processing limitations of the user equipment at the current computing level; The memory limitations of the user equipment based on available memory resources or based on currently used memory resources; Has the power saving mode been activated for the user equipment? The battery or power status of the user equipment, or the available battery resources of the user equipment; or The user equipment is in an overheated state.

4. The apparatus according to any one of claims 1 to 3, wherein the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for the machine learning model includes at least one of the following: Information describing the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration or related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration; The user equipment's runtime capabilities limit the instruction that the user equipment cannot implement continuous learning of the machine learning model according to the continuous learning configuration; An indication of one or more runtime capability limitations of the user equipment, the one or more runtime capability limitations preventing the user equipment from implementing continuous learning of the machine learning model according to the continuous learning configuration; One or more user equipment proposals for one or more parameters of the continuous learning configuration; or One or more user device proposals for one or more parameters associated with the continuous learning of the machine learning model.

5. The apparatus according to any one of claims 1 to 4, wherein the received continuous learning configuration update received from the network node comprises at least one of the following: The updated continuous learning configuration includes at least one parameter that has been changed or updated based on the runtime capability limitation information transmitted from the user equipment to the network node; The network node's instruction to cancel the continuous learning request; and / or The network node provides instructions to the user equipment regarding the current version of the machine learning model.

6. The apparatus according to any one of claims 1 to 5, wherein the runtime capability limitation information transmitted from the user equipment to the network node includes one or more user equipment proposals for one or more parameters configured for the continuous learning or for one or more parameters associated with the continuous learning of the machine learning model, comprising one or more of the following: A proposal for the adaptive ratio or the optimal number or percentage of layers to be adapted to the machine learning model; Proposals regarding the accuracy level of the machine learning model; A proposal regarding power limitations for continuous learning of the machine learning model; A proposal regarding the cycle of continuous learning for the machine learning model; A proposal addressing the complexity constraints of continuous learning for the aforementioned machine learning model; A proposal for a delay budget for continuous learning of the machine learning model; or A proposal regarding training time constraints for continuous learning of the machine learning model.

7. The apparatus according to any one of claims 1 to 6, wherein the runtime capability limitation information is transmitted as or via at least one of the following: Information or one or more parameters provided within a user equipment assistance message; or One or more power saving parameters, or information or one or more parameters provided within a power saving information element.

8. A method comprising: The user equipment receives a continuous learning request from a network node, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning for a machine learning model; The user equipment determines the runtime capability limitations of the user equipment; The user equipment determines, based on its runtime capability limitations, that it cannot perform continuous learning for the machine learning model according to the received continuous learning configuration; The user equipment transmits runtime capability limitation information to the network node, the runtime capability limitation information being related to the user equipment's inability to perform continuous learning for the machine learning model based on the received continuous learning configuration; as well as The user equipment receives continuous learning configuration updates from the network node based on the runtime capability limitation information transmitted to the network node.

9. The method of claim 8, wherein the continuous learning configuration comprises at least one of the following: The adaptive ratio, or the number or percentage of layers to be adapted to the machine learning model; Regarding the accuracy level of the machine learning model; Training periods used for continuous learning of the machine learning model; or The cycle used for continuous learning of the machine learning model.

10. The method according to any one of claims 8 to 9, wherein the runtime capability limitation of the user equipment reflects the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: Based on available processing resources or based on the processing limitations of the user equipment at the current computing level; The memory limitations of the user equipment based on available memory resources or based on currently used memory resources; Has the power saving mode been activated for the user equipment? The battery or power status of the user equipment, or the available battery resources of the user equipment; or The user equipment is in an overheated state.

11. The method according to any one of claims 8 to 10, wherein the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for the machine learning model includes at least one of the following: Information describing the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration or related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration; The user equipment's runtime capabilities limit the instruction that the user equipment cannot implement continuous learning of the machine learning model according to the continuous learning configuration; An indication of one or more runtime capability limitations of the user equipment, the one or more runtime capability limitations preventing the user equipment from implementing continuous learning of the machine learning model according to the continuous learning configuration; One or more user equipment proposals for one or more parameters of the continuous learning configuration; or One or more user device proposals for one or more parameters associated with the continuous learning of the machine learning model.

12. The method according to any one of claims 8 to 11, wherein the received continuous learning configuration update received from the network node comprises at least one of the following: The updated continuous learning configuration includes at least one parameter that has been changed or updated based on the runtime capability limitation information transmitted from the user equipment to the network node; The network node's instruction to cancel the continuous learning request; and / or The network node provides instructions to the user equipment regarding the current version of the machine learning model.

13. The method of any one of claims 8 to 12, wherein the runtime capability limitation information transmitted from the user equipment to the network node includes one or more user equipment proposals for one or more parameters configured for the continuous learning or for one or more parameters associated with the continuous learning of the machine learning model, comprising one or more of the following: A proposal for the adaptive ratio or the optimal number or percentage of layers to be adapted to the machine learning model; Proposals regarding the accuracy level of the machine learning model; A proposal regarding power limitations for continuous learning of the machine learning model; A proposal regarding the cycle of continuous learning for the machine learning model; A proposal addressing the complexity constraints of continuous learning for the aforementioned machine learning model; A proposal for a delay budget for continuous learning of the machine learning model; or A proposal regarding training time constraints for continuous learning of the machine learning model.

14. The method according to any one of claims 8 to 13, wherein the runtime capability limitation information is transmitted as or via at least one of the following: Information or one or more parameters provided within a user equipment assistance message; or One or more power saving parameters, or information or one or more parameters provided within a power saving information element.

15. An apparatus comprising: At least one processor; as well as At least one memory storing instructions that, when executed by the at least one processor, cause the device to at least: A continuous learning request is transmitted from a network node to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning on a machine learning model; The network node receives runtime capability limitation information from the user equipment, the runtime capability limitation information being related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration based on the user equipment's runtime capability limitations; The network node determines the continuous learning configuration update based on the received runtime capability limitation information; as well as The network node transmits the continuous learning configuration update to the user equipment.

16. The apparatus of claim 15, wherein the continuous learning configuration comprises at least one of the following: The adaptive ratio, or the number or percentage of layers to be adapted to the machine learning model; Regarding the accuracy level of the machine learning model; Training periods used for continuous learning of the machine learning model; or The cycle used for continuous learning of the machine learning model.

17. The apparatus of any one of claims 15 to 16, wherein the runtime capability limitation of the user equipment reflects the current state of the user equipment relative to one or more limitations of the user equipment, including at least one of the following: Based on available processing resources or based on the processing limitations of the user equipment at the current computing level; The memory limitations of the user equipment based on available memory resources or based on currently used memory resources; Has the power saving mode been activated for the user equipment? The battery or power status of the user equipment, or the available battery resources of the user equipment; or The user equipment is in an overheated state.

18. The apparatus of any one of claims 15 to 17, wherein the runtime capability limitation information relating to the user equipment's inability to perform continuous learning for the machine learning model includes at least one of the following: Information describing the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration or related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration; The user equipment's runtime capabilities limit the instruction that the user equipment cannot implement continuous learning of the machine learning model according to the continuous learning configuration; An indication of one or more runtime capability limitations of the user equipment, the one or more runtime capability limitations preventing the user equipment from implementing continuous learning of the machine learning model according to the continuous learning configuration; One or more user equipment proposals for one or more parameters of the continuous learning configuration; or One or more user device proposals for one or more parameters associated with the continuous learning of the machine learning model.

19. The apparatus according to any one of claims 15 to 18, wherein the continuous learning configuration update comprises at least one of the following: The updated continuous learning configuration includes at least one parameter that has been changed or updated based on the runtime capability limitation information transmitted from the user equipment to the network node; The network node's instruction to cancel the continuous learning request; and / or The network node provides instructions to the user equipment regarding the current version of the machine learning model.

20. The apparatus of any one of claims 15 to 19, wherein the runtime capability limitation information received by the network node from the user equipment includes one or more user equipment proposals for one or more parameters configured for the continuous learning or for one or more parameters associated with the continuous learning of the machine learning model, comprising one or more of the following: A proposal for the adaptive ratio or the optimal number or percentage of layers to be adapted to the machine learning model; Proposals regarding the accuracy level of the machine learning model; A proposal regarding power limitations for continuous learning of the machine learning model; A proposal regarding the cycle of continuous learning for the machine learning model; A proposal addressing the complexity constraints of continuous learning for the aforementioned machine learning model; A proposal for a delay budget for continuous learning of the machine learning model; or A proposal regarding training time constraints for continuous learning of the machine learning model.

21. The apparatus according to any one of claims 15 to 20, wherein the runtime capability limitation information is received as or via at least one of the following: Information or one or more parameters provided within a user equipment assistance message; or One or more power saving parameters, or information or one or more parameters provided within a power saving information element.

22. A method comprising: A continuous learning request is transmitted from a network node to a user equipment, the continuous learning request including a continuous learning configuration for the user equipment to perform continuous learning on a machine learning model; The network node receives runtime capability limitation information from the user equipment, the runtime capability limitation information being related to the user equipment's inability to perform continuous learning for the machine learning model according to the continuous learning configuration based on the user equipment's runtime capability limitations; The network node determines the continuous learning configuration update based on the received runtime capability limitation information; as well as The network node transmits the continuous learning configuration update to the user equipment.

Images