System and device for determining model accuracy and associated method

Abstract

A system (100), a device (102), and a method (300) for determining model accuracy are disclosed. The method (300) comprises an input step (302) that includes receiving at least one input signal associated with a model quality threshold range; and a processing step (304) that includes: obtaining a model quality performance value based on derived inference of a current model; and / or determining an accuracy of the current model by analyzing the model quality performance value with the model quality threshold range.

Description

Field of the invention

[0001] The present disclosure generally relates to a system or a device, or both, for determining model accuracy, for example, in connection with a user equipment (UE) usable for communication. The present disclosure further relates to a method that can be associated with the system and / or the device. Background of the invention

[0002] In general, energy efficiency and power saving would be helpful in communication networks, for example a telecommunications network based on the 5G (fifth generation) NR (New Radio) standard of the 3rd Generation Partnership Project (3GPP).

[0003] Current techniques may not address the issue of determining model accuracy by a base station or user equipment (UE) in a communications network. This can lead to problems such as inaccurate inference, unnecessary energy consumption when running an inaccurate model (e.g., an artificial intelligence / machine learning-based model or AI / ML model), and accumulated error in the model (e.g., an artificial intelligence / machine learning-based model or AI / ML model), leading to performance degradation. Thus, current techniques may not enable optimal energy efficiency and power savings.

[0004] The present disclosure contemplates that it would be helpful to address or at least mitigate one or more issues associated with conventional techniques to enable energy efficiency and power conservation when determining model accuracy. Brief description of the invention

[0005] According to a first aspect of the present invention, a method for determining model accuracy is provided, the method comprising: an input step comprising receiving at least one input signal associated with a model quality threshold range; and a processing step comprising: obtaining a model quality performance value based on derived inference of a current model; and / or determining an accuracy of the current model by analyzing the model quality performance value with the model quality threshold range.

[0006] Advantageously, the method as described herein can cause a user equipment (UE) to perform inference based on an accurate model. This can reduce the effort required to minimize the signaling of inference from a less accurate model.

[0007] In one embodiment, the method further comprises requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range; and configuring the model based on the model quality performance value if the model quality performance value is within the threshold range.

[0008] In one embodiment, the method further comprises determining at least one updated model parameter; communicating the at least one updated model parameter; and configuring the model based on the at least one updated model parameter.

[0009] In one embodiment, the model quality threshold range is associated with a model quality parameter range (MQRP).

[0010] In one embodiment, the model quality parameter range (MQRP) includes: time range, inference accuracy range, and / or number of data points in a dataset.

[0011] In one embodiment, the model is an artificial intelligence / machine learning (AIML) model.

[0012] In one embodiment, at least one base station is configured to: predetermine the model quality threshold range; and communicate the model quality threshold range.

[0013] In one embodiment, the at least one base station corresponds to at least one next generation node B (gNB).

[0014] In one embodiment, a user equipment (UE) is configured to perform the input step and the processing step, wherein the model quality threshold range is communicable from the gNB to the UE.

[0015] In one embodiment, the model quality threshold range is received by the UE from the gNB via: System Information Block (SIB), Master Information Block (MIB), and / or UE-specific message.

[0016] In one embodiment, a computer program is provided comprising instructions which, when executed by a computer, cause the computer to perform the input step and / or the processing step according to the method of the first aspect.

[0017] In one embodiment, a computer-readable storage medium is provided having stored therein data representing computer-executable software, the software comprising instructions which, when executed by the computer, perform the input step and / or the processing step according to the method of the first aspect.

[0018] In one embodiment, an apparatus for determining model accuracy is provided, comprising: a first module configured to receive at least one input signal associated with a model quality threshold range; a second module configured to process and / or enable the processing step according to the method of the first aspect to generate at least one output signal; and a third module configured to communicate at least one output signal, wherein the output signal corresponds to a control signal for requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range, and for configuring the model based on the model quality performance value if the model quality performance value is within the threshold range.

[0019] In one embodiment, the device (102) corresponds to a user equipment (UE) that is communicable with a device (104) corresponding to a base station, wherein the base station corresponds to a next generation node B (gNB) configured to communicate the at least one input signal to the UE.

[0020] In one embodiment, a system is provided comprising: (at least one) device(s); and (at least one) apparatus(es), wherein the device(s) and the apparatus(es) are connectable via wired coupling and / or wireless coupling.

[0021] Advantageously, the system as disclosed herein may exhibit energy efficiency by reducing accumulated errors and accurate model determination. Short description of the drawings

[0022] Embodiments of the disclosure are described below with reference to the following drawings. It shows: Fig. 1A is a schematic diagram illustrating a system for determining model accuracy, which may include at least one device, according to an embodiment of the invention; Fig. 1B an example scenario in connection with the system of Fig. 1A according to an embodiment of the invention; Fig. 2 a schematic diagram showing the device of Fig. 1A illustrates in more detail according to an embodiment of the invention; Fig. 3 a procedure in connection with the system of Fig. 1A according to an embodiment of the invention; Fig. 4A to Fig. 4C schematic diagrams showing the information flow associated with the process of Fig. 3 according to an embodiment of the invention. Detailed description

[0023] The present description discloses an apparatus for performing the acts of the methods. Such an apparatus may be specially constructed for the required purposes or may comprise a computer or other device selectively activated or reconfigured by a computer program stored in the computer. The algorithms and displays presented herein are not tied to any particular computer or other apparatus. Various machines with programs in accordance with the present teachings may be used. Alternatively, the construction of a more specialized apparatus for performing the required method steps may be indicated. The structure of a computer will be apparent from the description below.

[0024] Furthermore, the present description also implicitly discloses a computer program, as it is obvious to a person skilled in the art that the individual steps of the method described herein can be implemented by computer code. The computer program is not intended to be limited to a particular programming language and implementation thereof. It is understood that a variety of programming languages ​​and codings thereof can be used to implement the teachings of the disclosure contained herein. Furthermore, the computer program is not intended to be limited to a particular control sequence. There are many other variants of the computer program that can use other control sequences without departing from the spirit or scope of the disclosure.

[0025] Furthermore, one or more of the steps of the computer program may be performed in parallel rather than sequentially. Such a computer program may be stored on any computer-readable medium. The computer-readable medium may comprise storage devices such as magnetic or optical disks, memory chips, or other storage devices suitable for interoperating with a computer. The computer-readable medium may also comprise a hard-wired medium such as the Internet system or a wireless medium such as the mobile telephone system. The computer program, when loaded onto and executed on such a computer, effectively results in a device that implements the steps of the preferred method.

[0026] The present disclosure generally contemplates enabling and optimizing a network (e.g., in conjunction with a 3GPP-based standard / specification, etc.) and / or the efficiency and mobility (e.g., energy efficiency or power savings) of a user equipment (UE) according to an embodiment of the invention. Specifically, the present disclosure contemplates the possibility of determining a model accuracy of a UE in conjunction with a 3GPP standard(s).

[0027] This disclosure generally contemplates that data collection in communication networks may be performed for various life cycle management (LCM) purposes, such as model training, model inference, model monitoring, model selection, model updating, etc., each of which may have different requirements and impact on a potential specification. Some examples that may have a specification impact include measurement configuration and reporting, content, type, and format of data, including data pertaining to model input, data pertaining to ground truth, data quality, and / or other relevant information.

[0028] The present disclosure contemplates that collaboration levels may be defined by having network-UE collaboration levels as one aspect. Examples of network-UE collaboration include level x, where there is no collaboration, and level y, where collaboration is signaling-based without model transfer. Such a collaboration level (i.e., level y) may include cases without model deployment. Another example may be level z, where signaling-based collaboration with model transfer exists.

[0029] The present disclosure also contemplates that performing intermediate evaluations of the model performance, for example, of a model based on artificial intelligence / machine learning (AI / ML or AIML), may be considered to derive intermediate key performance inputs (KPIs), such as AI / ML model accuracy and channel status information (CSI) output, for the purpose of comparing solutions related to the model (e.g., AI / ML model). Furthermore, the present disclosure contemplates that having an accurate model (e.g., AI / ML model) for deriving inference may be highly important to determine the precise outputs based on the proposed use cases.

[0030] The present disclosure further contemplates that network management can be improved by utilizing three different models in an AI / ML communication link (e.g., air interface). The models can be network-side (e.g., a base station or a next-generation Node B (gNB)) models, UE-side models, and both UE-side and network-side models, where one model (e.g., AI / ML model) can be present at both the UE-side and / or network-side end devices.

[0031] The present disclosure considers how a network can configure a model quality parameter range (MQRP) for the UE to determine whether the model is accurate enough to derive an inference or whether a model update is required before executing the model (e.g., AI / ML model). In this way, energy consumption in the network can be optimized.

[0032] The present disclosure contemplates that the exact definition of model accuracy may not be described in current systems and techniques, which may result in inaccurate inference. This may also lead to accumulated error in the model (e.g., AI / ML model), resulting in degradation of network performance. Furthermore, unnecessary energy may be consumed to execute an inaccurate model (e.g., AI / ML model). The present disclosure contemplates that it may be important for the UE or base station (e.g., gNB) to determine the model accuracy when executing the model (e.g., AI / ML model).

[0033] The present disclosure contemplates that the base station (e.g., gNB) may configure an AI / ML MQRP for the UE. The accuracy range may be any or all of the quality parameters, such as, but not limited to, the time range, the inference accuracy range, and the number of data points in the dataset. The base station (e.g., gNB) may signal the parameter range(s) to the UE via, for example, but not limited to, a System Information Block (SIB), a Master Information Block (MIB), a UE-specific message, or RRC configuration messages.

[0034] The present disclosure contemplates that upon receiving the accuracy threshold, the UE may perform the following operations. In one embodiment, the UE may compare the derived inference if it is above the threshold range. If the model quality performance is within the model quality parameter range, the UE proceeds with the inference of the AI / ML model. If the model quality performance is outside the model quality parameter range, the UE requests an update of the model (e.g., AI / ML model) from the base station (e.g., gNB). Upon receiving the request to update the model (e.g., AI / ML model), the base station (e.g., gNB) may provide the updated model parameters (e.g., AI / ML model parameters) to the UE.

[0035] In the above manner, the UE can perform inference based on an accurate model, thus reducing the effort required to minimize signaling of inference from an inaccurate model. According to one embodiment of the invention, power savings and energy consumption efficiency may be enabled in the UE or the network.

[0036] The above is explained below with reference to Fig. 1 to Fig. 4 is discussed in more detail.

[0037] Referring to Fig. Figure 1A shows a schematic diagram illustrating a system 100 for determining model accuracy according to one embodiment of the invention. The system 100 may be suitable, for example, for enabling energy and power efficiency improvements according to one embodiment of the invention.

[0038] As shown, according to an embodiment of the invention, the system 100 may include one or more devices 102, at least one apparatus 104, and optionally a communications network 106.

[0039] The device(s) 102 may be coupled to the device(s) 104. Specifically, according to one embodiment of the invention, the device(s) 102 may be coupled to the device(s) 104, for example, via the communications network 106.

[0040] In one embodiment, the device(s) 102 may be coupled to the communication network 106, and the device(s) 104 may be coupled to the communication network 106. The coupling may be via a wired coupling or a wireless coupling, or both. The device(s) 102 may be generally configured to communicate with the device(s) 104 via the communication network 106, according to one embodiment of the invention.

[0041] For example, according to one embodiment of the invention, the device(s) 102 may be associated with, correspond to, or include one or more user equipment (UE) that may carry one or more computers. For example, according to one embodiment of the invention, a device 102 may correspond to a UE that carries at least one computer (e.g., an electronic device or module with computing capabilities, such as a mobile electronic device that may be brought into a vehicle or an electronic module that may be installed in a vehicle, according to one embodiment of the invention) that may be configured to perform one or more processing tasks associated with adaptive / dynamic / gradual control.

[0042] In one embodiment, the device(s) 102 may, for example, be configured to receive one or more input signals and to perform at least one processing task based on the input signal(s) in a manner that generates one or more output signals. The input signal(s) may, for example, be communicated by the device(s) 104 and received by the device(s) 102 according to one embodiment of the invention. The input signal may be associated with a model quality threshold range. As a possible option, the output signal(s) may, for example, be communicated by the device(s) 102 according to one embodiment of the invention. The output signal may correspond to a control signal for determining a model accuracy. The device(s) 102 according to one embodiment of the invention will be described later with reference to Fig. 2 is discussed in more detail.

[0043] The device(s) 104 may, for example, be associated with / correspond to at least one base station, wherein the at least one base station may be a Next Generation Node B (gNB). Furthermore, the device(s) 104 may, for example, be configured to carry / be associated with / include one or more computers (e.g., an electronic device / module with computing capabilities), which may, for example, be configured to perform one or more processing tasks in conjunction with the base station. The device(s) 104 may, according to an embodiment of the invention, be configured to generate one or more input signals that may be communicated to the device(s) 102. This will be discussed in more detail later in the context of an example scenario according to an embodiment of the invention.

[0044] The communication network 106 may, for example, correspond to an internet communication network, a cellular-based communication network, a wired communication network, a Global Navigation Satellite System (GNSS)-based communication network, a wireless communication network, or any combination thereof. Communication (e.g., between the devices 102 and / or between the device(s) 102 and the device(s) 104) via the communication network 106 may be via wired communication or wireless communication, or both.

[0045] As mentioned, the device(s) 102 may, for example, be configured to receive at least one input signal and to perform at least one processing task in connection with dynamic / adaptive / gradual control with respect to the input signal(s) in such a way that at least one output signal is generated. Furthermore, the device(s) 104 according to an embodiment of the invention may, for example, be configured to generate (and communicate) the input signal(s) to the device(s) 102. Accordingly, the device(s) 104 may predetermine the model quality threshold range (MQRP) and also communicate the MQRP to the device(s) 102. This will be described below according to an embodiment of the invention in the context of an example scenario with reference to Fig. 1B discussed.

[0046] Fig. 1B shows an example scenario in connection with the system of Fig. 1A according to an embodiment of the invention. Specifically, Fig. Figure 1B shows an example of a functional structure for a New Radio air interface based on artificial intelligence / machine learning (AI / ML). As shown in the figure, a data acquisition module can output data, such as training data, monitoring data, and inference data, to various modules, such as a model training module, a network management module, and an inference module.

[0047] The network management module may receive monitoring data from the data acquisition module and inference outputs from the inference module and transmit or communicate performance feedback or a refresh training request to the model training module. The network management module may also communicate management instructions to the inference module and a model transfer or submission request to a model storage module.

[0048] The inference module may receive the inference data from the data acquisition module, the management instruction from the network management module, and a model transfer or transmission from the model storage module to process them and generate the inference output that goes to the network management module.

[0049] The model training module receives the training data from the data acquisition module and the performance feedback or refresh training request from the network management module to process them and generate a trained or updated model, which goes to the model storage module.

[0050] The model storage module may receive the model transfer or submission request from the network management module and the trained or updated model from the model training module to process them and generate the model transfer or submission to be passed to the inference module.

[0051] The above-described aspect(s) of the system 100 of the present invention may also apply analogously to the aspect(s) of a device 102 of the present invention described below. Likewise, the below-described aspect(s) of the device 102 of the invention may also apply analogously to the aspect(s) of the system 100 of the invention described above.

[0052] The aforementioned device(s) 102 or user equipment(s) (UE) will be referred to below with reference to Fig. 2 is discussed in more detail.

[0053] Referring to Fig. 2 is a schematic diagram illustrating a device 102, shown in more detail in the context of an exemplary implementation 200 according to an embodiment of the invention.

[0054] In the example implementation 200, the device 102 may correspond to an electronic module 200a. In one example, the electronic module 200a may correspond to a mobile device, for example, that may be brought into the vehicle by a user, according to an embodiment of the invention. In another example, the electronic module 200a may correspond to an electronic device that may be installed / mounted in the vehicle, according to an embodiment of the invention. In this regard, the electronic module 200a may be considered to be carried by the vehicle (e.g., either brought into the vehicle by a user or installed / mounted in the vehicle).

[0055] It is contemplated that the electronic module 200a according to an embodiment of the invention may be capable of performing one or more processing tasks associated with processing related to adaptive / dynamic / gradual control.

[0056] The electronic module 200a may, for example, include a housing 200b. Furthermore, the electronic module 200a may, for example, support a first module 202, a second module 204, or a third module 206, or any combination thereof.

[0057] In one embodiment, the electronic module 200a may carry a first module 202, a second module 204, and / or a third module 206. In a specific example, the electronic module 200a according to an embodiment of the invention may carry a first module 202, a second module 204, and a third module 206.

[0058] In this regard, it will be appreciated that in one embodiment, the housing 200b may be shaped and sized to support the first module 202, the second module 204, or the third module 206, or any combination thereof.

[0059] The first module 202 may be coupled to the second module 204 or the third module 206, or both. The second module 204 may be coupled to the first module 202 or the third module 206, or both. The third module 206 may be coupled to the first module 202 or the second module 204, or both. In one example, according to an embodiment of the invention, the first module 202 may be coupled to the second module 204, and the second module 204 may be coupled to the third module 206. The coupling between the first module 202, the second module 204, and / or the third module 206 may, for example, be via wired coupling or wireless coupling, or both. The first module 202, the second module 204, and the third module 206 may each correspond to a hardware-based module or a software-based module, or both, according to an embodiment of the invention.

[0060] In one example, the first module 202 may correspond to a hardware-based receiver that may be configured to receive one or more input signals. The input signal(s) may be communicated, for example, by the device(s) 104 (or the base station, e.g., a gNB) according to an embodiment of the invention.

[0061] The second module 204 may, according to one embodiment of the invention, correspond, for example, to a hardware-based processor that may be configured to perform one or more processing tasks (e.g., in a manner that generates one or more output signals), as described later with reference to Fig. 3 is discussed in more detail.

[0062] The third module 206 may correspond to a hardware-based transmitter that may be configured to communicate one or more output signals from the electronic module 200a. The output signal(s) may, for example, comprise one or more instructions / commands / control signals in connection with the aforementioned dynamic / adaptive / gradual control configuration / determination strategy to enable efficiency (e.g., power / energy efficiency and / or communication efficiency) according to one embodiment of the invention. For example, the output signal(s) may be control signals for determining model accuracy.

[0063] The present disclosure contemplates the possibility that the first and second modules 202, 204 may be an integrated software / hardware-based module, for example, an electronic part capable of carrying a software program or algorithm associated with receiving and processing functions, or an electronic module programmed to perform the receiving and processing functions. The present disclosure further contemplates the possibility that the first and third modules 202, 206 may be an integrated software / hardware-based module, for example, an electronic part capable of carrying a software program or algorithm associated with receiving and transmitting functions, or an electronic module programmed to perform the receiving and transmitting functions.The present disclosure also further contemplates the possibility that the first and third modules 202, 206 may be an integrated hardware module, such as a hardware-based transceiver, capable of performing the functions of receiving and transmitting.

[0064] For example, according to an embodiment of the invention, the UE may be further configured to process the input signal(s) in a manner as described later with reference to Fig. 3, one or more output signals are generated in a manner that enables efficiency, such as power efficiency or energy efficiency. In a specific example, the output signal(s) according to an embodiment of the invention may comprise one or more control signals to enable some form of dynamic / adaptive / gradual control configuration / determination strategy to enable efficiency, such as power efficiency or energy efficiency. For example, the output signal(s) may be control signals for determining model accuracy.

[0065] The advantageous aspect(s) of the device 102 of the present invention described above may also apply analogously to the aspect(s) of a processing / communication method of the present invention described below. Likewise, the aspect(s) of the method of the invention described below may also apply analogously to the aspect(s) of the device 102 of the invention described above. It should be appreciated that these statements apply analogously to the previously discussed system 100 of the present disclosure.

[0066] Referring to Fig. 3, a method 300 (or a communication method) for determining model accuracy, such as an artificial intelligence / machine learning (AIML or AI / ML) model, is shown in conjunction with the system 100 according to an embodiment of the invention.

[0067] The method 300 may be suitable, for example, for enabling energy efficiency, grid optimization, and power savings according to one embodiment of the invention.

[0068] According to one embodiment of the invention, the method 300 may include an input step 302, a processing step 304, or an output step 306, or any combination thereof.

[0069] In one embodiment, processing method 300 may include input step 302. In another embodiment, processing method 300 may include input step 302 and processing step 304. In another embodiment, processing method 300 may include input step 302, processing step 304, and output step 306. In yet another embodiment, processing method 300 may include processing step 304 and input step 302 or output step 306, or both. In yet another embodiment, processing method 300 may include input step 302, processing step 304, and output step 306. In yet another additional embodiment, processing method 300 may include processing step 304.In yet another further additional embodiment, the processing method 300 may include the input step 302, the processing step 304, or the output step 306, or any combination thereof (ie, the input step 302, the processing step 304, and / or the output step 306).

[0070] With respect to input step 302, one or more input signals may be received. For example, according to one embodiment of the invention, the input signal(s) may be communicated by device(s) 104 and received by device(s) 102.

[0071] The input step 302 may include receiving at least one input signal associated with a model quality threshold range. The model quality threshold range may be associated with a model quality parameter range (MQRP), which may include: time range, inference accuracy range, and / or the number of data points in a data set. The model quality parameter range (MQRP) may be predetermined by at least one base station (or device 104). The at least one base station may correspond to at least one next-generation node B (gNB), which may communicate the model quality threshold range to the device 102 (or UE).The model quality threshold range can be received by the UE from the gNB via System Information Block (SIB), Master Information Block (MIB) and / or UE-specific message (e.g., Radio Resource Control (RRC) reconfiguration).

[0072] With respect to processing step 304, according to one embodiment of the invention, at least one processing task may be performed in connection with the received input signal(s) in a manner that generates one or more output signals.

[0073] Processing step 304 may include obtaining a model quality performance value based on derived inference of a current model; and / or determining an accuracy of the current model by analyzing the model quality performance value with the model quality threshold range. Processing step 304 may further include requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range; and configuring the model based on the model quality performance value if the model quality performance value is within the threshold range.

[0074] Processing step 304 may further include determining at least one updated model parameter; communicating the at least one updated model parameter; and configuring the model based on the at least one updated model parameter. In one embodiment, processing step 304 may be performed by the device 102 (UE) as well as the apparatus 104 (base station or gNB).

[0075] With respect to output step 306, the output signal(s) may be communicated as an option, for example, according to one embodiment of the invention. For example, the output signal(s) may optionally be communicated by the device 102. In a more specific example, the output signal(s) may optionally be communicated by the device 102 to at least one device 104 or another device 102, or both, according to one embodiment of the invention. In one embodiment, the device 102 (or UE) may also perform the input step 302, the processing step 304, and the output step 306.

[0076] The present disclosure further contemplates a computer program (not shown) that may include instructions that, when executed by a computer (not shown), cause the computer to perform input step 302, processing step 304, and / or output step 306, as discussed with reference to method 300. For example, according to one embodiment of the invention, the computer program may include instructions that, when executed by a computer, cause the computer to perform input step 302 and / or processing step 304.

[0077] The present disclosure further contemplates a computer-readable storage medium (not shown) having stored therein data representing software executable by a computer (not shown), the software comprising instructions that, when executed by the computer, cause the performance of input step 302, processing step 304, and / or output step 306, as discussed with reference to method 300. For example, according to one embodiment of the invention, the computer-readable storage medium may have stored therein data representing computer-executable software, the software comprising instructions that, when executed by the computer, cause the computer to perform input step 302 and / or processing step 304.

[0078] Further, in view of the foregoing, it will be appreciated that the present disclosure generally contemplates an apparatus 102 for determining model accuracy, which may include a first module 202, a second module 204, and / or a third module 206.

[0079] The first module 202 may be configured to receive one or more input signals. The input signal(s) may, for example, be associated with a model quality threshold range.

[0080] The second module 204 may be configured to process the input signal(s) according to the method 300, as previously discussed, and / or to enable such processing to generate one or more output signals.

[0081] The third module 206 may be configured to communicate one or more output signals. The output signal(s) may, for example, correspond to one or more control signals for requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range, and configuring the model based on the model quality performance value if the model quality performance value is within the threshold range.

[0082] In one embodiment, device 102 may correspond to a user equipment (UE) that can communicate with a device 104 that corresponds to a base station. The base station may, for example, correspond to a next-generation Node B (gNB), which may be configured to communicate one or more signals (e.g., an input signal(s)) to the UE.

[0083] Furthermore, in view of the foregoing, it will be appreciated that the present disclosure generally contemplates a system 100 that may include one or more devices 102 and one or more apparatuses 104. The device(s) 102 and the apparatus(es) 104 may be coupled, for example, via wired coupling and / or wireless coupling.

[0084] It should be appreciated that the embodiments described above may be combined in any manner (e.g., one or more embodiments as discussed in the "Detailed Description" section may be combined with one or more embodiments as described in the "Summary of the Invention" section).

[0085] Furthermore, those skilled in the art should recognize that variations and combinations of embodiments described above that are not alternatives or replacements may be combined to form still further embodiments.

[0086] In one example, the possibility that the output signal(s) is / are communicated by the device(s) 102 was discussed. It will be appreciated that the output signal(s) do not necessarily have to be communicated by the device(s) 102. Specifically, according to one embodiment of the invention, the possibility is contemplated that the output signal(s) do not necessarily have to be communicated outside of the device(s) 102. More specifically, according to one embodiment of the invention, the output signal(s) may, for example, correspond to internal command(s) / instruction(s) (e.g., communicated only within a device 102) for adaptively controlling the operating configuration of a device 102.

[0087] In another example, according to an embodiment of the invention, application(s) of the present disclosure may be possible in connection with / in the context of a wireless low power device(s) and / or ambient IoT (Internet of Things) device(s).

[0088] Fig. 4A to Fig. 4C show schematic representations illustrating the information flow associated with the method of Fig. 3 according to an embodiment of the invention.

[0089] In the example context, as in Fig. 4A, a gNB (or base station) according to an embodiment of the invention may, for example, be configured to generate / define / (pre)configure / set an AI / ML model accuracy threshold (e.g., model quality threshold range) for the UE and / or communicate one or more input signals that may correspond to, be associated with, or include the generated / defined / (pre)configured / set model quality threshold range(s). Furthermore, the gNB (or base station) according to an embodiment of the invention may, for example, be configured to provide or generate a model parameter update for the UE.

[0090] In the example context, as in Fig. For example, as shown in Figure 4B, a UE may be configured to receive a model quality parameter range (or model quality threshold range). Then, the UE determines an accuracy of derived inference by obtaining a model quality performance value based on derived inference of a current model and analyzing the model quality performance value with the model quality parameter range (or model quality threshold range). If the model quality performance value is not within the model quality parameter range (or model quality threshold range), the UE may request at least one model parameter update from the gNB (or the base station). On the other hand, if the model quality performance value is within the model quality parameter range (or model quality threshold range), the UE may use the derived inference by configuring the model based on the model quality performance value.

[0091] In the example context, as in Fig.4C, a user equipment (UE) may be configured to receive, at step 1, one or more input signals communicable from the gNB (or the base station). The gNB may, for example, predetermine or configure a model quality parameter range (or model quality threshold range), and the UE may be configured to process the input signal(s). The input signal may include or be associated with the model quality parameter range (or model quality threshold range). At step 2, the UE may, for example, be configured to determine (e.g., detect or measure) whether a model quality is satisfied by analyzing the model quality parameter range (or model quality threshold range) with a model quality performance value based on derived inference of a current model.If the model quality is not met or if the model quality performance value is not within the model quality parameter range (or model quality threshold range), the UE may request a model (or model parameter) update from the gNB (or base station) in step 3. Upon receiving the request, the gNB (or base station) provides the model update response to the UE in step 4. In step 5, the UE updates the model parameters using the updated model from the gNB (or base station).

[0092] In the foregoing manner, various embodiments of the disclosure have been described to address at least one of the above disadvantages. These embodiments are intended to be encompassed by the following claims and are not limited to the specific forms or arrangements of parts so described, and it will be apparent to those skilled in the art, in light of this disclosure, that numerous changes and / or modifications may be made, which are also intended to be encompassed by the following claims.

Claims

[1] A method (300) for determining a model accuracy, the method comprising: an input step (302) comprising receiving at least one input signal associated with a model quality threshold range; and a processing step (304) comprising: Obtaining a model quality performance value based on derived inference of a current model; and / or Determine an accuracy of the current model by analyzing the model quality performance value with the model quality threshold range. [2] The method (300) of claim 1, wherein the method step (304) further comprises: Requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range; and Configure the model based on the model quality performance value if the model quality performance value is within the threshold range. [3] The method (300) of claim 1, wherein the method step (304) further comprises: Determining at least one updated model parameter; Communicating the at least one updated model parameter; and Configuring the model based on the at least one updated model parameter. [4] The method (300) of claim 1, wherein the model quality threshold range is associated with a model quality parameter range (MQRP). [5] The method (300) of claim 4, wherein the model quality parameter range (MQRP) comprises: time range, inference accuracy range, and / or number of data points in a data set. [6] The method (300) of claim 1, wherein the model is an artificial intelligence / machine learning (AIML) based model. [7] The method (300) of claim 1, wherein at least one base station is configured to: Predetermining the model quality threshold range; and Communicating the model quality threshold range. [8] The method (300) of claim 7, wherein the at least one base station corresponds to at least one next generation node B (gNB). [9] The method (300) of claim 8, wherein a user equipment (UE) is configured to perform the input step (302) and the processing step (304), and wherein the model quality threshold range is communicable from the gNB to the UE. [10] The method (300) of claim 9, wherein the model quality threshold range is received by the UE from the gNB via: System Information Block (SIB), Master Information Block (MIB) and / or UE-specific message. [11] A computer program comprising instructions which, when executed by a computer, cause the computer to perform the input step (302) and / or the processing step (304) according to the method (300) of any one of the preceding claims. [12] A computer-readable storage medium having stored therein data representing computer-executable software, the software comprising instructions which, when executed by the computer, cause the input step (302) and / or the processing step (304) to be performed according to the method (300) of claims 1-10. [13] Device (102) for determining a model accuracy, comprising: a first module (202) configured to receive at least one input signal associated with a model quality threshold range; a second module (204) configured to process and / or enable the processing step (304) according to the method (300) of claim 1 to claim 10 to generate at least one output signal; and a third module (206) configured to communicate at least one output signal, wherein the output signal corresponds to a control signal for requesting at least one model parameter update if the model quality performance value is not within the model quality threshold range, and for configuring the model based on the model quality performance value if the model quality performance value is within the threshold range. [14] Device (102) according to claim 13, wherein the device (102) corresponds to a user equipment (UE) capable of communication with a device (104) corresponding to a base station, and wherein the base station corresponds to a next generation node B (gNB) configured to communicate the at least one input signal to the UE. [15] System (100) comprising: at least one device (102) according to one of claims 13 and 14; and at least one device (104) according to claim 14, wherein the device (102) and the apparatus (104) are connectable by wired coupling and / or wireless coupling.

Images