Electronic device, method for electronic device, computer readable storage medium, and computer program product

Abstract

The present disclosure provides an electronic device, a method for an electronic device, a computer readable storage medium, and a computer program product. The electronic device comprises at least one processor; and at least one memory including computer program code, wherein the at least one memory and the computer program code are configured to, with the at least one processor, cause the electronic device to perform: transmitting, to a network-side device, a data usage indication, the data usage indication being used to indicate whether a usage of measurement data transmitted by the terminal device to the network-side device is a first usage or a second usage, the first usage being used for obtaining an inferred decision result from a first artificial intelligence model arranged on the network-side device, the second usage being used for obtaining an inferred intermediate result from the first artificial intelligence model for updating a second artificial intelligence model that is a miniaturized version of the first artificial intelligence model arranged on the terminal device; and receiving, from the network-side device, the decision result or the intermediate result.

Description

Technical Field

[0001] This application relates to the fields of wireless communication and / or artificial intelligence (AI) technology, specifically to distributed intelligent reasoning and online model update technology. More specifically, it relates to electronic devices, methods for using electronic devices, computer-readable storage media, and computer program products. Background Technology

[0002] In wireless communication networks such as 5G networks, AI models deployed on network-side devices (such as base stations, cloud-side, or edge servers) can be used to perform specific applications, such as image recognition and beam prediction. Typically, AI models deployed on the network side are quite large and can be referred to as large-scale AI models (hereinafter also simply large models). These AI models are trained based on historical data or digital twin simulation data, enabling efficient inference and prediction within the service range of the base station.

[0003] In many applications, although AI models are deployed on the network side, their input data is collected by terminal devices. For example, when an AI model is applied to intelligent transportation scenarios and deployed on a base station, the terminal device can be a vehicle. The terminal device's camera collects real-time image data and uploads it to the base station, where the AI ​​model on the base station performs recognition and analysis. When an AI model is applied to wireless communication optimization and deployed on a base station, the terminal device can be a user equipment (UE). The UE collects wireless network signal characteristic data and uploads it to the base station, where the AI ​​model performs beam prediction or network resource optimization.

[0004] The above methods fully utilize the computing power of the base station to meet the intelligent inference needs of terminal devices, while improving the efficiency and accuracy of model inference. However, this requires users to continuously upload data and receive large model inference results. Therefore, to reduce data transmission overhead and improve inference real-time performance, lightweight AI models (hereinafter referred to as small models) can be deployed locally on the terminal device. The terminal device can perform small model inference based on local perception data, such as images captured by a vehicle's front-facing camera or wireless signals collected by a mobile phone. Summary of the Invention

[0005] A brief overview of this disclosure is given below to provide a basic understanding of certain aspects of it. It should be understood that this overview is not an exhaustive summary of this disclosure. It is not intended to identify key or essential parts of this disclosure, nor is it intended to limit its scope. Its purpose is merely to present certain concepts in a simplified form as a prelude to the more detailed description that follows.

[0006] According to one aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory including computer program code, wherein the at least one memory and the computer program code are configured, via the at least one processor, to cause the electronic device to perform: sending a data usage indication to a network-side device, the data usage indication indicating whether the usage of measurement data sent by a terminal device corresponding to the electronic device to the network-side device is a first usage or a second usage, wherein the first usage is for obtaining a decision result of inference from a first artificial intelligence model disposed on the network-side device, and the second usage is for obtaining an intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model disposed on the terminal device as a miniaturized version of the first artificial intelligence model; and receiving the decision result or the intermediate result from the network-side device.

[0007] According to another aspect of this disclosure, a method for an electronic device is provided, comprising: sending a data usage indication to a network-side device, the data usage indication indicating whether the usage of measurement data sent by a terminal device corresponding to the electronic device to the network-side device is a first usage or a second usage, wherein the first usage is used to obtain a decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second usage is used to obtain an intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and receiving the decision result or the intermediate result from the network-side device.

[0008] The electronic devices and methods described above can enable collaborative work between the large network-side model and the small terminal-side model through real-time interaction, as well as real-time and continuous model updates, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0009] According to one aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory including computer program code, wherein the at least one memory and the computer program code are configured, via the at least one processor, to cause the electronic device to perform: receiving a data usage indication from a terminal device, the data usage indication indicating whether the usage of measurement data sent by the terminal device to a network-side device corresponding to the electronic device is a first usage or a second usage, wherein the first usage is for obtaining a decision result of inference from a first artificial intelligence model disposed on the network-side device, and the second usage is for obtaining an intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model disposed on the terminal device as a miniaturized version of the first artificial intelligence model; and sending the decision result or the intermediate result to the terminal device.

[0010] According to another aspect of this disclosure, a method for an electronic device is provided, comprising: receiving a data usage indication from a terminal device, the data usage indication indicating whether the usage of measurement data sent by the terminal device to a network-side device corresponding to the electronic device is a first usage or a second usage, wherein the first usage is used to obtain a decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second usage is used to obtain an intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and sending the decision result or the intermediate result to the terminal device.

[0011] The electronic devices and methods described above can enable collaborative work between the large network-side model and the small terminal-side model through real-time interaction, as well as real-time and continuous model updates, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0012] According to one aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and at least one memory including computer program code, wherein the at least one memory and the computer program code are configured, via the at least one processor, to cause the electronic device to perform: comparing a first performance of inference of a first artificial intelligence model disposed on a network-side device, a second performance of inference of a second artificial intelligence model disposed on a terminal device corresponding to the electronic device, and a third performance in a truth case, wherein the second artificial intelligence model is a miniaturized version of the first artificial intelligence model; and determining an inference mode of the terminal device based on the result of the comparison.

[0013] According to another aspect of this disclosure, a method for an electronic device is provided, comprising: comparing a first performance of inference of a first artificial intelligence model disposed on a network-side device, a second performance of inference of a second artificial intelligence model disposed on a terminal device corresponding to the electronic device, and a third performance in a truth case, wherein the second artificial intelligence model is a miniaturized version of the first artificial intelligence model; and determining an inference mode of the terminal device based on the result of the comparison.

[0014] The electronic devices and methods described above can optimize inference decision-making, thereby enabling collaborative work between the large model on the network side and the small model on the terminal side. They can flexibly switch inference modes based on real-time data, ensuring the accuracy and real-time performance of operations. Furthermore, the electronic devices and methods described above can utilize feedback data to continuously learn and optimize the model, improving the overall performance and adaptability of the intelligent inference system.

[0015] In accordance with other aspects of this disclosure, computer program code and computer program products for implementing the above methods, as well as a computer-readable storage medium having the computer program code for implementing the above methods recorded thereon, are also provided.

[0016] These and other advantages of this disclosure will become more apparent from the following detailed description of preferred embodiments in conjunction with the accompanying drawings. Attached Figure Description

[0017] To further illustrate the above and other advantages and features of this disclosure, a more detailed description of specific embodiments of this disclosure is provided below with reference to the accompanying drawings. These drawings, together with the following detailed description, are included in and form a part of this specification. Elements having the same function and structure are denoted by the same reference numerals. It should be understood that these drawings only depict typical examples of this disclosure and should not be construed as limiting the scope of this disclosure. In the drawings:

[0018] Figure 1 This is a functional block diagram of an electronic device according to an embodiment of the present application;

[0019] Figure 2 This is another functional block diagram of an electronic device according to one embodiment of the present application;

[0020] Figure 3 This is another functional block diagram of an electronic device according to one embodiment of the present application;

[0021] Figure 4 This is a functional block diagram of an electronic device according to another embodiment of this application;

[0022] Figure 5 This is another functional block diagram of an electronic device according to another embodiment of the present application;

[0023] Figure 6 This is another functional block diagram of an electronic device according to another embodiment of the present application;

[0024] Figure 7 This is a functional block diagram of an electronic device according to another embodiment of this application;

[0025] Figure 8 This is another functional block diagram of an electronic device according to another embodiment of the present application;

[0026] Figure 9 A flowchart of a method for an electronic device according to an embodiment of this application is shown;

[0027] Figure 10 A flowchart of a method for an electronic device according to another embodiment of this application is shown;

[0028] Figure 11 A flowchart of a method for an electronic device according to another embodiment of this application is shown;

[0029] Figure 12 This is a block diagram illustrating a first example of a schematic configuration of an eNB or gNB to which the technologies of this disclosure can be applied;

[0030] Figure 13 This is a block diagram illustrating a second example of a schematic configuration of an eNB or gNB to which the technologies of this disclosure can be applied;

[0031] Figure 14 This is a block diagram illustrating an example of a schematic configuration of a smartphone to which the technologies of this disclosure can be applied;

[0032] Figure 15 This is a block diagram illustrating an example of a schematic configuration of a car navigation device to which the technology of this disclosure can be applied; and

[0033] Figure 16 This is a block diagram of an exemplary structure of a general-purpose personal computer in which methods and / or apparatus and / or systems according to embodiments of the present disclosure can be implemented. Detailed Implementation

[0034] Exemplary embodiments of the present disclosure will be described below with reference to the accompanying drawings. For clarity and brevity, not all features of actual implementations are described in the specification. However, it should be understood that many implementation-specific decisions must be made in the development of any such actual embodiment to achieve the developer’s specific goals, such as complying with constraints related to the system and business, and these constraints may vary from implementation to implementation. Furthermore, it should be understood that while development work can be very complex and time-consuming, such development work is merely a routine task for those skilled in the art who benefit from the present disclosure.

[0035] It should also be noted that, in order to avoid obscuring this disclosure with unnecessary details, only the equipment structure and / or processing steps closely related to the solution according to this disclosure are shown in the accompanying drawings, while other details that are not closely related to this disclosure are omitted.

[0036] <First Embodiment>

[0037] As mentioned earlier, besides deploying large-scale AI models on network-side devices, distributed intelligent inference can also be achieved by deploying lightweight AI models, i.e., small models, on terminal devices. In this case, the inference performance of small models may not meet the requirements. To improve the overall performance and adaptability of the intelligent inference system, this embodiment provides an electronic device 100 capable of optimizing inference decisions.

[0038] Figure 1 A functional block diagram of an electronic device 100 according to this embodiment is shown, such as... Figure 1 As shown, the electronic device 100 includes: a comparison unit 101 configured to compare a first performance of inference of a first AI model arranged on a network-side device, a second performance of inference of a second AI model arranged on a terminal device corresponding to the electronic device 100, and a third performance in the true case, wherein the second AI model is a miniaturized version of the first AI model; and a determination unit 102 configured to determine the inference mode of the terminal device based on the comparison result.

[0039] The comparison unit 101 and the determination unit 102 can be implemented by one or more processing circuits and at least one memory. The processing circuits can be implemented as chips, processors, etc., and the at least one memory can be any form of storage device such as RAM, ROM, or flash memory. The at least one memory is used, for example, to store computer program code and data required for the processing circuits to perform processing. Furthermore, it should be understood that... Figure 1The functional units in the electronic devices shown are logical modules divided according to the specific functions they perform, and are not used to limit the specific implementation methods. Furthermore, the above description also applies to other functional units mentioned later, and will not be repeated here.

[0040] Electronic device 100 can be located on the terminal device side or communicatively coupled to the terminal device. The terminal device may include, for example, various user equipment (UEs), vehicles in a vehicular network, roadside units (RSUs), etc. It should also be noted that electronic device 100 can be implemented at the chip level or at the device level. For example, electronic device 100 can function as the terminal device itself and may also include external devices such as memory and transceivers (not shown in the figure). The memory can be used to store programs and related data information that the terminal device needs to execute to implement various functions. The transceiver may include one or more communication interfaces to support communication with different devices (e.g., other terminal devices, base stations, cloud-side or edge-side servers, etc.), and the specific implementation form of the transceiver is not limited here. Furthermore, the network-side device here can be a base station, cloud-side, or edge-side server, etc.

[0041] The first AI model is relatively large and can be trained based on historical data or digital twin simulation data. For example, the first AI model can be trained by collecting measurement data and corresponding ground truth values ​​sent by terminal devices, and this collection can be targeted at different terminal devices. In other words, network-side devices can perform joint training based on data uploaded by multiple terminal devices to optimize the performance of the first AI model.

[0042] The second AI model can be obtained by training the first AI model. It is smaller in scale and has fewer parameters, allowing it to be deployed on the terminal device for rapid inference. For example, the second AI model can be obtained through knowledge distillation technology—learning key features from the first AI model—and can be trained and updated locally on the terminal device. The second AI model can be pre-installed in the device at the factory or obtained from the network.

[0043] The first and second AI models can be used in various applications, such as beam prediction, wireless network resource allocation, and image recognition, without any limitations.

[0044] In this embodiment, the terminal device can use either the inference result (decision result) of the first AI model or the inference result of the second AI model. Furthermore, for example, in specific applications such as beam prediction, the terminal device can also use traditional measurement results instead of using an AI model for inference; in this application, the traditional measurement results are considered the true value. On the other hand, for example, in image recognition applications, the user-input annotation results can be considered the true value. In other words, generally speaking, the third performance in the true value case is superior to the first performance corresponding to the first AI model and the second performance corresponding to the second AI model.

[0045] Each of the first through third performance metrics can be represented by one of the following: the decision result and truth value of the AI ​​model's inference; user experience or communication performance. In other words, performance can be measured by the AI ​​model's decision result and truth value itself, for example, by measuring the error rate; or it can be measured indirectly, such as by user experience and communication performance like bit error rate (BER) and signal-to-interference-plus-noise ratio (SINR). The specific method used depends on the actual application and / or user requirements and is not limiting. In this disclosure, the terminal device is capable of obtaining both the truth value and the output of the second AI model from the same input.

[0046] As an example, the determining unit 102 can make inference decisions based on at least one of the following rules to optimize the inference mode of the terminal device: if the first performance is within a predetermined degree lower than the third performance, determine that the terminal device will use the first AI model for inference; if the second performance is within a predetermined degree lower than the first or third performance, determine that the terminal device will use the second AI model for inference; if the second performance is more than a predetermined degree lower than the first or third performance, determine that the terminal device will use the first AI model for inference and update the second AI model; if both the first and second performances are more than a predetermined degree lower than the third performance, determine that the terminal device will not use the AI ​​model for inference and update both the first and second AI models.

[0047] The determining unit 102 can make inference decisions based on some or all of the rules mentioned above, which can be set according to the current state of the device, user preferences, user settings, etc. It should be understood that the rules mentioned above are merely exemplary and not restrictive. The degree of pre-determination involved in the rules mentioned above can be predetermined or obtained from the network-side device. Furthermore, the degree of pre-determination can be set to the same or different for different rules, which is not restrictive.

[0048] For example, if the first AI model's performance is within a predetermined range lower than the third AI model's performance, it indicates that the first AI model has sufficiently high inference accuracy and robustness in the current scenario. Therefore, the first AI model can be used for inference, and the terminal device can use its inference results to perform corresponding operations, such as predicting uplink and downlink beam pairs and their configurations. In this case, the terminal device reports the collected measurement data to the network side via signaling and should also inform the network side to perform inference based on the measurement data and provide the terminal device with the inference decision results.

[0049] If the second performance is within a predetermined range lower than the first or third performance, it indicates that the performance of the second AI model is close to that of the first AI model or the true value, and can also meet the requirements. Therefore, the terminal device can use the second AI model for inference to achieve rapid response and efficient control. In this case, the terminal device can inform the network side via signaling that it will adopt the decision result of the second AI model in subsequent decision-making processes, and since it does not need to report measurement data to the network side, bandwidth consumption is reduced.

[0050] If the second performance is significantly lower than the first or third performance by a predetermined degree, it indicates that the second AI model's performance is poor. In this case, the second AI model needs to be updated, and the terminal device can use the decision results from the inference of the first AI model to perform corresponding operations. Model updates may refer to retraining and optimization, or incremental training. In this situation, the terminal device reports the collected measurement data to the network side via signaling and should also inform the network side that the second AI model will be updated. For example, to enable the second AI model to be trained using the first AI model, the network side can provide the terminal device with the intermediate results of the first AI model's inference. These intermediate results correspond one-to-one with the measurement data and may include statistical values ​​of the model output, such as intermediate statistical features or activation statistical features. These statistical values ​​reflect the numerical distribution characteristics of the output of each layer during the model's inference process, such as mean, variance, standard deviation, skewness, kurtosis, and other parameter information, used to describe the statistical state of the activation data within the model. Intermediate results may be soft outputs, for example. Without limitation, intermediate results may also include decision results or inference results.

[0051] If both the first and second performance metrics are significantly lower than the third performance metrics, it indicates that both the first and second AI models are performing poorly and cannot meet the requirements, thus necessitating updates. In this case, the terminal device reports the collected measurement data to the network side via signaling and should also inform the network side to update the first AI model based on the measurement data.

[0052] As described above, after the determining unit 102 determines the inference mode that the terminal device will adopt, in many cases the terminal device needs to send measurement data to the network side, and this measurement data has different uses. Accordingly, such as Figure 2 As shown, the electronic device 100 also includes a communication unit 103, configured to send measurement data and data usage instructions to a network-side device, wherein the first usage is for obtaining the decision result of inference from the first AI model, the second usage is for obtaining the intermediate result of inference from the first AI model for updating the second AI model; and receiving the decision result or intermediate result from the network-side device.

[0053] Specifically, when the second AI model needs to be updated, the data usage is indicated as second usage; when the terminal device needs to directly use the decision results of the reasoning of the first AI model, the data usage is indicated as first usage.

[0054] To ensure data correspondence, at least one memory in the electronic device 100 can store measurement data and decision results or intermediate results in association.

[0055] Furthermore, the data usage indication is also used to indicate a third usage, which is to indicate that the measurement data is used to update the first AI model. In this case, the terminal device (e.g., communication unit 103) also sends the truth value corresponding to the measurement data to the network-side device. Similarly, at least one memory may also store the measurement data and the truth value in association.

[0056] In one example, the determining unit 102 is also configured to monitor a first performance, a second performance, and a third performance to periodically determine whether the second AI model or the first AI model needs to be updated. For example, at a periodic time point, the determining unit 102 determines that the second AI model needs to be updated when the second performance is lower than the first performance or the third performance by a predetermined amount.

[0057] like Figure 3 As shown, the electronic device 100 may further include an update unit 104 configured to update the second AI model. During the update of the second AI model, the terminal device collects and processes data, and uploads the collected and processed data to the network-side device. The network-side device uses this data as input for inference, generating corresponding soft outputs. As mentioned earlier, the soft outputs may include not only the statistical characteristics of the outputs at each layer, but also the inference results of the AI ​​model on the input data. The network-side device transmits the soft outputs to the terminal device via a downlink. That is, the communication unit 103 receives the soft outputs for the measurement data from the network-side device.

[0058] Furthermore, the second AI model performs inference on the measurement data to obtain its decision result and soft output. The update unit 104 can calculate an objective function based on the difference between the soft output of the first AI model and the soft output of the second AI model for the measurement data, and the difference between the true value of the measurement data and the decision result of the second AI model; and calculate the update gradient information of the model based on the objective function, and use the update gradient information to update the second AI model.

[0059] For example, the objective function, also known as the loss function, can be expressed as follows:

[0060] Loss=‖x1-x2‖+‖x3-x4‖ (1)

[0061] Where x1 represents the true value, x2 represents the decision result of the second AI model, x3 represents the soft output of the second AI model, and x4 represents the soft output of the first AI model.

[0062] The update unit 104 obtains update gradient information by calculating the objective function and uses this update gradient information to update the model. The update unit 104 can perform updates to the second AI model at predetermined intervals until the inference performance of the second AI model reaches predetermined conditions. This predetermined interval (i.e., the update interval) can be dynamically determined based on the data acquisition rate and / or movement speed of the terminal device. This is because the update interval is closely related to the data acquisition rate of the terminal device, which may be related to the movement speed of the terminal device. For example, when the terminal device is in a high-speed movement state, the data acquisition rate is higher, thus the update interval is shorter, and vice versa.

[0063] The aforementioned predetermined period can be determined by the network-side device. The communication unit 103 can be configured to send one or more of the following to the network-side device for determining the predetermined period: the computing power of the terminal device (e.g., the computing power of a GPU), data acquisition rate, inference time, movement speed, and the number of parameters of the second AI model. The communication unit 103 can also obtain information about the predetermined period set by the network-side device.

[0064] The update unit 104 repeats the calculation of the objective function and the update of the second AI model at a predetermined period, and stops updating when the performance of the inference of the second AI model reaches a predetermined condition, considering that it has met the performance requirements.

[0065] The predetermined conditions may be one of the following: the difference between the decision result of the inference of the second AI model and the decision result of the inference of the first AI model is below a first predetermined threshold; the difference between the decision result of the inference of the second AI model and the true value is below a second predetermined threshold; and the user experience or communication performance meets predetermined requirements when the decision result of the inference of the second AI model is applied.

[0066] The difference in decision results reflects the error rate of inference. The first and second predetermined thresholds can be the same or different. As an example of user experience or communication performance, for instance, in the case of applying an AI model for beam prediction, the performance of the second AI model's inference can be judged by whether the received signal SINR or BER obtained from the beam predicted by the second AI model meets predetermined requirements. Here, the first and second predetermined thresholds and predetermined requirements can be set by the network-side equipment or predetermined.

[0067] Furthermore, since updates require uplink and downlink data transmission, signal reception, and processing, network-side devices need to allocate communication resources to terminal devices accordingly. This includes uplink and downlink spectrum resources, as well as computing power resources required to process communication signals. For example, network-side devices can allocate communication resources based on the information reported by the terminal device used to determine a predetermined period, or they can allocate communication resources partially based on the predetermined period.

[0068] The above describes a detailed example of updating the second AI model. In practice, there are also situations where the performance of both the first and second AI models falls below requirements. In this case, the terminal device reports the measurement data to the network-side device and instructs that the measurement data be used to update (train) the first AI model. The first AI model can be updated using any existing method, which will not be elaborated upon here. Furthermore, the updated first AI model can, as described above, provide a soft output to the terminal device to update the second AI model. Alternatively, the first and second AI models can be updated simultaneously; this is not limiting.

[0069] In summary, the electronic device 100 according to this embodiment can optimize inference decision-making, thereby enabling collaborative work between the large model on the network side and the small model on the terminal side. It can flexibly switch inference modes based on real-time data, ensuring the accuracy and real-time performance of operations. Furthermore, the electronic device 100 can utilize feedback data to continuously learn and optimize the model, improving the overall performance and adaptability of the intelligent inference system.

[0070] <Second Embodiment>

[0071] This embodiment provides an electronic device 200, the functional module block diagram of which is shown below. Figure 4 As shown. Electronic device 200 includes: a communication unit 201 configured to send a data purpose indication to a network-side device, the data purpose indication indicating whether the purpose of measurement data sent by the terminal device corresponding to electronic device 200 to the network-side device is a first purpose or a second purpose, wherein the first purpose is used to obtain a decision result of inference from a first AI model deployed on the network-side device, and the second purpose is used to obtain intermediate results of inference from the first AI model for updating a second AI model, which is a miniaturized version of the first AI model, deployed on the terminal device; and to receive decision results or intermediate results from the network-side device.

[0072] The communication unit 201 can be implemented by one or more processing circuits and at least one memory. The processing circuits can be implemented as chips, processors, etc., and the at least one memory can be any form of storage device such as RAM, ROM, or flash memory. The at least one memory is used, for example, to store computer program code and data required for the processing circuits to perform processing. Furthermore, it should be understood that... Figure 4 The functional units in the electronic devices shown are logical modules divided according to the specific functions they perform, and are not used to limit the specific implementation methods. Furthermore, the above description also applies to other functional units mentioned later, and will not be repeated here.

[0073] Electronic device 200 can be located on the terminal device side or communicatively coupled to the terminal device. The terminal device may include, for example, various user equipment (UEs), vehicles in a vehicular network, roadside units (RSUs), etc. It should also be noted that electronic device 200 can be implemented at the chip level or at the device level. For example, electronic device 200 can function as the terminal device itself and may also include external devices such as memory and transceivers (not shown in the figure). The memory can be used to store programs and related data information that the terminal device needs to execute to implement various functions. The transceiver may include one or more communication interfaces to support communication with different devices (e.g., other terminal devices, base stations, cloud-side or edge-side servers, etc.), and the specific implementation form of the transceiver is not limited here. Furthermore, the network-side device here can be a base station, cloud-side, or edge-side server, etc.

[0074] Detailed descriptions of the first and second AI models have been given in the first embodiment and will not be repeated here. The second AI model can be obtained by training the first AI model, for example, through knowledge distillation. The first and second AI models can be used in various applications, such as beam prediction, wireless network resource allocation, and image recognition.

[0075] The aforementioned data usage indication can be sent via dedicated signaling, such as data channel signaling or control channel signaling (e.g., RRC, MAC CE). Alternatively, the data usage indication can be sent along with the measurement data. The terminal device uses this data usage indication to inform the network-side device of the intended use of the reported measurement data.

[0076] If the data usage indication indicates a primary purpose, it means that the first AI model should use the measurement data to infer and provide the decision results to the terminal device. For example, in a beam prediction scenario, the first AI model can determine the specific beam pair to be used by the base station and the terminal device based on the measurement data and notify the terminal device; in an image recognition scenario, the first AI model can identify objects in an image based on the measurement data and notify the terminal device of the recognition results. If the data usage indication indicates a secondary purpose, it means that the second AI model will use the measurement data for updates (incremental training), requiring the first AI model to provide intermediate results.

[0077] As described in the first embodiment, the intermediate results of the AI ​​model's inference may include statistical values ​​of the model's output; for example, the intermediate results may be soft outputs. Without limitation, soft outputs may also include the decision results or inference results of the AI ​​model's reasoning.

[0078] In addition, at least one memory of the electronic device 200 may also store measurement data and decision results or intermediate results in association to ensure the correspondence between measurement data and decision results or intermediate results.

[0079] like Figure 5 As shown, the electronic device 200 also includes an execution unit 202. This execution unit 202 is configured to perform a corresponding operation based on a decision result from a first AI model, provided that the purpose of the measurement data is for a first purpose. For example, in the case of beam prediction, the execution unit 202 configures the corresponding beam indicated in the decision result; in the case of image recognition, the execution unit 202 identifies objects in the image according to the decision result.

[0080] The execution unit 202 is also configured to update the second AI model based on the soft output from the first AI model, the true value of the measurement result, and the soft output and decision result of the second AI model when the purpose of the measurement data is a second purpose.

[0081] As an example, execution unit 202 can be configured to compute an objective function based on the difference between the soft output of the first AI model and the soft output of the second AI model, as well as the difference between the true value of the measurement data and the decision result of the second AI model; and compute update gradient information of the model based on the objective function, and use the update gradient information to update the second AI model.

[0082] The objective function is also called the loss function, for example, as shown in equation (1) in the first embodiment, and will not be repeated here.

[0083] Execution unit 202 obtains update gradient information by calculating the objective function and uses this update gradient information to update the model. Execution unit 202 can update the second AI model at predetermined intervals until the inference performance of the second AI model reaches predetermined conditions. This predetermined interval (i.e., update interval) can be dynamically determined based on the data acquisition rate and / or movement speed of the terminal device. This is because the update interval is closely related to the data acquisition rate of the terminal device, which may be related to the movement speed of the terminal device. For example, when the terminal device is moving at high speed, the data acquisition rate is higher, thus shortening the update interval; conversely, the update interval is longer.

[0084] The aforementioned predetermined period can be determined by the network-side device. The communication unit 201 can be configured to send one or more of the following to the network-side device for determining the predetermined period: the computing power of the terminal device (e.g., the computing power of a GPU), data acquisition rate, inference time, movement speed, and the number of parameters of the second AI model. The communication unit 201 can also obtain information about the predetermined period set by the network-side device.

[0085] The execution unit 202 repeats the calculation of the objective function and the update of the second AI model at a predetermined period, and stops updating when the performance of the inference of the second AI model reaches a predetermined condition.

[0086] The predetermined conditions can be, for example, one of the following: the difference between the decision result of the second AI model's inference and the decision result of the first AI model's inference is below a first predetermined threshold; the difference between the decision result of the second AI model's inference and the true value is below a second predetermined threshold; and the user experience or communication performance meets predetermined requirements when the decision result of the second AI model's inference is applied. The first predetermined threshold, the second predetermined threshold, and the predetermined requirements can be set by the network-side device or can be predetermined.

[0087] The difference in decision results reflects the error rate of inference. The first and second predetermined thresholds can be the same or different. As an example of user experience or communication performance, for instance, in the case of applying an AI model for beam prediction, the performance of the second AI model's inference can be judged by whether the received signal SINR or BER obtained by the beam predicted by the second AI model meets predetermined requirements. It can be seen that, similar to the first embodiment, performance can be represented by one of the following: the decision result and true value of the AI ​​model's inference; or user experience or communication performance.

[0088] Furthermore, since updates require uplink and downlink data transmission, signal reception, and processing, network-side devices need to allocate communication resources to terminal devices accordingly. This includes uplink and downlink spectrum resources, as well as computing power resources required to process communication signals. For example, network-side devices can allocate communication resources based on the information reported by the terminal device used to determine a predetermined period, or they can allocate communication resources partially based on the predetermined period.

[0089] like Figure 6 As shown, the electronic device 200 also includes a determination unit 203 configured to determine the use of the measurement data by performance comparison.

[0090] For example, the determining unit 203 is configured to compare the inference performance of the second AI model with at least one of the inference performance of the first AI model and the performance in the true-value case, and determine the use of the measurement data based at least on the result of the comparison. It is understood that when determining the use, the determining unit 203 may need to consider other factors besides the performance comparison result, such as communication status, latency requirements, and user preferences.

[0091] For example, if the inference performance of the second AI model is significantly lower than the inference performance of the first AI model or its performance in the true value case, the determining unit 203 determines the purpose of the measurement data as a second purpose, namely, to update the second AI model, which requires obtaining the soft output of the first AI model. During the update of the second AI model, the terminal device can operate based on the inference results of the first AI model or based on the true value (user annotation or conventional measurement results).

[0092] Conversely, when the inference performance of the second AI model is within a predetermined range lower than the inference performance of the first AI model or its performance in the true case, the determining unit 203 can determine that the terminal device will use the second AI model for inference. In this case, the measurement data may not be provided to the network device, therefore the determining unit 203 may not be certain about the purpose of the measurement data.

[0093] Furthermore, the determining unit 203 can also compare the performance of the inference of the first AI model with the performance in the true case, and when the performance of the inference of the first AI model is lower than the performance in the true case by a predetermined degree, the use of the measurement data is determined as the first use.

[0094] The data usage indication can also be used to indicate a third use, which is to indicate that the measurement data is used to update the first AI model. In this case, the terminal device also sends the truth value corresponding to the measurement data to the network-side device for updating (training) the first AI model.

[0095] For example, the determining unit 203 can be configured to determine the use of the measurement data as a third use when the performance of the inference of the first AI model is lower than the performance in the true case by a predetermined degree.

[0096] It should be noted that the above only lists examples of the rules for the determination operation performed by the determination unit 203. Various combinations and modifications can be made according to the actual situation, and these are not restrictive.

[0097] The communication unit 201 transmits measurement data and a data usage instruction based on the determined purpose of the measurement data.

[0098] On the other hand, the determining unit 203 can also be configured to: compare the inference performance of the first AI model (first performance), the inference performance of the second AI model (second performance), and the performance in the truth case (third performance), and determine the inference mode of the terminal device based on the comparison results.

[0099] The configuration of the determining unit 203 is similar to part of the configuration of the determining unit 102 in the first embodiment. For example, the determining unit 203 can optimize the inference mode of the terminal device by making inference decisions based on at least one of the following rules: if the first performance is within a predetermined degree lower than the third performance, determine that the terminal device will use the first AI model for inference; if the second performance is within a predetermined degree lower than either the first or third performance, determine that the terminal device will use the second AI model for inference; if the second performance is significantly lower than either the first or third performance, determine that the terminal device will use the first AI model for inference and update the second AI model; if both the first and second performances are significantly lower than the third performance, determine that the terminal device will not use the AI ​​model for inference and update both the first and second AI models. The predetermined degrees mentioned in each rule can be the same or different, and can be predetermined or obtained from the network-side device. A detailed description of this has been given in the first embodiment and will not be repeated here.

[0100] The determining unit 203 can also monitor the first performance, the second performance, and the third performance to periodically determine whether the second AI model or the first AI model needs to be updated. For example, if the determining unit 203 determines at a certain periodic time point that the second AI model needs to be updated, it determines the data usage indication of the relevant measurement data as the second usage and sends it to the network-side device along with the measurement data; if the determining unit 203 determines at a certain periodic time point that the first AI model needs to be updated, it determines the data usage indication of the relevant measurement data as the third usage and sends it to the network-side device along with the measurement data.

[0101] In summary, the electronic device 200 according to this embodiment can achieve collaborative work between the large model on the network side and the small model on the terminal side through real-time interaction, as well as real-time and continuous updates of the model, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0102] <Third Embodiment>

[0103] This embodiment provides an electronic device 300, the functional module block diagram of which is shown below. Figure 7 As shown. Electronic device 300 includes communication unit 301, configured to receive a data usage indication from a terminal device. The data usage indication is used to indicate whether the usage of measurement data sent by the terminal device to the network-side device corresponding to electronic device 300 is a first usage or a second usage. The first usage is used to obtain decision results of inference from a first AI model deployed on the network-side device, and the second usage is used to obtain intermediate results of inference from the first AI model for updating a second AI model, which is a miniaturized version of the first AI model, deployed on the terminal device; and to send the decision results or intermediate results to the terminal device.

[0104] The communication unit 301 can be implemented by one or more processing circuits and at least one memory. The processing circuits can be implemented as chips, processors, etc., and the at least one memory can be any form of storage device such as RAM, ROM, or flash memory. The at least one memory is used, for example, to store computer program code and data required for the processing circuits to perform processing. Furthermore, it should be understood that... Figure 7 The functional units in the electronic devices shown are logical modules divided according to the specific functions they perform, and are not used to limit the specific implementation methods. Furthermore, the above description also applies to other functional units mentioned later, and will not be repeated here.

[0105] Electronic device 300 is located on the network side, such as a base station, cloud-side server, or edge-side server. It should also be noted that electronic device 300 can be implemented at the chip level or at the device level. For example, electronic device 300 can function as the network-side device itself and may also include external devices such as memory and transceivers (not shown in the figure). The memory can be used to store programs and related data information that the network-side device needs to execute to implement various functions. The transceiver may include one or more communication interfaces to support communication with different devices (e.g., UE, other network-side devices, etc.), and the specific implementation form of the transceiver is not limited here.

[0106] Detailed descriptions of the first and second AI models have been given in the first embodiment and will not be repeated here. The second AI model can be obtained by training the first AI model, for example, through knowledge distillation. The first and second AI models can be used in various applications, such as beam prediction, wireless network resource allocation, and image recognition.

[0107] The aforementioned data usage indication can be sent via dedicated signaling, such as data channel signaling or control channel signaling (e.g., RRC, MAC CE). Alternatively, the data usage indication can be sent along with the measurement data. The network-side device uses this data usage indication to determine the intended use of the measurement data received from the terminal device.

[0108] If the data usage indication indicates a primary purpose, it means that the first AI model should use the measurement data to infer and provide the decision results to the terminal device. For example, in a beam prediction scenario, the first AI model can determine the specific beam pair to be used by the base station and the terminal device based on the measurement data and notify the terminal device. If the data usage indication indicates a secondary purpose, it means that the second AI model will use the measurement data for updates (incremental training), and the first AI model needs to provide intermediate results.

[0109] As described in the first embodiment, the intermediate results of the AI ​​model's inference may include statistical values ​​of the model's output; for example, the intermediate results may be soft outputs. Without limitation, soft outputs may also include the decision results or inference results of the AI ​​model's reasoning.

[0110] After the first AI model performs reasoning on the measurement data, the communication unit 301 sends the decision result or intermediate result to the terminal device accordingly.

[0111] The communication unit 301 can receive measurement data and data usage instructions from multiple terminal devices, that is, it can perform joint training based on the data uploaded by multiple terminal devices to optimize the performance of the first AI model.

[0112] In addition, the data usage indication can also be used to indicate a third use, which is to indicate that the measurement data is used to update the first AI model. In this case, the terminal device also sends the ground truth corresponding to the measurement data to the network-side device, and the first AI model is retrained based on the measurement data and the ground truth to improve its performance.

[0113] The communication unit 301 is also configured to receive one or more of the following information from the terminal device: the terminal device's computing power, data acquisition rate, inference time, movement speed, number of parameters of the second AI model, etc.

[0114] like Figure 8As shown, the electronic device 300 may further include a determining unit 302, configured to determine a predetermined period (i.e., update period) for updating the second AI model based on information received by the communication unit 301. The communication unit 301 provides the predetermined period to the terminal device, and when the second AI model needs to be updated, the terminal device sends the collected measurement data and a data usage indication indicating that the measurement data is used for a second purpose to the network-side device according to the predetermined period.

[0115] In addition, the determining unit 302 can also allocate communication resources to the terminal device based on the information received from the terminal device, such as uplink and downlink spectrum resources and computing power resources required to process communication signals, for the transmission and processing of measurement data and soft output, thereby ensuring the efficient execution of the update process.

[0116] Furthermore, the communication unit 301 can also provide the terminal device with a predetermined degree indicating the magnitude of the performance difference, so that the terminal device performs one or more of the following: determining the use of the measurement data as a second use based on the fact that the performance of the second AI model is lower than the performance of the first AI model or the performance in the true value case by a predetermined degree or more; determining the use of the measurement data as a first use based on the fact that the performance of the first AI model is within a predetermined degree of lower than the performance in the true value case; and determining the use of the measurement data as a third use based on the fact that the performance of the first AI model is lower than the performance in the true value case by a predetermined degree or more. Moreover, the predetermined degrees involved in the above items can be the same or different.

[0117] The communication unit 301 can also provide the terminal device with information on performance thresholds or requirements that serve as stopping conditions during the update of the second AI model. Details regarding the update of the second AI model, the determination of the purpose of measurement data, and the decision-making process of the terminal device's inference mode have been described in detail in the first and second embodiments, and these details also apply to this embodiment, and will not be repeated here.

[0118] In summary, the electronic device 300 according to this embodiment can achieve collaborative work between the large model on the network side and the small model on the terminal side, as well as real-time and continuous updates of the model through real-time interaction, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0119] <Fourth Embodiment>

[0120] In the process of describing the electronic device in the above embodiments, some processes or methods have obviously been disclosed. Hereinafter, without repeating some details already discussed above, a summary of these methods is given. However, it should be noted that although these methods are disclosed in the description of the electronic device, they do not necessarily employ or are performed by the components described. For example, the implementation of the electronic device may be partially or entirely implemented using hardware and / or firmware, while the methods discussed below may be implemented entirely by computer-executable programs, although these methods may also employ the hardware and / or firmware of the electronic device.

[0121] Additionally, it should be noted that the steps of the following methods do not necessarily have to be performed in the listed order, but can be performed in a different order, in parallel, or as necessary, without limitation.

[0122] Figure 9 A flowchart of a method for an electronic device according to an embodiment of this application is shown. The method includes: sending a data usage indication to a network-side device (S11), the data usage indication indicating whether the usage of measurement data sent by a terminal device corresponding to the electronic device to the network-side device is a first usage or a second usage, wherein the first usage is used to obtain a decision result of inference from a first AI model deployed on the network-side device, and the second usage is used to obtain intermediate results of inference from the first AI model for updating a second AI model, which is a miniaturized version of the first AI model, deployed on the terminal device; and receiving the decision result or intermediate result from the network-side device (S12). This method can be executed, for example, on the terminal device side.

[0123] For example, data usage indication can be sent via dedicated signaling. The first and second AI models can be used for beam prediction, wireless network resource allocation, image recognition, but are not limited to these. The second AI model is obtained by training the first AI model, and the intermediate result can be a soft output.

[0124] Measurement data and decision results or intermediate results can be stored together in at least one memory.

[0125] As shown in one of the dashed boxes in the figure, when the measurement data is used for a second purpose, the above method further includes step S13: updating the second AI model based on the soft output from the first AI model, the true value of the measurement data, and the soft output and decision result of the second AI model for the measurement data. As shown in another dashed box in the figure, when the measurement data is used for a first purpose, the above method further includes step S14: performing corresponding operations based on the decision result from the first AI model.

[0126] For example, step S13 may include: calculating an objective function based on the difference between the soft output of the first AI model and the soft output of the second AI model and the difference between the true value of the measurement data and the decision result of the second AI model; and calculating the update gradient information of the model based on the objective function, and using the update gradient information to update the second AI model.

[0127] The above method also includes executing step S13 at predetermined intervals until the inference performance of the second AI model reaches predetermined conditions (e.g., Figure 9 (See step S15). The predetermined period can be dynamically determined based on the data acquisition rate and / or movement speed of the terminal device. The terminal device can send one or more of the following to the network-side device for determining the predetermined period: the terminal device's computing power, data acquisition rate, inference time, movement speed, number of parameters of the second AI model, etc.

[0128] The predetermined conditions are, for example, one of the following: the difference between the decision result of the second AI model's inference and the decision result of the first AI model's inference is below a first predetermined threshold; the difference between the decision result of the second AI model's inference and the true value is below a second predetermined threshold; and the user experience or communication performance meets predetermined requirements when the decision result of the second AI model's inference is applied. These predetermined thresholds or requirements can be set by the network-side device or can be predetermined.

[0129] like Figure 9 As shown in another dashed box, the above method may further include step S10: comparing the inference performance of the second AI model with at least one of the inference performance of the first AI model and the performance in the true-value case, and determining the purpose of the measurement data based at least on the result of the comparison. In step S11, the measurement data and a data purpose instruction are sent based on the determined purpose of the measurement data.

[0130] For example, if the performance of the second AI model's inference is lower than the performance of the first AI model's inference or its performance in the true case by a predetermined degree, the use of the measurement data is determined to be a second use.

[0131] For example, in step S10, the performance of the inference of the first AI model can be compared with the performance in the true case. When the performance of the inference of the first AI model is lower than the performance in the true case by a predetermined degree, the use of the measurement data is determined as the first use.

[0132] Performance can be represented by one of the following: the decision results and truth values ​​of the AI ​​model's inference; user experience or communication performance.

[0133] Furthermore, the data usage indication can also indicate a third usage, which is used to indicate that the measurement data is used for updating the first AI model. In this case, the terminal device also sends the truth value corresponding to the measurement data to the network-side device. For example, step S10 may also include determining the usage of the measurement data as a third usage when the performance of the inference of the first AI model is lower than the performance in the truth value case by a predetermined degree.

[0134] On the other hand, step S10 may also include comparing the inference performance of the first AI model, the inference performance of the second AI model, and the performance in the true case, and determining the inference mode of the terminal device based on the comparison results.

[0135] The above method corresponds to the electronic device 200 of the second embodiment. For specific details, please refer to the description in the second embodiment, which will not be repeated here.

[0136] Figure 10 A flowchart of a method for an electronic device according to an embodiment of this application is shown. The method includes: receiving a data usage indication from a terminal device (S21), the data usage indication indicating whether the usage of measurement data sent by the terminal device to a network-side device corresponding to the electronic device is a first usage or a second usage, wherein the first usage is used to obtain a decision result of inference from a first AI model deployed on the network-side device, and the second usage is used to obtain intermediate results of inference from the first AI model for updating a second AI model, which is a miniaturized version of the first AI model, deployed on the terminal device; and sending the decision result or intermediate results to the terminal device (S22). This method can be performed, for example, at the network-side device.

[0137] For example, data usage indication can be sent via dedicated signaling. The first and second AI models can be used for beam prediction, wireless network resource allocation, image recognition, but are not limited to these. The second AI model is obtained by training the first AI model, and the intermediate result can be a soft output.

[0138] In addition, the data usage indication can also be used to indicate a third use, which is to indicate that the measurement data is used to update the first AI model. In this case, the terminal device also sends the truth value corresponding to the measurement data to the network-side device.

[0139] Furthermore, although not shown in the figure, the above method may also include: receiving one or more of the following information from the terminal device: the terminal device's computing power, data acquisition rate, inference time, movement speed, and the number of parameters of the second AI model; and determining a predetermined cycle for the second AI model to perform updates based on the received information. Additionally, communication resources may be allocated to the terminal device based on the received information, including uplink and downlink spectrum resources and computing power resources required for processing communication signals, for the transmission and processing of measurement data, decision results, or intermediate results.

[0140] The above method may further include providing a predetermined degree indicating the magnitude of the performance difference to the terminal device, such that the terminal device performs one or more of the following: determining the use of the measurement data as a second use when the performance of the inference based on the second AI model is lower than the performance of the inference based on the first AI model or the performance in the true value case by a predetermined degree or more; determining the use of the measurement data as a first use when the performance of the first AI model is within a predetermined degree lower than the performance in the true value case; and determining the use of the measurement data as a third use when the performance of the first AI model is lower than the performance in the true value case by a predetermined degree or more.

[0141] The above method corresponds to the electronic device 300 of the third embodiment. For specific details, please refer to the third embodiment, which will not be repeated here.

[0142] Figure 11 A flowchart of a method for an electronic device according to an embodiment of this application is shown. The method includes: comparing (S31) a first performance of inference of a first AI model deployed on a network-side device, a second performance of inference of a second AI model deployed on a terminal device corresponding to the electronic device, and a third performance in a truth case, wherein the second AI model is a miniaturized version of the first AI model; and (S32) determining the inference mode of the terminal device based on the comparison result. This method can, for example, be executed on the terminal device side.

[0143] Similarly, the first and second AI models can be used for beam prediction, wireless network resource allocation, image recognition, but are not limited to these. The second AI model is obtained by training the first AI model, and the intermediate result can be a soft output.

[0144] For example, step S32 may include at least one of the following: if the first performance is within a predetermined level lower than the third performance, determine that the terminal device will use the first AI model for inference; if the second performance is within a predetermined level lower than either the first or third performance, determine that the terminal device will use the second AI model for inference; if the second performance is significantly lower than either the first or third performance, determine that the terminal device will use the first AI model for inference and update the second AI model; and if both the first and second performances are significantly lower than the third performance, determine that the terminal device will not use the AI ​​model for inference and update both the first and second AI models. The predetermined level may be obtained from the network-side device or may be predetermined.

[0145] In addition, such as Figure 11 As shown in the dashed box, the above method may further include: sending measurement data and data usage indication to the network-side device (S33), wherein the data usage indication is used to indicate whether the purpose of the measurement data is a first purpose or a second purpose, wherein the first purpose is used to obtain the decision result of inference from the first AI model, and the second purpose is used to obtain the intermediate result of inference from the first AI model for updating the second AI model; and receiving the decision result or intermediate result from the network-side device (S34).

[0146] Although not shown in the figure, the above method may also include the step of storing measurement data and decision results or intermediate results in association in at least one memory.

[0147] Each of the first through third performance metrics can be represented by one of the following: the decision outcome and truth value of the AI ​​model's inference; user experience or communication performance.

[0148] In addition, the data usage indication can also be used to indicate a third use, which is to indicate that the measurement data is used to update the first AI model. In this case, the terminal device also sends the truth value corresponding to the measurement data to the network-side device.

[0149] like Figure 11 As shown in the additional dashed box, when the data usage indicator is used to indicate a second usage, the above method further includes step S35 of updating the second AI model. Step S35 can be performed at predetermined intervals until the inference performance of the second AI model reaches predetermined conditions (see [reference]). Figure 11 (Step S36 in the process). The predetermined period is dynamically determined, for example, based on the data acquisition rate and / or movement speed of the terminal device. The terminal device may send one or more of the following to the network-side device for determining the predetermined period: the computing power of the terminal device, the data acquisition rate, the inference time, the movement speed, and the number of parameters of the second AI model.

[0150] As an example, step S35 may include: calculating an objective function based on the difference between the soft output of the first AI model and the soft output of the second AI model for the measurement data and the difference between the true value of the measurement data and the decision result of the second AI model; and calculating the updated gradient information of the model based on the objective function, and using the updated gradient information to update the second AI model.

[0151] When the data usage indication is used to indicate a first usage, the above method further includes step S37: performing a corresponding operation based on the decision result of the first AI model.

[0152] On the other hand, the above method may also include the following steps: monitoring the first performance, the second performance, and the third performance to periodically determine whether the second AI model or the first AI model needs to be updated.

[0153] The above method corresponds to the electronic device 100 of the first embodiment. Specific details can be found in the description of the first embodiment and will not be repeated here. Note that the above methods can be used in combination or individually.

[0154] The technology disclosed herein can be applied to a variety of products.

[0155] Electronic devices 100 and 200 can be implemented as various user devices. User devices can be implemented as mobile terminals (such as smartphones, tablet PCs, laptop PCs, portable gaming terminals, portable / dongle-type mobile routers, and digital camera devices) or in-vehicle terminals (such as car navigation devices). User devices can also be implemented as terminals performing machine-to-machine (M2M) communication (also known as machine-type communication (MTC) terminals). Furthermore, user devices can be wireless communication modules (such as integrated circuit modules comprising a single chip) installed on each of the aforementioned terminals.

[0156] For example, electronic device 300 can be implemented as various base stations. A base station can be implemented as any type of evolved NodeB (eNB) or gNB (5G base station). eNBs include, for example, macro eNBs and small eNBs. A small eNB can be an eNB covering a cell smaller than a macro cell, such as a pico eNB, micro eNB, and femtocell eNB. A similar situation can occur with gNBs. Alternatively, a base station can be implemented as any other type of base station, such as a NodeB and a Base Transceiver Station (BTS). A base station can include: a subject configured to control wireless communication (also called base station equipment); and one or more remote radio heads (RRHs) located in a different location from the subject. Furthermore, various types of user equipment can operate as base stations by temporarily or semi-persistently performing base station functions.

[0157] [Application examples of base stations]

[0158] (First application example)

[0159] Figure 12 This is a block diagram illustrating a first example of a schematic configuration of an eNB or gNB to which the technologies of this disclosure can be applied. Note that the following description uses an eNB as an example, but it can also be applied to a gNB. The eNB 800 includes one or more antennas 810 and a base station device 820. The base station device 820 and each antenna 810 can be connected to each other via RF cables.

[0160] Each of the antennas 810 includes one or more antenna elements (such as multiple antenna elements included in a multiple-input multiple-output (MIMO) antenna) and is used by the base station equipment 820 to transmit and receive wireless signals. Figure 12 As shown, the eNB 800 may include multiple antennas 810. For example, the multiple antennas 810 may be compatible with multiple frequency bands used by the eNB 800. Although Figure 12 An example is shown in which the eNB 800 includes multiple antennas 810, but the eNB 800 may also include a single antenna 810.

[0161] The base station equipment 820 includes a controller 821, a memory 822, a network interface 823, and a wireless communication interface 825.

[0162] The controller 821 can be, for example, a CPU or a DSP, and operates various higher-level functions of the base station equipment 820. For example, the controller 821 generates data packets based on data in signals processed by the wireless communication interface 825, and transmits the generated packets via the network interface 823. The controller 821 can bundle data from multiple baseband processors to generate bundled packets and transmit the generated bundled packets. The controller 821 may have logical functions that perform controls such as radio resource control, radio bearer control, mobility management, admission control, and scheduling. This control can be performed in conjunction with nearby eNBs or core network nodes. The memory 822 includes RAM and ROM, and stores programs executed by the controller 821 and various types of control data (such as terminal lists, transmission power data, and scheduling data).

[0163] Network interface 823 is a communication interface used to connect base station equipment 820 to core network 824. Controller 821 can communicate with core network nodes or other eNBs via network interface 823. In this case, eNB 800 and core network nodes or other eNBs can be connected to each other through logical interfaces (such as S1 and X2 interfaces). Network interface 823 can also be a wired communication interface or a wireless communication interface for wireless backhaul. If network interface 823 is a wireless communication interface, it can use a higher frequency band for wireless communication compared to the frequency band used by wireless communication interface 825.

[0164] The wireless communication interface 825 supports any cellular communication scheme (such as LTE and LTE-Advanced) and provides wireless connectivity to terminals located in the cell of eNB 800 via antenna 810. The wireless communication interface 825 typically includes, for example, a baseband (BB) processor 826 and RF circuitry 827. The BB processor 826 can perform, for example, encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and performs various types of signal processing at layers such as L1, Media Access Control (MAC), Radio Link Control (RLC), and Packet Data Convergence Protocol (PDCP). Instead of controller 821, the BB processor 826 may have some or all of the above-described logical functions. The BB processor 826 may be a memory storing communication control programs, or a module including a processor and associated circuitry configured to execute programs. Updates can change the functionality of the BB processor 826. The module may be a card or blade inserted into a slot in base station equipment 820. Alternatively, the module may be a chip mounted on a card or blade. Meanwhile, the RF circuit 827 may include, for example, a mixer, a filter, and an amplifier, and transmits and receives wireless signals via the antenna 810.

[0165] like Figure 12 As shown, the wireless communication interface 825 may include multiple BB processors 826. For example, the multiple BB processors 826 may be compatible with multiple frequency bands used by the eNB 800. Figure 12 As shown, the wireless communication interface 825 may include multiple RF circuits 827. For example, the multiple RF circuits 827 may be compatible with multiple antenna elements. Although Figure 12 An example is shown in which the wireless communication interface 825 includes multiple BB processors 826 and multiple RF circuits 827, but the wireless communication interface 825 may also include a single BB processor 826 or a single RF circuit 827.

[0166] exist Figure 12In the eNB 800 shown, the communication unit 301 and transceiver of the electronic device 300 can be implemented by the wireless communication interface 825. At least a portion of the functions can also be implemented by the controller 821. For example, the controller 821 can receive data usage instructions and send decision results or intermediate results by executing the functions of the communication unit 301 and the determination unit 302, realize the collaborative work between the large model on the network side and the small model on the terminal side, and the real-time and continuous updating of the model, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0167] (Second application example)

[0168] Figure 13 This is a block diagram illustrating a second example of a schematic configuration of an eNB or gNB to which the technologies of this disclosure can be applied. Note that, similarly, the following description uses an eNB as an example, but it can also be applied to a gNB. The eNB 830 includes one or more antennas 840, a base station device 850, and an RRH 860. The RRH 860 and each antenna 840 can be connected to each other via RF cables. The base station device 850 and the RRH 860 can be connected to each other via high-speed lines such as fiber optic cables.

[0169] Each of the antennas 840 includes one or more antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used by the RRH 860 to transmit and receive wireless signals. Figure 13 As shown, the eNB 830 may include multiple antennas 840. For example, the multiple antennas 840 may be compatible with multiple frequency bands used by the eNB 830. Although Figure 13 An example is shown in which the eNB 830 includes multiple antennas 840, but the eNB 830 may also include a single antenna 840.

[0170] The base station equipment 850 includes a controller 851, a memory 852, a network interface 853, a wireless communication interface 855, and a connection interface 857. The controller 851, memory 852, and network interface 853 are connected to a reference... Figure 12 The controller 821, memory 822, and network interface 823 described are the same.

[0171] The wireless communication interface 855 supports any cellular communication scheme (such as LTE and LTE-Advanced) and provides wireless communication to terminals located in the sector corresponding to the RRH 860 via the RRH 860 and antenna 840. The wireless communication interface 855 may typically include, for example, a BB processor 856. In addition to the BB processor 856 being connected to the RF circuitry 864 of the RRH 860 via a connection interface 857, the BB processor 856 is connected to the reference... Figure 12The BB processor 826 is described as identical. Figure 13 As shown, the wireless communication interface 855 may include multiple BB processors 856. For example, the multiple BB processors 856 may be compatible with multiple frequency bands used by the eNB 830. Although Figure 13 An example is shown in which the wireless communication interface 855 includes multiple BB processors 856, but the wireless communication interface 855 may also include a single BB processor 856.

[0172] Connection interface 857 is an interface for connecting base station device 850 (wireless communication interface 855) to RRH 860. Connection interface 857 can also be a communication module for connecting base station device 850 (wireless communication interface 855) to the aforementioned high-speed line of RRH 860.

[0173] The RRH 860 includes a connectivity interface 861 and a wireless communication interface 863.

[0174] Connection interface 861 is an interface for connecting RRH 860 (wireless communication interface 863) to base station equipment 850. Connection interface 861 can also be a communication module for communication in the aforementioned high-speed line.

[0175] The wireless communication interface 863 transmits and receives wireless signals via antenna 840. The wireless communication interface 863 typically includes, for example, RF circuitry 864. RF circuitry 864 may include, for example, a mixer, a filter, and an amplifier, and transmits and receives wireless signals via antenna 840. Figure 13 As shown, the wireless communication interface 863 may include multiple RF circuits 864. For example, the multiple RF circuits 864 may support multiple antenna elements. Although Figure 13 An example is shown in which the wireless communication interface 863 includes multiple RF circuits 864, but the wireless communication interface 863 may also include a single RF circuit 864.

[0176] exist Figure 13 In the eNB 830 shown, the communication unit 301 and transceiver of the electronic device 300 can be implemented by the wireless communication interface 855 and / or the wireless communication interface 863. At least a portion of the functions can also be implemented by the controller 851. For example, the controller 851 can implement the sending of data usage instructions and the receiving of decision results or intermediate results by executing the functions of the communication unit 301 and the determination unit 302, realize the collaborative work between the large model on the network side and the small model on the terminal side, and the real-time and continuous updating of the model, thereby improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0177] [Application examples related to user equipment]

[0178] (First application example)

[0179] Figure 14 This is a block diagram illustrating an example of a schematic configuration of a smartphone 900 to which the technologies of this disclosure can be applied. The smartphone 900 includes a processor 901, a memory 902, a storage device 903, an external connection interface 904, a camera device 906, a sensor 907, a microphone 908, an input device 909, a display device 910, a speaker 911, a wireless communication interface 912, one or more antenna switches 915, one or more antennas 916, a bus 917, a battery 918, and an auxiliary controller 919.

[0180] The processor 901 can be, for example, a CPU or a system-on-a-chip (SoC), and controls the application layer and other functions of the smartphone 900. The memory 902 includes RAM and ROM, and stores data and programs executed by the processor 901. The storage device 903 can include storage media such as semiconductor memory and hard disks. The external connectivity interface 904 is an interface for connecting external devices, such as memory cards and Universal Serial Bus (USB) devices, to the smartphone 900.

[0181] The camera device 906 includes an image sensor (such as a charge-coupled device (CCD) and complementary metal-oxide-semiconductor (CMOS)) and generates captured images. The sensor 907 may include a set of sensors, such as a measurement sensor, a gyroscope sensor, a magnetometer sensor, and an accelerometer sensor. The microphone 908 converts sound input to the smartphone 900 into an audio signal. The input device 909 includes, for example, a touch sensor, keypad, keyboard, buttons, or switches configured to detect touches on the screen of the display device 910 and receives operations or information input from the user. The display device 910 includes a screen (such as a liquid crystal display (LCD) and an organic light-emitting diode (OLED) display) and displays the output image of the smartphone 900. The speaker 911 converts the audio signal output from the smartphone 900 into sound.

[0182] The wireless communication interface 912 supports any cellular communication scheme (such as LTE and LTE-Advanced) and performs wireless communication. The wireless communication interface 912 typically includes, for example, a BB processor 913 and RF circuitry 914. The BB processor 913 can perform, for example, encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and performs various types of signal processing for wireless communication. Meanwhile, the RF circuitry 914 can include, for example, mixers, filters, and amplifiers, and transmits and receives wireless signals via antenna 916. Note that although the figure shows a scenario where one RF link is connected to one antenna, this is only illustrative; scenarios where an RF link is connected to multiple antennas via multiple phase shifters are also included. The wireless communication interface 912 can be a single chip module on which the BB processor 913 and RF circuitry 914 are integrated. Figure 14 As shown, the wireless communication interface 912 may include multiple BB processors 913 and multiple RF circuits 914. Although Figure 14 An example is shown in which the wireless communication interface 912 includes multiple BB processors 913 and multiple RF circuits 914, but the wireless communication interface 912 may also include a single BB processor 913 or a single RF circuit 914.

[0183] In addition to cellular communication schemes, the wireless communication interface 912 can support other types of wireless communication schemes, such as short-range wireless communication schemes, near-field communication schemes, and wireless local area network (LAN) schemes. In this case, the wireless communication interface 912 may include a BB processor 913 and RF circuitry 914 for each wireless communication scheme.

[0184] Each of the antenna switches 915 switches the connection destination of the antenna 916 among multiple circuits (e.g., circuits for different wireless communication schemes) included in the wireless communication interface 912.

[0185] Each of the antennas 916 includes one or more antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used by the wireless communication interface 912 to transmit and receive wireless signals. Figure 14 As shown, the smartphone 900 may include multiple antennas 916. Although Figure 14 An example is shown in which the smartphone 900 includes multiple antennas 916, but the smartphone 900 may also include a single antenna 916.

[0186] Furthermore, the smartphone 900 may include an antenna 916 for each wireless communication scheme. In this case, the antenna switch 915 can be omitted from the configuration of the smartphone 900.

[0187] Bus 917 connects processor 901, memory 902, storage device 903, external connection interface 904, camera device 906, sensor 907, microphone 908, input device 909, display device 910, speaker 911, wireless communication interface 912, and auxiliary controller 919 to each other. Battery 918 supplies power to... Figure 14 The various blocks of the smartphone 900 shown are powered, and the feeders are partially shown as dashed lines in the figure. The auxiliary controller 919 operates the minimum necessary functions of the smartphone 900, for example, in sleep mode.

[0188] exist Figure 14 In the illustrated smart phone 900, the communication unit 103 and transceiver of the electronic device 100 can be implemented by the wireless communication interface 912. At least a portion of the functions can also be implemented by the processor 901 or the auxiliary controller 919. For example, the processor 901 or the auxiliary controller 919 can optimize the inference decision-making of the terminal device by executing the functions of the comparison unit 101, the determination unit 102, the communication unit 103, and the update unit 104, thereby realizing the collaborative work between the large model on the network side and the small model on the terminal side, flexibly switching the inference mode according to the real-time data situation, ensuring the accuracy and real-time performance of the operation execution, and continuously learning and optimizing the model using feedback data, thereby improving the overall performance and adaptability of the intelligent inference system.

[0189] exist Figure 14 In the illustrated smart phone 900, the communication unit 201 and transceiver of the electronic device 200 can be implemented by the wireless communication interface 912. At least a portion of the functions can also be implemented by the processor 901 or the auxiliary controller 919. For example, the processor 901 or the auxiliary controller 919 can implement the sending of data usage instructions and the receiving of decision results or intermediate results by executing the functions of the communication unit 201, the execution unit 202, and the determination unit 203, thereby realizing collaborative work between the large model on the network side and the small model on the terminal side, as well as the real-time and continuous updating of the model, improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0190] (Second application example)

[0191] Figure 15 This is a block diagram illustrating an example of a schematic configuration of a car navigation device 920 to which the technology of this disclosure can be applied. The car navigation device 920 includes a processor 921, a memory 922, a Global Positioning System (GPS) module 924, a sensor 925, a data interface 926, a content player 927, a storage medium interface 928, an input device 929, a display device 930, a speaker 931, a wireless communication interface 933, one or more antenna switches 936, one or more antennas 937, and a battery 938.

[0192] The processor 921 can be, for example, a CPU or a SoC, and controls the navigation functions and other functions of the car navigation device 920. The memory 922 includes RAM and ROM, and stores data and programs executed by the processor 921.

[0193] GPS module 924 uses GPS signals received from GPS satellites to measure the location (such as latitude, longitude, and altitude) of car navigation device 920. Sensor 925 may include a set of sensors, such as a gyroscope sensor, a geomagnetic sensor, and an air pressure sensor. Data interface 926 is connected to, for example, an in-vehicle network 941 via a terminal not shown, and acquires data generated by the vehicle (such as vehicle speed data).

[0194] Content player 927 reproduces content stored on storage media (such as CDs and DVDs), which is inserted into storage media interface 928. Input device 929 includes, for example, a touch sensor, button, or switch configured to detect touch on the screen of display device 930, and receives operations or information input from the user. Display device 930 includes a screen such as an LCD or OLED display and displays images or reproduced content for navigation functions. Speaker 931 outputs sound for navigation functions or reproduced content.

[0195] The wireless communication interface 933 supports any cellular communication scheme (such as LTE and LTE-Advanced) and performs wireless communication. The wireless communication interface 933 typically includes, for example, a BB processor 934 and RF circuitry 935. The BB processor 934 can perform, for example, encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and performs various types of signal processing for wireless communication. Meanwhile, the RF circuitry 935 can include, for example, a mixer, filters, and amplifiers, and transmits and receives wireless signals via an antenna 937. The wireless communication interface 933 can also be a chip module on which the BB processor 934 and RF circuitry 935 are integrated. Figure 15 As shown, the wireless communication interface 933 may include multiple BB processors 934 and multiple RF circuits 935. Although Figure 15 An example is shown in which the wireless communication interface 933 includes multiple BB processors 934 and multiple RF circuits 935, but the wireless communication interface 933 may also include a single BB processor 934 or a single RF circuit 935.

[0196] In addition to cellular communication schemes, the wireless communication interface 933 can support other types of wireless communication schemes, such as short-range wireless communication schemes, near-field communication schemes, and wireless LAN schemes. In this case, for each wireless communication scheme, the wireless communication interface 933 may include a BB processor 934 and an RF circuit 935.

[0197] Each of the antenna switches 936 switches the connection destination of the antenna 937 among multiple circuits (such as circuits for different wireless communication schemes) included in the wireless communication interface 933.

[0198] Each of the antennas 937 includes one or more antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used by the wireless communication interface 933 to transmit and receive wireless signals. Figure 15 As shown, the car navigation device 920 may include multiple antennas 937. Although Figure 15 An example is shown in which the car navigation device 920 includes multiple antennas 937, but the car navigation device 920 may also include a single antenna 937.

[0199] Furthermore, the car navigation device 920 may include an antenna 937 for each wireless communication scheme. In this case, the antenna switch 936 can be omitted from the configuration of the car navigation device 920.

[0200] Battery 938 via feeder to Figure 15 The various blocks of the car navigation device 920 shown are powered, and the feeders are partially shown as dashed lines in the figure. Battery 938 accumulates the power supplied from the vehicle.

[0201] exist Figure 15 In the illustrated car navigation device 920, the communication unit 103 and transceiver of the electronic device 100 can be implemented by the wireless communication interface 933. At least a portion of the functions can also be implemented by the processor 921. For example, the processor 921 can optimize the inference decision-making of the terminal device by executing the functions of the comparison unit 101, the determination unit 102, the communication unit 103, and the update unit 104, thereby enabling collaborative work between the large model on the network side and the small model on the terminal side, flexibly switching inference modes according to real-time data, ensuring the accuracy and real-time performance of operation execution, and continuously learning and optimizing the model using feedback data to improve the overall performance and adaptability of the intelligent inference system.

[0202] exist Figure 15In the illustrated car navigation device 920, the communication unit 201 and transceiver of the electronic device 200 can be implemented by the wireless communication interface 933. At least a portion of the functions can also be implemented by the processor 921. For example, the processor 921 can implement the sending of data usage instructions and the receiving of decision results or intermediate results by executing the functions of the communication unit 201, the execution unit 202, and the determination unit 203, thereby enabling collaborative work between the large model on the network side and the small model on the terminal side, as well as the real-time and continuous updating of the model, improving the performance of distributed intelligent reasoning and the overall performance and adaptability of the intelligent reasoning system.

[0203] The technology disclosed herein can also be implemented as an in-vehicle system (or vehicle) 940 comprising one or more of the following blocks: a car navigation device 920, an in-vehicle network 941, and a vehicle module 942. The vehicle module 942 generates vehicle data (such as vehicle speed, engine speed, and fault information) and outputs the generated data to the in-vehicle network 941.

[0204] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that those skilled in the art will understand that all or any step or component of the methods and apparatus of this disclosure can be implemented in any computing device (including processors, storage media, etc.) or network of computing devices, in the form of hardware, firmware, software or a combination thereof. This is something that those skilled in the art can achieve by using their basic circuit design knowledge or basic programming skills after reading the description of this disclosure.

[0205] Furthermore, this disclosure also proposes a program product storing machine-readable instruction code. When the instruction code is read and executed by a machine, the method described above according to embodiments of this disclosure can be performed.

[0206] Accordingly, the storage medium used to carry the program product storing machine-readable instruction code is also included in this disclosure. The storage medium includes, but is not limited to, floppy disks, optical disks, magneto-optical disks, memory cards, memory sticks, etc.

[0207] In the case of implementing this disclosure through software or firmware, transmission from a storage medium or network to a computer with a dedicated hardware architecture (e.g., Figure 16 The general-purpose computer 1600 shown is equipped with the programs that constitute the software, and when various programs are installed, the computer is able to perform various functions, etc.

[0208] exist Figure 16In this system, the Central Processing Unit (CPU) 1601 performs various processes based on programs stored in the Read-Only Memory (ROM) 1602 or programs loaded into the Random Access Memory (RAM) 1603 from the Storage Section 1608. The RAM 1603 also stores data required as needed when the CPU 1601 performs various processes. The CPU 1601, ROM 1602, and RAM 1603 are interconnected via a bus 1604. An input / output interface 1605 is also connected to the bus 1604.

[0209] The following components are connected to the input / output interface 1605: input section 1606 (including keyboard, mouse, etc.), output section 1607 (including display, such as cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.), storage section 1608 (including hard disk, etc.), and communication section 1609 (including network interface card, such as LAN card, modem, etc.). The communication section 1609 performs communication processing via a network, such as the Internet. If necessary, a drive 1610 may also be connected to the input / output interface 1605. Removable media 1611, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive 1610 as needed, so that computer programs read from them can be installed into the storage section 1608 as needed.

[0210] When the above series of processes are implemented by software, the program constituting the software is installed from a network such as the Internet or a storage medium such as removable media 1611.

[0211] Those skilled in the art will understand that such storage media are not limited to Figure 16 The illustration shows a removable medium 1611 containing a program, distributed separately from the device to provide the program to the user. Examples of removable media 1611 include disks (including floppy disks (registered trademark)), optical disks (including optical disc read-only memory (CD-ROM) and digital versatile disks (DVD)), magneto-optical disks (including mini-discs (MD) (registered trademark)), and semiconductor memory. Alternatively, the storage medium may be ROM 1602, a hard disk included in storage section 1608, etc., containing programs and distributed to the user along with the device containing them.

[0212] It should also be noted that in the apparatus, method, and system of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered equivalent solutions of this disclosure. Furthermore, the steps performing the above series of processes can naturally be executed in the order described, but are not necessarily required to be executed in chronological order. Some steps can be performed in parallel or independently of each other.

[0213] Finally, it should be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Furthermore, unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0214] While embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings, it should be understood that the embodiments described above are merely illustrative and do not constitute a limitation thereof. Those skilled in the art can make various modifications and alterations to the above embodiments without departing from the spirit and scope of the present disclosure. Therefore, the scope of the present disclosure is defined only by the appended claims and their equivalents.

[0215] This technology can also be configured as follows.

[0216] (1) An electronic device, comprising:

[0217] At least one processor; and

[0218] At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor:

[0219] Sending a data usage indication to a network-side device, the data usage indication being used to indicate whether the measurement data sent by the terminal device corresponding to the electronic device to the network-side device is for a first purpose or a second purpose, wherein the first purpose is used to obtain decision results of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain intermediate results of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and

[0220] Receive the decision result or the intermediate result from the network-side device.

[0221] (2) The electronic device according to (1), wherein the data usage indication is sent via dedicated signaling.

[0222] (3) The electronic device according to (1), wherein the second artificial intelligence model is obtained by training using the first artificial intelligence model, and the intermediate result is a soft output.

[0223] (4) The electronic device according to (3), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0224] When the measurement data is used for the second purpose, the second artificial intelligence model is updated based on the soft output from the first artificial intelligence model, the true value of the measurement data, and the soft output and decision results of the second artificial intelligence model for the measurement data; and

[0225] If the measurement data is used for the first purpose, the corresponding operation is performed based on the decision result from the first artificial intelligence model.

[0226] (5) The electronic device according to (4), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor, when the purpose of the measurement data is the second purpose:

[0227] Based on the difference between the soft output of the first artificial intelligence model and the soft output of the second artificial intelligence model, and the difference between the true value of the measurement data and the decision result of the second artificial intelligence model, the objective function is calculated; and

[0228] The updated gradient information of the model is calculated based on the objective function, and the updated gradient information is used to update the second artificial intelligence model.

[0229] (6) The electronic device according to (5), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0230] The second artificial intelligence model is updated at predetermined intervals until the inference performance of the second artificial intelligence model reaches predetermined conditions.

[0231] (7) The electronic device according to (6), wherein the predetermined period is dynamically determined based on the data acquisition rate and / or movement speed of the terminal device.

[0232] (8) The electronic device according to (7), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0233] Send one or more of the following to the network-side device for determining the predetermined period: the computing power of the terminal device, data acquisition rate, inference time, movement speed, and the number of parameters of the second artificial intelligence model.

[0234] (9) The electronic device according to (6), wherein the predetermined condition is one of the following: the difference between the decision result of the reasoning of the second artificial intelligence model and the decision result of the reasoning of the first artificial intelligence model is below a first predetermined threshold; the difference between the decision result of the reasoning of the second artificial intelligence model and the true value is below a second predetermined threshold; and the user experience or communication performance meets predetermined requirements when the decision result of the reasoning of the second artificial intelligence model is applied.

[0235] (10) The electronic device according to (1), wherein the first artificial intelligence model and the second artificial intelligence model are used for one of the following: beam prediction, wireless network resource allocation, and image recognition.

[0236] (11) The electronic device according to (1), wherein the at least one memory is configured to store the measurement data and the decision result or the intermediate result in association.

[0237] (12) The electronic device according to (1), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0238] The performance of the inference of the second artificial intelligence model is compared with at least one of the performance of the inference of the first artificial intelligence model and its performance in the true-value case, and the use of the measurement data is determined based at least on the result of the comparison; and

[0239] The measurement data and the data usage instruction are sent based on the determined purpose of the measurement data.

[0240] Wherein, when the performance of the inference of the second artificial intelligence model is lower than the performance of the inference of the first artificial intelligence model or the performance in the true value case by a predetermined degree or more, the use of the measurement data is determined to be the second use.

[0241] (13) The electronic device according to (1), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0242] The performance of the inference of the first artificial intelligence model is compared with the performance in the true value case. When the performance of the inference of the first artificial intelligence model is lower than the performance in the true value case by a predetermined degree, the use of the measurement data is determined as the first use.

[0243] (14) The electronic device according to (12), wherein the performance is represented by one of the following: the decision result and truth value of the reasoning of the artificial intelligence model; user experience or communication performance.

[0244] (15) The electronic device according to (1), wherein the data usage indication is further used to indicate a third usage, the third usage being used to indicate that the measurement data is used for updating the first artificial intelligence model, in which case the terminal device also sends the truth value corresponding to the measurement data to the network-side device.

[0245] (16) The electronic device according to (15), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0246] When the performance of the reasoning of the first artificial intelligence model is lower than the performance in the true case by a predetermined degree, the use of the measurement data is determined as the third use.

[0247] (17) The electronic device according to (15), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0248] The inference performance of the first AI model, the inference performance of the second AI model, and the performance in the truth case are compared; and

[0249] The reasoning mode of the terminal device is determined based on the results of the comparison.

[0250] (18) An electronic device comprising:

[0251] At least one processor; and

[0252] At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor:

[0253] Receive data usage indication from the terminal device, the data usage indication indicating whether the purpose of the measurement data sent by the terminal device to the network-side device corresponding to the electronic device is a first purpose or a second purpose, wherein the first purpose is used to obtain decision results of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain intermediate results of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and

[0254] The decision result or the intermediate result is sent to the terminal device.

[0255] (19) The electronic device according to (18), wherein the data usage indication is sent via dedicated signaling.

[0256] (20) The electronic device according to (18), wherein the second artificial intelligence model is obtained by training using the first artificial intelligence model, and the intermediate result is a soft output.

[0257] (21) The electronic device according to (18), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0258] Receive one or more of the following information from the terminal device: the terminal device's computing power, data acquisition rate, inference time, movement speed, and the number of parameters of the second artificial intelligence model; and

[0259] The predetermined cycle for the second artificial intelligence model to perform updates is determined based on the received information.

[0260] (22) The electronic device according to (21), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0261] The terminal device is allocated communication resources based on the received information.

[0262] (23) According to the electronic device of (18), wherein the data usage indication is further used to indicate a third usage, the third usage being used to indicate that the measurement data is used for updating the first artificial intelligence model, in which case the terminal device also sends the truth value corresponding to the measurement data to the network-side device.

[0263] (24) The electronic device according to (23), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0264] The terminal device is provided with a predetermined degree indicating the magnitude of the performance difference, such that the terminal device performs one or more of the following: if the performance of inference based on the second artificial intelligence model is lower than the performance of inference based on the first artificial intelligence model or the performance in the true value case by a predetermined degree or more, the measurement data is assigned to the second purpose; if the performance of the first artificial intelligence model is lower than the performance in the true value case by a predetermined degree or less, the measurement data is assigned to the first purpose; if the performance of the first artificial intelligence model is lower than the performance in the true value case by a predetermined degree or more, the measurement data is assigned to the third purpose.

[0265] (25) The electronic device according to (18), wherein the first artificial intelligence model and the second artificial intelligence model are used for one of the following: beam prediction, wireless network resource allocation, and image recognition.

[0266] (26) An electronic device comprising:

[0267] At least one processor; and

[0268] At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor:

[0269] The inference performance of a first AI model deployed on a network-side device, the inference performance of a second AI model deployed on a terminal device corresponding to the electronic device, and the performance in a truth case are compared, wherein the second AI model is a miniaturized version of the first AI model; and

[0270] The reasoning mode of the terminal device is determined based on the results of the comparison.

[0271] (27) The electronic device according to (26), wherein the at least one memory and the computer program code are further configured to cause the electronic device to perform at least one of the following via the at least one processor:

[0272] If the first performance is lower than the third performance by a predetermined degree, it is determined that the terminal device will use the first artificial intelligence model for reasoning.

[0273] If the second performance is lower than the first performance or the third performance by a predetermined degree, it is determined that the terminal device will use the second artificial intelligence model for reasoning;

[0274] If the second performance is significantly lower than the first or third performance by a predetermined degree, it is determined that the terminal device will use the first artificial intelligence model for inference and update the second artificial intelligence model; and

[0275] If both the first performance and the second performance are lower than the third performance by a predetermined degree, it is determined that the terminal device will not use the artificial intelligence model for reasoning and the first artificial intelligence model and the second artificial intelligence model will be updated.

[0276] (28) The electronic device according to (27), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0277] The measurement data and data usage indication are sent to the network-side device. The data usage indication specifies whether the measurement data is used for a first purpose or a second purpose, wherein the first purpose is used to obtain the decision result of inference from the first artificial intelligence model, and the second purpose is used to obtain the intermediate result of inference from the first artificial intelligence model for updating the second artificial intelligence model; and

[0278] Receive the decision result or the intermediate result from the network-side device.

[0279] (29) The electronic device according to (28), wherein the data usage indication is further used to indicate a third usage, the third usage being used to indicate that the measurement data is used for updating the first artificial intelligence model, in which case the terminal device also sends the truth value corresponding to the measurement data to the network-side device.

[0280] (30) The electronic device according to (26), wherein each of the first to third performance is represented by one of the following: the decision result and truth value of the reasoning of the artificial intelligence model; user experience or communication performance.

[0281] (31) The electronic device according to (28), wherein the at least one memory and the computer program code are further configured, via the at least one processor, to cause the electronic device to execute, in the event that the second artificial intelligence model needs to be updated:

[0282] Based on the difference between the soft output of the first artificial intelligence model and the soft output of the second artificial intelligence model for the measured data, and the difference between the true value of the measured data and the decision result of the second artificial intelligence model, a target function is calculated; and

[0283] The updated gradient information of the model is calculated based on the objective function, and the updated gradient information is used to update the second artificial intelligence model.

[0284] (32) The electronic device according to (31), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0285] The second artificial intelligence model is updated at predetermined intervals until the inference performance of the second artificial intelligence model reaches predetermined conditions.

[0286] (33) The electronic device according to (32), wherein the predetermined period is dynamically determined based on the data acquisition rate and / or movement speed of the terminal device.

[0287] (34) The electronic device according to (33), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0288] Send one or more of the following to the network-side device for determining the predetermined period: the computing power of the terminal device, data acquisition rate, inference time, movement speed, and the number of parameters of the second artificial intelligence model.

[0289] (35) The electronic device according to (28), wherein the at least one memory is configured to store the measurement data and the decision result or the intermediate result in association.

[0290] (36) The electronic device according to (27), wherein the at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor:

[0291] The first performance, the second performance, and the third performance are monitored to periodically determine whether the second AI model or the first AI model needs to be updated.

[0292] The predetermined degree is obtained from the network-side device.

[0293] (37) The electronic device according to (26), wherein the first artificial intelligence model and the second artificial intelligence model are used for one of the following: beam prediction, wireless network resource allocation, and image recognition.

[0294] (38) A method for an electronic device, comprising:

[0295] Sending a data usage indication to a network-side device, the data usage indication being used to indicate whether the measurement data sent by the terminal device corresponding to the electronic device to the network-side device is for a first purpose or a second purpose, wherein the first purpose is used to obtain decision results of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain intermediate results of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and

[0296] Receive the decision result or the intermediate result from the network-side device.

[0297] (39) A method for an electronic device, comprising:

[0298] Receive data usage indication from the terminal device, the data usage indication indicating whether the purpose of the measurement data sent by the terminal device to the network-side device corresponding to the electronic device is a first purpose or a second purpose, wherein the first purpose is used to obtain decision results of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain intermediate results of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; and

[0299] The decision result or the intermediate result is sent to the terminal device.

[0300] (40) A method for an electronic device, comprising:

[0301] The inference performance of a first AI model deployed on a network-side device, the inference performance of a second AI model deployed on a terminal device corresponding to the electronic device, and the performance in a truth case are compared, wherein the second AI model is a miniaturized version of the first AI model; and

[0302] The reasoning mode of the terminal device is determined based on the results of the comparison.

[0303] (41) A computer-readable storage medium having stored thereon computer-executable instructions that, when executed by a processor, cause the processor to perform the method according to any one of (38) to (40).

[0304] (42) A computer program product comprising a computer program / instructions, wherein the computer program / instructions, when executed by a processor, implement the steps of any one of (38) to (40) of the method.

Claims

1. An electronic device, comprising: At least one processor; as well as At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor: Sending a data usage indication to a network-side device, the data usage indication being used to indicate whether the measurement data sent by the terminal device corresponding to the electronic device to the network-side device is for a first purpose or a second purpose, wherein the first purpose is used to obtain the decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain the intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; as well as Receive the decision result or the intermediate result from the network-side device.

2. The electronic device according to claim 1, wherein, The at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor: If the measurement data is used for the second purpose, the second artificial intelligence model is updated based on the soft output from the first artificial intelligence model, the true value of the measurement data, and the soft output and decision results of the second artificial intelligence model for the measurement data; as well as If the measurement data is used for the first purpose, the corresponding operation is performed based on the decision result from the first artificial intelligence model.

3. The electronic device according to claim 2, wherein, The at least one memory and the computer program code are further configured, via the at least one processor, to cause the electronic device to execute, whereby the purpose of the measurement data is the second purpose: The objective function is calculated based on the difference between the soft output of the first artificial intelligence model and the soft output of the second artificial intelligence model, as well as the difference between the true value of the measurement data and the decision result of the second artificial intelligence model. as well as The updated gradient information of the model is calculated based on the objective function, and the updated gradient information is used to update the second artificial intelligence model.

4. The electronic device according to claim 3, wherein, The at least one memory and the computer program code are further configured to cause the electronic device to execute, via the at least one processor: The second artificial intelligence model is updated at predetermined intervals until the inference performance of the second artificial intelligence model reaches predetermined conditions.

5. The electronic device according to claim 1, wherein, The first artificial intelligence model and the second artificial intelligence model are used for one of the following: beam prediction, wireless network resource allocation, and image recognition.

6. An electronic device, comprising: At least one processor; as well as At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor: The terminal device receives a data usage indication, which indicates whether the measurement data sent by the terminal device to the network-side device corresponding to the electronic device is for a first purpose or a second purpose. The first purpose is used to obtain the decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain the intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device. as well as The decision result or the intermediate result is sent to the terminal device.

7. An electronic device, comprising: At least one processor; as well as At least one memory, including computer program code, wherein the at least one memory and the computer program code are configured to cause the electronic device to execute via the at least one processor: The first performance of inference of a first artificial intelligence model deployed on a network-side device, the second performance of inference of a second artificial intelligence model deployed on a terminal device corresponding to the electronic device, and the third performance in the truth case are compared, wherein the second artificial intelligence model is a miniaturized version of the first artificial intelligence model. as well as The reasoning mode of the terminal device is determined based on the results of the comparison.

8. A method for use in an electronic device, comprising: Sending a data usage indication to a network-side device, the data usage indication being used to indicate whether the measurement data sent by the terminal device corresponding to the electronic device to the network-side device is for a first purpose or a second purpose, wherein the first purpose is used to obtain the decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain the intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device; as well as Receive the decision result or the intermediate result from the network-side device.

9. A method for use in an electronic device, comprising: The terminal device receives a data usage indication, which indicates whether the measurement data sent by the terminal device to the network-side device corresponding to the electronic device is for a first purpose or a second purpose. The first purpose is used to obtain the decision result of inference from a first artificial intelligence model deployed on the network-side device, and the second purpose is used to obtain the intermediate result of inference from the first artificial intelligence model for updating a second artificial intelligence model, which is a miniaturized version of the first artificial intelligence model, deployed on the terminal device. as well as The decision result or the intermediate result is sent to the terminal device.

10. A method for an electronic device, comprising: The first performance of inference of a first artificial intelligence model deployed on a network-side device, the second performance of inference of a second artificial intelligence model deployed on a terminal device corresponding to the electronic device, and the third performance in the truth case are compared, wherein the second artificial intelligence model is a miniaturized version of the first artificial intelligence model. as well as The reasoning mode of the terminal device is determined based on the results of the comparison.

Images