Data processing method and related device

Abstract

Provided in the embodiments of the present application are a data processing method and a related device, which are applied to the field of communications. The method comprises: receiving monitoring data of each of M AI functions, the M AI functions comprising a first AI function and a second AI function, and the monitoring data of each AI function being used for determining a value of a performance indicator of the AI function; and determining a performance monitoring result, performance information of each AI function in the performance monitoring result being obtained on the basis of the value of the performance indicator of the AI function. Since an input of the second AI function is determined on the basis of an output of the first AI function, the performance of the second AI function may be affected by the performance of the first AI function. In the present application, each AI function has its own independent monitoring data, and the value of the performance indicator of each AI function can be obtained on the basis of the monitoring data of the AI function, thus helping to determine the actual performance of each AI function as rapidly as possible.

Description

A data processing method and related equipment

[0001] This application claims priority to Chinese Patent Application No. 202510056992.8, filed on January 10, 2025, entitled "A Data Processing Method and Related Equipment", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of communications, and more particularly to a data processing method and related equipment. Background Technology

[0003] Air interface artificial intelligence (AI) can significantly improve air interface performance by introducing AI technology. 3GPP has discussed use cases such as AI channel state indicator (CSI) feedback, AI beam management, and AI positioning. It also discussed a framework for air interface AI lifecycle management (LCM), which manages the processes of generating, deploying, and monitoring air interface AI functions or models. Since AI models are typically trained on data, the accuracy of predictions obtained from AI models has considerable uncertainty. Therefore, performance monitoring of AI models is necessary to measure their performance. For example, one performance monitoring method is to obtain the predicted output data of the AI ​​model based on its input data, compare this predicted data with the true value, and obtain the parameter values ​​of the AI ​​model's performance indicators, thus reflecting the performance of the AI ​​function.

[0004] Future air interface AI use cases may involve multiple AI models within a single AI use case. For example, the aforementioned AI use case could be an AI receiver, which might incorporate AI-based channel estimation, demodulation, or decoding modules. Alternatively, the aforementioned AI use case could be an AI transceiver, which might incorporate AI-based coding, modulation, reference signal determination, channel estimation, and demodulation or decoding modules. When an AI use case includes multiple AI models, how to monitor the performance of these AI models is a pressing issue that needs to be addressed. Summary of the Invention

[0005] This application provides a data processing method and related equipment for receiving monitoring data of each AI function among multiple AI functions, and then obtaining the performance index value of each AI function based on the monitoring data of each AI function, so as to facilitate the rapid detection of the performance of multiple AI modules.

[0006] The embodiments of this application provide the following technical solutions:

[0007] In a first aspect, embodiments of this application provide a data processing method applied to a first device. For example, the first device may be a network device, or it may be a component of the network device (e.g., a processor, circuit, chip, or chip system responsible for communication functions, including but not limited to a modem chip, a baseband chip, a system-on-a-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip, etc.). Alternatively, the first device may also be a logic module or software capable of implementing all or part of the functions of the network device. The following description uses a first device as an example.

[0008] The method includes: a first device receiving monitoring data for each of M AI functions, wherein the monitoring data for each of the M AI functions is used to determine the value of the performance index of each AI function, and M is an integer greater than or equal to 1; and then the first device determining the performance monitoring result, wherein the performance monitoring result includes the performance information of N AI functions among the M AI functions, wherein the performance information of each AI function is obtained based on the value of the performance index of each AI function, and N is an integer greater than or equal to 1 and N is less than or equal to M.

[0009] For example, each of the M AI functions can be implemented by one or more AI models. In this application, the AI ​​model can also be referred to as a machine learning model, or an AI module or a machine learning ML module.

[0010] Optionally, the M AI functions include a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function (which can be understood as determining the output of the second AI function requiring the output of the first AI function). In other words, there is a correlation between the first AI function and the second AI function, and the input of the second AI function depends on the output of the first AI function. Alternatively, determining the input of the second AI function based on the output of the first AI function can also be described as: there is a sequential relationship between the first AI function and the second AI function, with the first AI function being the preceding AI function of the second AI function. Alternatively, determining the input of the second AI function based on the output of the first AI function can also be described as: there is an upstream-downstream relationship between the first AI function and the second AI function, with the first AI function being the upstream AI function of the second AI function.

[0011] For example, the performance monitoring results of an AI function may include the values ​​of its performance metrics, or the results of whether those performance metrics meet a threshold. Furthermore, the performance information for each AI function may include the values ​​of its performance metrics, or the results of whether its performance metrics meet a third threshold. It should be noted that the third threshold here is a general term; that is, not all AI functions correspond to the same third threshold, but rather each AI function can have its own corresponding third threshold.

[0012] When processing data, the first device may need to sequentially implement different functions through M AI functions, including a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function. For example, in an AI receiver, the input of the demodulation AI function includes channel information output by the channel estimation AI function; similarly, the input of the channel decoding AI function includes prediction data output by the demodulation AI function. The performance of a particular AI function may be affected by the performance of upstream AI functions, making it difficult to distinguish the independent performance of each AI function during performance monitoring. In this implementation, the first device can receive monitoring data from each of the M AI functions. Each AI function has its own independent monitoring data. Therefore, after receiving the monitoring data for each AI function, the first device can obtain the performance index value of each AI function based on the monitoring data, facilitating a quick determination of whether the performance of each AI function is normal. For example, by calculating the performance index value of each AI function, the first device can quickly determine the impact of upstream AI functions on the performance of downstream AI functions, thus helping to quickly determine the true performance of each AI function. Furthermore, by simultaneously sending monitoring data from M AI functions, the latency and power consumption of the first device in acquiring the monitoring data can be reduced.

[0013] In one possible implementation, the first device can also send performance monitoring results to the second device, so that the second device can also obtain the performance status of N AI functions. This is also beneficial for the second device to obtain the real performance status of N AI functions in a timely manner, and also for the second device to identify the AI ​​functions that have performance problems in a timely manner.

[0014] In one possible implementation, the first device sends performance monitoring results to the second device, including: after receiving monitoring data from M AI functions P times, the first device sends performance monitoring results to the second device, where P is a preset value greater than or equal to 1 or a value indicated or configured by the second device. It should be noted that the value of M in different of the aforementioned P times can be the same or different. Alternatively, after receiving indication information from the second device, the first device sends performance monitoring results to the second device; for example, sending performance monitoring results to the second device in X time slots after receiving the indication information, where X is a non-negative integer. Alternatively, when the performance index value of any one of the M AI functions meets a first threshold, the first device sends performance monitoring results to the second device. Alternatively, when the performance index value of any one of the M AI functions does not meet the first threshold, the first device sends performance monitoring results to the second device.

[0015] This implementation provides multiple ways to provide feedback on performance monitoring results, which greatly improves the flexibility of the solution and makes it easier to choose the appropriate feedback method based on actual needs.

[0016] In one possible implementation, the M AI functions may include at least one of the following: channel estimation, demodulation, channel decoding, source decoding, modulation and demodulation, channel coding and decoding, or source coding and decoding.

[0017] This implementation method clarifies which AI functions can be used for the M AI functions, which not only improves the integration of this solution with specific application scenarios, but also expands the application scenarios of this solution.

[0018] In one possible implementation, the monitoring data for the M AI functions includes one or more of the following: a first reference signal used to determine the input for channel estimation; a second reference signal used to determine the reference value for channel estimation; modulation symbols used to determine the input for demodulation or modulation / demodulation; and coded codewords used to determine the input for channel decoding or channel coding / decoding. Optionally, the coded codewords can also be used to determine the input for joint source-channel decoding or joint source-channel coding / decoding.

[0019] Optionally, the coded codeword can also be used to determine the input for source decoding or source coding decoding. Optionally, the modulation symbol can also be used to determine the input for source-channel joint decoding and demodulation or source-channel joint coding decoding and modulation / demodulation.

[0020] In this implementation, when the M AI functions include at least one of the following: channel estimation, demodulation, channel decoding, modulation / demodulation, or channel coding / decoding, the first reference signal can determine the input for channel estimation, the reference value and / or modulation symbol for channel estimation can determine the input for demodulation or modulation / demodulation, and the coded codeword can determine the input for channel decoding or channel coding / decoding. That is, it clarifies which data are included in the monitoring data of the M AI functions, as well as the input of each AI function. By simultaneously sending the monitoring data of the M AI functions, the performance monitoring efficiency is further improved, and the power consumption and latency of the terminal device in performance monitoring are reduced.

[0021] In one possible implementation, the modulation symbols in the monitoring data of the aforementioned M AI functions can be obtained by modulating the first codeword. If the AI ​​function is demodulation or modulation-demodulation, the input of the AI ​​function includes the received modulation symbols, and the reference value corresponding to the output of the AI ​​function is the first codeword. The first codeword is determined according to a predefined method. For example, the aforementioned predefinition can be performed through a communication protocol, or it can be predefined by the base station through non-communication protocol forms such as radio resource control (RRC) configuration information or indication information. Determining the first codeword according to a predefined method can be understood as the determination process of the first codeword being predefined. For example, to ensure the determinism of the determination process of the first codeword, the determination process of the first codeword does not use the AI ​​model.

[0022] In this implementation, since the first codeword is determined according to a predefined method, errors, unknowns, or uncertainties are avoided in the process of determining the first codeword. This avoids the process of determining the first codeword from affecting the performance of the AI ​​demodulation function, which is beneficial to obtaining the performance of the AI ​​demodulation function.

[0023] In one possible implementation, if the AI ​​function is channel estimation, the reference value corresponding to the output of the AI ​​function is determined based on a second reference signal; or, if the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is a first codeword; or, if the AI ​​function is channel decoding or encoding / decoding, the reference value corresponding to the output of the AI ​​function is a second codeword, which is obtained in a predefined manner.

[0024] Optionally, determining the first codeword according to a predefined method can be further understood as: the first codeword is determined using a predefined first formula and the value of the first parameter of a predefined type. Optionally, determining the second codeword according to a predefined method can be further understood as: the second codeword is determined using a predefined second formula and the value of the second parameter of a predefined type.

[0025] In this implementation, the reference value corresponding to the output of the channel estimation AI function is obtained based on the second reference signal; the reference value corresponding to the output of the demodulation or modulation / demodulation AI function is the first codeword; and the reference value corresponding to the output of the channel decoding or channel coding / decoding AI function is the second codeword. Both the first and second codewords are obtained using a predefined method. The first device can directly generate the first and / or second codewords using a predefined method, thereby ensuring that the first device can obtain more accurate reference values. Since the performance index value of each AI function is obtained based on the output of that AI function and the reference value corresponding to the output, obtaining more accurate reference values ​​is beneficial for obtaining more accurate performance index values ​​for each AI function, thereby improving the accuracy of the performance monitoring process. At the same time, generating the aforementioned reference values ​​using a predefined method can also reduce transmission overhead and power consumption.

[0026] In one possible implementation, the monitoring data of the M AI functions further includes a first codeword and / or a second codeword, wherein if the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword; if the AI ​​function is channel decoding or code decoding, the reference value corresponding to the output of the AI ​​function is the second codeword. If the aforementioned M AI functions simultaneously include demodulation and channel decoding, or if the aforementioned M AI functions simultaneously include modulation / demodulation and code decoding, then the monitoring data of the M AI functions includes the first codeword and the second codeword.

[0027] In this implementation, the monitoring data of the M AI functions also includes a first codeword and / or a second codeword. In other words, the reference value corresponding to the output of the demodulation or modulation / demodulation AI function and / or the reference value corresponding to the output of the channel decoding or channel coding / decoding AI function can be included in the monitoring data of the M AI functions and sent by the second device to the first device. This provides another way to obtain the reference value corresponding to the output of the AI ​​function. The second device can independently determine the first codeword and / or the second codeword, which improves the randomness of the monitoring data.

[0028] In one possible implementation, when the number of resource elements (REs) occupied by the second reference signal is less than the second threshold, the reference value of the channel estimation determined based on the second reference signal is invalid.

[0029] In this implementation, since the number of REs occupied by the second reference signal is usually known if no puncturing or rate matching is performed on the second reference signal during resource mapping, the number of REs occupied by the second reference signal is often known. When the number of REs occupied by the second reference signal is less than a second threshold, it means that some data in the second reference signal was missing when the second device performed resource mapping. Therefore, the reference value of the channel estimation determined based on the second reference signal is inaccurate, and the first device cannot obtain accurate channel information. In this case, the reference value of the channel estimation determined based on the second reference signal is considered invalid. A valid performance indicator value cannot be obtained based on an invalid channel estimation reference value, thus avoiding erroneous monitoring of the AI ​​function's performance and improving the accuracy of the AI ​​function performance monitoring process.

[0030] In one possible implementation, a first channel can be added between the first device and the second device. The first device and the second device use this first channel to transmit monitoring data for M AI functions. This first channel is independent of physical channels used for transmitting user plane or control plane data, such as physical shared channels, physical control channels, and physical broadcast channels. In other words, the first channel only carries data for AI function monitoring and does not carry other data. The first device receives monitoring data for each of the M AI functions, including receiving the monitoring data for each of the M AI functions from the first channel. For example, the first channel can be a physical monitoring channel (PMCH). For instance, if the first device is a terminal device and the second device is a base station, the first channel can be a physical downlink monitoring channel (PDMCH); if the first device is a base station and the second device is a terminal device, the first channel can be a physical uplink monitoring channel (PUMCH).

[0031] In this implementation, since the monitoring data belongs to neither the control plane nor the user plane, a first channel is added to send the monitoring data of each of the M AI functions. This decouples the content carried by each physical channel, making it easier to manage the process of sending monitoring data.

[0032] In one possible implementation, the resource mapping location information of the monitoring data in the first channel is predefined or preconfigured. The aforementioned predefinition refers to predefinition by the communication protocol; for example, preconfiguration can be preconfigured by the base station using Radio Resource Control (RRC) configuration information or indication information. The resource mapping location information includes the resource mapping start position and the resource mapping density.

[0033] In this implementation, the resource mapping location information of the monitoring data in the first channel can be specified by predefinition or preconfiguration. The resource mapping location information includes the resource mapping start position and resource mapping density. By specifying the source of the resource mapping location information of the monitoring data, the overhead required to indicate the monitoring data can be reduced.

[0034] In one possible implementation, the monitoring data in the first channel is transmitted using semi-persistent scheduling (SPS). The SPS configuration information or activation indication specifies the resource mapping location between AI function information and monitoring data. Specifically, transmitting monitoring data for M AI functions using SPS means that during the period when monitoring data is transmitted using SPS, the first device will periodically receive monitoring data for M AI functions on the resources indicated in the SPS configuration information or activation indication.

[0035] The AI ​​function information includes information for each of the M AI functions. This information allows the first device to determine which AI functions or AI models to monitor for performance. For example, the information for each AI function may include one or more of the following: the function type of the AI ​​function, the ID of the AI ​​model used to implement the AI ​​function, the ID or associated ID of the training dataset of the AI ​​model, or other information that allows the first device to determine which AI models to monitor for performance. The SPS's RRC configuration information or activation indication may also include resource mapping location information, which indicates the resource mapping location of the monitoring data.

[0036] In this implementation, the monitoring data of M AI functions are periodically sent multiple times using the SPS method, which helps to reduce the control overhead during the transmission of monitoring data.

[0037] In one possible implementation, the monitoring data in the first channel is sent using a dynamically triggered method. The dynamically triggered indication information indicates the resource mapping position of the AI ​​function information and the monitoring data. Alternatively, the resource mapping position of the AI ​​function information and the monitoring data can be indicated in the RRC configuration information. The RRC configuration information or the dynamically triggered indication information may include resource mapping position information, which indicates the resource mapping position of the monitoring data.

[0038] In this implementation, the monitoring data of M AI functions is sent by dynamic triggering, which improves the flexibility of the monitoring data sending process and allows the monitoring data to be sent flexibly according to actual needs.

[0039] In one possible implementation, the dynamically triggered indication information uses a dedicated first downlink control information (DCI) format. And / or, the dynamically triggered indication information is transmitted on a dedicated second channel, for example, the second channel may be a physical downlink monitoring control channel (PDMCCH). And / or, the dynamically triggered indication information uses a second DCI format, which is the DCI format corresponding to a physical downlink shared channel (PDSCH) or a physical uplink shared channel (PUSCH).

[0040] This implementation provides two methods for sending dynamically triggered indications, improving the flexibility of the solution. The dynamically triggered indication information indicates that monitoring data is to be sent. This information uses a dedicated first DCI format and / or is sent on a dedicated second channel. This allows the first device to easily detect that the indication information indicates the need to send monitoring data by utilizing the first DCI format and / or the second channel, reducing coupling with other indication information.

[0041] In one possible implementation, the monitoring data is transmitted along with the data in the shared channel (SCH). This can also be understood as the monitoring data for the M AI functions being transmitted along with the data in the SCH. In this case, the data in the SCH is carried in the physical shared channel (PSCH). The first device receives the monitoring data for each of the M AI functions, including: the first device receiving the monitoring data for each of the M AI functions from the physical shared channel PSCH and / or from the first channel.

[0042] Furthermore, in one scenario, a first channel independent of the PSCH may not exist; that is, monitoring data for the M AI functions is transmitted within the PSCH. In another scenario, monitoring data for the M AI functions is transmitted within the first channel, and the monitoring data in the first channel is transmitted simultaneously with the data in the PSCH. Both the first channel and the PSCH simultaneously occupy the radio resources of the current communication, such as the radio resources of the current downlink or uplink communication.

[0043] In this implementation, monitoring data can be transmitted along with data in the SCH. In other words, monitoring data can utilize the available communication resources of the SCH. Since the amount of monitoring data is relatively small, its impact on SCH transmission is minimal. Compared to transmitting monitoring data through a separate first channel, this method reuses the available communication resources of the SCH, eliminating the need for independent scheduling and resource allocation for the first channel, thus reducing the overhead of resource allocation for the first channel. Furthermore, it reduces the need for terminal devices to receive monitoring data independently, thereby shutting down radio frequency and other receiving modules when there is no SCH data transmission, reducing the power consumption of the terminal devices.

[0044] In one possible implementation, whether the monitoring data of M AI functions are transmitted via an independent first channel or transmitted along with data from a shared channel, resource mapping conflicts may occur during resource mapping of the monitoring data of the M AI functions. A resource mapping conflict means that different data are mapped to the same RE. Optionally, the resource mapping priority of the monitoring data is lower than the resource mapping priority of the third reference signal. In other words, when the monitoring data of the M AI functions conflicts with the third reference signal, the third reference signal is mapped to the location of the conflicting RE. For example, the third reference signal includes the reference signal in the physical broadcasting channel (PBCH) and the reference signal in the PSCH.

[0045] In this implementation, the resource mapping priority of the monitoring data is lower than that of the resource mapping priority of the third reference signal. The third reference signal includes the reference signal in PBCH and the reference signal in PSCH. Since the resource mapping priority of the monitoring channel is lower, the transmission of the reference signal that needs to be transmitted normally is guaranteed, thus avoiding the impact of the performance monitoring operation of the AI ​​function on the normal data transmission in the communication network.

[0046] Secondly, embodiments of this application provide a data processing method applied to a second device. For example, the second device may be a terminal device, or it may be a component of the terminal device (e.g., a processor, circuit, chip, or chip system responsible for communication functions, including but not limited to a modem chip, a baseband chip, a system-on-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip, etc.). Alternatively, the first device may be a logic module or software capable of implementing all or part of the terminal device's functions. The following description uses a second device as an example.

[0047] The method includes: a second device sending monitoring data of each of M AI functions to a first device, wherein the monitoring data of each AI function is used to determine the value of the performance index of each AI function, M is an integer greater than 1, the M AI functions include a first AI function and a second AI function, and the input of the second AI function is determined based on the output of the first AI function.

[0048] In one possible implementation, the method further includes: a second device receiving performance monitoring results from a first device, the performance monitoring results including performance monitoring information for N AI functions out of M AI functions, the performance monitoring information for each AI function being obtained based on the value of the performance index of each AI function, where N is an integer greater than or equal to 1 and N is less than or equal to M.

[0049] In one possible implementation, the second device sends monitoring data of each of the M AI functions to the first device, including: the second device sending monitoring data of each of the M AI functions to the first device through a first channel.

[0050] In one possible implementation, monitoring data is transmitted along with data in the shared channel (SCH).

[0051] The meanings of the terms used in the various possible implementations of the second aspect of this application, as well as the beneficial effects they bring, can be found in the first aspect, and will not be repeated here.

[0052] Thirdly, embodiments of this application provide an apparatus that performs the functions described in the first aspect. For example, the apparatus includes modules, units, or means corresponding to the operations involved in the first aspect. These modules, units, or means can be implemented in software, hardware, or a combination of both. For instance, the apparatus includes a processing unit and a transceiver unit. The transceiver unit receives monitoring data from each of M AI functions, where M is an integer greater than 1. The M AI functions include a first AI function and a second AI function, where the input of the second AI function is determined based on the output of the first AI function. The monitoring data for each AI function is used to determine the value of a performance indicator for each AI function. The processing unit determines a performance monitoring result, which includes performance information for N AI functions out of the M AI functions. The performance information for each AI function is obtained based on the value of a performance indicator for each AI function, where N is an integer greater than or equal to 1 and less than or equal to M.

[0053] In the third aspect of this application, the apparatus is also used to perform the steps performed by the first device in the first aspect and various possible implementations of the first aspect. The specific implementations of the steps in the third aspect, the meanings of the terms, and the beneficial effects thereon can all be found in the first aspect, and will not be repeated here.

[0054] Fourthly, embodiments of this application provide an apparatus that performs the functions described in the second aspect above. For example, the apparatus includes modules, units, or means corresponding to the operations involved in the second aspect. These modules, units, or means can be implemented in software, hardware, or a combination of both. For instance, the apparatus includes a transceiver unit; this transceiver unit is used to send monitoring data for each of M AI functions to a first device. The monitoring data for each AI function is used to determine the value of a performance indicator for each AI function. M is an integer greater than 1, and the M AI functions include a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function.

[0055] In the fourth aspect of this application, the data processing apparatus is also used to perform the steps performed by the second device in the second aspect and various possible implementations of the second aspect. The specific implementations of the steps in the fourth aspect, the meanings of the terms, and the beneficial effects thereon can all be found in the second aspect, and will not be repeated here.

[0056] Fifthly, embodiments of this application provide an apparatus including a processor and a memory, the processor being coupled to the memory, the memory being used to store a program; the processor being used to execute the program in the memory, causing the apparatus to perform the data processing method described in the first or second aspect above.

[0057] In a sixth aspect, embodiments of this application provide a system including a first device and a second device, wherein the first device performs the data processing method described in the first aspect, and the second device performs the data processing method described in the second aspect.

[0058] In a seventh aspect, embodiments of this application provide a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the data processing methods described in the first or second aspect.

[0059] Eighthly, embodiments of this application provide a computer program product, which includes a program that, when run on a computer, causes the computer to perform the data processing method described in the first or second aspect.

[0060] Ninthly, this application provides a chip or chip system including at least one processor for supporting a communication device in implementing any possible implementation of the method described in the first or second aspect above. For example, the chip may be a baseband chip, a modem chip, a system-on-a-chip (SoC) chip containing a modem core, a system-in-package (SIP) chip, or a communication module, etc.

[0061] In one possible design, the chip or chip system may further include a memory for storing program instructions and data necessary for the communication device. The chip system may be composed of chips or may include chips and other discrete devices. Optionally, the chip system may also include interface circuitry that provides program instructions and / or data to the at least one processor. Attached Figure Description

[0062] Figure 1 is a schematic diagram of an architecture of a wireless communication system provided in an embodiment of this application;

[0063] Figure 2 is a schematic diagram of a data processing method provided in an embodiment of this application;

[0064] Figure 3 is a schematic diagram of determining the first codeword using the first formula according to an embodiment of this application;

[0065] Figure 4 is a schematic diagram of resource mapping of monitoring data of M AI functions provided in an embodiment of this application;

[0066] Figure 5 shows two schematic diagrams of resource mapping of monitoring data of M AI functions provided in the embodiments of this application;

[0067] Figure 6 is a schematic diagram of another data processing method provided in an embodiment of this application;

[0068] Figure 7 is a schematic diagram of obtaining the performance index value of each AI function based on the monitoring data of each AI function in M ​​AI functions according to an embodiment of this application;

[0069] Figure 8 is a schematic diagram of transmission monitoring data and performance monitoring results provided in an embodiment of this application;

[0070] Figure 9 is another schematic diagram of obtaining the performance index value of each AI function based on the monitoring data of each AI function in M ​​AI functions according to an embodiment of this application;

[0071] Figure 10 is a schematic diagram of another data processing method provided in an embodiment of this application;

[0072] Figure 11 is a schematic diagram of transmitting monitoring data of each of the M AI functions using SPS according to an embodiment of this application;

[0073] Figure 12 is a schematic diagram of sending performance monitoring results to a base station according to an embodiment of this application;

[0074] Figure 13 is another flowchart illustrating the data processing method provided in an embodiment of this application;

[0075] Figure 14 is a schematic diagram of sending monitoring data of each of the M AI functions in a dynamic triggering manner according to an embodiment of this application;

[0076] Figure 15 is another flowchart illustrating the data processing method provided in an embodiment of this application;

[0077] Figure 16 is another schematic diagram of the monitoring data of M AI functions provided in the embodiment of this application;

[0078] Figure 17 is a schematic diagram of a data processing device provided in an embodiment of this application;

[0079] Figure 18 is a schematic diagram of a device provided in an embodiment of this application. Detailed Implementation

[0080] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0081] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0082] In the embodiments of this application, "send" and "receive" refer to the direction of signal transmission. For example, "send information to device XX" can be understood as the destination of the information being device XX, which may include direct transmission via the air interface or indirect transmission by other units or modules via the air interface. "Receive information from device YY" can be understood as the source of the information being device YY, which may include direct reception from device YY via the air interface or indirect reception from device YY via other units or modules via the air interface. "Send" can also be understood as the "output" of the chip interface, and "receive" can also be understood as the "input" of the chip interface. In other words, sending and receiving can occur between devices or within devices, for example, through buses, traces, or interfaces between components, modules, chips, software modules, or hardware modules within a device. It is understood that information may undergo necessary processing, such as encoding and modulation, between the source and destination of information transmission, but the destination can understand the valid information from the source. Similar expressions in this application can be understood in a similar way and will not be elaborated further.

[0083] In the embodiments of this application, "instruction" can include direct and indirect instructions, as well as explicit and implicit instructions. The information indicated by a certain piece of information (hereinafter referred to as instruction information) is called the information to be instructed. In specific implementation, there are many ways to indicate the information to be instructed, such as, but not limited to, directly indicating the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly indicate the information to be instructed by indicating other information, where there is a correlation between the other information and the information to be instructed; or it can only indicate a part of the information to be instructed, while the other parts are known or pre-agreed upon. For example, the instruction can be implemented by using a pre-agreed (e.g., a predefined communication protocol) arrangement of various information, thereby reducing the instruction overhead to some extent. This application does not limit the specific method of instruction. It is understood that for the sender of the instruction information, the instruction information can be used to indicate the information to be instructed; for the receiver of the instruction information, the instruction information can be used to determine the information to be instructed.

[0084] Before providing a detailed description of the method provided in this application, the application scenario of the method provided in the embodiments of this application will be introduced first. Please refer to Figure 1, which is a schematic diagram of the architecture of a wireless communication system provided in an embodiment of this application. The method provided in this application can be applied to a wireless communication system. As shown in Figure 1, the wireless communication system includes a network device 101 and terminal devices (mobile stations, MS) 102. Wireless connections can be established between the network device 101 and each terminal device 102, and wireless connections can also be established between each terminal device 102.

[0085] Network device 101 can refer to a device that provides wireless access services in a wireless network. For example, network device 101 can be a device that connects terminal device 102 to the wireless network, and can also be called a base station; the aforementioned base station can be various forms of macro base stations, micro base stations, relay stations, or access points, etc. In wireless communication systems employing different wireless access technologies, the name of network device 101 with base station functionality may differ. For example, a base station can be called an evolved Node B (eNB), Node B (NB), the next generation Node B (gNB) in a 5th generation (5G) communication system, a home base station (e.g., home evolved Node B, or home Node B (HNB), a base band unit (BBU), a wireless fidelity (Wi-Fi) access point (AP), a transmission reception point (TRP), or a radio network controller (RNC), etc. In another possible scenario, network device 101 may include multiple network nodes that cooperate to achieve wireless access, with each network node implementing a portion of the base station's functions. For example, network nodes may be central units (CUs), distributed units (DUs), CU-control plane (CPs), CU-user plane (UPs), or radio units (RUs), etc. CUs and DUs may be separate entities or included in the same network element, such as a baseband unit (BBU). RUs may be included in radio frequency equipment or radio frequency units, such as remote radio units (RRUs), active antenna units (AAUs), or remote radio heads (RRHs). In different systems, CUs (or CU-CPs and CU-UPs), DUs, or RUs may have different names, but their meanings will be understood by those skilled in the art.For example, in an ORAN system, a CU can also be called an open CU (O-CU), a DU can also be called an open DU (O-DU), a CU-CP can also be called an open CU-CP (O-CU-CP), a CU-UP can also be called an open CU-UP (O-CU-UP), and an RU can also be called an open RU (O-RU). Any of the CU (or CU-CP, CU-UP), DU, and RU units can be implemented through software modules, hardware modules, or a combination of software and hardware modules. This application embodiment does not limit the specific device form of network device 101.

[0086] Terminal device 102 refers to a wireless terminal device capable of receiving scheduling information and instruction information sent by network device 101. Terminal device 102 can be a device with wireless communication capabilities, and can be implemented through software modules, hardware modules, or a combination of software and hardware modules; for example, terminal device can be a handheld device, vehicle-mounted device, wearable device, computing device, or other processing device, etc., which are not exhaustively listed here.

[0087] Terminal device 102 can communicate with one or more core networks or the Internet via a wireless access network (RAN). For example, terminal device 102 can be a portable, pocket-sized, handheld, computer-embedded, or vehicle-mounted mobile device that exchanges data with the RAN. Exemplarily, terminal device 102 can be a user agent, cellular phone, smartphone, personal digital assistant (PDA), tablet PC, modem, handset, laptop computer, personal communication service (PCS) phone, remote station, access point (AP), remote terminal, access terminal, customer premises equipment (CPE), terminal, user equipment (UE), or mobile terminal (MT), etc.; or, terminal device 102 can also be a chip or software in the aforementioned mobile devices.

[0088] For example, terminal device 102 can also be a wearable device or a chip or software within a wearable device. Wearable devices are a general term for devices that utilize wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that are worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not merely hardware devices; they achieve powerful functions through software support, data interaction, and cloud interaction. Broadly defined, wearable smart devices include those with comprehensive functions, large sizes, and the ability to perform complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses. They also include devices focused on a specific application function that require the use of other devices, such as smart bracelets, smart helmets, and smart jewelry for vital sign monitoring.

[0089] For example, terminal device 102 can also be a drone, robot, terminal device in device-to-device (D2D) communication, terminal device in vehicle-to-everything (V2X) communication, virtual reality (VR) device, augmented reality (AR) device, wireless terminal in industrial control, terminal device in self-driving, terminal device in remote medical care, terminal device in smart grid, wireless terminal in smart city, terminal device in smart home, etc.; or, terminal device 102 can also be the chip or software in the aforementioned devices.

[0090] In addition, terminal device 102 may also be a terminal device in a future communication system or a terminal device in a future evolved public land mobile network (PLMN), etc. The embodiments of this application do not limit the device form of terminal device 102.

[0091] For example, network device 101 and / or terminal device 102 may each deploy an AI receiver. The AI ​​receiver may include multiple AI functions, which may include: channel estimation, demodulation, channel decoding, source decoding, and joint source-channel decoding. The method provided in this application can be used to monitor the performance of at least one of the aforementioned AI functions. The demodulation in this application can also be understood as equalization and detection, and can be used in multi-antenna scenarios (such as multiple-input multiple-output (MIMO)).

[0092] For example, an AI transceiver can be deployed on network device 101 and terminal device 102. The AI ​​transceiver can include multiple AI functions, such as source coding, channel coding, joint source-channel coding, modulation, channel estimation, demodulation, channel decoding, source decoding, and joint source-channel decoding. The method provided in this application can be used to monitor the performance of at least one of the aforementioned AI functions. It should be noted that the AI ​​receiver and AI transceiver can also include more AI functions, which are not exhaustively listed here. It is understood that the deployment of the AI ​​receiver or AI transceiver in this application refers to a scenario where at least one AI function is deployed in the network device and / or terminal device using the AI ​​receiver use case or AI transceiver use case, i.e., in the links or functional modules involved in the receiving process or the transmission and reception process. The AI ​​functions in the AI ​​receiver in this application can also be understood as the AI ​​functions in the AI ​​receiver use case.

[0093] Besides the scenario described in Figure 1 above, the method provided in this application embodiment can also be applied to other communication system scenarios. For example, network elements in the core network and base stations can also be connected via wireless networks and exchange data with each other via wireless networks. The method provided in this application can be used to monitor the performance of AI functions used in communication between network elements in the core network and base stations. This application embodiment does not limit the specific scenario in which the data processing method is applied.

[0094] It should be noted that the wireless communication systems mentioned in the embodiments of this application include, but are not limited to: 5th Generation Mobile Communication Technology (5G) communication systems, future communication systems, satellite communication systems, short-range communication systems, narrowband Internet of Things (NB-IoT) systems, and Long Term Evolution (LTE) systems. The embodiments of this application do not limit the specific architecture of the wireless communication systems.

[0095] Based on the above description, the detailed implementation process of the data processing method provided in this application will be described below. Specifically, please refer to Figure 2, which is a schematic diagram of a data processing method provided in an embodiment of this application. As shown in Figure 2, the data processing method provided in this application may include:

[0096] 201. The first device receives monitoring data for each of the M AI functions. The monitoring data for each AI function is used to determine the value of the performance index of each AI function.

[0097] For example, the second device sends monitoring data for each of the M AI functions to the first device. Correspondingly, the first device can receive monitoring data for each of the M AI functions from the second device. The term "monitoring" in this application can also be replaced with "performance monitoring," "model monitoring," "use case monitoring," "feature monitoring," or "configuration monitoring." Similarly, the term "monitoring data" can be replaced with "performance monitoring data," "model monitoring data," "use case monitoring data," "feature monitoring data," or "configuration monitoring data," where M is an integer greater than or equal to 1. When used in a wireless communication system, the data in this application can also be understood as measurement results, signals, sequences, etc.

[0098] Optionally, if performance monitoring is being performed on the AI ​​function in an AI receiver or AI transceiver, the first device can be a terminal device and the second device can be a base station; alternatively, the first device can be a base station and the second device can be a terminal device. For examples of the product forms of base stations and terminal devices, please refer to the above description, which will not be repeated here. If performance monitoring is being performed on the AI ​​function within other devices in the communications field, the first and second devices can also be replaced with other devices, depending on the specific application scenario.

[0099] In this application, an AI function refers to an AI feature / feature group enabled by one or more configurations (such as radio resource control (RRC) configurations), which can be determined by applicable conditions supported by the first device and / or the second device. For example, functional modules such as channel estimation and demodulation can also be referred to as different features. When a feature can be enabled by AI under a specific configuration, it can be referred to as the corresponding AI function. For example, if channel estimation can be performed using AI under bandwidth configuration 1, then channel estimation under bandwidth configuration 1 can be referred to as an AI function. As another example, if demodulation can be performed using AI under modulation scheme configuration 1, then demodulation under modulation scheme configuration 1 can be referred to as an AI function. Optionally, an AI function may include one or more AI models that can implement the AI ​​function. For example, each of the multiple AI models in the UE (i.e., an example of the first device) can implement the target AI function. The UE (i.e., an example of the first device) can choose which of the aforementioned AI models to use to implement the target AI function. In this scenario, the AI ​​model can be transparent to the network device (i.e., an example of the second device), while the AI ​​function is not transparent to the network device. In this application, it is not limited to whether the network device configures or instructs the UE based on the AI ​​model or the AI ​​function. In this application, the AI ​​function can also be replaced by AI model, AI module, AI functional unit, AI functional module, AI use case, AI configuration, or AI feature, and these descriptions are interchangeable.

[0100] For example, each of the M AI functions can be implemented by one or more AI models. In this application, the AI ​​model can also be referred to as a machine learning model. Optionally, the M AI functions may include at least one of the following: channel estimation, demodulation, channel decoding, source decoding, modulation and demodulation, channel coding and decoding, or source coding and decoding. For example, if the performance monitoring of the AI ​​functions in an AI receiver is being performed, the M AI functions may include at least one of the following: channel estimation, demodulation, channel decoding, source decoding, or joint source-channel decoding. As another example, if the performance monitoring of the AI ​​functions in an AI transceiver is being performed, the M AI functions may include at least one of the following: channel estimation, modulation and demodulation, channel coding and decoding, source coding and decoding, or joint source-channel coding and decoding. In this implementation, by simultaneously transmitting the performance monitoring data corresponding to the M AI functions, the power consumption and latency required for the first device to receive the monitoring data are reduced. It clarifies which AI functions can be used for the M AI functions, not only improving the integration of this solution with specific application scenarios but also expanding the application scenarios of this solution.

[0101] For example, the source-channel joint coding can be source-channel joint coding of CSI information, and the source-channel joint decoding can be source-channel joint decoding of CSI information. For example, the source coding can be source coding of CSI information (e.g., compression of CSI information), and the source decoding can be source decoding of CSI information (e.g., recovery of CSI information).

[0102] Optionally, M is an integer greater than 1, and the M AI functions include a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function (which can be understood as determining the output of the second AI function requires the output of the first AI function). In other words, there is a dependency relationship between the first AI function and the second AI function, and the input of the second AI function depends on the output of the first AI function (which can be understood as determining the output of the second AI function depends on the output of the first AI function). Alternatively, there is an order relationship between the first AI function and the second AI function, and the first AI function can be the preceding AI function of the second AI function. Or, there is an upstream and downstream relationship between the first AI function and the second AI function, and the first AI function can be the upstream AI function of the second AI function.

[0103] For example, if the performance of the AI ​​function in an AI receiver is being monitored, for instance, the first AI function is channel estimation and the second AI function is demodulation. The output of the first AI function includes channel information, and the input of the second AI function includes channel information and the received modulation symbols. Alternatively, the first AI function is demodulation and the second AI function is channel decoding. The output of the first AI function includes the coded codeword obtained after demodulating the modulation symbols using the first AI function, and the input of the second AI function includes the aforementioned coded codeword. Another example is that the first AI function is channel decoding and the second AI function is source decoding. The output of the first AI function includes the codeword obtained after channel decoding the coded codeword using the first AI function, and the input of the second AI function includes the aforementioned codeword, and so on.

[0104] If the performance of AI functions in an AI transceiver is being monitored, for example, the first AI function is channel estimation and the second AI function is modulation and demodulation. The output of the first AI function includes channel information, and the input of the second AI function includes channel information and received modulation symbols; or, for example, the first AI function is modulation and demodulation and the second AI function is channel coding and decoding. The output of the first AI function includes coded codewords, and the input of the second AI function includes the aforementioned coded codewords; or, for example, the first AI function is channel coding and decoding and the second AI function is source coding and decoding. The output of the first AI function includes the codewords obtained after channel decoding by the first AI function, and the input of the second AI function includes the aforementioned codewords, and so on.

[0105] It should be noted that this example only illustrates the performance monitoring of AI functions in AI receivers or AI transceivers. When monitoring the performance of AI functions in other devices in the communication field, the first and second AI functions may also manifest as other functions, which can be determined based on the actual application scenario. This application does not exhaustively list them.

[0106] In this application, monitoring data for each of the M AI functions is used to determine the performance metric value for each AI function. The performance metric in this application can also be referred to as performance monitoring metric, AI function monitoring metric, AI model monitoring metric, monitoring metric, or model monitoring metric, etc. The performance metric for each AI function can be obtained based on the output of that AI function and the reference value corresponding to its output. The reference value corresponding to the output of the AI ​​function can be the ground truth value corresponding to the output of the AI ​​function. When the ground truth value is unavailable (e.g., the ground truth value in channel estimation is the actual channel information, which cannot be accurately obtained through estimation), the reference value corresponding to the output of the AI ​​function can be understood as an approximation of the ground truth value corresponding to the output of the AI ​​function; that is, the reference value can be used as the ground truth value. Therefore, in this application, for the sake of simplicity, reference value and ground truth value can be interchanged.

[0107] For example, if the AI ​​function is channel estimation, the performance metrics of the AI ​​function may include at least one of the following: normalized mean square error (NMSE), mean square error (MSE), generalized cosine similarity (GCS), squared generalized cosine similarity (SGCS), or other performance metrics. For instance, if the AI ​​function is channel estimation, the performance metrics of the AI ​​function can be obtained by calculating the NMSE and / or SGCS between the output of the AI ​​function and the reference value.

[0108] For example, if the AI ​​function is demodulation or modulation / demodulation, the performance metrics of the AI ​​function can be bit error rate (BER), cross-entropy, or other performance metrics. For instance, if the AI ​​function is demodulation or modulation / demodulation, the BER between the output of the AI ​​function and the reference value can be calculated to obtain the value of the performance metric of the AI ​​function.

[0109] For example, if the AI ​​function is channel decoding or channel coding decoding, the performance metrics of the AI ​​function may include at least one of the following: block error rate (BLER), BER, cross entropy, or other performance metrics; for instance, if the AI ​​function is channel decoding or channel coding decoding, the performance metrics of the AI ​​function can be obtained by calculating the BLER and / or BER between the output of the AI ​​function and the reference value.

[0110] It should be understood that the above examples are only for the convenience of understanding this solution. When monitoring the performance of other AI functions, other performance indicators can be used. The specific ones can be determined in combination with the actual application scenario. This application does not exhaustively list them.

[0111] For example, monitoring data from M AI functions can be used to obtain the input of each of the M AI functions, and the input of each AI function is used to obtain the output of each AI function. Optionally, the monitoring data from the M AI functions may include one or more of the following: a first reference signal (RS), used to determine the input of channel estimation; a second reference signal, used to determine the reference value of channel estimation; modulation symbols, used to determine the input of demodulation or modulation / demodulation; or coded codewords, used to determine the input of channel decoding or channel coding / decoding. In this application, decoding can also be understood as decoding. Optionally, the coded codewords may also be used to determine the input of source-channel joint decoding or source-channel joint coding / decoding. Optionally, the coded codewords may also be used to determine the input of source decoding or source coding / decoding. Optionally, the modulation symbols may also be used to determine the input of source-channel joint decoding and demodulation or source-channel joint coding / decoding and modulation / demodulation. It should be understood that the above monitoring data may be data and / or signals transmitted through or without a channel. In this embodiment, when the M AI functions include at least one of the following: channel estimation, demodulation, channel decoding, modulation / demodulation, or channel coding / decoding, the first reference signal can determine the input for channel estimation, the reference value and / or modulation symbol for channel estimation can determine the input for demodulation or modulation / demodulation, and the coded codeword can determine the input for channel decoding or channel coding / decoding. That is, it clarifies which data are included in the monitoring data of the M AI functions, as well as the input of each AI function. By simultaneously sending the monitoring data of the M AI functions, the performance monitoring efficiency is further improved, and the power consumption and latency of the terminal device in performance monitoring are reduced.

[0112] For example, if the AI ​​function is channel estimation, the input of this AI function can be obtained based on a first reference signal from the monitoring data of M AI functions. For instance, the input of this AI function could be the received signal at the first reference signal receiving location and / or the first reference signal itself. The first reference signal can be a reference signal determined or generated by the second device using an AI model. Alternatively, the first reference signal can also be a reference signal determined by the second device using a non-AI method. For example, the first reference signal can be determined using a protocol-defined sequence (such as a Gold sequence, a Zadoff-Chu (ZC) sequence, an M sequence, etc.). In other words, the generation process of the first reference signal can also be without using or relying on an AI model. In this application, the reference signal determined using a non-AI method can be a legacy reference signal or other conventional reference signals.

[0113] For example, if the AI ​​function is demodulation or modulation / demodulation, the input to the AI ​​function can include channel information and received modulation symbols. The input to the AI ​​function can be obtained based on the second reference signal and modulation symbols from the monitoring data of M AI functions. The second reference signal is used to determine the reference value for channel estimation. This reference value for channel estimation can be understood as accurate channel information. The second reference signal can also be a reference signal obtained by the second device using a non-AI method. For example, the second reference signal can be a traditional RS, such as a demodulation reference signal (DMRS) generated based on Gold or ZC sequences as defined by the protocol, or other types of reference signals, etc., which will not be exhaustively listed here. The reference value for the channel estimation obtained based on the second reference signal is also determined using a non-AI method, such as least squares (LS), minimum mean square error (MMSE), Wiener filtering, etc. In other words, no AI model is used when performing channel estimation based on the second reference signal and obtaining the reference value for the channel estimation. For example, the first device can perform MMSE estimation based on the received second reference signal and the known transmitted second reference signal to obtain the reference value for the channel estimation. It should be noted that the aforementioned MMSE can be replaced by other algorithms that do not use AI models for channel estimation. This example is only to demonstrate the feasibility of this scheme. Exemplarily, the input to the aforementioned demodulation or modulation / demodulation can include the aforementioned reference value for the channel estimation determined based on the second reference signal.

[0114] For example, if the AI ​​function is channel decoding or channel coding, the input of the AI ​​function can include coded code blocks from the monitoring data of M AI functions.

[0115] Optionally, if one of the M AI functions is channel estimation, the reference value corresponding to the output of the AI ​​function can be determined based on the second reference signal. For example, the reference value corresponding to the output of the aforementioned channel estimation includes the reference value of the channel estimation determined based on the second reference signal.

[0116] Optionally, the modulation symbols in the monitoring data of the aforementioned M AI functions can be obtained by modulating the first codeword using a second device. If the AI ​​function is demodulation or modulation-demodulation, the input of the AI ​​function includes the received modulation symbols, and the reference value corresponding to the output of the AI ​​function is the first codeword. For example, the first codeword can be understood as a first bit stream or a first sequence. The first codeword is determined according to a predefined method. For example, the aforementioned predefined method can be predefined through a communication protocol, or it can be predefined by the base station through non-communication protocol forms such as radio resource control (RRC) configuration information or indication information.

[0117] The determination of the first codeword according to a predefined method can be understood as the determination process of the first codeword being predefined. For example, in order to ensure the determinism of the determination process of the first codeword, no AI model is used in the determination process of the first codeword.

[0118] In this embodiment of the application, since the first codeword is determined according to a predefined method, errors or unknowns are avoided in the process of determining the first codeword, thereby avoiding the impact of the process of determining the first codeword on the performance of the demodulation AI function, which is conducive to obtaining the true performance of the demodulation AI function.

[0119] Optionally, determining the first codeword according to a predefined method can be further understood as follows: the first codeword is determined using a predefined first formula and the value of a first parameter of a predefined type. The value of the first parameter of the predefined type can be understood as the value of the parameter included in the predefined first formula, or it can also be understood as the seed of the predefined first formula. The predefined first formula and the first parameter of the predefined type can be predefined through a communication protocol, or they can be predefined by the base station through non-communication protocol forms such as RRC configuration information or indication information. For example, the first formula is a formula for generating a Gold sequence, and the Gold sequence generated using the first formula can be used as the first codeword, or the first formula is a formula for generating the initial value of a Gold sequence. The first parameter can be a parameter in the formula for generating a Gold sequence, or the first parameter can be a parameter in the formula for determining the initial value of a Gold sequence. For example, taking the first formula as a formula for generating the initial value of a Gold sequence, where the first formula contains one formula and the first parameter has three types, the first formula can be: C init =(2 17 (N t (2*N) AI +1)+2*N AI +N1)mod 2 31

[0120] Where, N tN AI N1 is the first parameter, and its type is predefined, for example, N t For time-related parameters (such as time slot number), N AI N1 represents parameters related to AI functions (such as AI function identifiers), while N1 represents other predefined parameters (such as base station configuration parameters). It should be noted that N... t N AI The specific values ​​corresponding to N1 are the values ​​of the first parameter. `mod` is the modulo operation, or remainder operation. After determining the initial value of the Gold sequence using the first formula, the terminal device (an example of the first device) and the network device (an example of the second device) can determine the target Gold sequence based on the initial value. Based on the target Gold sequence, the first codeword can be further determined; for example, the target Gold sequence is used as the first codeword. It should be noted that the aforementioned first formula can have other expressions, and the first parameter involved in the first formula is not limited to the above-mentioned types. The examples here are only to demonstrate the feasibility of this scheme.

[0121] The first formula can also be an AI model. For example, the first formula can be a data generator determined by an AI model or other formulas, etc. This application does not exhaustively list them. Both the first device and the second device can determine the first codeword using a predefined first formula and the value of a first parameter of a predefined type.

[0122] Optionally, the first parameter of the predefined type may include at least one of the following: the identifier (ID) of the first AI function, the time information of the transmission time of the monitoring data (e.g., time slot number), or other network device configuration parameters. The first AI function is demodulation or modulation-demodulation. For example, the ID of the first AI function may be the ID of the AI ​​model used to implement the aforementioned first AI function, or the ID of the first AI function may be the ID of the configuration corresponding to the first AI function (e.g., RRC configuration). Optionally, in order to generate the modulation symbols in the monitoring data of M AI functions, the second device may first determine the first codeword based on the ID of the first AI function and / or the time slot number of the transmission time of the monitoring data using the first formula, and then modulate the first codeword to obtain the aforementioned modulation symbols. For the second device, the time slot number of the transmission time of the monitoring data is the time slot number of the transmission time of the monitoring data. Optionally, if the monitoring data of M AI functions is transmitted using SPS, the time slot number of the transmission time of the monitoring data can also be understood as the time slot number of the transmission time indicated by the SPS trigger, which can be understood as the transmission time when the monitoring data of M AI functions is first transmitted.

[0123] Optionally, if the first codeword is determined according to a predefined method, it can be further understood as follows: the first codeword is determined using a predefined first formula and the value of a first parameter of a predefined type. In order to obtain the truth value corresponding to the output of the first AI function, demodulation or modulation / demodulation, the first device can determine the first codeword based on the ID of the first AI function and / or the time information of the transmission time of the monitoring data (e.g., timeslot number). For example, when the first device and the second device are time-synchronized, for the first device, the timeslot number of the transmission time of the monitoring data is the timeslot number of the reception time of the monitoring data.

[0124] Optionally, the first formula may include at least two sub-formulas, and the first codeword may be generated based on at least two sequences determined by the at least two sub-formulas. For example, the first codeword is obtained by scrambling a sequence different from the first bit sequence on the basis of the first bit sequence. The first bit sequence is determined based on the first sub-formula in the first formula, and the sequences different from the first bit sequence are determined based on other sub-formulas besides the first sub-formula in the first formula. To understand this scheme more intuitively, please refer to Figure 3. Figure 3 is a schematic diagram of determining the first codeword using the first formula according to an embodiment of this application. As shown in Figure 3, the first device (optionally, also including a second device) generates the first bit sequence using the first sub-formula based on the ID of the first AI function; and generates the second bit sequence using the second sub-formula based on the timeslot number of the transmission time of the monitoring data. Optionally, the first bit sequence is scrambled using the second bit sequence to obtain the first codeword. It should be understood that the example in Figure 3 is only for the convenience of understanding this scheme and is not intended to limit this scheme.

[0125] Optionally, the coded codewords in the monitoring data of the aforementioned M AI functions are obtained by modulating the second codewords using a second device. If the AI ​​function is channel decoding or channel coding decoding, the input of the AI ​​function includes the aforementioned coded codewords, and the reference value corresponding to the output of the AI ​​function is the second codeword. For example, the second codeword can be understood as a second bitstream. The second codeword is determined according to a predefined method, the meaning of which can be found in the above description and will not be repeated here.

[0126] Optionally, determining the second codeword according to a predefined method can be further understood as follows: the second codeword is determined using a predefined second formula and the values ​​of a predefined type of second parameter. The values ​​of the predefined type of second parameter can be understood as the values ​​of the parameters included in the predefined second formula, or as the seed of the predefined second formula. The predefined second formula and the predefined type of second parameter can be predefined through a communication protocol, or by the base station through non-communication protocol forms such as RRC configuration information or indication information. The second formula can be an AI model or a non-model algorithm, for example, a data generator determined by an AI model or other formulas, etc., which are not exhaustively listed in this application. Both the second device and the second device can determine the second codeword using the predefined second formula and the values ​​of the predefined type of second parameter.

[0127] Optionally, the predefined type of second parameter may include at least one of the following: the ID of the second AI function, the time information of the transmission time of the monitoring data (e.g., the timeslot number), or other network device configuration parameters. The second AI function is channel decoding or channel coding decoding. For example, the ID of the second AI function may be the ID of the AI ​​model used to implement the aforementioned second AI function, or the ID of the second AI function may be the ID of the configuration corresponding to the second AI function (e.g., RRC configuration). Optionally, in order to generate coded code blocks in the monitoring data of M AI functions, the second device may first determine the second codeword using the second formula based on the ID of the second AI function and / or the timeslot number of the transmission time of the monitoring data (e.g., the transmission time of the monitoring data), and then encode the second codeword to obtain the aforementioned coded code blocks.

[0128] Optionally, if the second codeword is determined according to a predefined method, it can be further understood as follows: the second codeword is determined by using a predefined second formula and the value of a predefined type of second parameter. In order to obtain the truth value corresponding to the output of the second AI function, which is channel decoding or channel coding decoding, the first device can determine the second codeword based on the ID of the second AI function and the time information of the transmission time of the monitoring data (e.g., the time slot number). For example, when the first device and the second device are time-synchronized, the time slot number of the transmission time of the monitoring data is the same as the time slot number of the reception time of the monitoring data for the first device.

[0129] For example, the second codeword can be understood as a second bitstream or a second sequence. For instance, the second formula is a formula for generating the second sequence, or a formula for generating the initial value of the second sequence. The second parameter can be a parameter in the formula for generating the second sequence, or a parameter in the formula for determining the initial value of the second sequence. Further, the second formula can include one formula, or the second formula can include at least two sub-formulas, and the second codeword can be generated based on at least two sequences determined by the at least two sub-formulas.

[0130] Optionally, the first formula and the second formula can use the same formula. The second parameter of the predefined type includes the ID of the second AI function and the timeslot number of the transmission time of the monitoring data. The first device (optionally, also includes the second device) determines the second codeword based on the ID of the second AI function and the timeslot number of the transmission time of the monitoring data using the second formula. The specific implementation method can be found in the above description. The difference is that the ID of the first AI function is replaced with the ID of the second AI function. This will not be described again here.

[0131] In this embodiment, the reference value corresponding to the output of the channel estimation AI function is obtained based on a second reference signal; the reference value corresponding to the output of the demodulation or modulation / demodulation AI function is a first codeword; and the reference value corresponding to the output of the channel decoding or channel coding / decoding AI function is a second codeword. Both the first and second codewords are obtained using a predefined method. Therefore, the first device can directly generate the first and / or second codewords using a predefined method, thereby ensuring that the first device obtains a more accurate reference value. Since the performance index value of each AI function is obtained based on the output of that AI function and the corresponding reference value, obtaining a more accurate reference value is beneficial for obtaining a more accurate performance index value for each AI function, thereby improving the accuracy of the performance monitoring process. Simultaneously, generating the aforementioned reference values ​​using a predefined method can also reduce transmission overhead and power consumption.

[0132] Optionally, in another scenario, the monitoring data of the M AI functions may further include a first codeword and / or a second codeword. If the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword, and the first device can obtain the first codeword from the monitoring data of the M AI functions. For example, the base station device transmits the first codeword in a non-AI manner, and the terminal device determines the first codeword in a non-AI manner (where the first codeword can be correctly demodulated or decoded using a non-AI method). "Transmitting the first codeword in a non-AI manner" can be understood as the modulation process of the first codeword not employing an AI model. If the AI ​​function is channel decoding or encoding / decoding, the reference value corresponding to the output of the AI ​​function is the second codeword, and the first device can obtain the second codeword from the monitoring data of the M AI functions. For example, the base station device transmits the second codeword in a non-AI manner, and the terminal device determines the second codeword in a non-AI manner (where the second codeword can be correctly demodulated or decoded using a non-AI method). "Transmitting the second codeword in a non-AI manner" can be understood as the encoding and modulation process of the first codeword not employing an AI model.

[0133] In this embodiment, the monitoring data of the M AI functions further includes a first codeword and / or a second codeword. In other words, the reference value corresponding to the output of the demodulation or modulation / demodulation AI function and / or the reference value corresponding to the output of the channel decoding or channel coding / decoding AI function can be included in the monitoring data of the M AI functions and sent by the second device to the first device. This provides another way to obtain the reference value corresponding to the output of the AI ​​function. The second device can independently determine the first codeword and / or the second codeword, which improves the randomness of the monitoring data and thus improves the generalization of the monitoring scheme provided in this application.

[0134] For example, in one scenario, a first channel can be added between the first device and the second device. The first device and the second device use this first channel to transmit monitoring data for M AI functions. This first channel is independent of physical channels used for transmitting user plane or control plane data, such as physical shared channels, physical control channels, and physical broadcast channels. In other words, the first channel only carries data for AI function monitoring and not other data. The physical shared channel may include physical downlink shared channels and physical uplink shared channels; the physical control channel may include physical downlink control channels. For example, step 201 may include: the second device sending monitoring data for each of the M AI functions to the first device through the first channel; correspondingly, the first device receiving monitoring data for each of the M AI functions from the first channel. For example, the first channel may be a physical monitoring channel (PMCH). For example, if the first device is a terminal device and the second device is a base station, the first channel may be a physical downlink monitoring channel (PDMCH); if the first device is a base station and the second device is a terminal device, the first channel may be a physical uplink monitoring channel (PUMCH). In this embodiment, since the monitoring data belongs to neither the control plane nor the user plane, a first channel is added to send the monitoring data of each of the M AI functions, thereby decoupling the content carried by each physical channel and facilitating accurate management of the process of sending monitoring data.

[0135] For example, before the second device sends the monitoring data of each of the M AI functions to the first device through the first channel, it needs to map the monitoring data of the M AI functions onto resource elements (REs), i.e., perform physical resource mapping. For example, the resource mapping location information of the monitoring data in the first channel is predefined or preconfigured, wherein the aforementioned predefinition refers to the predefinition by the communication protocol, for example, the preconfiguration can be preconfigured by the base station using RRC configuration information or indication information.

[0136] Optionally, the resource mapping location information includes the resource mapping start position and the resource mapping density. Further, the resource mapping start position may include the start position of the monitoring data in at least one of the time, frequency, and spatial domains. Optionally, in this implementation, the monitoring data of the M AI functions are transmitted through an independent first channel, and the aforementioned start position may refer to the start position of the monitoring data within the first channel. The resource mapping density may include the mapping density of the monitoring data in at least one of the time, frequency, and spatial domains. The mapping density can be the interval between the RE positions where the monitoring data is located; for example, monitoring data is mapped every P REs in the frequency domain. The mapping density can also be the number of REs occupied by the monitoring data within a given resource range; for example, each resource block RB has Q REs, and these Q REs carry the monitoring data.

[0137] Furthermore, in one scenario, as described above, the monitoring data for the M AI functions can include various types of data, and the resource mapping location information can include the resource mapping start position and resource mapping density of each of the aforementioned various types of data. Alternatively, in another scenario, the monitoring data for each of the M AI functions can be considered as a set of data, and the resource mapping location information can include the resource mapping start position and resource mapping density of each of the M sets of data. The specific implementation method can be determined based on the actual application scenario.

[0138] It should be noted that if the first channel does not exist, transmission can also be made between the first device and the second device through other physical channels (such as physical shared channels). In this case, the configuration information involved in the transmission process based on the pre-configured resource mapping can also be applied to the relevant configurations of other physical channels.

[0139] To understand this solution more intuitively, please refer to Figure 4. Figure 4 is a schematic diagram of resource mapping of monitoring data of M AI functions provided by an embodiment of this application. Figure 4 shows the resource mapping positions of the first reference signal, the second reference signal, the modulation symbol, and the coded code block in the first channel. It should be understood that the example in Figure 4 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0140] In this embodiment, the resource mapping location information of the monitoring data in the first channel can be specified by predefined or preconfigured method. The resource mapping location information includes the resource mapping start position and resource mapping density. By specifying the source of the resource mapping location information of the monitoring data, the overhead required to indicate the monitoring data can be reduced, and the completeness and feasibility of this solution can be improved.

[0141] In one implementation where monitoring data for M AI functions is transmitted between the first device and the second device via a first channel, the monitoring data in the first channel is transmitted using semi-persistent scheduling (SPS). Transmitting the monitoring data for M AI functions using SPS means that during the period when the monitoring data is transmitted using SPS, the first device periodically receives the monitoring data for M AI functions on the resources indicated in the SPS configuration information or activation indication.

[0142] For example, the SPS’s RRC configuration information or SPS’s activation indication indicates the resource mapping location of AI function information and monitoring data. The aforementioned SPS’s activation indication can also be replaced by a trigger indication called SPS.

[0143] The AI ​​function information includes information for each of the M AI functions. This information allows the first device to determine which AI functions or AI models to monitor for performance. For example, the information for each AI function may include one or more of the following: the function type of the AI ​​function, the ID of the AI ​​model used to implement the AI ​​function, the ID of the training dataset or associated ID of the AI ​​model, or other information that allows the first device to determine which AI models to monitor for performance. The specific information can be determined based on the actual application scenario.

[0144] The SPS RRC configuration information or activation indication may also include resource mapping location information. The resource mapping location information indicates the resource mapping location of the monitoring data. The meaning of the resource mapping location information can be found in the above description and will not be repeated here.

[0145] Optionally, if the monitoring data of the M AI functions includes a second reference signal, the SPS's RRC configuration information or activation indication may also indicate the second reference signal. For example, the second reference signal may be generated in a predefined manner. For instance, the second reference signal may be determined using a predefined third formula and the value of a predefined type of third parameter.

[0146] Optionally, the SPS's RRC configuration information or activation indication also indicates the performance metrics used when monitoring the performance of each of the M AI functions. Examples of performance metrics for each AI function are provided above and will not be repeated here.

[0147] Optionally, the SPS RRC configuration information or activation indication also indicates the feedback method of the performance monitoring results. For example, the SPS RRC configuration information or activation indication indicates that after the first device receives monitoring data of M AI functions P times, it sends the performance monitoring results to the second device, where P is an integer greater than or equal to 1; or, the SPS RRC configuration information or activation indication indicates that the performance monitoring results are sent on the Xth uplink time slot after receiving the SPS deactivation indication, where X is an integer value greater than or equal to 1; or, the SPS RRC configuration information or activation indication indicates that when the value of the performance index of any AI function among the M AI functions meets the first threshold corresponding to the performance index of that AI function, the performance monitoring results are sent to the second device.

[0148] For example, the SPS RRC configuration information may also include: the radio network tempory identity (RNTI) associated with the SPS activation indication; and the SPS period, which indicates how often the monitoring data transmission operation of M AI functions is performed during the period when monitoring data is transmitted using the SPS method.

[0149] It should be noted that if a first channel does not exist, transmission can also be made between the first device and the second device through other physical channels (such as a physical shared channel). In this case, the configuration information involved in the SPS-based transmission process described above can also be applied to the relevant configurations of other physical channels.

[0150] It should be noted that the SPS RRC configuration information or activation indication may include more or less information. The example here is only for the convenience of understanding this solution. The specific information included can be determined based on the actual application scenario.

[0151] In this embodiment, the monitoring data of M AI functions are periodically sent multiple times using the SPS method, which helps to reduce the control overhead during the transmission of monitoring data.

[0152] In another implementation, the monitoring data in the first channel is sent using a dynamically triggered method. The dynamic trigger indication information indicates the resource mapping position of the AI ​​function information and the monitoring data. Alternatively, the resource mapping position of the AI ​​function information and the monitoring data can be indicated in the RRC configuration information. The RRC configuration information or the dynamic trigger indication information may include resource mapping position information, which indicates the resource mapping position of the monitoring data. The meanings of the AI ​​function information and the resource mapping position information can be found in the above description and will not be repeated here. In this embodiment, the use of a dynamically triggered method to send monitoring data for M AI functions is beneficial to improving the flexibility of the monitoring data transmission process and to enabling flexible transmission of monitoring data according to actual needs.

[0153] Optionally, the dynamically triggered indication information may also indicate a second reference signal, and / or the performance metrics used when monitoring the performance of each of the M AI functions. Optionally, the dynamically triggered indication information may also indicate whether the first device needs to send all performance monitoring results that have not yet been sent to the second device, etc. It should be noted that the dynamically triggered indication information may include more or less information; the example here is only for ease of understanding of this solution, and the specific information included can be determined based on the actual application scenario.

[0154] It should be noted that if the first channel does not exist, transmission can also be made between the first device and the second device through other physical channels (such as physical shared channels). In this case, the configuration information involved in the above-mentioned transmission process based on dynamic triggering can also be applied to the relevant configurations of other physical channels.

[0155] For example, the dynamically triggered indication information uses a dedicated first downlink control information (DCI) format; optionally, when the first device detects the first DCI in the first time slot, it indicates that monitoring data for M AI functions will be transmitted in the nth time slot after the first time slot, where n is a non-negative integer. And / or, the dynamically triggered indication information is transmitted in a dedicated second channel, for example, the second channel can be a physical downlink monitoring control channel (PDMCCH); optionally, the monitoring period of the PDMCCH is longer than that of the physical downlink control channel (PDCCH), in other words, if the PDCCH is monitored for indication information every first time interval and the PDMCCH is monitored for indication information every second time interval, the second time interval is longer than the first time interval. And / or, the dynamically triggered indication information uses a second DCI format, which is the DCI format corresponding to the physical downlink shared channel (PDSCH) or physical uplink shared channel (PDSCH), etc. For example, the second DCI format is the DCI 1_1 format in the new radio (NR) protocol. The second DCI format has corresponding fields to indicate the transmission of monitoring data. The transmission method adopted by the dynamically triggered indication information can be determined in combination with the actual application scenario.

[0156] In this embodiment, two methods for transmitting dynamically triggered indications are provided, improving the implementation flexibility of this solution. The dynamically triggered indication information indicates that monitoring data is to be transmitted. The dynamically triggered indication information uses a dedicated first DCI format, and / or is transmitted in a dedicated second channel. This allows the first device to easily detect that the indication information indicates the transmission of monitoring data using the first DCI format and / or the second channel, reducing coupling with other indication information.

[0157] In another scenario, the monitoring data of the M AI functions are transmitted along with the data in the shared channel (SCH), which can also be understood as the monitoring data of the M AI functions being transmitted following the data in the SCH. In this case, the data in the SCH is carried in the physical shared channel (PSCH), where the physical shared channel may include a physical downlink shared channel and a physical uplink shared channel. For example, step 201 may include: the second device transmitting the monitoring data of each of the M AI functions in the PSCH and / or the first channel; correspondingly, the first device receiving the monitoring data of each of the M AI functions in the PSCH and / or the first channel. Further, in one scenario, there may be no first channel independent of the PSCH, i.e., the monitoring data of the M AI functions is transmitted in the PSCH. In another scenario, the monitoring data of the M AI functions is transmitted in the first channel, and the monitoring data in the first channel and the data in the PSCH are transmitted simultaneously. The first channel and the PSCH simultaneously occupy the radio resources of the current communication, such as the radio resources of the current downlink communication or the radio resources of the current uplink communication.

[0158] For example, the resource mapping location information of the monitoring data in the PSCH and / or the resource mapping location information of the monitoring data in the first channel are predefined or preconfigured, wherein the aforementioned predefinition refers to the predefinition by the communication protocol, for example, the preconfiguration can be preconfigured by the base station using RRC configuration information or indication information.

[0159] The resource mapping location information includes the resource mapping start position and resource mapping density. The resource mapping start position and density can be understood in conjunction with the above description; repeated parts will not be elaborated here. It should be noted that, in this case, since the monitoring data of the M AI functions are sent along with the data in the SCH, the aforementioned start position can also refer to the relative position of the monitoring data in the PSCH. For example, the aforementioned start position can include the offset of the monitoring data relative to the start position of the PSCH in at least one of the time, frequency, and spatial domains. Further, in one case, as can be seen from the above description, the monitoring data of the M AI functions can include multiple types of data, and the resource mapping location information can include the relative position of each of the aforementioned multiple types of data in the PSCH. Alternatively, in another case, the monitoring data of each AI function in the M AI functions can be regarded as a group of data, and the resource mapping location information can include the relative position of the resource mapping start position of each group of data in the PSCH, etc. The specific implementation method can be determined based on the actual application scenario.

[0160] To more intuitively understand this solution, please refer to Figure 5. Figure 5 shows two schematic diagrams of resource mapping of monitoring data for M AI functions provided in this application embodiment. Figure 5 includes sub-schematic diagram (a) and sub-schematic diagram (b). In both sub-schematic diagrams (a) and (b) of Figure 5, the monitoring data includes a second reference signal, modulation symbol, and coded code block, and the monitoring data is transmitted along with the data in the SCH. Sub-schematic diagram (a) of Figure 5 shows the example of transmitting monitoring data in the PSCH, and sub-schematic diagram (b) of Figure 5 shows the example of transmitting monitoring data in the first channel. Figure 5 shows the radio resources occupied by the second reference signal, modulation symbol, and coded code block, respectively. Offset 1 indicates the relative position of the radio resources occupied by the second reference signal in the PSCH, offset 2 indicates the relative position of the radio resources occupied by the coded code block in the PSCH, and offset 3 indicates the relative position of the radio resources occupied by the modulation symbol in the PSCH. It should be understood that the examples in Figure 5 are only for the convenience of understanding this solution and are not intended to limit this solution.

[0161] In this embodiment, monitoring data can be transmitted along with data in the SCH. In other words, monitoring data can utilize the available communication resources of the SCH. Since the amount of monitoring data is small, its impact on SCH transmission is minimal. Compared to transmitting monitoring data through a separate first channel, this method reuses the available communication resources of the SCH, eliminating the need for independent scheduling and resource allocation for the first channel, thus reducing the overhead of resource allocation for the first channel. Furthermore, it reduces the need for terminal devices to receive monitoring data independently, thereby shutting down radio frequency and other receiving modules when there is no SCH data transmission, reducing the power consumption of the terminal device.

[0162] Whether the monitoring data of M AI functions is transmitted using an independent first channel, or the monitoring data of M AI functions is transmitted along with data in a shared channel, resource mapping conflicts may occur when performing resource mapping on the monitoring data of M AI functions. A resource mapping conflict means that different data are mapped to the same RE. Optionally, if a resource mapping conflict occurs between the monitoring data of M AI functions and the third reference signal, the resource mapping priority of the monitoring data of M AI functions is lower than the resource mapping priority of the third reference signal. In other words, when a resource mapping conflict occurs between the monitoring data of M AI functions and the third reference signal, the third reference signal is mapped to the location of the conflicting RE. For example, the third reference signal includes the reference signal in the physical broadcasting channel (PBCH) and the reference signal in the PSCH.

[0163] When the second device performs resource mapping on the monitoring data of M AI functions, if a resource mapping conflict occurs, the number of REs available for the monitoring data of the M AI functions will be reduced because the third reference signal is mapped to the conflicting RE position. For example, the second device can perform puncturing or rate matching on each type of data included in the monitoring data of the M AI functions. Punching refers to covering the monitoring data with the third reference signal, resulting in missing monitoring data at the RE position where the resource mapping conflict occurs; rate matching refers to knowing in advance that certain RE positions will be occupied by the third reference signal, and therefore not using the aforementioned RE positions when performing resource mapping on the monitoring data.

[0164] For example, when the modulation symbols in the monitoring data of M AI functions are resource-mapped and a resource mapping conflict occurs with the third reference signal, the data at the conflicting position can be punctured; as another example, when the coded code blocks in the monitoring data of M AI functions are resource-mapped and a resource mapping conflict occurs with the third reference signal, the data at the conflicting position can be rate-matched; as yet another example, when the second reference signal and the third reference signal in the monitoring data of M AI functions have a resource mapping conflict, the signal at the conflicting position can be punctured, and so on. The specific situation can be determined based on the actual application scenario.

[0165] In this embodiment, the resource mapping priority of the monitoring data is lower than that of the resource mapping priority of the third reference signal. The third reference signal includes the reference signal in the PBCH and the reference signal in the PSCH. Since the resource mapping priority of the monitoring channel is lower, the transmission of the reference signal that needs to be transmitted normally is guaranteed, thus avoiding the impact of the performance monitoring of the AI ​​function on the normal data transmission in the communication network.

[0166] 202. The first device determines the performance monitoring results. The performance monitoring results include the performance information of N AI functions out of M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function. M is an integer greater than 1, and N is an integer greater than or equal to 1 and N is less than or equal to M.

[0167] For example, the performance monitoring results of the AI ​​function may include the values ​​of the performance indicators of the AI ​​function, or the results of whether the values ​​of the performance indicators of the AI ​​function meet the threshold.

[0168] For example, the performance monitoring results include performance information for N AI functions out of M AI functions. The performance information for each of the N AI functions may include the value of the performance metric monitored for each AI function, or it may be the result of whether the value of the performance metric monitored for each AI function meets a third threshold. It should be noted that the third threshold here is a general term, meaning that not all AI functions correspond to the same third threshold, but rather that each AI function has its own corresponding third threshold.

[0169] In one implementation, after receiving monitoring data from each of the M AI functions, the first device obtains the input of each AI function from the monitoring data of each of the M AI functions, processes the input using each of the M AI functions to obtain the output of each of the M AI functions, and obtains the performance index value of each of the M AI functions based on the output of each of the M AI functions and the reference value corresponding to the output of each AI function, thereby determining the performance monitoring result.

[0170] For example, in this implementation, the N AI functions can include each of the M AI functions, or the N AI functions can include N AI functions whose performance metric values ​​meet or do not meet the third threshold. For example, an AI function whose performance metric value meets the third threshold represents a poorly performing AI function, or vice versa. In the following description, we will use the example of an AI function whose performance metric value meets the third threshold representing a poorly performing AI function. That is, if the performance monitoring metric of an AI function meets the third threshold, it means that the performance of the corresponding AI function does not meet the requirements.

[0171] In another implementation, after receiving monitoring data from each of the M AI functions, the first device can, according to the dependencies between the M AI functions, first determine the performance index value of the upstream target AI function. If the performance index value of the target AI function indicates that it meets a third threshold, performance monitoring of the downstream AI functions can be stopped. If the performance index value of the target AI function indicates that it does not meet the third threshold, the performance index value of the AI ​​function that depends on the target AI function is then determined, and so on, to determine the performance monitoring result. For example, the target AI function is represented by the upstream AI function among the M1 AI functions. This can be understood as follows: other AI functions among the M1 AI functions depend on the target AI function, but the target AI function does not depend on any of the M1 AI functions. In other words, the input of other AI functions among the M1 AI functions is obtained based on the output of the target AI function, but the input of the target AI function is not obtained based on the output of any of the M1 AI functions. Where M is greater than or equal to M1, in more detail, M AI functions include M1 AI functions, M1 AI functions can include all of the AI ​​functions in M ​​AI functions, or M1 AI functions can also include some of the AI ​​functions in M ​​AI functions.

[0172] For example, if M AI functions include AI function 1, AI function 2, and AI function 3, where AI function 1 is channel estimation, AI function 2 is demodulation, and AI function 3 is channel decoding, the input of AI function 2 is determined based on the output of AI function 1, and the input of AI function 3 is obtained based on the input of AI function 2. Taking the example that an AI function's performance is considered poor when its performance index meets a third threshold, when the second device monitors the performance of the three AI functions, it can first determine the performance index of AI function 1. If, based on the performance index of AI function 1, AI function 1 meets the third threshold 1, performance monitoring of AI function 2 and AI function 3 can be stopped. If, based on the performance index of AI function 1, the target AI function does not meet the third threshold 1, then the performance index of AI function 2 is determined. If, based on the performance index of AI function 2, AI function 2 meets the third threshold 2, performance monitoring of AI function 3 can be stopped. If, based on the performance index of AI function 2, the target AI function does not meet the third threshold 2, then the performance index of AI function 3 is determined. It should be understood that this example is only for ease of understanding of this scheme and is not intended to limit this scheme.

[0173] For example, in this implementation, the N AI functions may include each of the M AI functions that has performed performance monitoring, or the N AI functions may include the N AI functions whose performance metric values ​​satisfy a third threshold among the M AI functions.

[0174] Optionally, if the monitoring data of the M AI functions includes a second reference signal, after receiving the monitoring data of each of the M AI functions, the first device can also determine the number of REs occupied by the second reference signal. If the number of REs occupied by the second reference signal is less than a second threshold, the reference value of the channel estimation determined based on the second reference signal is invalid. In this embodiment, since the number of REs occupied by the second reference signal is often known if no puncturing or rate matching is performed on the second reference signal during resource mapping, when the second reference signal is less than the second threshold, it means that some data in the second reference signal was missing when the second device performed resource mapping. Therefore, the reference value of the channel estimation determined based on the second reference signal is inaccurate, and the first device cannot obtain accurate channel information. In this case, the reference value of the channel estimation determined based on the second reference signal is considered invalid, and thus, a valid performance index value cannot be obtained based on the invalid channel estimation reference value. This avoids erroneous monitoring of the performance of AI functions and helps improve the accuracy of the performance monitoring process of AI functions.

[0175] For example, if the reference value for channel estimation determined based on the second reference signal is invalid, then the performance metric value of the channel estimation AI function is invalid, and the performance metric value of the demodulation AI function is also invalid. Optionally, if the first device cannot obtain a valid performance metric value, the performance monitoring result may include a predefined invalid value.

[0176] 203. The first device sends the performance monitoring results to the second device.

[0177] Step 203 is optional. If step 203 is not executed, the first device can determine the performance of the AI ​​function on the first device based on the performance monitoring results after executing step 202. Optionally, after determining the performance of the AI ​​function on the first device, it can further determine whether to use the corresponding AI function on the first device. Optionally, the first device can send the decision on using the AI ​​function to the second device.

[0178] If step 203 is executed, for example, step 203 may include: after receiving monitoring data for P times of M AI functions, the first device sends performance monitoring results to the second device, where P is an integer greater than or equal to 1, and P is a preset value or a value indicated or configured by the second device. Optionally, the P value corresponding to each AI function may be the same or different.

[0179] Alternatively, after receiving the instruction information from the second device, the first device sends the performance monitoring results to the second device; for example, the performance monitoring results are sent to the second device in X time slots after receiving the instruction information, where X is a non-negative integer.

[0180] Alternatively, when the performance metric of any one of the M AI functions meets a first threshold, the first device sends a performance monitoring result to the second device. Here, AI functions whose performance metric meets the first threshold represent poor-performing AI functions. The third threshold can be the same as or different from the first threshold; the specific implementation can be determined based on the actual application scenario. In another scenario, AI functions whose performance metric meets the first threshold represent good-performing AI functions.

[0181] Alternatively, if the performance metric of any of the M AI functions fails to meet the first threshold, the first device sends a performance monitoring result to the second device. Here, AI functions whose performance metric meets the first threshold represent poor-performing AI functions. The third threshold can be the same as or different from the first threshold; the specific implementation can be determined based on the actual application scenario. In another scenario, AI functions whose performance metric meets the first threshold represent high-performing AI functions.

[0182] In this embodiment, performance monitoring results can also be sent to a second device, so that the second device can also know the performance status of N AI functions. This is beneficial for the second device to understand the true performance status of N AI functions in a timely manner, and also beneficial for the second device to identify AI functions with performance problems in a timely manner. In addition, multiple implementation methods for providing feedback on performance monitoring results are provided, which greatly improves the implementation flexibility of this solution and is conducive to selecting appropriate feedback methods based on actual needs.

[0183] When processing data, the first device may need to sequentially implement different functions through M AI functions, including a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function. For example, in an AI receiver, the input of the demodulation AI function includes channel information output by the channel estimation AI function; similarly, the input of the channel decoding AI function includes prediction data output by the demodulation AI function. The performance of a particular AI function may be affected by the performance of upstream AI functions, making it difficult to distinguish the independent performance of each AI function during performance monitoring. In this embodiment, the first device can receive monitoring data for each of the M AI functions. Each AI function has its own independent monitoring data. Therefore, after receiving the monitoring data for each AI function, the first device can obtain the performance index value of each AI function based on the monitoring data, facilitating a quick determination of whether the performance of each AI function is normal. For example, by calculating the performance index value of each AI function, the first device can quickly determine the impact of upstream AI functions on the performance of downstream AI functions, thus helping to quickly determine the true performance of each AI function. Furthermore, by simultaneously sending monitoring data from M AI functions, the latency and power consumption of the first device in acquiring the monitoring data can be reduced.

[0184] To more intuitively understand this solution, several specific embodiments are provided below. Please refer to Figure 6 first. Figure 6 is another flowchart illustrating the data processing method provided in this application embodiment. Figure 6 uses the first device as the terminal device, the second device as the base station, and M AI functions including channel estimation and demodulation. The monitoring data of the M AI functions in the first channel are transmitted along with the data in the downlink shared channel (Downlink-SCH, DL-SCH), and the data in the DL-SCH is carried on the PDSCH. Taking this as an example, the data processing method provided in this application may include:

[0185] 601. The base station sends the first configuration information to the terminal device.

[0186] For example, the first configuration information includes first RRC configuration information, which is the information sent by the base station when configuring the first channel with RRC. The first RRC configuration information may include a first identifier, AI function information, and resource mapping location information when the monitoring data of M AI functions are transmitted along with the data in the PDSCH. Optionally, the first RRC configuration information may also indicate the configuration information of the second reference signal, a predefined first formula, a predefined type of the first parameter, the value of the first parameter, and the performance index and performance index threshold of at least one of the M AI functions. The specific meanings of the aforementioned terms can be found in the description in the above embodiments, and will not be elaborated here.

[0187] 602. The base station sends the second RRC configuration information to the terminal device.

[0188] The second RRC configuration information is the information sent by the base station when configuring the PDSCH with RRC. The second RRC configuration information includes a second identifier and indicates the association between the second identifier and the first identifier, so as to indicate that the monitoring data of M AI functions in the first channel are along with the data in the PDSCH.

[0189] It should be noted that step 602 is an optional step.

[0190] 603. The base station sends monitoring data for each of the M AI functions to the terminal device. The monitoring data in the first channel and the data in the PDSCH are sent along with the channel. The monitoring data for each AI function is used to determine the value of the performance index of each AI function.

[0191] For example, when the base station sends data to the terminal device through the PDSCH, it can perform resource mapping on the monitoring data of the M AI functions in the first channel to embed the monitoring data of the M AI functions in the first channel into the resources shared with the PDSCH. Thus, when the base station sends data to the terminal device through the PDSCH, it can simultaneously send the monitoring data of the M AI functions to the terminal device. Correspondingly, the terminal device can receive the monitoring data of each of the M AI functions. The information types contained in the monitoring data of the M AI functions can be referred to the description in the above embodiments, and will not be described in detail here.

[0192] Optionally, when the base station performs resource mapping on the monitoring data of M AI functions in the first channel, if the monitoring data of the M AI functions conflicts with the third reference signal, it can perform puncturing and / or rate matching on the monitoring data of the M AI functions. The resource mapping conflict resolution method adopted for each type of data in the monitoring data of the M AI functions can be referred to the description in the above embodiments, and will not be described in detail here.

[0193] Optionally, the DCI corresponding to the PDSCH can indicate whether the monitoring data in the first channel was transmitted along with the PDSCH during each PDSCH transmission. For example, one bit can be used in the DCI of the PDSCH to indicate whether the monitoring data in the first channel was transmitted along with the PDSCH during the PDSCH transmission.

[0194] 604. The terminal device determines the performance monitoring results based on the monitoring data of each of the M AI functions. The performance monitoring results include the performance information of N AI functions out of the M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function.

[0195] The specific implementation of step 604 and the specific meanings of the terms in step 604 can be found in the description of step 202 in the embodiment corresponding to Figure 2 above, and will not be repeated here. For example, the monitoring data for each of the M AI functions may include one or more of a first reference signal, a second reference signal, and a modulation symbol. For instance, the terminal device determines the performance monitoring results of M=2 AI functions, namely, a channel estimation AI function and a demodulation AI function. The terminal device performs channel estimation based on the second reference signal to obtain a reference value for channel estimation. When monitoring the channel estimation AI function, the reference value output by this AI function includes the aforementioned reference value for channel estimation determined based on the second reference signal. Similarly, when monitoring the demodulation AI function, this AI function can be based on the aforementioned reference value for channel estimation determined based on the second reference signal. For example, the aforementioned reference value for channel estimation determined based on the second reference signal can be used as input when monitoring the demodulation AI function.

[0196] Furthermore, in one implementation, the terminal device uses a parallel approach to determine the performance index values ​​of the channel estimation and demodulation AI functions based on the first reference signal and the modulation symbol, respectively. Specifically, the terminal device can obtain the output of the channel estimation AI function based on the received first reference signal (e.g., inputting the received first reference signal and / or the first reference signal into the channel estimation AI function), and obtain the performance index value of the channel estimation AI function based on the output of the channel estimation AI function and the aforementioned reference value of the channel estimation determined based on the second reference signal; and obtain the output of the demodulation AI function based on the aforementioned reference value of the channel estimation determined based on the second reference signal and the received modulation symbol (e.g., inputting the reference value of the channel estimation determined based on the second reference signal and the received modulation symbol into the demodulation AI function), determine the first codeword based on the value of the first parameter of the predefined first formula and the predefined type, and obtain the performance index value of the demodulation AI function based on the output of the demodulation AI function and the first codeword, thereby determining the performance monitoring result.

[0197] In another implementation, the terminal device can first obtain the output of the channel estimation AI function based on the received first reference signal. Based on the output of the channel estimation AI function and the reference value of the channel estimation determined based on the second reference signal, it can obtain the performance index value of the channel estimation AI function. If the performance index value of the channel estimation AI function meets a third threshold, it indicates that the performance of the channel estimation AI function is poor, and performance monitoring of the demodulation AI function can be discontinued. For example, the performance monitoring result of the demodulation AI function can be determined as "unable to determine". If the performance index value of the channel estimation AI function does not meet the third threshold, the terminal device can obtain the output of the demodulation AI function based on the reference value of the channel estimation determined based on the second reference signal and the received modulation symbols. It can then determine the first codeword based on a predefined first formula and the value of the first parameter of a predefined type. Based on the output of the demodulation AI function and the first codeword, it can obtain the performance index value of the demodulation AI function, and thus determine the performance monitoring result.

[0198] To more intuitively understand this solution, please refer to Figure 7. Figure 7 is a schematic diagram illustrating how the performance index of each AI function is obtained based on the monitoring data of each of the M AI functions provided in this application embodiment. Figure 7 uses the transmission of monitoring data of the M AI functions via a first channel and PSCH as an example. As shown in Figure 7, the monitoring data of the M AI functions is transmitted along with the data in the PDSCH. The monitoring data of the M AI functions includes a first reference signal, a second reference signal, and modulation symbols (for example, the number of REs occupied by the first reference signal in the time / frequency / spatial domain can be less than the number of REs occupied by the second reference signal in the same time / frequency / spatial domain range). The second reference signal and modulation symbols in the monitoring data of the M AI functions are transmitted through the first channel. The first reference signal in the monitoring data of the M AI functions (i.e., DMRS in Figure 7) multiplexes the DMRS in the PSCH. The DMRS in Figure 7 is an example of a first reference signal obtained using a non-AI method, such as a reference signal determined based on a Gold sequence or a ZC sequence. The modulation symbols in Figure 7 are obtained by modulating the first codeword using a non-AI method, such as using regular constellation modulation methods like quadrature phase shift keying (QPSK) or quadrature amplitude modulation (QAM). The terminal device can determine the performance index of each AI function in the channel estimation and demodulation AI functions in parallel. For example, the terminal device performs channel estimation using a non-AI method based on the second reference signal to obtain a reference value for channel estimation; based on the received first reference signal, it obtains the output of the channel estimation AI function; and calculates the NMSE (an example of a performance index) based on the output of the channel estimation AI function and the aforementioned reference value for channel estimation. The terminal device obtains the output of the demodulation AI function based on the received modulation symbols and the reference value for channel estimation; and calculates the BER (another example of a performance index) based on the output of the demodulation AI function and the first codeword. The first codeword is determined using a predefined method. It should be understood that the examples in Figure 7 are only for illustrative purposes and are not intended to limit the scope of this scheme. For example, in another possible implementation, the DMRS in Figure 7 can also be overlaid with user data in the PSCH, i.e., the use case of superimposed pilot (SIP).

[0199] 605. The terminal device sends the performance monitoring results to the base station.

[0200] For example, the terminal device can send performance monitoring results to the base station at the feedback time of the PDSCH along with the PDSCH. For instance, the performance monitoring results can be included in the uplink control information (UCI) containing the hybrid automatic repeat request acknowledge (HAQR ACK) / negative acknowledge (NACK) of the PDSCH, or in other information sent simultaneously with the UCI. To more intuitively understand this solution, please refer to Figure 8. Figure 8 is a schematic diagram of transmission monitoring data and performance monitoring results provided in an embodiment of this application. The gray rectangles in Figure 8 represent the radio resources occupied by the monitoring data of M AI functions. As shown in Figure 8, the monitoring data of the M AI functions are sent along with the data in the PDSCH. The performance monitoring results are carried in the UCI containing the HAQR ACK / NACK of the PDSCH, or in other information sent simultaneously with the UCI. It should be understood that the example in Figure 8 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0201] It should be noted that, as shown in Figure 6, which takes the performance monitoring of the channel estimation and demodulation functions in the AI ​​receiver of the terminal device as an example, the base station uses a non-AI method to determine the reference signal to obtain the first reference signal and the second reference signal in the monitoring data of M AI functions. It also uses a non-AI method to perform modulation to obtain the modulation symbols in the monitoring data of M AI functions. For example, when using a non-AI method for modulation, traditional non-AI constellation modulation methods such as regular quadrature phase shift keying (QPSK), quadrature amplitude modulation (QAM), or 64QAM can be used. For example, the reference signal obtained using a non-AI method is generated based on a pseudo-random sequence such as a Gold sequence, m-sequence, or ZC sequence. The reference signal can be a DMRS, sounding reference signal (SRS), phase-tracking reference signal (PTRS), positioning reference signal (PRS), channel state information reference signal (CSIRS), etc. This application does not exhaustively list all non-AI modulation methods, pseudo-random sequences, and non-AI reference signals; the specific methods can be determined based on the actual application scenario. In another embodiment, if the performance of the AI ​​function (channel decoding) in the AI ​​receiver of the terminal device is also monitored, the base station will also use a non-AI method for channel coding to obtain coded code blocks from the monitoring data of M AI functions. For example, when using a non-AI method for channel coding, traditional non-AI channel coding methods such as low-density parity check (LDPC) coding, polar coding, or Turbo coding can be used.

[0202] Optionally, based on the embodiment corresponding to Figure 6, performance monitoring can also be performed on the M AI functions in the AI ​​transceiver. For example, the M AI functions include channel estimation and modulation / demodulation. Compared with the embodiment shown in Figure 6, the differences may include: the first reference signal and modulation symbol in Figure 6 are determined using a non-AI method. If performance monitoring is performed on the AI ​​functions in the AI ​​transceiver, the first reference signal can be generated by the base station using an AI model or determined by AI. For example, multiple reference signal identifiers can be deployed on both the terminal device and the base station. The base station can input one of the aforementioned identifiers into the AI ​​model used to determine the reference signal to obtain the first reference signal generated by the aforementioned AI model. Another example is that the reference signal set is used as a trainable parameter. The reference signal set includes multiple reference signals. During the training phase of the AI ​​model used for channel estimation, the reference signal set is determined after joint training with the AI ​​model used for channel estimation. After the training operation of the AI ​​model used for channel estimation is completed, the reference signal set can be deployed on the terminal device. The terminal device can obtain the first reference signal from the reference signal set. In this application, the trainable parameter can also be replaced by the learnable parameter. The modulation symbol can be obtained by the base station processing the first codeword through the modulation AI function or determined by AI. In addition, the demodulation AI function in Figure 6 above can be replaced by modulation and demodulation. The specific implementation process of this implementation method will not be described in detail here. Please refer to the above description for understanding.

[0203] To understand this solution more intuitively, please refer to Figure 9. Figure 9 is another schematic diagram of obtaining the performance index value of each AI function based on the monitoring data of each AI function in the M AI functions provided in the embodiment of this application. Figure 9 can be understood in conjunction with the above description of Figure 7. Repeated parts will not be repeated. The difference is that in Figure 9, the first reference signal in the monitoring data of the M AI functions is used as an example to obtain the reference signal by using the AI ​​method. In other words, the reference signal can be determined by the AI ​​model. The AI ​​modulation symbol in Figure 9 represents the first codeword after being modulated by the AI ​​method. For example, the first codeword is input into the modulation of the AI ​​function to obtain the AI ​​modulation symbol in Figure 9. It should be understood that the example in Figure 9 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0204] Please refer to Figure 10. Figure 10 is another flowchart illustrating the data processing method provided in this application embodiment. Figure 10 uses the first device as the terminal device and the second device as the base station, taking the transmission of monitoring data of M AI functions through an independent first channel using SPS as an example. The data processing method provided in this application may include:

[0205] 1001. The base station sends the first RRC configuration information to the terminal device.

[0206] For example, the first RRC configuration information is the information sent by the base station when configuring the first channel with RRC. The first RRC configuration information may include a first identifier, AI function information, and resource mapping location information. The first RRC configuration information may also include SPS configuration information, for example, the first RRC configuration information may also include the RNTI associated with the SPS and the period of the SPS. Optionally, the first RRC configuration information may also indicate the feedback method of performance monitoring results. Optionally, the first RRC configuration information may also indicate the information of the second reference signal, the predefined first formula and the predefined type of the first parameter, and the performance index of each AI function among the M AI functions. It should be noted that the meaning of the various information included in the first RRC configuration information can be referred to the description in the above embodiments, and will not be repeated here.

[0207] 1002. The base station sends an SPS trigger indication to the terminal device.

[0208] For example, the SPS trigger indication can be a DCI. When the base station performs channel coding on the DCI carrying the SPS trigger indication, it can use the SPS-associated RNTI for scrambling. Correspondingly, the terminal device can use the SPS-associated RNTI for descrambling to obtain the SPS trigger indication.

[0209] 1003. The base station sends monitoring data of each of the M AI functions to the terminal device through the first channel. The monitoring data of each AI function is used to determine the value of the performance index of each AI function.

[0210] For example, after SPS activation, the base station periodically sends monitoring data for each of the M AI functions to the terminal device through an independent first channel. Monitoring data for each of the M AI functions may be sent multiple times within the same SPS cycle. Optionally, the same monitoring data may be sent multiple times. In other words, sending monitoring data for each of the M AI functions multiple times within an SPS cycle can be considered as multiple retransmissions of the monitoring data. Therefore, the true value corresponding to the output of each of the M AI functions may remain unchanged within an SPS cycle.

[0211] Optionally, when the terminal device determines the first codeword and / or the second codeword based on the ID of the AI ​​function and the time information of the transmission time of the monitoring data, the time slot number of the transmission time of the monitoring data can adopt the time information of the reception time indicated by the SPS trigger.

[0212] It should be noted that the specific implementation of step 1003 can be understood in conjunction with the above embodiments, and will not be described in detail here.

[0213] 1004. The terminal device determines the performance monitoring results based on the monitoring data of each of the M AI functions. The performance monitoring results include the performance information of N AI functions out of the M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function.

[0214] The specific implementation of step 604 and the specific meanings of the terms in step 604 can be found in the descriptions in the above embodiments, and will not be elaborated here. Optionally, since the terminal device may not be able to obtain a valid performance indicator value based on the monitoring data of each of the M AI functions sent in a certain instance (e.g., the performance indicator value is invalid, or the performance indicator value cannot be determined), the terminal device may also determine a first number of monitoring data participating in the valid calculation (e.g., the number of times the monitoring data participating in the valid calculation is received). The monitoring data participating in the valid calculation represents that a valid performance indicator value has been obtained based on the monitoring data of each of the M AI functions sent in that instance.

[0215] 1005. The terminal device receives the SPS deactivation instruction from the base station.

[0216] For example, after receiving the SPS deactivation instruction from the base station, the terminal device stops receiving monitoring data.

[0217] To understand this solution more intuitively, please refer to Figure 11. Figure 11 is a schematic diagram of transmitting monitoring data of each of the M AI functions using SPS according to an embodiment of this application. In Figure 11, the monitoring data of the M AI functions is transmitted through an independent first channel. As shown in Figure 11, the base station sends an SPS trigger indication to the terminal device, indicating the start of periodic transmission of monitoring data of the M AI functions. The base station sends an SPS deactivation indication to the terminal device, indicating the end of one SPS cycle. The base station stops transmitting the monitoring data of the M AI functions to the terminal device. Figure 11 uses the transmission of monitoring data of the M AI functions 4 times in one SPS cycle as an example. It should be understood that the example in Figure 11 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0218] 1006. The terminal device sends the performance monitoring results to the base station.

[0219] For example, if the SPS RRC configuration information indicates that the first device sends performance monitoring results to the second device after receiving monitoring data for P times of M AI functions, then step 1006 may include: the terminal device sending performance monitoring results to the second device after receiving monitoring data for P times of M AI functions. And / or, if the SPS RRC configuration information indicates that performance monitoring results are sent over X time slots after receiving the SPS deactivation instruction, then step 1006 may include: the terminal device sending performance monitoring results on the uplink (UL) channel over X time slots after receiving the SPS deactivation instruction.

[0220] Optionally, the terminal device sends the performance monitoring results and the first quantity to the base station.

[0221] To understand this solution more intuitively, please refer to Figure 12. Figure 12 is a schematic diagram of sending performance monitoring results to the base station according to an embodiment of this application. As shown in Figure 12, the terminal device can send the performance monitoring results to the second device after receiving monitoring data of M AI functions 3 times, and / or can send the performance monitoring results on the UL channel of X time slots after receiving the SPS deactivation indication. It should be understood that the example in Figure 12 is only for the convenience of understanding this solution and is not intended to limit this solution.

[0222] It should be noted that the embodiments of this application do not limit the order between steps 1005 and 1006. For example, if the terminal device sends the performance monitoring result to the second device after receiving monitoring data of P times M AI functions, it is possible to execute step 1006 first and then step 1005. Or, for example, if the terminal device sends the performance monitoring result to the second device after receiving monitoring data of P times M AI functions, and the terminal device sends the performance monitoring result on the UL channel of X time slots after receiving the SPS deactivation instruction, it is possible to execute step 1006 once, then step 1005, then step 1006 again, etc. The specific situation can be determined according to the actual application scenario.

[0223] Please refer to Figure 13. Figure 13 is another flowchart illustrating the data processing method provided in this application embodiment. Figure 13 uses the first device as the terminal device and the second device as the base station, taking the dynamic triggering method to send monitoring data of M AI functions through an independent first channel as an example. The data processing method provided in this application may include:

[0224] 1301. The base station sends the first RRC configuration information to the terminal device.

[0225] For example, the first RRC configuration information is the information sent by the base station when configuring the first channel using RRC. The first RRC configuration information may include a first identifier and AI function information. Optionally, the first RRC configuration information may also indicate information about the second reference signal, a predefined first formula and a predefined type of the first parameter, and the performance index of each of the M AI functions. It should be noted that the meaning of the various information included in the first RRC configuration information can be found in the description in the above embodiments, and will not be repeated here.

[0226] 1302. The base station sends a dynamic triggering indication message to the terminal device.

[0227] The specific method for sending the dynamically triggered indication information can be found in the description of the above embodiments, and will not be repeated here. For example, the dynamically triggered indication information indicates the resource mapping location of the monitoring data for M AI functions; optionally, the dynamically triggered indication information also indicates the feedback method of the performance monitoring results. For example, the dynamically triggered indication information indicates that, optionally, when the value of the performance indicator of any one of the M AI functions meets a first threshold, the base station sends the performance monitoring result to the terminal device.

[0228] 1303. The base station sends monitoring data of each of the M AI functions to the terminal device through the first channel. The monitoring data of each AI function is used to determine the value of the performance index of each AI function.

[0229] After the base station sends the dynamic triggering indication information to the terminal device, it can send the monitoring data of each of the M AI functions to the terminal device through the first channel. The specific implementation of step 1303 can be understood in conjunction with the above embodiments, and will not be described in detail here.

[0230] 1304. The terminal device determines the performance monitoring results based on the monitoring data of each of the M AI functions. The performance monitoring results include the performance information of N AI functions out of the M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function.

[0231] 1305. When the performance index value of any one of the M AI functions meets the first threshold, the terminal device sends the performance monitoring result to the base station.

[0232] The specific implementation methods of steps 1304 and 1305, as well as the specific meanings of the terms in steps 1304 and 1305, can be found in the descriptions in the above embodiments, and will not be elaborated here.

[0233] To better understand this solution, please refer to Figure 14. Figure 14 is a schematic diagram of a method for transmitting monitoring data of each of the M AI functions using dynamic triggering, as provided in this embodiment of the application. Figure 14 shows the transmission of monitoring data for the M AI functions through an independent first channel. As shown in Figure 14, after the base station sends the DCI indicating dynamic triggering to the terminal device for the first time, it can send monitoring data for each of the M AI functions to the terminal device. The monitoring data for the M AI functions sent this time includes channel estimation and modulation monitoring data. The monitoring data for the M AI functions includes: a first reference signal, a second reference signal, and modulation symbols. Since the performance index value of any one of the M AI functions is determined to be non-compliant with the first threshold based on the monitoring data of the M AI functions sent this time, the terminal device does not send performance monitoring results to the base station. The base station sends the dynamic triggering indication information without monitoring data to the terminal device and sends user data to the base station through the PDSCH. After the base station sends the DCI indicating dynamic triggering to the terminal device for the second time, it can send monitoring data for each of the M AI functions to the terminal device. The monitoring data for the M AI functions sent this time includes monitoring data for channel estimation, modulation, and channel decoding. The monitoring data for the M AI functions includes: a first reference signal, a second reference signal, a modulation symbol, and a coded code block. Since the performance index value of channel decoding in the M AI functions is determined to meet the first threshold based on the monitoring data of the M AI functions sent this time, the terminal device sends the performance monitoring result to the base station. The performance monitoring result includes the performance index value of each of the three AI functions: channel estimation, modulation, and channel decoding. Optionally, the terminal device sends the performance monitoring result through signaling or information such as UCI, MAC-control element (MAC-CE), RRC, or UE assistance information (UAI). As shown in Figure 14, the first transmission of monitoring data is used to monitor the performance of the channel estimation and modulation AI functions. The second transmission of monitoring data is used to monitor the performance of the channel estimation, modulation, and channel decoding AI functions. The two transmissions of monitoring data can monitor the performance of different AI functions, and the resource mapping positions of the two transmissions of monitoring data can also be different. The dynamically triggered indication information can indicate which AI functions are monitored each time and the resource mapping position of each monitoring data. Optionally, the dynamically triggered indication information can also indicate that the performance monitoring result is sent when the performance index value of any AI function among the M AI functions meets the first threshold. It should be understood that the example in Figure 14 is only for the convenience of understanding this scheme and is not intended to limit this scheme.

[0234] Please refer to Figure 15. Figure 15 is another flowchart illustrating the data processing method provided in this application embodiment. Figure 15 uses a first device as the terminal device and a second device as the base station, employing a dynamic triggering method to send monitoring data of M AI functions through an independent first channel. The monitoring data of the M AI functions also includes a first codeword and a second codeword. Taking this example, the data processing method provided in this application may include:

[0235] 1501. The base station sends the first RRC configuration information to the terminal device.

[0236] 1502. The base station sends a dynamic triggering indication message to the terminal device.

[0237] 1503. The base station sends monitoring data of each of the M AI functions to the terminal device through the first channel. The monitoring data of the M AI functions includes: a first reference signal, a second reference signal, a modulation symbol, a first codeword, an encoded code block, and a second codeword.

[0238] 1504. The terminal device determines the performance monitoring results based on the monitoring data of each of the M AI functions. The performance monitoring results include the performance information of N AI functions out of the M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function.

[0239] 1505. When the performance index value of any one of the M AI functions meets the first threshold, the terminal device sends the performance monitoring result to the base station.

[0240] The specific implementation of the embodiment corresponding to Figure 15 is similar to that of the embodiment corresponding to Figure 13. The difference is that in the embodiment corresponding to Figure 13, the terminal device obtains the first codeword and the second codeword based on a predefined method. In the embodiment corresponding to Figure 15, the monitoring data of M AI functions includes the first codeword and the second codeword. The first codeword and the second codeword can be determined by the base station itself. For example, the base station determines the first codeword and the second codeword by random generation.

[0241] To more intuitively understand this solution, please refer to Figure 16. Figure 16 is another schematic diagram of the monitoring data of M AI functions provided in this application embodiment. As shown in Figure 16, the monitoring data of M AI functions includes a reference signal obtained using AI (i.e., an example of the first reference signal), a second reference signal, a first codeword, an AI modulation symbol, a second codeword, and an AI coding block. The first codeword is determined by the base station itself, the AI ​​modulation symbol is obtained by modulating the first codeword using the modulation AI function, the second codeword is determined by the base station itself, and the AI ​​coding block is obtained by channel coding the second codeword using the channel coding AI function. The terminal device determines the first and second codewords using a non-AI method. Optionally, if the terminal device cannot determine the first and / or second codewords using a non-AI method (e.g., demodulation and / or channel decoding of the first and / or second codewords fails using a non-AI method), then the performance monitoring results of the modulation / demodulation AI function and / or channel coding / decoding AI function corresponding to this monitoring data are invalid or cannot be determined. It should be understood that the examples in Figure 16 are for the convenience of understanding this solution only and are not intended to limit this solution.

[0242] It is understood that the steps or situations in different embodiments of the above multiple embodiments can be combined to provide more embodiments. In other words, the examples in the above multiple embodiments are for the convenience of understanding this solution and do not mean that the method provided by this application is limited to the above embodiments. In addition, in the specific implementation of this application, the collection, use and processing of relevant data may be involved. When the above embodiments of this application are applied to specific products or technologies, the collection, use and processing of relevant data involved need to comply with the relevant laws, regulations and standards of relevant countries and regions.

[0243] Based on the embodiments corresponding to Figures 1 to 16, in order to better implement the above-mentioned solutions of the embodiments of this application, related devices for implementing the above-mentioned solutions are also provided below. Specifically, refer to Figure 17, which is a schematic diagram of a data processing device provided in an embodiment of this application. In one implementation, the data processing device 1700 is used to implement the functions of the first device in each method embodiment corresponding to Figures 1 to 16. For example, the data processing device 1700 includes a transceiver unit 1701 and a processing unit 1702. The transceiver unit 1701 is used to receive monitoring data for each of M artificial intelligence (AI) functions, where M is an integer greater than 1, the M AI functions include a first AI function and a second AI function, the input of the second AI function is determined based on the output of the first AI function, and the monitoring data for each AI function is used to determine the value of the performance index of each AI function. The processing unit 1702 is used to determine the performance monitoring result, which includes performance information for N AI functions out of the M AI functions. The performance information for each AI function is obtained based on the value of the performance index of each AI function, where N is an integer greater than or equal to 1 and N is less than or equal to M.

[0244] Optionally, the transceiver unit 1701 is also used to send performance monitoring results to the second device.

[0245] Optionally, the transceiver unit 1701 is specifically used to: after receiving monitoring data of M AI functions P times, send performance monitoring results to the second device, where P is a value greater than or equal to 1; or, after receiving instruction information from the second device, send performance monitoring results to the second device; or, when the value of the performance index of any one of the M AI functions meets the first threshold, send performance monitoring results to the second device.

[0246] Optionally, the M AI functions include at least one of the following: channel estimation, demodulation, channel decoding, modulation / demodulation, or coding / decoding.

[0247] Optionally, the monitoring data for the M AI functions includes one or more of the following: a first reference signal used to determine the input for channel estimation; a second reference signal used to determine the reference value for channel estimation; modulation symbols; and coded codewords.

[0248] Optionally, the modulation symbol is obtained by modulating the first codeword, and the first codeword is determined according to a predefined method.

[0249] Optionally, if the AI ​​function is channel estimation, the reference value corresponding to the output of the AI ​​function is determined based on the second reference signal; or, if the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword; or, if the AI ​​function is channel decoding or encoding / decoding, the reference value corresponding to the output of the AI ​​function is the second codeword, which is obtained in a predefined manner.

[0250] Optionally, the monitoring data of the M AI functions may also include a first codeword and / or a second codeword, wherein if the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword; if the AI ​​function is channel decoding or code decoding, the reference value corresponding to the output of the AI ​​function is the second codeword.

[0251] Optionally, when the number of resource units (REs) occupied by the second reference signal is less than the second threshold, the reference value of the channel estimation determined based on the second reference signal is invalid.

[0252] Optionally, the transceiver unit 1701 is specifically used to receive monitoring data of each of the M AI functions from the first channel.

[0253] Optionally, the resource mapping location information of the monitoring data in the first channel is predefined or preconfigured, and the resource mapping location information includes the resource mapping start position and resource mapping density.

[0254] Optionally, the monitoring data in the first channel is transmitted using a semi-persistent scheduling (SPS) method, and the configuration information or activation indication of the SPS indicates the resource mapping location of the AI ​​function information and the monitoring data.

[0255] Optionally, the monitoring data in the first channel is sent in a dynamically triggered manner, and the dynamic triggering indication information indicates the resource mapping location of the AI ​​function information and the monitoring data.

[0256] Optionally, the indication information for dynamic triggering uses the first downlink control information (DCI) format, and / or the indication information for dynamic triggering is transmitted in a dedicated second channel.

[0257] Optionally, the transceiver unit 1701 is specifically used to receive monitoring data of each of the M AI functions from the physical shared channel PSCH and / or the first channel, and the monitoring data is transmitted along with the data in the shared channel SCH.

[0258] Optionally, the resource mapping priority of the monitoring data is lower than the resource mapping priority of the third reference signal, which includes the reference signal in the Physical Broadcast Channel (PBCH) and the reference signal in the Physical Shared Channel (PSCH).

[0259] In another implementation, the data processing device 1700 is used to implement the functions of the second device in the various method embodiments corresponding to Figures 1 to 16. For example, the data processing device 1700 includes a transceiver unit 1701; the transceiver unit 1701 is used to send monitoring data of each of the M AI functions to the first device, wherein the monitoring data of each AI function is used to determine the value of the performance index of each AI function, M is an integer greater than 1, the M AI functions include a first AI function and a second AI function, and the input of the second AI function is determined based on the output of the first AI function.

[0260] Optionally, the transceiver unit 1701 is further configured to receive performance monitoring results from the first device. The performance monitoring results include performance monitoring information for N AI functions out of M AI functions. The performance monitoring information for each AI function is obtained based on the value of the performance index of each AI function, where N is an integer greater than or equal to 1 and N is less than or equal to M.

[0261] Optionally, the transceiver unit 1701 is specifically used to send monitoring data of each of the M AI functions to the first device through the first channel.

[0262] Optionally, monitoring data is transmitted along with data in the shared channel (SCH).

[0263] It should be noted that the information interaction and execution process between the modules / units in the data processing device 1700 are based on the same concept as the various method embodiments corresponding to Figures 1 to 16 in this application. For details, please refer to the description in the method embodiments shown above in this application, which will not be repeated here.

[0264] Please refer to Figure 18, which is a schematic diagram of a device provided in an embodiment of this application. The device 1800 may include at least one processor 1801 and a communication interface 1802. Optionally, the device 1800 may also include at least one accelerator 1803 and a memory 1804.

[0265] Optionally, the processor 1801 implements the method in the above embodiments by reading program instructions stored in the memory 1804; or, the processor 1801 reads program instructions stored in the memory 1804 and implements the steps executed by the machine learning model in the method in the above embodiments through the accelerator 1803; or, the processor 1801 may also implement the method in the above embodiments by reading program instructions stored internally; or, the processor 1801 may also read program instructions stored internally and implement the steps executed by the machine learning model in the method in the above embodiments through the accelerator 1803.

[0266] When the processor 1801 reads the program instructions stored in the memory 1804 to implement the method in the above embodiments, the memory 1804 stores the program instructions that implement the method provided in the above embodiments of this application.

[0267] Optionally, at least one processor 1801 is one or more CPUs, either a single-core CPU or a multi-core CPU. For example, the memory 1804 includes, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, or optical memory. The memory 1804 stores program instructions for the operating system. For example, at least one accelerator 1803 may include at least one of the following: GPU, NPU, TPU, ASIC, FPGA, or other types of accelerators. After the program instructions stored in the memory 1804 are read by the at least one processor 1801, the device 1800 executes the corresponding operations in the foregoing embodiments.

[0268] For example, the communication interface 1802 can be a wired interface or a wireless interface, and the communication interface 1802 is used to perform data transmission and reception in the various method embodiments corresponding to Figures 1 to 16.

[0269] It should be understood that the communication interface 1802 has the functions of receiving and sending data. The functions of "receiving data" and "sending data" can be integrated into the same transceiver interface, or the functions of "receiving data" and "sending data" can be implemented in different interfaces, without limitation here. In other words, the communication interface 1802 may include one or more interfaces for implementing the functions of "receiving data" and "sending data".

[0270] After the processor 1801 reads the program instructions from the memory 1804, other functions that the device 1800 can perform are described in the preceding method embodiments.

[0271] Optionally, the device 1800 also includes a bus 1805, through which the processor 1801 and the communication interface 1802 are typically interconnected, but can also be interconnected in other ways.

[0272] The device 1800 provided in this application embodiment is used to execute the methods executed by the first device or the second device in the above-described method embodiments, and to achieve the corresponding beneficial effects. The specific implementation of the device 1800 shown in FIG18 can be referred to the descriptions in the foregoing method embodiments, and will not be repeated here.

[0273] This application also provides a system comprising a first device and a second device, wherein the first device performs the steps performed by the first device in the method described in the embodiments shown in Figures 1 to 16 above, and the second device performs the steps performed by the second device in the method described in the embodiments shown in Figures 1 to 16 above.

[0274] This application also provides a computer-readable storage medium storing a program for signal processing. When the program is run on a computer, it causes the computer to perform the steps executed by the first device in the method described in the embodiments shown in Figures 1 to 16, or causes the computer to perform the steps executed by the second device in the method described in the embodiments shown in Figures 1 to 16.

[0275] This application also provides a computer program product that, when run on a computer, causes the computer to perform the steps executed by the first device in the methods described in the embodiments shown in Figures 1 to 16, or causes the computer to perform the steps executed by the second device in the methods described in the embodiments shown in Figures 1 to 16.

[0276] The first or second device provided in this application embodiment can specifically be a chip, which includes a processing unit and a communication unit. The processing unit can be, for example, a processor, and the communication unit can be, for example, an input / output interface, pins, or circuits. The processing unit can execute computer execution instructions stored in the storage unit to cause the chip to execute the data processing method described in the embodiments shown in Figures 1 to 16. Optionally, the storage unit is a storage unit within the chip, such as a register or cache. The storage unit can also be a storage unit located outside the chip in the wireless access device, such as a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, such as random access memory (RAM).

[0277] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this application are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video optical disc; or it can be a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or non-volatile storage medium, or may include both types of storage media.

[0278] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the accompanying drawings of the device embodiments provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0279] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0280] In the above embodiments, the implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, in the form of a computer program product.

[0281] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A data processing method, characterized in that, The method includes: Receive monitoring data for each of M artificial intelligence (AI) functions, where M is an integer greater than 1. The M AI functions include a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function. The monitoring data for each AI function is used to determine the value of the performance index of each AI function. The performance monitoring results are determined, which include the performance information of N AI functions out of the M AI functions. The performance information of each AI function is obtained based on the value of the performance index of each AI function, where N is an integer greater than or equal to 1 and N is less than or equal to M.

2. The method according to claim 1, characterized in that, The method further includes sending the performance monitoring results to a second device.

3. The method according to claim 2, characterized in that, Sending the performance monitoring results to the second device includes: After receiving monitoring data for the M AI functions P times, the performance monitoring result is sent to the second device, where P is a value greater than or equal to 1; or, After receiving the instruction information from the second device, the performance monitoring results are sent to the second device; or, When the performance index value of any one of the M AI functions meets the first threshold, the performance monitoring result is sent to the second device.

4. The method according to any one of claims 1 to 3, characterized in that, The M AI functions include at least one of the following: channel estimation, demodulation, channel decoding, modulation / demodulation, or coding / decoding.

5. The method according to claim 4, characterized in that, The monitoring data for the M AI functions includes one or more of the following: A first reference signal is used to determine the input for channel estimation; The second reference signal is used to determine a reference value for channel estimation; Modulation symbols; Encoding codewords.

6. The method according to claim 5, characterized in that, The modulation symbol is obtained by modulating the first codeword, which is determined according to a predefined method.

7. The method according to claim 6, characterized in that, If the AI ​​function is channel estimation, the reference value corresponding to the output of the AI ​​function is determined based on the second reference signal; or... If the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword; or... If the AI ​​function is channel decoding or code decoding, the reference value corresponding to the output of the AI ​​function is the second codeword, which is obtained in a predefined manner.

8. The method according to claim 6, characterized in that, The monitoring data of the M AI functions also includes the first codeword and / or the second codeword, wherein if the AI ​​function is demodulation or modulation / demodulation, the reference value corresponding to the output of the AI ​​function is the first codeword; if the AI ​​function is channel decoding or code decoding, the reference value corresponding to the output of the AI ​​function is the second codeword.

9. The method according to claim 5, characterized in that, When the number of resource units (REs) occupied by the second reference signal is less than the second threshold, the reference value of the channel estimation determined based on the second reference signal is invalid.

10. The method according to claim 1 or 2, characterized in that, Receiving monitoring data for each of the M AI functions includes: receiving monitoring data for each of the M AI functions from a first channel.

11. The method according to claim 10, characterized in that, The resource mapping location information of the monitoring data in the first channel is predefined or preconfigured, and the resource mapping location information includes the resource mapping start position and the resource mapping density.

12. The method according to claim 10, characterized in that, The monitoring data in the first channel is transmitted using a semi-persistent scheduling (SPS) method, and the configuration information or activation indication of the SPS indicates the resource mapping location of the AI ​​function information and the monitoring data.

13. The method according to claim 10, characterized in that, The monitoring data in the first channel is sent in a dynamically triggered manner, and the dynamic triggering indication information indicates the resource mapping location of the AI ​​function information and the monitoring data.

14. The method according to claim 13, characterized in that, The indication information for dynamic triggering uses the first downlink control information (DCI) format, and / or the indication information for dynamic triggering is transmitted in a dedicated second channel.

15. The method according to any one of claims 1 to 3, characterized in that, The receiving of monitoring data for each of the M AI functions includes: receiving monitoring data for each of the M AI functions from the Physical Shared Channel (PSCH) and / or the first channel, wherein the monitoring data is transmitted along with the data in the Shared Channel (SCH).

16. The method according to any one of claims 1 to 3, characterized in that, The resource mapping priority of the monitoring data is lower than that of the resource mapping priority of the third reference signal, which includes the reference signal in the Physical Broadcast Channel (PBCH) and the reference signal in the Physical Shared Channel (PSCH).

17. A data processing method, characterized in that, The method includes: Monitoring data for each of the M AI functions is sent to the first device. The monitoring data for each AI function is used to determine the value of the performance index of each AI function. M is an integer greater than 1. The M AI functions include a first AI function and a second AI function. The input of the second AI function is determined based on the output of the first AI function.

18. The method according to claim 17, characterized in that, The method further includes: Receive performance monitoring results from the first device. The performance monitoring results include performance monitoring information for N of the M AI functions. The performance monitoring information for each AI function is obtained based on the value of the performance index of each AI function. N is an integer greater than or equal to 1 and less than or equal to M.

19. The method according to claim 17 or 18, characterized in that, Sending monitoring data for each of the M AI functions to the first device includes: The monitoring data of each of the M AI functions is sent to the first device through the first channel.

20. The method according to claim 17 or 18, characterized in that, The monitoring data is transmitted along with the data in the shared channel SCH.

21. An apparatus, characterized in that, The apparatus includes a transceiver unit and a processing unit, the transceiver unit and the processing unit being used to perform the method as described in any one of claims 1 to 16.

22. An apparatus, characterized in that, The apparatus includes a transceiver unit for performing the method as described in any one of claims 17 to 20.

23. A device, characterized in that, It includes a processor and a memory, wherein the processor is coupled to the memory. The memory is used to store programs; The processor is configured to execute a program in the memory, causing the device to perform the method as described in any one of claims 1 to 20.

24. A system, characterized in that, The system includes a first device and a second device, the first device performing the method as described in any one of claims 1 to 16, and the second device performing the method as described in any one of claims 17 to 20.

25. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when run on a computer, causes the computer to perform the method as described in any one of claims 1 to 20.

26. A computer program product, characterized in that, The computer program product includes a program that, when run on a computer, causes the computer to perform the method as described in any one of claims 1 to 20.

27. A chip system, characterized in that, The chip system includes a processor that performs the method as described in any one of claims 1 to 20.

Images