Performance determination method and device, equipment and storage medium

Abstract

The invention discloses a performance determination method and device, equipment and a storage medium, and relates to the field of communication. The method comprises the following steps: determining the performance of a second AI scheme based on a first AI scheme; wherein the second AI scheme is used for executing a first communication task. The performance of the second AI scheme is determined through the first AI scheme, the execution result of the second AI scheme does not need to be depended on, and the flexibility of performance determination of the communication scheme based on the AI is improved.

Description

Performance determination method, device, equipment and storage medium Technical Field

[0001] The present application relates to the field of communications, and in particular to a performance determination method, apparatus, device, and storage medium. Background Art

[0002] In related technologies, performance evaluation of a target communication solution typically relies on the execution results of the target communication solution. This means that performance evaluation results are only available after the target communication solution has completed execution. This performance evaluation method has limitations.

[0003] Summary of the Invention

[0004] The present application provides a method, apparatus, device, and storage medium for determining performance. The technical solution is as follows:

[0005] According to one aspect of the present application, a performance determination method is provided, the method being executed by a terminal device, the method comprising:

[0006] Based on the first artificial intelligence (AI) solution, performance of a second AI solution is determined; wherein the second AI solution is used to perform the first communication task.

[0007] According to one aspect of the present application, a performance determination method is provided, the method being performed by a network device, the method comprising:

[0008] Based on the first AI solution, performance of a second AI solution is determined; wherein the second AI solution is used to perform the first communication task.

[0009] According to one aspect of the present application, a performance determination device is provided, the device comprising:

[0010] The first determination module is configured to determine performance of a second AI solution based on the first AI solution; wherein the second AI solution is configured to perform the first communication task.

[0011] According to one aspect of the present application, a performance determination device is provided, the device comprising:

[0012] The second determination module is configured to determine the performance of a second AI solution based on the first AI solution; wherein the second AI solution is used to perform the first communication task.

[0013] According to one aspect of the present application, a communication device is provided, which terminal includes: a processor; a transceiver connected to the processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to load and execute the executable instructions, and the communication device implements the performance determination method as described in the above aspects.

[0014] According to one aspect of the present application, a computer-readable storage medium is provided, in which executable instructions are stored. The executable instructions are loaded and executed by a processor, and the computer-readable storage medium is used to implement the performance determination method described in the above aspect.

[0015] According to one aspect of the present application, a computer program product is provided, which includes computer instructions, the computer instructions are stored in a computer-readable storage medium, a processor of a communication device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the communication device executes to implement the performance determination method as described in the above aspect.

[0016] According to one aspect of the present application, a chip is provided, which includes a programmable logic circuit and / or program instructions, and is used to implement the performance determination method described in the above aspects when the chip is running.

[0017] According to one aspect of the present application, a computer program is provided, which includes computer instructions. A processor of a communication device executes the computer instructions, so that the communication device performs the performance determination method as described in the above aspect.

[0018] The technical solutions provided by the embodiments of the present application include at least the following beneficial effects:

[0019] The performance of the second AI solution is determined by the first AI solution without relying on the execution result of the second AI solution, thereby improving the flexibility of performance determination of AI-based communication solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] FIG1 shows a schematic diagram of a communication solution in the related art;

[0022] FIG2 shows a schematic diagram of a communication solution in the related art;

[0023] FIG3 shows a schematic diagram of a communication solution in the related art;

[0024] FIG4 shows a schematic diagram of a communication solution in the related art;

[0025] FIG5 shows a schematic diagram of a communication system provided by some exemplary embodiments of the present application;

[0026] FIG6 is a schematic flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0027] FIG7 shows a flow chart of a performance determination method provided by some exemplary embodiments of the present application;

[0028] FIG8 is a flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0029] FIG9 is a schematic flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0030] FIG10 is a schematic flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0031] FIG11 is a schematic flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0032] FIG12 is a schematic flow chart showing a performance determination method provided by some exemplary embodiments of the present application;

[0033] FIG13 shows a structural block diagram of a performance determination device provided by some exemplary embodiments of the present application;

[0034] FIG14 shows a structural block diagram of a performance determination device provided by some exemplary embodiments of the present application;

[0035] FIG15 shows a schematic structural diagram of a communication device provided by some exemplary embodiments of the present application. DETAILED DESCRIPTION

[0036] To make the objectives, technical solutions, and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the accompanying drawings. Exemplary embodiments will be described in detail herein, with examples shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of devices and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0037] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0038] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining."

[0039] First, the relevant technologies involved in the embodiments of this application are introduced:

[0040] Artificial Intelligence (AI):

[0041] AI is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also studies the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0042] AI technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. AI foundational technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning (ML) / deep learning.

[0043] AI-based wireless communication solutions:

[0044] Currently, AI-based solutions are increasingly being used in wireless communication systems. For example, AI-based implementations include: Channel-State Information (CSI) feedback, as shown in Figure 1, which uses an AI encoder and an AI decoder to achieve AI-based CSI compression and feedback; channel estimation, as shown in Figure 2, which uses an AI channel estimator to achieve high-performance estimation of a given channel, thereby providing the required CSI for subsequent coherent demodulation; positioning, as shown in Figure 3, which uses an AI-based positioning algorithm to rely on positioning channel information to obtain high-precision positioning results, where positioning channel information refers to channel information used for positioning; and beam management, as shown in Figure 4, which uses an AI-based beam management algorithm to obtain preferred or more refined beam information based on known beam information, or to obtain predictions of beam information for future moments.

[0045] Performance evaluation of AI / ML solutions:

[0046] In the related art, the performance evaluation of AI / ML solutions is mainly based on the inference performance of AI / ML solutions on test data sets. For example, for CSI compression and recovery solutions, the CSI recovery accuracy obtained by a specific AI / ML solution under specific compression feedback bit conditions is used as the performance evaluation indicator of the solution. For example, the difference between the ideal CSI information, or the CSI information to be compressed, and the CSI information obtained through compression and recovery is used as the CSI recovery accuracy. For CSI prediction solutions, similar to CSI compression and recovery solutions, the obtained CSI prediction accuracy can be used as the performance evaluation indicator of the solution. For example, the difference between the ideal CSI information, or the target CSI information, and the CSI information obtained through prediction is used as the CSI prediction accuracy.

[0047] However, such performance evaluation methods rely heavily on test datasets, labeled data, target results, and system execution efficiency. For example, for CSI compression and recovery schemes, performance evaluation can use the difference between the achievable CSI recovery accuracy (derived by the difference between the CSI recovery results and the ideal CSI recovery results on the test dataset) and the target recovery accuracy (the target recovery accuracy corresponds to the CSI recovery accuracy of the comparison scheme; if the CSI recovery accuracy is lower than the target recovery accuracy, the AI-based CSI recovery scheme is not necessary) as the performance evaluation metric for the scheme under specific compression feedback bit conditions. Therefore, in addition to knowing the CSI recovery results of the currently running scheme, performance evaluation requires a test dataset for performance evaluation, or an ideal or target CSI recovery reference result, and an ideal or target CSI recovery accuracy. Furthermore, it should be noted that these ideal or target CSI recovery reference results and accuracy vary across samples and scenarios. Relying on these ideal or target CSI recovery reference results and CSI recovery accuracy for determination results can lead to significant computational, resource, and transmission overhead.

[0048] Furthermore, the performance evaluation of AI-based wireless communication solutions in related technologies is often performed post-hoc. For example, many performance evaluations of AI-based wireless communication solutions rely on the gap between the current calculated and inferred results and the ideal, target results. This means that performance evaluations must be conducted after the AI-based wireless communication solution has been in operation for a period of time, and then the results of the operation must be used to determine whether the performance is poor. Performance evaluations cannot be completed before the AI-based wireless communication solution is used.

[0049] Based on the above problems, this application proposes a performance determination method to support performance evaluation of AI-based communication solutions.

[0050] Figure 5 shows a schematic diagram of a communication system provided by some exemplary embodiments of the present application. The communication devices included in the communication system are a network device 510 and a terminal device 520, or a terminal device 520 and a terminal device 530, which are not limited in the present application.

[0051] The network device 510 in the present application is a device with wireless transceiver functions, and the network device 510 includes but is not limited to: Evolved Node B (eNB), Radio Network Controller (RNC), Node B (NB), Base Station Controller (BSC), Base Transceiver Station (BTS), Home Base Station (e.g., Home Evolved Node B, or Home Node B, HNB), Baseband Unit (BBU), Access Point (AP) in Wireless Fidelity (Wi-Fi) system, Wireless Relay Node, Wireless Backhaul Node, Transmission Point (TP) or Transmission and Reception Point (TRP), etc. It can also be the Next Generation Node B (NGNB) in the fifth generation (5G) mobile communication system. B, gNB) or transmission point (TRP or TP), or one or a group of (including multiple antenna panels) antenna panels of a base station in a 5G system, or a network node constituting a gNB or transmission point, such as a baseband unit (BBU) or a distributed unit (DU), or a base station in a Beyond Fifth Generation (B5G) or a sixth generation (6G) mobile communication system, or a core network (CN), fronthaul, backhaul, radio access network (RAN), network slicing, etc. The network device may be a device including one or more of a centralized unit (CU) node, a DU node, and an active antenna unit (AAU) node. In addition, the CU can be divided into a network device in the RAN, or the CU can be divided into a network device in the CN, and this application does not limit this.

[0052] The terminal device 520 and the terminal device 530 in this application are devices with wireless transceiver functions, or are called user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, and user device. The terminals include, but are not limited to, handheld devices, wearable devices, vehicle-mounted devices, and Internet of Things devices, such as mobile phones, tablet computers, e-book readers, laptop computers, desktop computers, televisions, game consoles, mobile Internet devices (MIDs), augmented reality (AR) terminals, virtual reality (VR) terminals, and mixed reality (MR) terminals, wearable devices, handles, electronic tags, controllers, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical care, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, wireless terminals in remote medical surgery, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loops (WLANs), and wireless terminals in industrial control. Loop (WLL) stations, personal digital assistants (PDA), TV set-top boxes (STB), customer premises equipment (CPE), etc.

[0053] The network device 510 and the terminal device 520 communicate with each other via some air interface technology, such as a Uu interface.

[0054] Exemplarily, there are two communication scenarios between the network device 510 and the terminal device 520: an uplink communication scenario and a downlink communication scenario. Uplink communication refers to sending signals to the network device 510; downlink communication refers to sending signals to the terminal device 520.

[0055] The terminal device 520 and the terminal device 530 communicate with each other via some air interface technology, such as a Uu interface.

[0056] In some embodiments, there are two communication scenarios between terminal device 520 and terminal device 530: a first sideline communication scenario and a second sideline communication scenario. The first sideline communication refers to sending signals to terminal device 530, and the second sideline communication refers to sending signals to terminal device 520.

[0057] Terminal device 520 and terminal device 530 are both within the network coverage and located in the same cell, or terminal device 520 and terminal device 530 are both within the network coverage but located in different cells, or terminal device 520 is within the network coverage but terminal device 530 is outside the network coverage.

[0058] In some embodiments, the network device 510 is an access point (AP) or a station (STA). In some scenarios, an AP can be referred to as an AP STA, meaning that, in a sense, an AP is also a type of STA. In some scenarios, a STA is also referred to as a non-AP STA.

[0059] In some embodiments, the network device 510 is an AP or a STA.

[0060] An AP acts as a bridge between a wired network and a wireless network. Its primary function is to connect wireless network clients together and then connect the wireless network to the Ethernet. An AP device can be a terminal device or a network device equipped with a Wi-Fi chip.

[0061] It should be understood that the role of STA in the communication system is not absolute. For example, in some scenarios, when a mobile phone is connected to a router, the mobile phone is a non-AP STA. When the mobile phone serves as a hotspot for other mobile phones, the mobile phone plays the role of AP.

[0062] In some embodiments, a non-AP STA may support the 802.11be standard. A non-AP STA may also support various current and future 802.11 family WLAN standards, such as 802.11ax, 802.11ac, 802.11n, 802.11g, 802.11b, and 802.11a. A non-AP STA may also be used in a network environment that supports a next-generation WLAN system. The next-generation WLAN system is a WLAN system that has evolved from the 802.11ax system and is backward compatible with the 802.11ax system. The next-generation Wi-Fi communication is any new generation of Wi-Fi communication after Wi-Fi 7 based on the IEEE 802.11be specification, such as Ultra High Reliability (UHR) communication. For example, a non-AP STA is a UHR STA.

[0063] In some embodiments, the AP may be a device supporting the 802.11be standard. The AP may also be a device supporting various current and future 802.11 family WLAN standards, such as 802.11ax, 802.11ac, 802.11n, 802.11g, 802.11b, and 802.11a. The AP may also be used in a network environment supporting a next-generation WLAN system, which is a WLAN system evolved from the 802.11ax system and can meet backward compatibility with the 802.11ax system. Next-generation Wi-Fi communication refers to any new generation of Wi-Fi communication after Wi-Fi 7 based on the IEEE 802.11be specification, such as UHR communication.

[0064] The technical solutions provided in the embodiments of the present application can be applied to various communication systems, such as: Global System of Mobile communication (GSM) system, Orthogonal Frequency Division Multiplexing (OFDM) system, Code Division Multiple Access (CDMA) system, Wideband Code Division Multiple Access (WCDMA) system, General Packet Radio Service (GPRS), Long Term Evolution (LTE) system, LTE Frequency Division Duplex (FDD) system, LTE Time Division Duplex (TDD) system, Advanced Long Term Evolution (LTE-A) system, Universal Mobile Telecommunication System (UMTS), Worldwide Interoperability for Microwave Access (WiMAX) communication system, 5G mobile communication system, New Radio (NR) system, NR system evolution system, LTE on unlicensed spectrum (LTE-based access to unlicensed spectrum). spectrum, LTE-U) system, NR (NR-based access to unlicensed spectrum, NR-U) system on unlicensed spectrum, terrestrial communication network (Terrestrial Networks, TN) system, non-terrestrial communication network (Non-Terrestrial Networks, NTN) system, wireless local area network (Wireless Local Area Networks, WLAN), wireless fidelity (Wireless Fidelity, Wi-Fi), cellular Internet of Things system, cellular passive Internet of Things system, and may also be applicable to subsequent evolution systems of 5G NR system, and may also be applicable to B5G, 6G and subsequent evolution systems. In some embodiments of the present application, "NR" may also be referred to as 5G NR system or 5G system.The 5G mobile communication system may include non-standalone (NSA) and / or standalone (SA) networking. Furthermore, terminal devices 520 and 530 may also include smart printers, train detectors, gas station sensors, etc., with primary functions including at least one of collecting data, receiving control information and downlink data from network device 510, and transmitting uplink data to network device 510.

[0065] The terminal device 530 and the network device 510 may communicate directly or indirectly.

[0066] The technical solutions provided in the embodiments of the present application can also be applied to machine type communication (MTC), long term evolution technology for machine-to-machine communication (LTE-M), device-to-device (D2D) network, machine-to-machine (M2M) network, Internet of Things (IoT) network or other networks. Among them, the IoT network can include, for example, the Internet of Vehicles. Among them, the communication mode in the Internet of Vehicles system is collectively referred to as vehicle to other devices (Vehicle to X, V2X, X can represent anything), for example, the V2X can include: vehicle to vehicle (V2V) communication, vehicle to infrastructure (V2I) communication, vehicle to pedestrian communication (V2P) or vehicle to network (V2N) communication, etc.

[0067] The network device 510, the terminal device 520, and the terminal device 530 can each be configured with multiple antennas, which may include at least one transmit antenna for sending signals and at least one receive antenna for receiving signals. In addition, each communication device also includes a transmitter chain and a receiver chain. Those skilled in the art will understand that they may include multiple components related to signal transmission and reception (such as processors, modulators, multiplexers, demodulators, demultiplexers, or antennas). Therefore, the network device 510 and the terminal device 520, and the terminal device 520 and the terminal device 530 can communicate using multi-antenna technology.

[0068] The communication system provided in the present application can be applied to but not limited to at least one of the following communication scenarios: uplink communication scenario, downlink communication scenario, and sidelink communication scenario.

[0069] FIG6 is a flow chart illustrating a performance determination method provided by some exemplary embodiments of the present application. The method is executed by the terminal device shown in FIG5 . The method includes at least some of the following steps:

[0070] Step 610: Determine the performance of the second AI solution based on the first AI solution.

[0071] AI solutions refer to AI / ML-based, communication-related solutions, such as communication-related AI models and communication-related AI datasets.

[0072] Communication-related refers to solving communication-related problems, improving communication-related performance, or performing communication-related tasks.

[0073] The second AI solution is used to perform a first communication task. The first communication task refers to a communication-related task, such as at least one of CSI compression, CSI recovery, CSI prediction, beam selection, beam prediction, and positioning.

[0074] Performance refers to performance related to communication, such as whether the second AI solution is used to perform the first communication task, and / or the difference between the second AI solution and the comparison solution, and / or the credibility of the second AI solution.

[0075] Determining the performance of the second AI solution based on the first AI solution can be understood as using the first AI solution to determine the performance of the second AI solution, or as obtaining the performance of the second AI solution after executing the first AI solution. Optionally, the performance of the second AI solution is determined based on the first AI solution before using the second AI solution to perform the first communication task.

[0076] In summary, the method provided in this application determines the performance of the second AI solution through the first AI solution without relying on the execution result of the second AI solution, thereby improving the flexibility of performance determination of AI-based communication solutions.

[0077] FIG7 is a flow chart showing a performance determination method provided by some exemplary embodiments of the present application. The method is executed by the terminal device shown in FIG5 . The method includes at least some of the following steps:

[0078] Step 710: Obtain a second AI solution;

[0079] The second AI solution is used to perform a first communication task. The first communication task refers to a communication-related task, such as at least one of CSI compression, CSI recovery, CSI prediction, beam selection, beam prediction, and positioning.

[0080] In some embodiments, step 710 may be implemented as “receiving a second AI solution.” The second AI solution is trained by a network device, and the network device sends the trained second AI solution to the terminal device, which receives the second AI solution sent by the network device.

[0081] In this application, training, constructing and generating have the same meaning.

[0082] In some embodiments, step 710 may be implemented as “training a second AI solution”.

[0083] In some embodiments, the second AI solution is trained by the terminal device. The terminal device receives the second data set sent by the network device, and the terminal device trains the second AI solution based on the second data set.

[0084] In some embodiments, a second AI solution is trained by the network device and the terminal device. The second AI solution includes a first federated sub-solution and a second federated sub-solution. The network device trains the first federated sub-solution. The terminal device receives a second data set sent by the network device, and the terminal device trains the second federated sub-solution based on the second data set.

[0085] Step 720: Receive a first AI solution;

[0086] In some embodiments, step 720 refers to the terminal device receiving the first AI solution sent by the network device. The first AI solution is trained by the network device, and the network device sends the trained first AI solution to the terminal device.

[0087] The network device sends the first AI solution through a Radio Resource Control (RRC) message, a Physical Downlink Shared Channel (PDSCH), or a specific channel for transmitting the AI / ML model.

[0088] The network device instructs the transmission of the first AI solution through downlink control information (DCI), a media access control layer (MAC) control element (CE), an RRC message, and a broadcast message.

[0089] In some embodiments, step 720 may be implemented as steps 721b and 723b, as shown in FIG. 8 .

[0090] Step 721b: Sending a first request message to the network device;

[0091] The terminal device sends a first request message to the network device through uplink control information (UCI) or an RRC message, where the first request message is used to request the network device to send a first AI solution.

[0092] Step 723b: Receive the first AI solution sent by the network device;

[0093] The first AI solution is trained by the network device, and the network device sends the constructed first AI solution to the terminal device.

[0094] The network device sends the first AI solution through an RRC message, or a PDSCH, or a specific channel for transmitting an AI / ML model.

[0095] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0096] In some embodiments, step 720 may be implemented as steps 721 c , 723 c , and 725 c , as shown in FIG. 9 .

[0097] Step 721c: Receive first indication information sent by the network device;

[0098] The network device sends the first indication information through at least one of DCI, MAC CE, RRC signaling, RRC reconfiguration message, system broadcast, Master Information Block (MIB), System Information Block (SIB), SIB1, etc., where the first indication information is used to indicate or notify the first AI scheme to determine the performance of the second AI scheme.

[0099] Step 723c: Sending a first request message to the network device;

[0100] The terminal device sends a first request message to the network device through a UCI or RRC message, where the first request message is used to request the network device to send a first AI solution.

[0101] Step 725c: Receive the first AI solution sent by the network device;

[0102] The first AI solution is trained by the network device, and the network device sends the constructed first AI solution to the terminal device.

[0103] The network device sends the first AI solution through an RRC message, or a PDSCH, or a specific channel for transmitting an AI / ML model.

[0104] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0105] Step 730: Receive a first data set, where the first data set is used to train a first AI solution;

[0106] In some embodiments, step 730 refers to the terminal device receiving the first data set sent by the network device.

[0107] The network device sends the data set via RRC messages, PDSCH, or a specific channel for transmitting AI / ML models.

[0108] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0109] In some embodiments, step 730 may be implemented as steps 731b and 733b, as shown in FIG. 10 .

[0110] Step 731b: Sending a second request message to the network device;

[0111] The terminal device sends a second request message to the network device through a UCI or RRC message, where the second request message is used to request the network device to send the first data set.

[0112] Step 733b: Receive a first data set sent by the network device;

[0113] The network device sends the first data set through an RRC message, a PDSCH, or a specific channel for transmitting the AI / ML model.

[0114] The network device instructs the transmission of the first data set through DCI, MAC CE, RRC message, or broadcast message.

[0115] In some embodiments, step 730 may be implemented as steps 731 c , 733 c , and 735 c , as shown in FIG. 11 .

[0116] Step 731c: Receive second indication information sent by the network device;

[0117] The network device sends the second indication information through at least one of DCI, MAC CE, RRC signaling, RRC reconfiguration message, system broadcast, MIB, SIB, SIB2, etc. The second indication information is used to indicate or notify that the first AI solution is used to determine the performance of the second AI solution, or the second indication information is used to indicate or notify that the first data set is used to determine the performance of the second AI solution.

[0118] Step 733c: Sending a second request message to the network device;

[0119] The terminal device sends a second request message to the network device through a UCI or RRC message, where the second request message is used to request the network device to send the first data set.

[0120] Step 735c: Receive a first data set sent by the network device;

[0121] The network device sends the first data set through an RRC message, a PDSCH, or a specific channel for transmitting the AI / ML model.

[0122] The network device instructs the transmission of the first data set through DCI, MAC CE, RRC message, or broadcast message.

[0123] Step 740: Training a first AI solution based on the first data set;

[0124] Step 750: Input the first input information into the first AI solution to obtain first output information;

[0125] The first input information of the first AI solution includes feature information and / or data information corresponding to the feature information;

[0126] The feature information refers to feature information corresponding to at least one of the input interface, output interface, and intermediate interface corresponding to the second AI solution. In other words, the feature information is feature information on some or all interfaces of the second AI solution. The data information refers to data information obtained after data processing of the feature information.

[0127] In some embodiments, the characteristic information includes at least one of the following: channel state information reference signal CSI-RS; time domain channel; frequency domain channel; eigenvector; CSI reporting information; precoding matrix indicator (PMI) information; beam measurement results; beam identification (ID); beam configuration scheme; beam quality prediction results; channel impulse response (CIR) information; reference signal receiving power (RSRP) information; time difference of arrival (TDOA) information; direction of arrival (DOA) information; line-of-sight (LOS) information; non-line-of-sight (NLOS) information; angle of arrival (AOA) information.

[0128] In some embodiments, data processing includes at least one of averaging, merging, distribution statistics, Fast Fourier Transform (FFT) calculation, Inverse Fast Fourier Transform (IFFT) calculation, and Singular Value Decomposition (SVD).

[0129] In some embodiments, the first output information includes at least one of the following:

[0130] Use a second AI solution;

[0131] Not using a second AI solution;

[0132] The performance of the second AI solution is better than that of the comparison solution;

[0133] The performance of the second AI solution is not better than that of the comparison solution;

[0134] The probability that the second AI solution performs better than the comparison solution;

[0135] The credibility of the second AI solution;

[0136] The comparison solution is a solution for performing the first communication task other than the second AI solution.

[0137] In some embodiments, the second AI solution has M interfaces, where M is a positive integer, such as at least one of the following: the input interface and output interface of the second AI solution trained by the network device, the input interface and output interface of the second AI solution trained by the terminal device and the network device, the input interface and output interface of the first federated sub-solution, the input interface and output interface of the second federated sub-solution, or a model intermediate layer other than the above interfaces. These interface information, or processing information of these interface information (such as at least one of averaging, merging, distribution statistics, FFT of a specific dimension, IFFT calculation, SVD decomposition, etc.), can be used as the first input information of the first AI solution. The first output information of the first AI solution is information used to determine whether to use the second AI solution to perform the first communication task, such as whether the second AI solution is available, whether the second AI solution is better than the comparison solution, the probability that the second AI solution is recommended for use, and the probability that the second AI solution is better than the comparison solution.

[0138] The following is an example:

[0139] 1. Take the first communication task of CSI compression, and / or CSI recovery, and / or CSI prediction as an example.

[0140] Then, the first input information includes at least one of the following:

[0141] Input information for the second AI scheme for CSI compression;

[0142] For example, the input information of the CSI encoder (compression model) includes the Channel State Information-Reference Signal (CSI-RS), the time domain channel, the frequency domain channel H, the eigenvector W, or the information after data processing of the CSI-RS, channel, and eigenvector (such as at least one of averaging, merging, distribution statistics, FFT calculation in a specific dimension, IFFT calculation, and SVD decomposition).

[0143] Output information of the second AI scheme for CSI compression;

[0144] For example, the output information of the CSI encoder (compression model), such as CSI reporting information, PMI information, etc., or the processed information (such as at least one of averaging, merging, distribution statistics, FFT calculation of a specific dimension, IFFT calculation, SVD decomposition, etc.).

[0145] Input information for the second AI solution for CSI recovery;

[0146] For example, the input information of the CSI decoder (recovery model), such as CSI reporting information, PMI information, etc., or the processed information (such as at least one of averaging, merging, distribution statistics, FFT or IFFT calculation in a specific dimension, SVD decomposition, etc.).

[0147] Output information of the second AI solution for CSI recovery;

[0148] For example, the output information of the CSI decoder (recovery model), such as the time domain channel, the frequency domain channel H, the eigenvector W, or the processed information of the above information, channels, and eigenvectors (such as at least one of averaging, merging, distribution statistics, FFT of a specific dimension, IFFT calculation, SVD decomposition, etc.).

[0149] Input information for the second AI solution for CSI prediction;

[0150] For example, the input information of a CSI prediction scheme (e.g., a model), such as CSI-RS, time domain channel, frequency domain channel H, eigenvector W, or information after processing of the above reference signal, channel, and eigenvector (e.g., at least one of averaging, merging, distribution statistics, FFT of a specific dimension, IFFT calculation, and SVD decomposition).

[0151] Output information of the second AI solution for CSI prediction;

[0152] For example, the output information of a CSI prediction scheme (e.g., a model), such as CSI-RS, time domain channel, frequency domain channel H, eigenvector W, or information after processing of the above reference signal, channel, and eigenvector (e.g., at least one of averaging, merging, distribution statistics, FFT in a specific dimension, IFFT calculation, and SVD decomposition).

[0153] It should be noted that the first input information may be one or more, or may be one or more, such as multiple times of information in the time domain as the first input information. For example, the CSI-RS may be periodically used as the first input information in the time domain, or the output information of multiple CSI encoders may be used as the first input information.

[0154] Then, the first output information includes at least one of the following:

[0155] The judgment results that can be used by the second AI solution;

[0156] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution can be used to perform the first communication task.

[0157] Not using a second AI solution;

[0158] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution cannot be used to perform the first communication task.

[0159] The performance of the second AI solution is better than that of the comparison solution;

[0160] For example, based on the input feature vector W, when W is used as input, it is determined that the performance of the second AI solution is better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is better than the effect of using the comparison solution to perform the first communication task.

[0161] For example, based on the input CSI reporting information S, when S is used as input, it is determined that the performance of the second AI solution is better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is better than the effect of using the comparison solution to perform the first communication task.

[0162] The comparison scheme may be CSI feedback of type 1 (TYPE1) or type 2 (TYPE2) or eTYPE2, or a CSI compression scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a CSI recovery scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0163] The performance of the second AI solution is not better than that of the comparison solution;

[0164] For example, based on the input feature vector W, when W is used as input, it is determined that the performance of the second AI solution is not better than the performance of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is not better than the effect of using the comparison solution to perform the first communication task.

[0165] For example, based on the input CSI reporting information S, when S is used as input, it is determined that the performance of the second AI solution is not better than the performance of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is not better than the effect of using the comparison solution to perform the first communication task.

[0166] The comparison scheme may be CSI feedback of type 1 (TYPE1) or type 2 (TYPE2) or eTYPE2, or a CSI compression scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a CSI recovery scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0167] The probability that the second AI solution performs better than the comparison solution;

[0168] For example, based on the input CSI reporting information S, when S is used as input, a probability P1 is determined that the performance of the second AI solution is better than that of the comparison solution, or a probability P2 is determined that the performance of the second AI solution is not better than that of the comparison solution.

[0169] The comparison scheme may be CSI feedback of type 1 (TYPE1) or type 2 (TYPE2) or eTYPE2, or a CSI compression scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a CSI recovery scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0170] The credibility of the second AI solution;

[0171] For example, based on the input feature vector W, when W is used as input, the credibility probability of the result of the second AI solution is judged to be Q.

[0172] In some embodiments, the first input signal is quantized. The first input signal refers to a signal input to the first AI scheme. Quantization refers to the process of approximating a continuous value (or a large number of possible discrete values) of a signal to a finite number (or a small number) of discrete values.

[0173] The quantification method includes at least one of the following:

[0174] Uniform quantization; non-uniform quantization; quantization based on Int4 format; quantization based on Int8 format; quantization based on floating point 16 (float16) format; quantization based on float32 format; quantization based on float64 format.

[0175] Uniform quantization can be N-bit uniform quantization within a given range, such as within [-1, 1] or [0, 1]. Uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device. Non-uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device.

[0176] 2. Take the first communication task of beam selection and / or beam prediction as an example.

[0177] Then, the first input information includes at least one of the following:

[0178] Channel quality information;

[0179] For example, CSI-RS, time domain channel, frequency domain channel H, eigenvector W, or information after data processing of the above CSI-RS, channel, and eigenvector (for example, at least one of averaging, merging, distribution statistics, FFT of a specific dimension, IFFT calculation, SVD decomposition, etc.).

[0180] Input information for the second AI scheme for beam selection;

[0181] For example, input information for a beam selection scheme (e.g., a model), such as measurement results for K beams, and / or beam IDs, and / or a beam configuration scheme, such as measurement results for four transmit beams, their beam IDs, and their angles and widths. Alternatively, the information may be processed (e.g., by performing at least one of averaging, merging, distribution statistics, FFT calculations in a specific dimension, IFFT calculations, and SVD decomposition).

[0182] Input information for the second AI solution for beam prediction;

[0183] For example, input information for a beam prediction scheme (e.g., a model) can include measurement results for K beams, and / or beam IDs, and / or a beam configuration scheme, where K is a positive integer. For example, this information includes measurement results for four transmit beams, their beam IDs, and their angles and widths. Alternatively, this information can be processed (e.g., by performing at least one of averaging, merging, distribution statistics, FFT calculations in a specific dimension, IFFT calculations, and SVD decomposition).

[0184] Output information of the second AI scheme for beam selection;

[0185] For example, output information from a beam selection scheme (e.g., a model), such as measurement results for K beams, and / or beam IDs, and / or beam configuration schemes, such as measurement results for four transmit beams, their beam IDs, and their angles and widths. Alternatively, information derived from the aforementioned beam information may be processed (e.g., by performing at least one of averaging, merging, distribution statistics, FFT calculations or IFFT calculations in a specific dimension, or SVD decomposition).

[0186] Output information of the second AI solution for beam prediction;

[0187] For example, output information from a beam prediction solution (e.g., a model), such as measurement results for L beams, and / or beam IDs, and / or beam configurations, where L is a positive integer. For example, measurement results for 64 transmit beams, beam angles, and widths, etc., or processed information from the aforementioned beam information (e.g., at least one of averaging, merging, distribution statistics, FFT calculations or IFFT calculations in a specific dimension, and SVD decomposition).

[0188] It should be noted that the first input information may be one or more, or may be one or more, such as multiple times of information in the time domain as the first input information. For example, the CSI-RS may be periodically used as the first input information in the time domain, or the output information of multiple CSI encoders may be used as the first input information.

[0189] Then, the first output information includes at least one of the following:

[0190] The judgment results that can be used by the second AI solution;

[0191] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution can be used to perform the first communication task.

[0192] Not using a second AI solution;

[0193] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution cannot be used to perform the first communication task.

[0194] The performance of the second AI solution is better than that of the comparison solution;

[0195] For example, based on the beam measurement results, it is determined that the performance of the second AI solution is better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is better than the effect of using the comparison solution to perform the first communication task.

[0196] The comparison scheme is a beam selection scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a beam prediction scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0197] The performance of the second AI solution is not better than that of the comparison solution;

[0198] For example, based on the beam measurement results, it is determined that the performance of the second AI solution is not better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is not better than the effect of using the comparison solution to perform the first communication task.

[0199] The comparison scheme is a beam selection scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a beam prediction scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0200] The probability that the second AI solution performs better than the comparison solution;

[0201] For example, based on the input historical beam measurement result S, when S is used as input, the probability P1 of judging that the performance of the second AI solution is better than that of the comparison solution, or the probability P2 of judging that the performance of the second AI solution is not better than that of the comparison solution.

[0202] The comparison scheme is a beam selection scheme (such as a model) based on an AI / ML model other than the second AI scheme, or a beam prediction scheme (such as a model) based on an AI / ML model other than the second AI scheme.

[0203] The credibility of the second AI solution;

[0204] For example, based on the input beam measurement results, the credibility probability of the result of the second AI solution is judged to be Q.

[0205] In some embodiments, the first input signal is quantized. The first input signal refers to a signal input to the first AI scheme. Quantization refers to the process of approximating a continuous value (or a large number of possible discrete values) of a signal to a finite number (or a small number) of discrete values.

[0206] The quantification method includes at least one of the following:

[0207] Uniform quantization; non-uniform quantization; quantization based on Int4 format; quantization based on Int8 format; quantization based on float16 format; quantization based on float32 format; quantization based on float64 format; quantization based on RSRP value; quantization based on RSRP interval; quantization based on Reference Signal Receiving Quality (RSRQ) value; quantization based on RSRQ interval; quantization based on Signal to Interference plus Noise Ratio (SINR) value; quantization based on SINR interval.

[0208] Uniform quantization can be N-bit uniform quantization within a given range, such as within [-1, 1] or [0, 1]. Uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device. Non-uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device.

[0209] In some embodiments, the R bit is used to indicate at least one of an RSPR value, an RSPR interval, an RSRQ value, an RSRQ interval, an SINR value, and an SINR interval with a quantization step size of k decibels (dB).

[0210] 3. Take positioning as an example for the first communication task.

[0211] Then, the first input information includes at least one of the following:

[0212] Channel quality information;

[0213] For example, CSI-RS, time domain channel, frequency domain channel H, eigenvector W, or information after data processing of the above CSI-RS, channel, and eigenvector (for example, at least one of averaging, merging, distribution statistics, FFT of a specific dimension, IFFT calculation, SVD decomposition, etc.).

[0214] Input information for the second AI solution used for positioning;

[0215] For example, the input information of the positioning solution (e.g., model) may include at least one of the CIR information, RSRP information, TDOA information, and DOA information of the positioning base station, or the processed information (e.g., at least one of averaging, merging, distribution statistics, FFT calculation in a specific dimension, IFFT calculation, SVD decomposition, etc.).

[0216] Output information of the second AI solution for positioning;

[0217] For example, the output information of an indirect positioning solution (e.g., a model), such as at least one of the TOA, TDOA, LOS / NLOS determination, and AOA information corresponding to multiple positioning nodes or network devices. Or the processed information (e.g., at least one of averaging, merging, distribution statistics, FFT or IFFT calculation in a specific dimension, SVD decomposition, etc.)

[0218] It should be noted that the first input information may be one or more pieces of information; it may also be one or more pieces of information, such as multiple pieces of information in the time domain as the first input information.

[0219] Then, the first output information includes at least one of the following:

[0220] The judgment results that can be used by the second AI solution;

[0221] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution can be used to perform the first communication task.

[0222] Not using a second AI solution;

[0223] For example, based on the input channel H or feature vector W, it is determined that when H or W is used as input, the second AI solution cannot be used to perform the first communication task.

[0224] The performance of the second AI solution is better than that of the comparison solution;

[0225] For example, based on the positioning information (CIR, RSRP, etc.) of multiple base stations, it is determined that the performance of the second AI solution is better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is better than the effect of using the comparison solution to perform the first communication task.

[0226] The comparison solution is a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model), or a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model). The comparison solution is an AI / ML-based solution or a non-AI / ML-based solution.

[0227] The performance of the second AI solution is not better than that of the comparison solution;

[0228] For example, based on the positioning information (CIR, RSRP, etc.) of multiple base stations, it is determined that the performance of the second AI solution is not better than that of the comparison solution. In other words, the effect of using the second AI solution to perform the first communication task is not better than the effect of using the comparison solution to perform the first communication task.

[0229] The comparison solution is a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model), or a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model). The comparison solution is an AI / ML-based solution or a non-AI / ML-based solution.

[0230] The probability that the second AI solution performs better than the comparison solution;

[0231] For example, based on at least one of CIR, RSRP, RSRQ, TOA, TDOA, and AOA information of at least one positioning node (or network device), a probability P1 is determined that the performance of the second AI solution is better than that of the comparison solution, or a probability P2 is determined that the performance of the second AI solution is not better than that of the comparison solution.

[0232] The comparison solution is a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model), or a positioning solution based on an AI / ML model other than the second AI solution (e.g., a model). The comparison solution is an AI / ML-based solution or a non-AI / ML-based solution.

[0233] The credibility of the second AI solution;

[0234] For example, based on the positioning measurement results, the credibility probability of the result of the second AI solution is judged to be Q.

[0235] In some embodiments, the first input signal is quantized. The first input signal refers to a signal input to the first AI scheme. Quantization refers to the process of approximating a continuous value (or a large number of possible discrete values) of a signal to a finite number (or a small number) of discrete values.

[0236] The quantification method includes at least one of the following:

[0237] Uniform quantization; non-uniform quantization; quantization based on Int4 format; quantization based on Int8 format; quantization based on float16 format; quantization based on float32 format; quantization based on float64 format; quantization based on RSRP value; quantization based on RSRP interval; quantization based on Reference Signal Receiving Quality (RSRQ) value; quantization based on RSRQ interval; quantization based on Signal to Interference plus Noise Ratio (SINR) value; quantization based on SINR interval.

[0238] Uniform quantization can be N-bit uniform quantization within a given range, such as within [-1, 1] or [0, 1]. Uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device. Non-uniform quantization schemes are predefined by the communication protocol, determined by the network device, or determined by the terminal device.

[0239] In some embodiments, the R bit is used to indicate at least one of an RSPR value, an RSPR interval, an RSRQ value, an RSRQ interval, an SINR value, and an SINR interval with a quantization step size of k decibels (dB).

[0240] In some embodiments, the first AI scheme and the second AI scheme can be combined into one scheme. The input information, partial input information, output information, partial output information, auxiliary information, and partial auxiliary information of this combined scheme can be the input information (i.e., first input information), output information (first output information), and auxiliary information (information that assists or participates in the execution of the first AI scheme) of the above-mentioned first AI scheme.

[0241] Step 760: Determine the performance of the second AI solution based on the first output information.

[0242] Steps 750 and 760 are to determine the performance of the second AI solution based on the first AI solution.

[0243] In some embodiments, before using the second AI solution to perform the first communication task, the performance of the second AI solution is determined based on the first AI solution.

[0244] The performance of AI / ML-based wireless communication solutions is directly related to the scenarios and data used to train the model. If the data to be tested is unrelated or significantly different from the data used to train the model, the reliability of the current AI / ML model, the reliability of the results, and the advantages of the results compared with the comparison solution can be directly determined from the data to be tested. In this embodiment, when the second AI solution is trained on one end (for example, by the network device, or a portion of the second AI solution is trained by the network device side), the network device side can, during the training and acquisition of the second AI solution, also obtain a first AI solution for determining the performance of the second AI solution based on the target data set and non-target data set data, and send the first AI solution to the terminal device for use, or send the corresponding training data of the first AI solution to the terminal device for use.

[0245] In summary, the method provided by this application, compared to related art performance evaluation schemes that use information such as test datasets, labeled data, target task model inference results, and usage results, allows the performance evaluation of the second AI solution to be completed using an AI / ML performance evaluation scheme without excessive reliance on this information. Furthermore, when a network device constructs a first AI solution, it can directly rely on the dataset used to construct the second AI solution and the local performance evaluation results of the second AI solution, thereby improving the flexibility and simplicity of constructing the first AI solution. Because this information is unavailable to the terminal device, transmitting the first AI solution, or the dataset corresponding to the first AI solution, over the air interface ensures the validity of the first AI solution or its dataset, while also enabling the terminal device to directly use the first AI solution or construct the first AI solution using the first dataset, enabling the terminal device to implement performance monitoring, model selection, and model updates for the second AI solution based on AI / ML.

[0246] Taking into account the need for the terminal device to quickly and effectively determine the performance of the second AI solution when the second AI solution or part of the second AI solution needs to interact in an AI-based wireless communication solution, such as when it needs to be transmitted from a network device to a terminal device. A method is provided for forming a first AI solution on the network device side and transmitting it to the terminal device through the air interface for the terminal device to monitor the performance of the second AI solution. This method can be used to determine whether the second AI solution can be used based on data and environmental characteristics before the terminal device uses the second AI solution, thereby achieving the effect of prior performance evaluation. In addition, this method also avoids the reliance on monitoring and evaluation of test data sets, label data, and target task model usage results, and can use a pre-built performance evaluation model to determine the usability of the second AI solution.

[0247] FIG12 is a flow chart showing a performance determination method provided by some exemplary embodiments of the present application. The method is executed by the network device shown in FIG5 . The method includes at least some of the following steps:

[0248] Step 1210: training a second AI solution;

[0249] The second AI solution is used to perform a first communication task. The first communication task refers to a communication-related task, such as at least one of CSI compression, CSI recovery, CSI prediction, beam selection, beam prediction, and positioning.

[0250] Step 1220: Send a second AI solution or a second data set;

[0251] In some embodiments, the second AI solution is trained by a network device, and the network device sends the second AI solution to the terminal device.

[0252] In some embodiments, a second AI solution is trained by a network device and a terminal device. The second AI solution includes a first federated sub-solution and a second federated sub-solution. The network device trains the first federated sub-solution. The network device sends a second data set to the terminal device, and the terminal device trains the second federated sub-solution based on the second data set.

[0253] Step 1230: Send the first AI solution;

[0254] In some embodiments, step 1230 refers to the network device sending a first AI solution to the terminal device. The first AI solution is trained by the network device, and the network device sends the trained first AI solution to the terminal device.

[0255] The network device sends the first AI solution through an RRC message, or a PDSCH, or a specific channel for transmitting an AI / ML model.

[0256] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0257] In some embodiments, step 1230 may be implemented as step (a) and step (b).

[0258] Step (a): receiving a first request message sent by a terminal device;

[0259] The terminal device sends a first request message to the network device through a UCI or RRC message, where the first request message is used to request the network device to send a first AI solution.

[0260] Step (b): sending the first AI solution to the terminal device;

[0261] The first AI solution is trained by the network device, and the network device sends the constructed first AI solution to the terminal device.

[0262] The network device sends the first AI solution through an RRC message, or a PDSCH, or a specific channel for transmitting an AI / ML model.

[0263] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0264] In some embodiments, step 1230 may be implemented as step (i), step (ii), and step (iii).

[0265] Step (i): sending first indication information to the terminal device;

[0266] The network device sends the first indication information through at least one of DCI, MAC CE, RRC signaling, RRC reconfiguration message, system broadcast, MIB, SIB, SIB1, etc., where the first indication information is used to indicate or notify the first AI solution to determine the performance of the second AI solution.

[0267] Step (ii): receiving a first request message sent by the terminal device;

[0268] The terminal device sends a first request message to the network device through a UCI or RRC message, where the first request message is used to request the network device to send a first AI solution.

[0269] Step (iii): sending the first AI solution to the terminal device;

[0270] The first AI solution is trained by the network device, and the network device sends the constructed first AI solution to the terminal device.

[0271] The network device sends the first AI solution through an RRC message, or a PDSCH, or a specific channel for transmitting an AI / ML model.

[0272] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0273] Step 1240: Send a first data set, where the first data set is used to train a first AI solution;

[0274] In some embodiments, step 1240 refers to the network device sending the first data set to the terminal device.

[0275] The network device sends the data set via RRC messages, PDSCH, or a specific channel for transmitting AI / ML models.

[0276] The network device instructs the transmission of the first AI solution through DCI, MAC CE, RRC message, and broadcast message.

[0277] In some embodiments, step 1240 may be implemented as step (c) and step (d).

[0278] Step (c): receiving a second request message sent by the terminal device;

[0279] The terminal device sends a second request message to the network device through a UCI or RRC message, where the second request message is used to request the network device to send the first data set.

[0280] Step (d): sending the first data set to the terminal device;

[0281] The network device sends the first data set through an RRC message, a PDSCH, or a specific channel for transmitting the AI / ML model.

[0282] The network device instructs the transmission of the first data set through DCI, MAC CE, RRC message, or broadcast message.

[0283] In some embodiments, step 1240 may be implemented as step (e), step (f), and step (g).

[0284] Step (e): sending second indication information to the terminal device;

[0285] The network device sends the second indication information through at least one of DCI, MAC CE, RRC signaling, RRC reconfiguration message, system broadcast, MIB, SIB, SIB2, etc. The second indication information is used to indicate or notify that the first AI solution is used to determine the performance of the second AI solution, or the second indication information is used to indicate or notify that the first data set is used to determine the performance of the second AI solution.

[0286] Step (f): receiving a second request message sent by the terminal device;

[0287] The terminal device sends a second request message to the network device through a UCI or RRC message, where the second request message is used to request the network device to send the first data set.

[0288] Step (g): sending the first data set to the terminal device;

[0289] The network device sends the first data set through an RRC message, a PDSCH, or a specific channel for transmitting the AI / ML model.

[0290] The network device instructs the transmission of the first data set through DCI, MAC CE, RRC message, or broadcast message.

[0291] Step 1250: Determine the performance of the second AI solution based on the first AI solution.

[0292] Determining the performance of the second AI solution based on the first AI solution can be understood as using the first AI solution to determine the performance of the second AI solution, or as obtaining the performance of the second AI solution after executing the first AI solution.

[0293] In some embodiments, before using the second AI solution to perform the first communication task, the performance of the second AI solution is determined based on the first AI solution.

[0294] For details, please refer to step 750 and step 760, which will not be repeated here.

[0295] In summary, the method provided by this application, compared to related art performance evaluation schemes that use information such as test datasets, labeled data, target task model inference results, and usage results, allows the performance evaluation of the second AI solution to be completed using an AI / ML performance evaluation scheme without excessive reliance on this information. Furthermore, when a network device constructs a first AI solution, it can directly rely on the dataset used to construct the second AI solution and the local performance evaluation results of the second AI solution, thereby improving the flexibility and simplicity of constructing the first AI solution. Furthermore, the first AI solution, or the dataset corresponding to the first AI solution, is sent to the terminal device, ensuring the validity of the first AI solution or its dataset while enabling the terminal device to directly use the first AI solution or construct the first AI solution using the first dataset, enabling the terminal device to implement performance monitoring, model selection, and model updates for the second AI solution based on AI / ML.

[0296] Figure 13 shows a block diagram of a performance determination device provided by some exemplary embodiments of the present application. The device includes at least some of the following modules: a first determination module 1320, a first receiving module 1340, a first sending module 1360, and a first training module 1380:

[0297] A first determination module 1320 is configured to determine the performance of a second AI solution based on the first AI solution;

[0298] The second AI solution is used to perform the first communication task.

[0299] In some embodiments, the apparatus further includes a first receiving module 1340 configured to receive the first AI solution sent by a network device.

[0300] In some embodiments, the apparatus further includes a first sending module 1360, configured to send first request information to the network device, where the first request information is used to request the network device to send the first AI solution.

[0301] In some embodiments, the first receiving module 1340 is configured to receive first indication information sent by the network device, where the first indication information is used to indicate or notify the first AI solution to determine performance of the second AI solution.

[0302] In some embodiments, the apparatus further includes a first training module 1380 for training the first AI solution based on a first data set.

[0303] In some embodiments, the first receiving module 1340 is configured to receive the first data set sent by a network device.

[0304] In some embodiments, the first sending module 1360 is configured to send a second request message to the network device, where the second request message is used to request the network device to send the first data set.

[0305] In some embodiments, the first receiving module 1340 is configured to receive second indication information sent by the network device, where the second indication information is used to indicate or notify that the first data set is used to determine the performance of the second AI solution.

[0306] In some embodiments, the first determining module 1320 is configured to input first input information into the first AI solution to obtain first output information; and determine the performance of the second AI solution based on the first output information;

[0307] Wherein, the first input information includes feature information and / or data information corresponding to the feature information;

[0308] The feature information refers to feature information corresponding to at least one of the input interface, the output interface, and the intermediate interface corresponding to the second AI solution, and the data information refers to data information obtained after data processing of the feature information.

[0309] In some embodiments, the characteristic information includes at least one of the following: channel state information reference signal CSI-RS; time domain channel; frequency domain channel; characteristic vector; CSI reporting information; PMI information; beam measurement results; beam ID; beam configuration scheme; beam quality prediction results; CIR information; RSRP information; TDOA information; DOA information; LOS information; NLOS information; AOA information.

[0310] In some embodiments, the first output information includes at least one of the following:

[0311] Use the second AI solution;

[0312] Not using the second AI solution;

[0313] The performance of the second AI solution is better than that of the comparison solution;

[0314] The performance of the second AI solution is not better than that of the comparison solution;

[0315] The probability that the performance of the second AI solution is better than the performance of the comparison solution;

[0316] The credibility of the second AI solution;

[0317] The comparison solution is a solution for performing the first communication task other than the second AI solution.

[0318] In some embodiments, the first determining module 1320 is configured to quantize the first input signal.

[0319] In some embodiments, the quantization method includes at least one of the following: uniform quantization; non-uniform quantization; a quantization method based on the Int4 format; a quantization method based on the Int8 format; a quantization method based on the float16 format; a quantization method based on the float32 format; a quantization method based on the float64 format; a quantization method based on the RSRP value; a quantization method based on the RSRP interval; a quantization method based on the RSRQ value; a quantization method based on the RSRQ interval; a quantization method based on the SINR value; a quantization method based on the SINR interval.

[0320] In some embodiments, the first training module 1380 is used to train the second AI solution.

[0321] In some embodiments, the first receiving module 1340 is configured to receive the second AI solution sent by the network device.

[0322] In some embodiments, the first receiving module 1340 is used to receive a second data set sent by the network device; and the first training module 1380 is used to train the second AI solution based on the second data set.

[0323] In some embodiments, the first receiving module 1340 is configured to receive a second data set sent by the network device; the first training module 1380 is configured to train a second federation sub-scheme based on the second data set;

[0324] The second AI solution includes the second federation sub-solution and a first federation sub-solution, and the first federation sub-solution is trained by the network device.

[0325] In some embodiments, the first communication task is used for at least one of the following: CSI compression; CSI recovery; CSI prediction; beam selection; beam prediction; positioning.

[0326] In summary, the apparatus provided in this embodiment determines the performance of a second AI solution using a first AI solution, without relying on the execution results of the second AI solution, thereby increasing the flexibility of determining the performance of AI-based communication solutions. Furthermore, a terminal device can directly use the first AI solution sent by a network device, or a first data set sent by the network device, to construct the first AI solution. This ensures the validity of the first AI solution or the data set of the first AI solution, while also increasing the flexibility and simplicity of constructing the first AI solution. This enables the terminal device to implement performance monitoring, model selection, and model updates for the second AI solution based on AI / ML.

[0327] Figure 14 shows a block diagram of a performance determination device provided by some exemplary embodiments of the present application. The device includes at least some of the following modules: a second determination module 1420, a second training module 1440, a second sending module 1460, and a second receiving module 1480:

[0328] A second determination module 1420 is configured to determine the performance of a second AI solution based on the first AI solution;

[0329] The second AI solution is used to perform the first communication task.

[0330] In some embodiments, the apparatus further includes a second training module 1440 for training the first AI solution.

[0331] In some embodiments, the apparatus further includes a second sending module 1460, configured to send the first AI solution to a terminal device.

[0332] In some embodiments, the apparatus further includes a second receiving module 1480, configured to receive first request information sent by the terminal device, where the first request information is used to request the network device to send the first AI solution.

[0333] In some embodiments, the second sending module 1460 is configured to send first indication information to the terminal device, where the first indication information is used to indicate or notify that the first AI solution is used to determine the performance of the second AI solution.

[0334] In some embodiments, the second sending module 1460 is configured to send the first data set to the terminal device.

[0335] In some embodiments, the second receiving module 1480 is used to receive second request information sent by the terminal device, where the second request information is used to request the network device to send the first data set.

[0336] In some embodiments, the second sending module 1460 is configured to send second indication information to the terminal device, where the second indication information is used to indicate or notify that the first data set is used to determine the performance of the second AI solution.

[0337] In some embodiments, the second training module 1440 is configured to input first input information into the first AI solution to obtain first output information; and determine the performance of the second AI solution based on the first output information;

[0338] Wherein, the first input information includes feature information and / or data information corresponding to the feature information;

[0339] The feature information refers to feature information corresponding to at least one of the input interface, the output interface, and the intermediate interface corresponding to the second AI solution, and the data information refers to data information obtained after data processing of the feature information.

[0340] In some embodiments, the characteristic information includes at least one of the following: channel state information reference signal CSI-RS; time domain channel; frequency domain channel; characteristic vector; CSI reporting information; PMI information; beam measurement results; beam ID; beam configuration scheme; beam quality prediction results; CIR information; RSRP information; TDOA information; DOA information; LOS information; NLOS information; AOA information.

[0341] In some embodiments, the first output information includes at least one of the following:

[0342] Use the second AI solution;

[0343] Not using the second AI solution;

[0344] The performance of the second AI solution is better than that of the comparison solution;

[0345] The performance of the second AI solution is not better than that of the comparison solution;

[0346] The probability that the performance of the second AI solution is better than the performance of the comparison solution;

[0347] The credibility of the second AI solution;

[0348] The comparison solution is a solution for performing the first communication task other than the second AI solution.

[0349] In some embodiments, the second determining module 1420 is configured to quantize the first input signal.

[0350] In some embodiments, the quantization method includes at least one of the following: uniform quantization; non-uniform quantization; a quantization method based on the Int4 format; a quantization method based on the Int8 format; a quantization method based on the float16 format; a quantization method based on the float32 format; a quantization method based on the float64 format; a quantization method based on the RSRP value; a quantization method based on the RSRP interval; a quantization method based on the RSRQ value; a quantization method based on the RSRQ interval; a quantization method based on the SINR value; a quantization method based on the SINR interval.

[0351] In some embodiments, the second AI solution is trained by the terminal device and / or trained by the network device.

[0352] In some embodiments, the second sending module 1460 is configured to send the second AI solution to the terminal device.

[0353] In some embodiments, the second sending module 1460 is configured to send a second data set to the terminal device, where the second data set is used to train the second AI solution.

[0354] In some embodiments, the second sending module 1460 is configured to send a second data set to the terminal device, where the second data set is used to train a second federation sub-scheme;

[0355] The second training module 1440 is used to train the first federation sub-scheme;

[0356] The second AI solution includes the second federation sub-solution and the first federation sub-solution.

[0357] In some embodiments, the first communication task is used for at least one of the following:

[0358] Channel state information CSI compression;

[0359] CSI recovery;

[0360] CSI forecast;

[0361] beam selection;

[0362] Beam prediction;

[0363] position.

[0364] In summary, the apparatus provided in this embodiment determines the performance of a second AI solution using a first AI solution, without relying on the execution results of the second AI solution. This improves the flexibility of determining the performance of AI-based communication solutions. Furthermore, it supports the terminal device directly using the first AI solution sent by the network device, or the first data set sent by the network device, to construct the first AI solution. This not only ensures the validity of the first AI solution or the data set of the first AI solution, but also improves the flexibility and simplicity of constructing the first AI solution, enabling the terminal device to implement performance monitoring, model selection, and model updates for the second AI solution based on AI / ML.

[0365] It should be noted that the device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0366] Regarding the device in this embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method and will not be elaborated here.

[0367] FIG15 shows a schematic structural diagram of a communication device provided in some exemplary embodiments of the present application. The communication device 1500 includes: a processor 1501 , a receiver 1502 , a transmitter 1503 , a memory 1504 and a bus 1505 .

[0368] Processor 1501 includes one or more processing cores. Processor 1501 executes various functional applications and information processing by running software programs and modules. In some embodiments, processor 1501 can be used to implement the functions and steps of first determination module 1320, and / or first training module 1380, and / or second determination module 1420, and / or second training module 1440 described above.

[0369] Receiver 1502 and transmitter 1503 can be implemented as a communication component, which can be a communication chip. In some embodiments, receiver 1502 can be used to implement the functions and steps of first receiving module 1340 and / or second receiving module 1480 described above. In some embodiments, transmitter 1503 can be used to implement the functions and steps of first transmitting module 1360 and / or second transmitting module 1460 described above.

[0370] The memory 1504 is connected to the processor 1501 via a bus 1505. The memory 1504 may be used to store at least one instruction, and the processor 1501 may be used to execute the at least one instruction to implement each step in the above method embodiment.

[0371] In addition, the memory 1504 can be implemented by any type of volatile or non-volatile storage device or a combination thereof. Volatile or non-volatile storage devices include but are not limited to: magnetic disks or optical disks, electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), static random-access memory (SRAM), read-only memory (ROM), magnetic memory, flash memory, and programmable read-only memory (PROM).

[0372] In some embodiments, the receiver 1502 receives signals / data independently, or the processor 1501 controls the receiver 1502 to receive signals / data, or the processor 1501 requests the receiver 1502 to receive signals / data, or the processor 1501 cooperates with the receiver 1502 to receive signals / data.

[0373] In some embodiments, the transmitter 1503 independently sends signals / data, or the processor 1501 controls the transmitter 1503 to send signals / data, or the processor 1501 requests the transmitter 1503 to send signals / data, or the processor 1501 cooperates with the transmitter 1503 to send signals / data.

[0374] In an exemplary embodiment of the present application, a computer-readable storage medium is also provided, in which at least one program is stored. The at least one program is loaded and executed by a processor, and the computer-readable storage medium is used to implement the performance determination method provided by the above-mentioned various method embodiments.

[0375] In an exemplary embodiment of the present application, a chip is also provided, which includes a programmable logic circuit and / or program instructions. When the chip runs on a communication device, it is used to implement the performance determination method provided by the above-mentioned various method embodiments.

[0376] In an exemplary embodiment of the present application, a computer program product is further provided. When the computer program product is run on a processor of a communication device, the communication device is enabled to perform the above-mentioned performance determination method.

[0377] In an exemplary embodiment of the present application, a computer program is further provided. The computer program includes computer instructions. A processor of a communication device executes the computer instructions, so that the communication device performs the above-mentioned performance determination method.

[0378] Those skilled in the art will appreciate that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any media that facilitates the transmission of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0379] The above description is merely an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A performance determination method, characterized in that: The method is executed by a terminal device, and includes: Determining performance of a second AI solution based on the first AI solution; The second AI solution is used to perform the first communication task.

2. The method according to claim 1, characterized in that Before determining the performance of the second AI solution based on the first AI solution, the method further includes at least one of the following: receiving the first AI solution sent by the network device; Sending a first request message to the network device, where the first request message is used to request the network device to send the first AI solution; First indication information sent by the network device is received, where the first indication information is used to instruct or notify the first AI solution to be used for determining performance of the second AI solution.

3. The method according to claim 1, characterized in that Before determining the performance of the second AI solution based on the first AI solution, the method further includes: The first AI solution is trained based on a first data set.

4. The method according to claim 3, characterized in that Before constructing the first AI solution based on the first data set, the method further includes at least one of the following: receiving the first data set sent by a network device; Sending a second request message to the network device, where the second request message is used to request the network device to send the first data set; Second indication information sent by the network device is received, where the second indication information is used to indicate or notify that the first data set is used to determine performance of the second AI solution.

5. The method according to any one of claims 1 to 4, characterized in that: The determining, based on the first AI solution, the performance of the second AI solution includes: Inputting first input information into the first AI solution to obtain first output information; and determining performance of the second AI solution based on the first output information; Wherein, the first input information includes feature information and / or data information corresponding to the feature information; The feature information refers to feature information corresponding to at least one of the input interface, the output interface, and the intermediate interface corresponding to the second AI solution, and the data information refers to data information obtained after data processing of the feature information.

6. The method according to claim 5, characterized in that The characteristic information includes at least one of the following: Channel State Information Reference Signal CSI-RS; Time domain channel; Frequency domain channel; Eigenvector; CSI reporting information; Precoding matrix indicator PMI information; Beam measurement results; Beam identification ID; Beam configuration scheme; Beam quality prediction results; Channel impulse response CIR information; Reference signal received power RSRP information; Time difference of arrival (TDOA) information; Direction of arrival (DOA) information; Line of sight transmission LOS information; Non-line-of-sight transmission of NLOS information; Angle of arrival (AOA) information.

7. The method according to claim 5, characterized in that The first output information includes at least one of the following: Use the second AI solution; The second AI solution is not used; The performance of the second AI solution is better than that of the comparison solution; The performance of the second AI solution is not better than the performance of the comparison solution; the probability that the performance of the second AI solution is better than the performance of the comparison solution; The credibility probability of the second AI solution; The comparison solution is a solution for performing the first communication task other than the second AI solution.

8. The method according to any one of claims 5 to 7, characterized in that: The method further comprises: quantizing the first input signal; The quantification method includes at least one of the following: Uniform quantization; Non-uniform quantization; Quantization method based on Int4 format; Quantization method based on Int8 format; Quantization method based on float16 format; Quantization method based on float32 format; Quantization method based on float64 format; Quantization method based on reference signal received power RSRP value; Quantization method based on RSRP interval; Quantification method based on reference signal reception quality RSRQ value; Quantization method based on RSRQ interval; Quantization method based on signal-to-interference-and-noise ratio SINR value; Quantization method based on SINR interval.

9. The method according to any one of claims 1 to 8, characterized in that: The second AI solution is trained by the terminal device and / or trained by a network device.

10. The method according to claim 9, characterized in that The second AI solution is trained by the network device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: Receive the second AI solution sent by the network device.

11. The method according to claim 9, characterized in that The second AI solution is trained by the terminal device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: receiving a second data set sent by the network device; The second AI solution is trained based on the second data set.

12. The method according to claim 9, characterized in that The second AI solution is trained by the terminal device and the network device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: receiving a second data set sent by the network device; training a second federated sub-scheme based on the second data set; The second AI solution includes the second federation sub-solution and a first federation sub-solution, and the first federation sub-solution is trained by the network device.

13. The method according to any one of claims 1 to 12, characterized in that: The first communication task is used for at least one of the following: Channel state information CSI compression; CSI recovery; CSI forecast; beam selection; Beam prediction; position.

14. A performance determination method, characterized in that: The method is performed by a network device, and includes: Determining performance of a second AI solution based on the first AI solution; The second AI solution is used to perform the first communication task.

15. The method according to claim 14, characterized in that Before determining the performance of the second AI solution based on the first AI solution, the method further includes: Train the first AI solution.

16. The method according to claim 15, characterized in that The method further comprises at least one of the following: Sending the first AI solution to the terminal device; receiving a first request message sent by the terminal device, where the first request message is used to request the network device to send the first AI solution; First indication information is sent to the terminal device, where the first indication information is used to instruct or notify the first AI solution to be used to determine performance of the second AI solution.

17. The method according to claim 15, characterized in that The method further comprises at least one of the following: sending the first data set to the terminal device; receiving a second request message sent by the terminal device, where the second request message is used to request the network device to send the first data set; Sending second indication information to the terminal device, where the second indication information is used to indicate or notify that the first data set is used to determine the performance of the second AI solution.

18. The method according to any one of claims 14 to 17, characterized in that The determining, based on the first AI solution, the performance of the second AI solution includes: Inputting first input information into the first AI solution to obtain first output information; and determining performance of the second AI solution based on the first output information; Wherein, the first input information includes feature information and / or data information corresponding to the feature information; The feature information refers to feature information corresponding to at least one of the input interface, the output interface, and the intermediate interface corresponding to the second AI solution, and the data information refers to data information obtained after data processing of the feature information.

19. The method according to claim 18, characterized in that The characteristic information includes at least one of the following: Channel State Information Reference Signal CSI-RS; Time domain channel; Frequency domain channel; Eigenvector; CSI reporting information; Precoding matrix indicator PMI information; Beam measurement results; Beam identification ID; Beam configuration scheme; Beam quality prediction results; Channel impulse response CIR information; Reference signal received power RSRP information; Time difference of arrival (TDOA) information; Direction of arrival (DOA) information; Line of sight transmission LOS information; Non-line-of-sight transmission of NLOS information; Angle of arrival (AOA) information.

20. The method according to claim 18, wherein The first output information includes at least one of the following: Use the second AI solution; The second AI solution is not used; The performance of the second AI solution is better than that of the comparison solution; The performance of the second AI solution is not better than the performance of the comparison solution; the probability that the performance of the second AI solution is better than the performance of the comparison solution; The credibility probability of the second AI solution; The comparison solution is a solution for performing the first communication task other than the second AI solution.

21. The method according to any one of claims 18 to 20, characterized in that The method further comprises: quantizing the first input signal; The quantification method includes at least one of the following: Uniform quantization; Non-uniform quantization; Quantization method based on Int4 format; Quantization method based on Int8 format; Quantization method based on float16 format; Quantization method based on float32 format; Quantization method based on float64 format; Quantization method based on reference signal received power RSRP value; Quantization method based on RSRP interval; Quantification method based on reference signal reception quality RSRQ value; Quantization method based on RSRQ interval; Quantization method based on signal-to-interference-and-noise ratio SINR value; Quantization method based on SINR interval.

22. The method according to any one of claims 14 to 21, characterized in that The second AI solution is trained by the terminal device and / or trained by the network device.

23. The method according to claim 22, characterized in that The second AI solution is trained by the network device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: Send the second AI solution to the terminal device.

24. The method according to claim 22, characterized in that The second AI solution is trained by the terminal device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: Sending a second data set to the terminal device, where the second data set is used to train the second AI solution.

25. The method according to claim 22, wherein The second AI solution is trained by the terminal device and the network device; Before determining the performance of the second AI solution based on the first AI solution, the method further includes: Sending a second data set to the terminal device, where the second data set is used to train a second federated sub-scheme; training the First Federal Sub-Program; The second AI solution includes the second federation sub-solution and the first federation sub-solution.

26. The method according to any one of claims 14 to 25, characterized in that The first communication task is used for at least one of the following: Channel state information CSI compression; CSI recovery; CSI forecast; beam selection; Beam prediction; position.

27. A performance determination device, characterized in that The device comprises: A first determination module, configured to determine performance of a second AI solution based on the first AI solution; The second AI solution is used to perform the first communication task.

28. A performance determination device, characterized in that The device comprises: a second determination module, configured to determine performance of a second AI solution based on the first AI solution; The second AI solution is used to perform the first communication task.

29. A communication device, characterized in that: The communication device comprises: processor; a transceiver connected to the processor; a memory for storing executable instructions for the processor; The processor is configured to load and execute the executable instructions, and the communication device implements the performance determination method according to any one of claims 1 to 13 or claims 14 to 26.

30. A computer-readable storage medium, characterized in that The computer-readable storage medium stores executable instructions, which are loaded and executed by a processor. The computer-readable storage medium is used to implement the performance determination method according to any one of claims 1 to 13 or claims 14 to 26.

31. A chip, characterized in that: The chip includes a programmable logic circuit or a program, and the chip is used to implement the performance determination method according to any one of claims 1 to 13 or claims 14 to 26.

32. A computer program product, characterized in that The computer program product includes computer instructions, which are stored in a computer-readable storage medium. The processor of the communication device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the communication device performs the performance determination method as described in any one of claims 1 to 13 or claims 14 to 26.

33. A computer program, characterized in that The computer program includes computer instructions, and the processor of the communication device executes the computer instructions, so that the communication device performs the performance determination method according to any one of claims 1 to 13 or claims 14 to 26.

Images