Model performance monitoring method, device and storage medium

Abstract

The present disclosure relates to a model performance monitoring method, a device and a storage medium. The method comprises: receiving first instruction information sent by a network device; on the basis of first information, determining, from a first model and / or a second model, a third model for which model performance monitoring needs to be performed; and on the basis of the third model, reporting first CSI to the network device, wherein the first model comprises a first-part model deployed in a terminal device and a second-part model deployed in the network device, and the second model comprises a third-part model deployed in the terminal device and a fourth-part model deployed in the network device, the first-part model being used for compressing CSI, the second-part model being used for performing CSI recovery on the CSI compressed by the first-part model, the third-part model being used for compressing and / or predicting the CSI, and the fourth-part model being used for performing CSI recovery and / or CSI prediction on the CSI compressed by the third-part model. In this way, configuration signaling overheads of model performance monitoring are reduced.

Description

Model performance monitoring methods, equipment and storage media Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a model performance monitoring method, device, and storage medium. Background Technology

[0002] With the advancement of communication technology, artificial intelligence (AI) models have been introduced into communication systems. AI models can reduce terminal feedback overhead or improve the accuracy of Channel Status Information (CSI) feedback. However, the effectiveness of AI models depends on their performance. Therefore, how to monitor the performance of AI models has become an urgent problem to be solved.

[0003] Summary of the Invention

[0004] This disclosure presents a model performance monitoring method, device, and storage medium.

[0005] According to a first aspect of the embodiments of this disclosure, a model performance monitoring method is proposed, executed by a terminal device, the method comprising:

[0006] Receive first indication information sent by the network device, the first indication information being used to instruct model performance monitoring;

[0007] Based on the first information, a third model that needs to be monitored for model performance is determined from the first model and / or the second model. The first information includes information corresponding to the first channel state information (CSI) that the terminal device needs to report. The first CSI is used to monitor the performance of the third model.

[0008] The first CSI is reported to the network device according to the third model;

[0009] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0010] According to a second aspect of the embodiments of this disclosure, a model performance monitoring method is proposed, executed by a network device, the method comprising:

[0011] Send a first instruction message to the terminal device, the first instruction message being used to instruct model performance monitoring;

[0012] The terminal device receives first channel state information (CSI) reported according to a third model. The first CSI is used to monitor the performance of the third model. The third model is a model that needs to be monitored for model performance, which is determined by the terminal device from the first model and / or the second model based on first information. The first information includes information corresponding to the first CSI.

[0013] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0014] According to a third aspect of the embodiments of this disclosure, a terminal device is provided, comprising:

[0015] The transceiver module is configured to receive first indication information sent by the network device, the first indication information being used to instruct model performance monitoring;

[0016] The processing module is configured to determine a third model that needs to be monitored for model performance from a first model and / or a second model based on first information. The first information includes information corresponding to the first channel state information (CSI) that the terminal device needs to report. The first CSI is used to monitor the performance of the third model.

[0017] The transceiver module is also configured to report the first CSI to the network device according to the third model;

[0018] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0019] According to a fourth aspect of the embodiments of this disclosure, a network device is provided, comprising:

[0020] The transceiver module is configured to send a first indication information to the terminal device, the first indication information being used to indicate that model performance monitoring is to be performed.

[0021] The transceiver module is further configured to receive first channel state information (CSI) reported by the terminal device according to the third model. The first CSI is used to monitor the performance of the third model. The third model is a model that needs to be monitored for model performance, which is determined by the terminal device from the first model and / or the second model according to first information. The first information includes information corresponding to the first CSI.

[0022] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0023] According to a fifth aspect of the embodiments of this disclosure, a communication device is provided, comprising:

[0024] One or more processors; wherein the communication device may be used to execute an optional implementation of the first aspect or the second aspect.

[0025] According to a sixth aspect of the present disclosure, a communication system is provided, including a terminal device and a network device, wherein the terminal device is configured to perform the method described in the optional implementation of the first aspect, and the network device is configured to perform the method described in the optional implementation of the second aspect.

[0026] According to a seventh aspect of the present disclosure, a storage medium is provided that stores instructions that, when executed on a communication device, cause the communication device to perform the method as described in an optional implementation of the first or second aspect.

[0027] The technical solution provided by this disclosure can produce the following beneficial effects: receiving first indication information sent by a network device, the first indication information being used to indicate model performance monitoring; determining a third model that needs to be monitored from a first model and / or a second model according to the first information, the first information including information corresponding to the first channel state information (CSI) that the terminal device needs to report, the first CSI being used to monitor the performance of the third model; reporting the first CSI to the network device according to the third model; wherein, the first model includes a first part model deployed on the terminal device and a second part model deployed on the network device, the second model includes a third part model deployed on the terminal device and a fourth part model deployed on the network device, the first part model being used to compress the CSI, the second part model being used to recover the CSI after compression of the first part model, the third part model being used to compress and / or predict the CSI, and the fourth part model being used to recover and / or predict the CSI after compression of the third part model. In other words, the terminal device can determine the third model that needs performance monitoring from the first model and / or the second model based on the information corresponding to the first CSI that needs to be reported, and report the first CSI to the network device so that the performance of the third model can be monitored. In this way, performance monitoring of the first and second models can be achieved using unified parameter configuration, reducing the overhead of configuration signaling used for model performance monitoring and reducing the LCM complexity of the model.

[0028] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the accompanying drawings required for the description of the embodiments are introduced below. The following drawings are only some embodiments of this disclosure and do not impose specific limitations on the protection scope of this disclosure.

[0030] Figure 1A is a schematic diagram of the architecture of a communication system according to an embodiment of the present disclosure.

[0031] Figure 1B is a schematic diagram illustrating a CSI compression feedback and recovery based on a bilateral AI / ML model according to an embodiment of the present disclosure.

[0032] Figure 1C is a schematic diagram illustrating a CSI compression feedback based on a bilateral CSI prediction compression model according to an embodiment of the present disclosure.

[0033] Figure 2A is an interactive schematic diagram of a model performance monitoring method according to an embodiment of the present disclosure.

[0034] Figure 2B is a schematic diagram of CSI reporting according to an embodiment of the present disclosure.

[0035] Figure 2C is an interactive schematic diagram illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0036] Figure 3A is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0037] Figure 3B is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0038] Figure 3C is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0039] Figure 3D is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0040] Figure 4A is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0041] Figure 4B is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0042] Figure 4C is a schematic flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0043] Figure 4D is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure.

[0044] Figure 5 is an interactive schematic diagram of a model performance monitoring method according to an embodiment of the present disclosure.

[0045] Figure 6A is a schematic diagram of the structure of a terminal device proposed in an embodiment of this disclosure.

[0046] Figure 6B is a schematic diagram of the structure of a network device proposed in an embodiment of this disclosure.

[0047] Figure 7A is a schematic diagram of the structure of the communication device proposed in an embodiment of this disclosure.

[0048] Figure 7B is a schematic diagram of the chip structure proposed in an embodiment of this disclosure. Detailed Implementation

[0049] This disclosure presents a model performance monitoring method, device, and storage medium.

[0050] In a first aspect, embodiments of this disclosure propose a model performance monitoring method, executed by a terminal device, the method comprising:

[0051] Receive first indication information sent by the network device, the first indication information being used to instruct model performance monitoring;

[0052] Based on the first information, a third model that needs to be monitored for model performance is determined from the first model and / or the second model. The first information includes information corresponding to the first channel state information (CSI) that the terminal device needs to report. The first CSI is used to monitor the performance of the third model.

[0053] The first CSI is reported to the network device according to the third model;

[0054] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0055] In the above embodiments, the terminal device can determine the third model that needs performance monitoring from the first model and / or the second model based on the first indication information, and report the first CSI to the network device so as to perform performance monitoring on the third model. In this way, performance monitoring of the first and second models can be achieved using a unified parameter configuration, reducing the overhead of configuration signaling for model performance monitoring and reducing the LCM complexity of the model.

[0056] In conjunction with some embodiments of the first aspect, in some embodiments, the first CSI includes measured CSI and / or compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI is the CSI at the M times output by the first part model and / or the third part model, where M is a positive integer.

[0057] In the above embodiments, when the terminal device performs performance monitoring on the third model, it may only report compressed CSI, while the network device may report both measured CSI and compressed CSI.

[0058] In conjunction with some embodiments of the first aspect, in some embodiments, the first information includes at least one of the following: the value of M, a first numerical value, and the M time points, wherein the first numerical value is used to determine the M time points.

[0059] In the above embodiments, M time points can be determined based on at least one of the first pieces of information, making the determination of the M time points more flexible.

[0060] In conjunction with some embodiments of the first aspect, in some embodiments, determining the third model requiring model performance monitoring from the first model and / or the second model based on the first information includes at least one of the following:

[0061] If the value of M is greater than 1, the third model is determined to be the second model;

[0062] The value of M is equal to 1, the first value is less than 0, and the M times determined according to the first value are the same as one of the transmission times of the Channel State Information Reference Signal (CSI-RS) resource periodically transmitted by the network device, and the third model is determined to be the first model.

[0063] The value of M is equal to 1, the first value is less than 0, and the M times determined according to the first value are different from each transmission time of the network device periodically sending CSI-RS resources, so the third model is determined to be the second model;

[0064] Each of the M time points is the same as one of the time points when the network device periodically sends CSI-RS resources, thus determining the third model as the first model;

[0065] The M time points are different from each transmission time of the CSI-RS resources periodically sent by the network device, thus determining the third model as the second model.

[0066] In conjunction with some embodiments of the first aspect, in some embodiments, the value of M is determined by at least one of the following methods:

[0067] The terminal device determines the value of M;

[0068] The value of M is configured for the network device.

[0069] In the above embodiments, the value of M can be determined by the terminal device or configured by the network device.

[0070] In conjunction with some embodiments of the first aspect, in some embodiments, the terminal device determines the value of M to include at least one of the following:

[0071] The model deployed on the terminal device does not include the fourth model, and the value of M is determined to be 1. The fourth model is a model used to predict CSI.

[0072] The model deployed on the terminal device includes the fourth model, and the number of times corresponding to CSI that the fourth model can predict is used as the value of M.

[0073] In the above embodiments, the value of M can be determined based on the model deployed on the terminal device.

[0074] In conjunction with some embodiments of the first aspect, in some embodiments, the M time points are determined by at least one of the following methods:

[0075] The M time points are determined based on the downlink time point corresponding to the CSI reference resource and the first value;

[0076] Based on the CSI reporting time and the first value, the M time points are determined;

[0077] The M times are determined based on the time corresponding to the first indication information and the first value.

[0078] In the above embodiments, the M time points can be determined in multiple ways, making the determination of the M time points more flexible.

[0079] In conjunction with some embodiments of the first aspect, in some embodiments, the time corresponding to the first indication information includes at least one of the following:

[0080] The moment when the network device sends the first indication information;

[0081] The time at which the terminal device receives the first instruction information.

[0082] In conjunction with some embodiments of the first aspect, in some embodiments, the determination of the first value includes at least one of the following:

[0083] The terminal device determines the first value;

[0084] The network device is configured with the first value;

[0085] The first value is determined based on an initial value and the time interval between two adjacent moments, wherein the initial value and the time interval are configured or predefined by the network device;

[0086] The first value is predefined.

[0087] In the above embodiments, the first value can be determined in multiple ways, making the determination of the first value more flexible.

[0088] In conjunction with some embodiments of the first aspect, in some embodiments, the first value is predefined and includes at least one of the following:

[0089] The first value is predefined as the downlink time corresponding to the CSI reference resource;

[0090] The first value is predefined as the time when the fourth model deployed on the terminal device predicts CSI, and the fourth model is a model used to predict CSI;

[0091] The first value is predefined as the time when the network device sends the downlink pilot.

[0092] In the above embodiments, the first value can be predefined in a variety of ways.

[0093] In conjunction with some embodiments of the first aspect, in some embodiments, the first CSI includes the measurement CSI and the compressed CSI, the first CSI is used by the network device to determine model performance criteria, the model performance criteria are used to determine the performance of the third model; wherein, the model performance criteria include at least one of the following: squared cosine similarity (SGCS) and normalized mean square error (NMSE).

[0094] In the above embodiments, measurement CSI and compression CSI can be reported to the network device so that the network device can monitor the performance of the third model.

[0095] In conjunction with some embodiments of the first aspect, in some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0096] Receive the second CSI sent by the network device, wherein the second CSI is the CSI at the M time points recovered by the network device based on the compressed CSI, through the second part model and / or the fourth part model;

[0097] The model performance criteria are determined based on the measured CSI and the second CSI, and the model performance criteria are used to determine the performance of the third model; wherein the model performance criteria include at least one of the following: SGCS, NMSE.

[0098] In the above embodiments, a compressed CSI can be reported to the network device so that the network device can send a second CSI based on the compressed CSI. In this way, the terminal device can determine the performance of the third model based on the measured CSI and the second CSI.

[0099] In conjunction with some embodiments of the first aspect, in some embodiments, receiving the second CSI sent by the network device includes at least one of the following:

[0100] Receive all second CSIs sent by the network device at the same time;

[0101] Receive the CSIs at each time point in the second CSI, which are sent sequentially by the network device in a first order.

[0102] In the above embodiments, the network device can send all the second CSIs at once, or it can send the CSIs at each time step by step in the first order.

[0103] In conjunction with some embodiments of the first aspect, in some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0104] Receive pilot signals sent by the network device through precoding and shaping, wherein the precoding is a second CSI, and the second CSI is the CSI at the M time points recovered by the network device based on the compressed CSI through the second part model and / or the fourth part model;

[0105] The performance of the third model is determined based on the precoded and shaped pilot signal.

[0106] In the above embodiments, compressed CSI can be sent to the network device, and precoded and shaped pilot signals sent by the network device can be received. The performance of the third model can be determined based on the precoded and shaped pilot signals.

[0107] In conjunction with some embodiments of the first aspect, in some embodiments, receiving the precoded and shaped pilot signal sent by the network device includes at least one of the following:

[0108] Receive M precoded and shaped pilot signals sent by the network device at the same time;

[0109] Receive M precoded and shaped pilot signals sent sequentially by the network device in a first order.

[0110] In the above embodiments, the network device can send M precoded and shaped pilot signals at once, or it can send M precoded and shaped pilot signals sequentially in a first order.

[0111] In conjunction with some embodiments of the first aspect, in some embodiments, reporting the first CSI to the network device according to the third model includes:

[0112] The compressed CSI is determined according to the third model;

[0113] The first CSI is reported to the network device.

[0114] In conjunction with some embodiments of the first aspect, in some embodiments, determining the compressed CSI according to the third model includes at least one of the following:

[0115] The third model is the first model, and the compressed CSI is determined to be the CSI at the M times output by the first part of the model;

[0116] The third model is the second model, and the compressed CSI is determined to be the CSI at the M times output by the third part of the model.

[0117] In the above embodiments, a third model for performance monitoring can be used to determine the first CSI that needs to be reported to the network device.

[0118] In conjunction with some embodiments of the first aspect, in some embodiments, reporting the first CSI to the network device includes at least one of the following:

[0119] Report all first CSIs to the network device at the same time;

[0120] The network device shall report the CSI at each time point of the first CSI in the second order.

[0121] In the above embodiments, the terminal device can report all first CSIs at once, or it can report CSIs at each time step by step in the second order.

[0122] In conjunction with some embodiments of the first aspect, in some embodiments, the time for reporting the first CSI is determined by at least one of the following methods:

[0123] The network device provides instructions via downlink signaling;

[0124] Predefined;

[0125] The time when the terminal device compresses and reports the CSI is taken as the time when the first CSI is reported.

[0126] Secondly, this disclosure provides a model performance monitoring method, executed by a network device, the method comprising:

[0127] Send a first instruction message to the terminal device, the first instruction message being used to instruct model performance monitoring;

[0128] The terminal device receives first channel state information (CSI) reported according to a third model. The first CSI is used to monitor the performance of the third model. The third model is a model that needs to be monitored for model performance, which is determined by the terminal device from the first model and / or the second model based on first information. The first information includes information corresponding to the first CSI.

[0129] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0130] In conjunction with some embodiments of the second aspect, in some embodiments, the first CSI includes measured CSI and / or compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI is the CSI at the M times output by the first part model and / or the third part model, where M is a positive integer.

[0131] In conjunction with some embodiments of the second aspect, in some embodiments, the first information includes at least one of the following: the value of M, a first numerical value, and the M time points, wherein the first numerical value is used to determine the M time points.

[0132] In conjunction with some embodiments of the second aspect, in some embodiments, the value of M is determined by at least one of the following methods:

[0133] The terminal device determines the value of M;

[0134] The value of M is configured for the network device.

[0135] In conjunction with some embodiments of the second aspect, in some embodiments, the terminal device determines the value of M to include at least one of the following:

[0136] The model deployed on the terminal device does not include the fourth model, and the value of M is determined to be 1. The fourth model is a model used to predict CSI.

[0137] The model deployed on the terminal device includes the fourth model, and the number of times corresponding to CSI that the fourth model can predict is used as the value of M.

[0138] In conjunction with some embodiments of the second aspect, in some embodiments, the M time points are determined by at least one of the following methods:

[0139] The M time points are determined based on the downlink time point corresponding to the CSI reference resource and the first value;

[0140] Based on the CSI reporting time and the first value, the M time points are determined;

[0141] The M times are determined based on the time corresponding to the first indication information and the first value.

[0142] In conjunction with some embodiments of the second aspect, in some embodiments, the time corresponding to the first indication information includes at least one of the following:

[0143] The moment when the network device sends the first indication information;

[0144] The time at which the terminal device receives the first instruction information.

[0145] In conjunction with some embodiments of the second aspect, in some embodiments, the determination of the first value includes at least one of the following:

[0146] The terminal device determines the first value;

[0147] The network device is configured with the first value;

[0148] The first value is determined based on an initial value and the time interval between two adjacent moments, wherein the initial value and the time interval are configured or predefined by the network device;

[0149] The first value is predefined.

[0150] In conjunction with some embodiments of the second aspect, in some embodiments, the first value is predefined and includes at least one of the following:

[0151] The first value is predefined as the downlink time corresponding to the CSI reference resource;

[0152] The first value is predefined as the time when the fourth model deployed on the terminal device predicts CSI, and the fourth model is a model used to predict CSI;

[0153] The first value is predefined as the time when the network device sends the downlink pilot.

[0154] In conjunction with some embodiments of the second aspect, in some embodiments, the first CSI includes the measurement CSI and the compressed CSI, and the method further includes:

[0155] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0156] The model performance criteria are determined based on the measured CSI and the second CSI, and the model performance criteria are used to determine the performance of the third model; wherein the model performance criteria include at least one of the following: squared cosine similarity (SGCS) and normalized mean square error (NMSE).

[0157] In conjunction with some embodiments of the second aspect, in some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0158] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0159] The second CSI is sent to the terminal device. The second CSI is used by the terminal device to determine the model performance criteria, which are used to determine the performance of the third model. The model performance criteria include at least one of the following: SGCS and NMSE.

[0160] In conjunction with some embodiments of the second aspect, in some embodiments, sending the second CSI to the terminal device includes at least one of the following:

[0161] Send all second CSIs to the terminal device at the same time;

[0162] The CSIs for each moment in the second CSI are sent to the terminal device in the first order.

[0163] In conjunction with some embodiments of the second aspect, in some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0164] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0165] A precoded and shaped pilot signal is sent to the terminal device, wherein the precoding is the second CSI, and the precoded and shaped pilot signal is used by the terminal device to determine the performance of the third model.

[0166] In conjunction with some embodiments of the second aspect, in some embodiments, sending the precoded and shaped pilot signal to the terminal device includes at least one of the following:

[0167] At the same time, M precoded and shaped pilot signals are sent to the terminal device;

[0168] M precoded and shaped pilot signals are sent to the terminal device in the first order.

[0169] In conjunction with some embodiments of the second aspect, in some embodiments, receiving the first CSI reported by the terminal device includes at least one of the following:

[0170] Receive all first CSIs reported by the terminal device at the same time;

[0171] The terminal device receives the CSIs at each time point in the first CSI, reported sequentially in a second order.

[0172] In conjunction with some embodiments of the second aspect, in some embodiments, the time when the terminal device reports the first CSI is determined by at least one of the following methods:

[0173] The network device provides instructions via downlink signaling;

[0174] Predefined;

[0175] The time when the terminal device compresses and reports the CSI is taken as the time when the first CSI is reported.

[0176] Thirdly, embodiments of this disclosure propose a model performance monitoring method, the method comprising:

[0177] The network device sends a first indication message to the terminal device, the first indication message being used to instruct the model performance monitoring to be performed.

[0178] The terminal device determines a third model that needs to be monitored for model performance from the first model and / or the second model based on the first information. The first information includes information corresponding to the first channel state information (CSI) that the terminal device needs to report. The first CSI is used to monitor the performance of the third model.

[0179] The terminal device reports the first CSI to the network device;

[0180] The first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device. The second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device. The first part of the model is used to compress CSI. The second part of the model is used to recover CSI from the compressed CSI of the first part of the model. The third part of the model is used to compress and / or predict CSI. The fourth part of the model is used to recover and / or predict CSI from the compressed CSI of the third part of the model.

[0181] Fourthly, embodiments of this disclosure provide a terminal device, which may include at least one of a transceiver module and a processing module; wherein the terminal device may be used to execute an optional implementation of the first aspect.

[0182] Fifthly, embodiments of this disclosure provide a network device that may include at least one of a transceiver module and a processing module; wherein the network device may be used to perform an optional implementation of the second aspect.

[0183] In a sixth aspect, embodiments of this disclosure provide a terminal device, which may include one or more processors; wherein the communication device may be used to execute optional implementations such as the first aspect or the second aspect.

[0184] In a seventh aspect, embodiments of this disclosure provide a communication system that may include: a terminal device and a network device; wherein the terminal device is configured to perform the method described in the optional implementation of the first aspect, and the network device is configured to perform the method described in the optional implementation of the second aspect.

[0185] Eighthly, embodiments of this disclosure provide a storage medium storing instructions that, when executed on a communication device, cause the communication device to perform the method as described in an optional implementation of the first or second aspect.

[0186] In a ninth aspect, embodiments of this disclosure provide a program product that, when executed by a communication device, causes the communication device to perform the method as described in an optional implementation of the first or second aspect.

[0187] In a tenth aspect, embodiments of this disclosure provide a computer program that, when run on a computer, causes the computer to perform the methods described in an optional implementation of the first or second aspect.

[0188] Eleventhly, embodiments of this disclosure provide a chip or chip system. The chip or chip system includes processing circuitry configured to perform the methods described in optional implementations of the first or second aspect.

[0189] It is understood that the aforementioned terminal devices, network devices, communication devices, communication systems, storage media, program products, computer programs, chips, or chip systems can all be used to execute the methods proposed in the embodiments of this disclosure. Therefore, the beneficial effects that can be achieved can be referred to the beneficial effects in the corresponding methods, and will not be repeated here.

[0190] This disclosure provides a model performance monitoring method, device, and storage medium. In some embodiments, the terms "model performance monitoring method" and "information processing method," "communication method," etc., can be used interchangeably; the terms "model performance monitoring device" and "information processing device," "communication device," "communication equipment," etc., can be used interchangeably; and the terms "model performance monitoring system," "communication system," etc., can be used interchangeably.

[0191] This disclosure is not exhaustive, but merely illustrative of some embodiments, and is not intended to limit the scope of protection of this disclosure. Unless otherwise specified, each step in a particular embodiment can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment can be arbitrarily interchanged. Furthermore, the optional implementation methods in a particular embodiment can be arbitrarily combined; moreover, the embodiments can be arbitrarily combined, for example, some or all steps of different embodiments can be arbitrarily combined, and a particular embodiment can be arbitrarily combined with the optional implementation methods of other embodiments.

[0192] In each of the disclosed embodiments, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions of the embodiments are consistent and can be referenced by each other. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0193] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the scope of this disclosure.

[0194] In this embodiment of the disclosure, unless otherwise stated, elements expressed in the singular form, such as "a," "an," "the," "the," "the," "the," "the," "the," "this," etc., can mean "one and only one," or "one or more," "at least one," etc. For example, when using articles such as "a," "an," "the," etc. in translation, the noun following the article can be understood as either a singular expression or a plural expression.

[0195] In some embodiments, "multiple" can refer to two or more.

[0196] In some embodiments, the terms “at least one of”, “one or more”, “a plurality of”, “multiple”, etc., may be used interchangeably.

[0197] In some embodiments, the notation "at least one of A and B", "A and / or B", "A in one case, B in another", "in response to one case A, in response to another case B", etc., may include the following technical solutions depending on the situation: in some embodiments, A (execute A regardless of B); in some embodiments, B (execute B regardless of A); in some embodiments, execution is selected from A and B (A and B are selectively executed); in some embodiments, A and B (both A and B are executed). The same applies when there are more branches such as A, B, C, etc.

[0198] In some embodiments, the notation "A or B" may include the following technical solutions, depending on the situation: in some embodiments, A (execution of A regardless of B); in some embodiments, B (execution of B regardless of A); in some embodiments, execution is selected from A and B (A and B are selectively executed). The same applies when there are more branches such as A, B, C, etc.

[0199] The prefixes "first," "second," etc., used in the embodiments of this disclosure are merely for distinguishing different descriptive objects and do not impose restrictions on the position, order, priority, quantity, or content of the descriptive objects. The description of the descriptive objects is found in the claims or the context of the embodiments, and the use of prefixes should not constitute unnecessary restrictions. For example, if the descriptive object is a "field," the ordinal numbers preceding "field" in "first field" and "second field" do not restrict the position or order of the "fields." "First" and "second" do not restrict whether the "fields" they modify are in the same message, nor do they restrict the order of "first field" and "second field." Similarly, if the descriptive object is a "level," the ordinal numbers preceding "level" in "first level" and "second level" do not restrict the priority between "levels." Furthermore, the number of descriptive objects is not limited by ordinal numbers and can be one or more. For example, in "first device," the number of "devices" can be one or more. Furthermore, the objects modified by different prefixes can be the same or different. For example, if the object being described is "device", then "first device" and "second device" can be the same device or different devices, and their types can be the same or different. Similarly, if the object being described is "information", then "first information" and "second information" can be the same information or different information, and their content can be the same or different.

[0200] In some embodiments, “including A,” “containing A,” “for indicating A,” and “carrying A” can be interpreted as directly carrying A or indirectly indicating A.

[0201] In some embodiments, the terms “in response to…”, “in response to determining…”, “in the case of…”, “when…”, “if…”, “if…”, etc., can be used interchangeably.

[0202] In some embodiments, the terms “greater than,” “greater than or equal to,” “not less than,” “more than,” “more than or equal to,” “not less than,” “higher than,” “higher than or equal to,” “not lower than,” and “above” can be used interchangeably, as can the terms “less than,” “less than or equal to,” “not greater than,” “less than,” “less than or equal to,” “not more than,” “lower than,” “lower than or equal to,” “not higher than,” and “below”.

[0203] In some embodiments, devices, etc., may be interpreted as physical or virtual, and their names are not limited to those described in the embodiments. Terms such as “device,” “equipment,” “circuit,” “network element,” “node,” “function,” “unit,” “section,” “system,” “network,” “chip,” “chip system,” “entity,” and “subject” are interchangeable.

[0204] In some embodiments, "network" can be interpreted as devices included in a network (e.g., access network devices, core network devices, etc.).

[0205] In some embodiments, the terms "Access Network Device (AN Device)," "Radio Access Network Device (RAN Device)," "Base Station (BS)," "Radio Base Station," "Fixed Station," "Node," "Access Point," "Transmission Point (TP)," "Reception Point (RP)," "Transmission / Reception Point (TRP)," "Panel," "Antenna Panel," "Antenna Array," "Cell," "Macro Cell," "Small Cell," "Femto Cell," "Pico Cell," "Sector," "Cell Group," "Serving Cell," "Carrier," "Component Carrier," and "Bandwidth Part (BWP)" can be used interchangeably.

[0206] In some embodiments, the terms "terminal", "terminal device", "user equipment (UE)", "user terminal", "mobile station (MS)", "mobile terminal (MT)", "subscriber station", "mobile unit", "subscriber unit", "wireless unit", "remote unit", "mobile device", "wireless device", "wireless communication device", "remote device", "mobile subscriber station", "access terminal", "mobile terminal", "wireless terminal", "remote terminal", "handset", "user agent", "mobile client", and "client" can be used interchangeably.

[0207] In some embodiments, access network devices, core network devices, or network devices can be replaced with terminals. For example, embodiments of this disclosure can also be applied to structures where communication between access network devices, core network devices, or network devices and terminals is replaced with communication between multiple terminals (e.g., device-to-device (D2D), vehicle-to-everything (V2X), etc.). In this case, the structure can also be configured such that the terminal has all or part of the functions of the access network device. Furthermore, terms such as "uplink" and "downlink" can be replaced with terms corresponding to communication between terminals (e.g., "sidelink"). For example, uplink channel, downlink channel, etc., can be replaced with sidelink channel or direct channel, and uplink link, downlink, etc., can be replaced with sidelink link or direct link.

[0208] In some embodiments, the terminal may be replaced by an access network device, a core network device, or a network device. In this case, the access network device, core network device, or network device may also be configured to have all or some of the functions of the terminal.

[0209] In some embodiments, the acquisition of data, information, etc., may comply with the laws and regulations of the country where the location is situated.

[0210] In some embodiments, data, information, etc., may be obtained with the user's consent.

[0211] Furthermore, each element, each row, or each column in the table of this disclosure can be implemented as an independent embodiment, and any combination of any element, any row, or any column can also be implemented as an independent embodiment.

[0212] Figure 1A is a schematic diagram of the architecture of a communication system according to an embodiment of the present disclosure. As shown in Figure 1A, the communication system 100 may include a terminal device 101 and a network device 102.

[0213] In some embodiments, terminal device 101 may include at least one of the following: mobile phone, wearable device, Internet of Things device, car with communication function, smart car, tablet computer, computer with wireless transceiver function, virtual reality (VR) terminal device, augmented reality (AR) terminal device, wireless terminal device in industrial control, wireless terminal device in self-driving, wireless terminal device in remote medical surgery, wireless terminal device in smart grid, wireless terminal device in transportation safety, wireless terminal device in smart city, and wireless terminal device in smart home, but is not limited thereto.

[0214] In some embodiments, network device 102 may include at least one of access network device and core network device.

[0215] In some embodiments, the access network device may be a node or device that connects a terminal device to a wireless network. The access network device may include, but is not limited to, at least one of the following in a 5G communication system: evolved Node B (eNB), next-generation eNB (ng-eNB), next-generation Node B (gNB), node B (NB), home node B (HNB), home evolved node B (HeNB), radio backhaul device, radio network controller (RNC), base station controller (BSC), base transceiver station (BTS), base band unit (BBU), mobile switching center, base station in a 6G communication system, open RAN, cloud RAN, base station in other communication systems, and access node in a Wi-Fi system.

[0216] In some embodiments, the technical solutions of this disclosure can be applied to the Open RAN architecture. In this case, the interfaces between or within access network devices involved in the embodiments of this disclosure can be transformed into internal interfaces of Open RAN. The processes and information interactions between these internal interfaces can be implemented by software or programs.

[0217] In some embodiments, the access network device may be composed of a central unit (CU) and a distributed unit (DU). The CU may also be called a control unit. The CU-DU structure can separate the protocol layer of the access network device. Some protocol layer functions are centrally controlled by the CU, while the remaining part or all protocol layer functions are distributed in the DU and centrally controlled by the CU. However, this is not the only possibility.

[0218] In some embodiments, the core network equipment may be a single device, multiple devices, or a group of devices. The core network may include at least one of the following: Evolved Packet Core (EPC), 5G Core Network (5GCN), and Next Generation Core (NGC).

[0219] It is understood that the communication system described in this disclosure is for the purpose of more clearly illustrating the technical solutions of this disclosure, and does not constitute a limitation on the technical solutions proposed in this disclosure. As those skilled in the art will know, with the evolution of system architecture and the emergence of new business scenarios, the technical solutions proposed in this disclosure are also applicable to similar technical problems.

[0220] The following embodiments of this disclosure can be applied to the communication system 100 shown in FIG1A, or to some of the main bodies, but are not limited thereto. The main bodies shown in FIG1A are examples. The communication system may include all or some of the main bodies in FIG1A, or it may include other main bodies outside of FIG1A. The number and form of each main body are arbitrary. Each main body may be physical or virtual. The connection relationship between the main bodies is an example. The main bodies may not be connected or may be connected. The connection can be in any way, it can be a direct connection or an indirect connection, it can be a wired connection or a wireless connection.

[0221] The embodiments disclosed herein can be applied to Long Term Evolution (LTE), LTE-Advanced (LTE-A), LTE-Beyond (LTE-B), SUPER 3G, IMT-Advanced, 4th generation mobile communication system (4G), 5th generation mobile communication system (5G), 5G new radio (NR), Future Radio Access (FRA), New-Radio Access Technology (RAT), New Radio (NR), New Radio Access (NX), Future generation radio access (FX), Global System for Mobile communications (GSM), CDMA2000, Ultra Mobile Broadband (UMB), IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), and IEEE 802.20, Ultra-Wideband (UWB), Bluetooth (a registered trademark), Public Land Mobile Network (PLMN) networks, Device-to-Device (D2D) systems, Machine-to-Machine (M2M) systems, Internet of Things (IoT) systems, Vehicle-to-Everything (V2X) systems, systems utilizing other communication methods, and next-generation systems built upon them, etc. Furthermore, multiple systems can be combined (e.g., a combination of LTE or LTE-A with 5G).

[0222] In some embodiments of this disclosure, AI technology can be used to reduce terminal feedback overhead or improve CSI feedback accuracy. Furthermore, 3GPP standardization research has been conducted on bilateral AI / Machine Learning (ML) models based on a terminal-side CSI generation model and a network-side CSI recovery model to achieve compressed feedback and recovery of CSI, respectively. Figure 1B is a schematic diagram illustrating CSI compressed feedback and recovery based on a bilateral AI / ML model according to an embodiment of this disclosure. As shown in Figure 1B, the UE side compresses the downlink channel information H using the CSI generation model (defined as encoder) and quantizes it into a binary bit stream before sending it to the gNB. The gNB side recovers H', which is approximately the same as the original downlink information, using the CSI recovery model (defined as decoder).

[0223] In some embodiments, channel temporal correlation can be used to improve system performance, and the specific implementation methods may include the following two:

[0224] Method 1: When compressing the CSI at the current time t, historical CSI before time t is used.

[0225] Method 2: First, predict the CSI at future times based on the estimated historical channel information, and then compress and report the predicted CSI using an AI model, or use an AI model to predict the CSI at future times and compress and report it.

[0226] For method 2, CSI prediction and CSI compression can be implemented using a corresponding UE-side one-sided CSI prediction AI / ML model and a two-sided CSI compression model, respectively, or a single two-sided model can be used to implement joint CSI prediction and CSI compression. If a single two-sided CSI prediction and compression model is used to implement joint CSI prediction and compression, this model still includes a CSI prediction and compression part model deployed on the UE side and a recovery prediction CSI part model deployed on the NW side. Figure 1C is a schematic diagram of CSI compression feedback based on a two-sided CSI prediction and compression model according to an embodiment of this disclosure. As shown in Figure 1C, the input information of the AI / ML-based CSI prediction and compression part model is the channel information at a historical time, and the output information is the compressed CSI or the binary bit stream obtained by quantizing the compressed CSI. The UE reports the binary bit rate obtained by quantizing and compressing the CSI to the NW side, and uses it as the input of the recovery prediction CSI part model on the NW side. After passing through the AI / ML-based recovery prediction CSI part model, the recovered channel information for M>=1 future time periods is obtained.

[0227] In some embodiments, the CSI generation part of the bilateral CSI compressed AI / ML model is also located on the UE side, and its input information is the channel information at the current measurement time. The CSI recovery part of the bilateral CSI compressed AI / ML model is located on the NW side, and its output information is the channel information at the current measurement time. Compared with the aforementioned bilateral CSI prediction compressed model, the input information of the generation part model on the UE side is different, and the output information of the recovery part model on the NW side is also different. Therefore, the bilateral models trained in these two directions are also different, and different models may use the same or different performance monitoring methods. If different models use independent model monitoring methods, an independent set of parameter configurations is introduced for each monitoring method, which not only increases the complexity of model lifecycle management (LCM) but also increases signaling configuration overhead. When the UE and gNB deploy multiple models as described in Figures 1B and 1C above, how to design a unified bilateral AI / ML model performance monitoring method to reduce LCM complexity or reduce related configuration signaling overhead is an urgent problem to be solved.

[0228] In some embodiments, bilateral AI / ML model performance monitoring may include the following methods:

[0229] Method 1: Monitoring model performance on the NW side:

[0230] Based on the target CSI reported by the UE (the encoder input in Figure 1B), the target CSI can be indicated by the eType II codebook or a higher-precision eType II codebook.

[0231] Method 2: UE-side monitoring model performance:

[0232] Monitor model performance based on decoder output information deployed on the UE side;

[0233] Note: The decoder deployed on the UE side can be the same as or different from the decoder deployed on the NW side, or it can be a reference model provided by the NW, or a proxy model developed on the UE side.

[0234] Monitoring model performance based on estimated intermediate KPIs;

[0235] Model monitoring is based on the estimation of monitoring results other than intermediate KPIs;

[0236] Based on the precoded reference signal transmitted from the NW side, such as the precoded CSI-RS or DMRS, the precoding is obtained from the decoder output information on the NW side;

[0237] Monitoring is performed based on the decoder output information on the NW side. The decoder output information is indicated to the UE by the NW through the eType II codebook or a high-precision eType II codebook.

[0238] In some embodiments, the current performance monitoring method is only applicable to CSI compression feedback that does not include CSI prediction functionality, and not applicable to AI / ML model monitoring with joint CSI prediction compression as described in the background art.

[0239] Figure 2A is an interactive schematic diagram of a model performance monitoring method according to an embodiment of the present disclosure. This method can be executed by the aforementioned communication system. As shown in Figure 2A, the method may include:

[0240] Step S2101: The network device sends the first instruction information to the terminal device.

[0241] In some embodiments, the terminal device may receive first indication information. For example, the terminal device may receive first indication information sent by a network device. As another example, the terminal device may also receive first indication information sent by other entities.

[0242] In some embodiments, the first indication information may be used to instruct model performance monitoring to be performed.

[0243] In some embodiments, the first indication information may be used to instruct the terminal device to report CSI at M (M>=1) times.

[0244] In some embodiments, the CSI data reported by the terminal device at M time points can be used for model performance monitoring.

[0245] Optionally, the first indication information can be used to indicate that the terminal device includes M CSIs in a single report.

[0246] Optionally, the first indication information can be used to indicate that the terminal device includes CSI at M times in multiple reports. For example, if M=4, the network device can indicate that the terminal device includes CSI at 2 times in the first report and CSI at 2 times in the second report.

[0247] In some embodiments, the network device may instruct the terminal device to report CSI at M time points via downlink signaling.

[0248] In some embodiments, the downlink signaling may include at least one of the following: Radio Resource Control (RRC), Media Access Control-Control Element (MAC-CE), and Downlink Control Information (DCI).

[0249] Step S2102: The terminal device determines the third model that needs to be monitored for model performance from the first model and / or the second model based on the first information.

[0250] In some embodiments, the first model may include a first part model deployed on a terminal device and a second part model deployed on a network device. The second model includes a third part model deployed on the terminal device and a fourth part model deployed on the network device. The first part model is used to compress CSI. The second part model is used to recover CSI from the compressed CSI of the first part model. The third part model is used to compress and / or predict CSI. The fourth part model is used to recover and / or predict CSI from the compressed CSI of the third part model.

[0251] In some embodiments, the terminal device may deploy both the first part model and the third part model simultaneously, and correspondingly, the network device may deploy both the second part model and the fourth part model simultaneously.

[0252] In some embodiments, the first information may include at least one of the following: the value of M, a first numerical value, and M time points, wherein the first numerical value is used to determine the M time points.

[0253] In some embodiments, the value of M indicates how many times CSI needs to be reported. For example, if M = 4, it means that the terminal device needs to report CSI at 4 times.

[0254] In some embodiments, M times represent which times of CSI need to be reported. For example, if the M times include time t0, time t1, time t2, and time t3, it means that the terminal device needs to upload the CSI at time t0, time t1, time t2, and time t3.

[0255] In some embodiments, the value of M can be determined in at least one of the following ways:

[0256] The terminal device determines the value of M;

[0257] The value of M in the network device configuration.

[0258] In some embodiments, the terminal device may predefine the value of M.

[0259] In some embodiments, the terminal device may determine the value of M in at least one of the following ways:

[0260] The model deployed on the terminal device does not include the fourth model, so the value of M is determined to be 1;

[0261] The model deployed on the terminal device includes this fourth model, and the number of times corresponding to the CSI that the fourth model can predict is used as the value of M.

[0262] In some embodiments, the fourth model can be a model for predicting CSI. For example, the fourth model can predict the CSI at future times based on the CSI at historical times.

[0263] In some embodiments, if the model deployed by the terminal device does not include the fourth model, it means that the terminal device cannot perform CSI prediction. In this case, the terminal device can only report the CSI measured at one time. That is, the terminal device can only report one CSI, i.e., M=1.

[0264] In some embodiments, if the model deployed by the terminal device includes the fourth model, indicating that the terminal device can perform CSI prediction, then the terminal device can report CSI measured at a single moment or CSI predicted at multiple moments by the fourth model. In other words, the terminal device can report CSI at multiple moments. Furthermore, the CSI reported by the terminal device at multiple moments can be CSI predicted by the fourth model. If the fourth model can predict CSI at N moments, then M takes the value N, meaning the terminal device can report CSI at N moments.

[0265] In some embodiments, the network device may configure the value of M and indicate the value of M to the terminal device.

[0266] In some embodiments, the method for determining the first value may include at least one of the following:

[0267] The terminal device determines this first value;

[0268] Configure the network device with this first value;

[0269] The first value is determined based on an initial value and the time interval between two adjacent moments, where the initial value and the time interval are configured or predefined by the network device.

[0270] The first value is predefined.

[0271] In some embodiments, each of the M time points can correspond to a first value, for example, the first value can be represented as l. i, i = 0, 1, 2, ..., M-1.

[0272] In some embodiments, the network device can configure the first value for the terminal device.

[0273] In some embodiments, the network device can configure the initial value and the time interval for the terminal device, and the terminal device can determine the first value based on the initial value and the time interval. For example, if the initial value is represented by l0, the time interval is represented by d, M = 4, and the four times are represented by t0, t1, t2, and t3, then the first value corresponding to time t0 can be l0, the first value corresponding to time t1 can be l0+d, the first value corresponding to time t2 can be l1+d, and the first value corresponding to time t3 can be l2+d.

[0274] In some embodiments, the first value being predefined may include at least one of the following:

[0275] This first value is predefined as the downlink time corresponding to the CSI reference resource;

[0276] The first value is predefined as the time when the fourth model deployed on the terminal device predicts CSI, and the fourth model is the model used to predict CSI;

[0277] The first value is predefined as the time when the network device sends the downlink pilot.

[0278] In some embodiments, the downlink time corresponding to the CSI reference resource can be used as the first value.

[0279] In some embodiments, the time of CSI predicted by the fourth model can be used as the first value. For example, if the fourth model predicts the CSI at times k1 to k4, then k1 to k4 can be used as the first value.

[0280] In some embodiments, the downlink pilot can be a Channel Status Information-Reference Signal (CSI-RS) resource.

[0281] In some embodiments, the time at which CSI-RS resources are sent to the network device can be used as the first value.

[0282] In some embodiments, the M time points can be determined in at least one of the following ways:

[0283] Based on the downlink time and first value corresponding to the CSI reference resource, M time points are determined;

[0284] Based on the CSI reporting time and the first value, determine M time points;

[0285] Based on the time corresponding to the first instruction information and the first value, determine the M times.

[0286] In some embodiments, the sum of the downlink time corresponding to the CSI reference resource and the first value can be used as M time points. For example, if M = 4, the downlink time point corresponding to the CSI reference resource is n. ref The first value is l i If i = 0, 1, 2, 3, then the four time points are n ref +l0、n ref +l1、n ref +l2、n ref +l3.

[0287] In some embodiments, the sum of the CSI reporting time and the first value can be used as M times. For example, if M = 4, the CSI reporting time is n, and the first value is l. i If i = 0, 1, 2, 3, then the four time points are n+l0, n+l1, n+l2, and n+l3, respectively.

[0288] In some embodiments, the time corresponding to the first indication information may include at least one of the following:

[0289] The moment when the network device sends the first indication information;

[0290] The moment when the terminal device receives the first instruction information.

[0291] For example, the sum of the time when the network device sends the first indication information and the first value can be used as M times. For instance, if M = 4, the CSI reporting time is n′, and the first value is l. i If i = 0, 1, 2, 3, then the four time points are n′+l0, n′+l1, n′+l2, and n′+l3, respectively.

[0292] In some embodiments, the network device can configure the value of M and the first value to the terminal device via RRC signaling, or configure M time points.

[0293] In some embodiments, the value of M, the first value, and the M time points can be configured according to the model for which performance monitoring is required.

[0294] For example, if performance monitoring of the second model is required, M=1 or M=2 can be configured, and the first value l0, l1=l0+T can be configured. Figure 2B is a schematic diagram of CSI reporting according to an embodiment of this disclosure. As shown in Figure 2B, n represents the time when CSI reporting occurs. The value of l0 configured by the network device makes l0+n satisfy an integer multiple of T and is the same as one of the times when the network device periodically sends CSI-RS.

[0295] For example, if performance monitoring of the first model is required, M=1 can be configured, and the first value l0 can be configured, l0<0.

[0296] In some embodiments, after receiving the first indication information sent by the network device, the terminal device can determine a third model that needs to be monitored for model performance from the first model and / or the second model.

[0297] In some embodiments, a third model requiring model performance monitoring can be determined through at least one of the following methods:

[0298] If the value of M is greater than 1, the third model is determined to be the second model;

[0299] The value of M is equal to 1. This first value is less than 0 and the M times determined by this first value are the same as one of the times when the network device periodically sends CSI-RS resources. The third model is determined to be the first model.

[0300] The value of M is equal to 1. This first value is less than 0, and the M times determined by this first value are different from each transmission time when the network device periodically sends CSI-RS resources. Therefore, the third model is determined to be the second model.

[0301] Each of the M time points is the same as one of the time points when the network device periodically sends CSI-RS resources, thus determining the third model as the first model;

[0302] The M time points are different from each transmission time of the CSI-RS resources periodically sent by the network device, so the third model is determined to be the second model.

[0303] In some embodiments, if the value of M is greater than 1, it means that the terminal device needs to report CSI at multiple times. In this case, the terminal device needs to predict the CSI at multiple times through the model, that is, the terminal device needs to use the second model. In this case, the performance of the second model can be monitored, that is, the third model is determined to be the second model.

[0304] In some embodiments, if the value of M is equal to 1, it can be determined whether the terminal device needs to perform CSI prediction based on the time when the terminal device needs to report the first CSI. For example, if the configured M times are determined based on the CSI reporting time and the first value, and the CSI reporting time is time n, then the terminal device needs to report the first CSI at time n+10. If time n+10 coincides with one of the transmission times of the network device periodically sending CSI-RS resources, then the terminal device can measure the CSI-RS resources received at time n+10 and report the first CSI based on the measurement result. In this case, the terminal device does not need to predict CSI through a model; it can perform performance monitoring on the first model, i.e., determine the third model as the first model.

[0305] In some embodiments, if the value of M is equal to 1, it can be determined whether the terminal device needs to perform CSI prediction based on the time when the terminal device needs to report the first CSI. For example, if the configured M times are determined based on the CSI reporting time and the first value, and the CSI reporting time is time n, then the terminal device needs to report the first CSI at time n+10. If time n+10 is different from each transmission time of the network device's periodic transmission of CSI-RS resources, then the terminal device cannot directly measure the CSI at time n+10 that needs to be reported, and needs to predict the CSI at time n+10 through a model. In this case, the performance of the second model can be monitored, i.e., the third model can be determined as the second model.

[0306] In some embodiments, the third model can be determined directly based on M time points.

[0307] For example, if each of the M time points coincides with one of the time points when the network device periodically sends CSI-RS resources, it means that the CSI at each of the M time points can be measured based on the corresponding CSI-RS resources, and the terminal device does not need to perform CSI prediction through a model. In this case, the performance of the first model can be monitored, i.e., the third model can be determined as the first model.

[0308] For example, if the M time points are different from each transmission time of the CSI-RS resources periodically sent by the network device, it means that the CSI at any of the M time points cannot be measured based on the CSI-RS resources at the corresponding time points, and the terminal device needs to perform CSI prediction using a model. In this case, the performance of the second model can be monitored, that is, the third model can be determined as the second model.

[0309] Step S2103: The terminal device determines the compressed CSI based on the third model.

[0310] In some embodiments, determining compressed CSI based on a third model may include at least one of the following:

[0311] The third model is the first model, and it is determined that the compressed CSI is the CSI at M time points output by the first part of the model;

[0312] The third model is the same as the second model, and the compressed CSI is determined to be the CSI at M time points output by the third part of the model.

[0313] In some embodiments, the terminal device can receive CSI-RS periodically transmitted by the network device, estimate downlink channel information based on the received CSI-RS, perform eigenvalue decomposition on the estimated downlink channel information, and obtain the eigenvector corresponding to the largest eigenvalue. If the third model is the first model, the eigenvector is input into the first part of the model to obtain the compressed CSI; if the third model is the second model, the eigenvector is input into the third part of the model to obtain the compressed CSI.

[0314] Step S2104: The terminal device reports the compressed CSI to the network device.

[0315] In some embodiments, the network device may receive compressed CSI. For example, the network device may receive compressed CSI sent by a terminal device. As another example, the network device may also receive compressed CSI sent by other entities.

[0316] In some embodiments, the terminal device may report the compressed CSI at M time points output by the first part model or the third part model to the network device.

[0317] In some embodiments, the timing of the terminal device reporting compressed CSI can be determined by at least one of the following methods:

[0318] Network devices provide instructions via downlink signaling;

[0319] Predefined;

[0320] The time when the terminal device compresses and reports CSI is taken as the time when the compressed CSI is reported.

[0321] In some embodiments, the network device may instruct the terminal device when to report compressed CSI via downlink signaling.

[0322] In some embodiments, the timing of reporting compressed CSI can be predefined; for example, the timing of reporting compressed CSI can be agreed upon through a protocol.

[0323] In some embodiments, the time when the terminal device reports the CSI can be taken as the time when the first CSI is reported.

[0324] In some embodiments, the time when the terminal device reports CSI can be understood as the time when the terminal device reports the measured CSI, or the time when the terminal device reports the CSI predicted by the fourth model.

[0325] In some embodiments, the terminal device may report the compressed CSI at the reporting time determined above.

[0326] For example, continuing with Figure 2B, if the third model is the second model, the terminal device can obtain the measured CSI (also called the target CSI) based on the received CSI-RS at times n+l0 (n3 in Figure 2B) and n+l1 (n4 in Figure 2B), and then report it at times n1 (n1 = n + δ1) and n2 (n2 = n + δ2) as indicated by the network device. δ1 and δ2 are the first values, which can be predefined or configured by the network device.

[0327] If the third model is the first model, the terminal device can report the measured CSI at time n+l0 (also known as the target CSI) at time n or report the measured CSI at the time closest to the time of the CSI reference resource.

[0328] In some embodiments, the terminal device reporting the compressed CSI to the network device may include at least one of the following:

[0329] Report all compressed CSIs to the network device at the same time;

[0330] Compressed CSI for each of the M time points is reported to the network device in the second order.

[0331] In some embodiments, the second order may be a protocol agreement or a network device instruction. For example, the second order may be a front-to-back order or a back-to-front order.

[0332] In some embodiments, after acquiring compressed CSI at M time points, the terminal device can combine all the compressed CSI in a second order and send the combined compressed CSI to the network device at the same time. For example, after measuring the compressed CSI at time t1 and time t2, the terminal device can combine the two compressed CSI in the order that the compressed CSI at time t1 comes first and the compressed CSI at time t2 comes last, and send the combined compressed CSI at time t1 and time t2 to the network device at the same time.

[0333] In some embodiments, the terminal device may also send the compressed CSI at each time step to the network device sequentially. For example, the terminal device may first send the compressed CSI at time t1 to the network device, and then send the compressed CSI at time t2 to the network device.

[0334] In some embodiments, the terminal device may transmit compressed CSI and / or measurement CSI on the Physical Uplink Control Channel (PUCCH) or the Physical Uplink Shared Channel (PUSCH).

[0335] Step S2105: The network device inputs the compressed CSI into the second part model and / or the fourth part model to obtain the second CSI.

[0336] In some embodiments, after receiving the compressed CSI sent by the terminal device, the network device can input the compressed CSI into the two-part model or the fourth-part model to obtain the second CSI.

[0337] For example, if the third model is the first model, the network device can perform dequantization on the compressed CSI and input the dequantized information into the second part of the model to obtain the CSI at M time points recovered by the second part of the model, which is the second CSI.

[0338] For example, if the third model is the second model, the network device can perform dequantization on the compressed CSI and input the dequantized information into the fourth model to obtain the CSI at M time points recovered by the fourth model, which is the second CSI.

[0339] In some embodiments, the second CSI may also be referred to as the recovery CSI.

[0340] In some embodiments, after obtaining the second CSI, the network device may execute steps S2106a and S2107a, or execute steps S2106b and S2107b.

[0341] Step S2106a: The network device sends a second CSI to the terminal device.

[0342] In some embodiments, the terminal device may receive a second CSI. For example, the terminal device may receive a second CSI sent by a network device. As another example, the terminal device may also receive a second CSI sent by another entity.

[0343] In some embodiments, after obtaining the second CSI, the network device may send the second CSI to the terminal device.

[0344] In some embodiments, the network device sending the second CSI to the terminal device may include at least one of the following:

[0345] Send all second CSIs to the terminal device at the same time;

[0346] The CSIs for each moment in the second CSI are sent to the terminal device in the first order.

[0347] In some embodiments, the first order may be a protocol agreement or a network device instruction. For example, the first order may be a front-to-back order or a back-to-front order.

[0348] In some embodiments, after acquiring the recovered CSIs at M time points, the network device can combine the multiple recovered CSIs in a first order and send the combined recovered CSIs to the terminal device. Taking M=2, with the M time points being t1 and t2, as an example, the network device can combine the recovered CSIs at time t1 first and t2 last, sending the recovered CSIs at times t1 and t2, i.e., the second CSI, to the terminal device. After receiving the second CSI sent by the network device, the terminal device can determine the recovered CSIs at different time points according to the first order.

[0349] In some embodiments, the network device may also send the recovery CSI at each of the M time points in the first order. For example, the network device may first send the recovery CSI at time t1 to the terminal device, and then send the recovery CSI at time t2 to the terminal device.

[0350] In some embodiments, the network device may transmit the second CSI via the Physical Downlink Control Channel (PDCCH) or the Physical Downlink Shared Channel (PDSCH).

[0351] Step S2107a: The terminal device determines the model performance criteria based on the measured CSI and the second CSI.

[0352] In some embodiments, the measured CSI may be the CSI at M times obtained by the terminal device based on the CSI-RS sent by the network device, or the CSI at M times predicted by the terminal device through the fourth model.

[0353] In some embodiments, the model performance criteria may include at least one of the following: Square Generalized Cosine Similarity (SGCS) and Normalized Mean Square Error (NMSE).

[0354] In some embodiments, the terminal device can calculate the SGCS based on the measured CSI and the second CSI using the following formula:

[0355] Where K1 is SGCS, i represents the i-th frequency domain unit, such as a sub-band, N3 represents the number of frequency domain units, and v i,t Let e ​​represent the measurement CSI at time t in the i-th frequency domain cell. i,t This represents the second CSI at time t in the i-th frequency domain unit.

[0356] It should be noted that the specific method for determining the performance of the third model based on NMSE can be found in the description of existing protocols, and will not be repeated here.

[0357] Step S2106b: The network device sends a pilot signal with pre-coded shape to the terminal device.

[0358] In some embodiments, the terminal device may receive pilot signals that have been precoded and shaped. For example, the terminal device may receive pilot signals that have been precoded and shaped sent by a network device. As another example, the terminal device may also receive pilot signals that have been precoded and shaped sent by other entities.

[0359] In some embodiments, the precoding may be the CSI at M time points recovered by the network device based on the compressed CSI through the second part model and / or the fourth part model, i.e., the second CSI.

[0360] In some embodiments, the network device sending precoded and shaped pilot signals to the terminal device may include at least one of the following:

[0361] At the same time, M precoded and shaped pilot signals are sent to the terminal device;

[0362] M precoded and shaped pilot signals are sent to the terminal device in the first order.

[0363] In some embodiments, each precode may correspond to a recovery CSI at one time point, and M precodes may correspond to M recovery CSIs at M time points, which is the second CSI.

[0364] It should be noted that the specific implementation methods of the above two methods for transmitting precoded and shaped pilot signals can be referred to the record of transmitting the second CSI in step S2106a, and will not be repeated here.

[0365] Step S2107b: The pilot signal pre-coded and shaped by the terminal device determines the performance of the third model.

[0366] In some embodiments, the terminal device may refer to existing protocols to determine the performance of the third model based on the precoded and shaped pilot signal, which will not be elaborated here.

[0367] Using the above method, the terminal device can determine the third model requiring performance monitoring from the first and / or second models based on the information corresponding to the first CSI reported as needed, and report a compressed CSI to the network device. It then receives a second CSI sent by the network device based on the compressed CSI, and determines the performance of the third model based on the measured CSI and the second CSI. This allows for performance monitoring of both the first and second models using a unified parameter configuration, reducing the overhead of configuration signaling for model performance monitoring and lowering the LCM complexity of the models.

[0368] The method involved in the embodiments of this disclosure may include at least one of the above steps S2101 to S2107a or steps S2101 to S2107b. For example, step S2101 can be implemented as an independent embodiment, step S2102 can be implemented as an independent embodiment, step S2103 can be implemented as an independent embodiment, step S2104 can be implemented as an independent embodiment, step S2105 can be implemented as an independent embodiment, step S2106a can be implemented as an independent embodiment, step S2106b can be implemented as an independent embodiment, step S2102 + step S2103 can be implemented as an independent embodiment, step S2103 + step S2104 can be implemented as an independent embodiment, step S2104 + step S2105 can be implemented as an independent embodiment, step S2105 + step S2106 can be implemented as an independent embodiment, step S2106a + step S2107a can be implemented as an independent embodiment, and step S2106b + step S2107b can be implemented as an independent embodiment, but are not limited thereto.

[0369] In some embodiments, the order of any two steps S2101 to S2107a or steps S2101 to S2107b can be interchanged or performed simultaneously.

[0370] In some embodiments, steps S2101 to S2107a or steps S2101 to S2107b are optional, and one or more of these steps may be omitted or substituted in different embodiments. For example, step S2101 may be omitted, and steps S2106b and S2107b may be omitted.

[0371] In some embodiments, other alternative implementations may be described before or after the specification corresponding to FIG2A.

[0372] Figure 2C is an interactive schematic diagram illustrating a model performance monitoring method according to an embodiment of the present disclosure. This method can be executed by the aforementioned communication system. As shown in Figure 2C, the method may include:

[0373] Step S2301: The network device sends the first instruction information to the terminal device.

[0374] The optional implementation of step S2301 can be found in the optional implementation of step S2101 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0375] Step S2302: The terminal device determines the third model that needs to be monitored for model performance from the first model and / or the second model based on the first information.

[0376] The optional implementation of step S2302 can be found in the optional implementation of step S2102 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0377] Step S2303: The terminal device determines the compressed CSI based on the third model.

[0378] The optional implementation of step S2303 can be found in the optional implementation of step S2103 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0379] Step S2304: The terminal device reports the measurement CSI and compressed CSI to the network device.

[0380] In some embodiments, the network device may receive both measurement CSI and compressed CSI. For example, the network device may receive measurement CSI and compressed CSI sent by a terminal device. As another example, the network device may also receive measurement CSI and compressed CSI sent by other entities.

[0381] In some embodiments, the terminal device may report both the measured CSI and the compressed CSI to the network device together, or it may report the measured CSI and the compressed CSI to the network device separately.

[0382] It should be noted that the specific methods for reporting measurement CSI and compressed CSI on the terminal device can be found in step S2104, which will not be repeated here.

[0383] Step S2305: The network device inputs the compressed CSI into the second part of the model and / or the fourth part of the model to obtain the second CSI.

[0384] The optional implementation of step S2305 can be found in the optional implementation of step S2105 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0385] Step S2306: The network device determines the model performance criteria based on the measured CSI and the second CSI.

[0386] In some embodiments, the model performance criteria may include at least one of the following: SGCS, NMSE.

[0387] In some embodiments, the network device can calculate the SGCS based on the measured CSI and the second CSI using formula (1). The specific calculation method can be found in step S2107a, and will not be repeated here.

[0388] Using the above method, the terminal device can determine the third model requiring performance monitoring from the first and / or second models based on the information corresponding to the first CSI reported as needed, and report the measurement CSI and compressed CSI to the network device. The network device can then recover the second CSI from the compressed CSI and determine the performance of the third model based on the measurement CSI and the second CSI. This allows for performance monitoring of both the first and second models using a unified parameter configuration, reducing the overhead of configuration signaling for model performance monitoring and lowering the LCM complexity of the models.

[0389] The methods involved in the embodiments of this disclosure may include at least one of the steps S2301 to S2306 described above. For example, step S2301 may be implemented as a standalone embodiment, step S2302 may be implemented as a standalone embodiment, step S2303 may be implemented as a standalone embodiment, and step S2302 + step S2303 may be implemented as a standalone embodiment, but are not limited thereto.

[0390] In some embodiments, the order of any two steps in steps S2301 to S2306 can be interchanged or they can be performed simultaneously.

[0391] In some embodiments, steps S2301 to S2303 are all optional steps. For example, steps S2301 and S2302 are optional, and one or more of these steps may be omitted or substituted in different embodiments.

[0392] In some embodiments, the names of information, etc., are not limited to the names described in the embodiments. Terms such as "information", "message", "signal", "signaling", "report", "configuration", "indication", "instruction", "command", "channel", "parameter", "domain", "field", "symbol", "symbol", "codebook", "codeword", "codepoint", "bit", "data", "program", and "chip" can be used interchangeably.

[0393] In some embodiments, “get,” “obtain,” “receive,” “transmit,” “bidirectional transmission,” and “send and / or receive” can be used interchangeably and can be interpreted as receiving from other entities, obtaining from protocols, obtaining from higher layers, obtaining through self-processing, or autonomous implementation, among other meanings.

[0394] In some embodiments, terms such as “send,” “transmit,” “report,” “distribute,” “transfer,” “bidirectional transmission,” “send and / or receive” can be used interchangeably.

[0395] In some embodiments, terms such as "certain," "preset," "default," "set," "indicated," "a certain," "any," and "first" can be used interchangeably. "Certain A," "preset A," "default A," "set A," "indicated A," "a certain A," "any A," and "first A" can be interpreted as A pre-defined in a protocol or the like, or as A obtained through setting, configuration, or instruction, or as specific A, a certain A, any A, or first A, but are not limited thereto.

[0396] Figure 3A is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 3A, the present disclosure relates to a model performance monitoring method, which can be executed by a terminal device. The method may include:

[0397] Step S3101: Receive the first instruction information.

[0398] The optional implementation of step S3101 can be found in the optional implementation of step S2101 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0399] Step S3102: Based on the first information, determine the third model that needs to be monitored for model performance from the first model and / or the second model.

[0400] The optional implementation of step S3102 can be found in the optional implementation of step S2102 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0401] Step S3103: Determine the compressed CSI based on the third model.

[0402] The optional implementation of step S3103 can be found in the optional implementation of step S2103 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0403] Step S3104: Report the compressed CSI.

[0404] The optional implementation of step S3104 can be found in the optional implementation of step S2105 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0405] Step S3105: Receive the second CSI.

[0406] The optional implementation of step S3105 can be found in the optional implementation of step S2106 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0407] Step S3106: Determine the model performance criteria based on the measured CSI and the second CSI.

[0408] The optional implementation of step S3106 can be found in the optional implementation of step S2107a in Figure 2A, and other related parts in the embodiment involved in Figure 2A, which will not be repeated here.

[0409] The methods involved in the embodiments of this disclosure may include at least one of the steps S3101 to S3106 described above. For example, step S3101 may be implemented as an independent embodiment, step S3102 may be implemented as an independent embodiment, step S3103 may be implemented as an independent embodiment, step S3104 may be implemented as an independent embodiment, step S3105 may be implemented as an independent embodiment, step S3106 may be implemented as an independent embodiment, step S3102 + step S3103 may be implemented as an independent embodiment, step S3103 + step S3104 may be implemented as an independent embodiment, and step S3105 + step S3106 may be implemented as an independent embodiment, but are not limited thereto.

[0410] In some embodiments, the order of any two steps in steps S3101 to S3106 can be interchanged or they can be performed simultaneously.

[0411] In some embodiments, steps S3101 to S3106 are optional, and one or more of these steps may be omitted or substituted in different embodiments. For example, step S3101.

[0412] Figure 3B is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 3B, the present disclosure relates to a model performance monitoring method, which can be executed by a terminal device. The method may include:

[0413] Step S3201: Receive the first instruction information.

[0414] The optional implementation of step S3201 can be found in the optional implementation of step S2101 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0415] Step S3202: Based on the first information, determine the third model that needs to be monitored for model performance from the first model and / or the second model.

[0416] The optional implementation of step S3202 can be found in the optional implementation of step S2102 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0417] Step S3203: Determine the compressed CSI based on the third model.

[0418] The optional implementation of step S3203 can be found in the optional implementation of step S2103 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0419] Step S3204: Report the compressed CSI.

[0420] The optional implementation of step S3204 can be found in the optional implementation of step S2105 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0421] Step S3205: Receive the pilot signal that has been precoded and shaped.

[0422] The optional implementation of step S3205 can be found in the optional implementation of step S2106b in Figure 2A, and other related parts in the embodiment involved in Figure 2A, which will not be repeated here.

[0423] Step S3206: Determine the performance of the third model based on the precoded and shaped pilot signal.

[0424] The optional implementation of step S3206 can be found in the optional implementation of step S2107b in Figure 2A, and other related parts in the embodiment involved in Figure 2A, which will not be repeated here.

[0425] The methods involved in the embodiments of this disclosure may include at least one of the steps S3201 to S3206 described above. For example, step S3201 may be implemented as an independent embodiment, step S3202 may be implemented as an independent embodiment, step S3203 may be implemented as an independent embodiment, step S3204 may be implemented as an independent embodiment, step S3205 may be implemented as an independent embodiment, step S3206 may be implemented as an independent embodiment, step S3202 + step S3203 may be implemented as an independent embodiment, step S3203 + step S3204 may be implemented as an independent embodiment, and step S3205 + step S3206 may be implemented as an independent embodiment, but are not limited thereto.

[0426] In some embodiments, the order of any two steps in steps S3201 to S3206 can be interchanged or they can be performed simultaneously.

[0427] In some embodiments, steps S3201 to S3206 are optional, and one or more of these steps may be omitted or substituted in different embodiments. For example, step S3201.

[0428] Figure 3C is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 3C, the present disclosure relates to a model performance monitoring method, which can be executed by a terminal device. The method may include:

[0429] Step S3301: Receive the first instruction information.

[0430] The optional implementation of step S3301 can be found in the optional implementation of step S2301 in Figure 2C, and other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0431] Step S3302: Based on the first information, determine the third model that needs to be monitored for model performance from the first model and / or the second model.

[0432] The optional implementation of step S3302 can be found in the optional implementation of step S2302 in Figure 2C, as well as other related parts in the embodiment involved in Figure 2C, which will not be repeated here.

[0433] Step S3303: Determine the compressed CSI based on the third model.

[0434] The optional implementation of step S3303 can be found in the optional implementation of step S2303 in Figure 2C, as well as other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0435] Step S3304: Report the measurement CSI and compressed CSI.

[0436] The optional implementation of step S3304 can be found in the optional implementation of step S2304 in Figure 2C, as well as other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0437] The methods involved in the embodiments of this disclosure may include at least one of the steps S3301 to S3304 described above. For example, step S3301 may be implemented as a standalone embodiment, step S3302 may be implemented as a standalone embodiment, step S3303 may be implemented as a standalone embodiment, and step S3302 + step S3303 may be implemented as a standalone embodiment, but are not limited thereto.

[0438] In some embodiments, the order of any two steps in steps S3301 to S3304 can be interchanged or they can be performed simultaneously.

[0439] In some embodiments, steps S3301 to S3304 are all optional. For example, step S3301 is optional, and one or more of these steps may be omitted or substituted in different embodiments.

[0440] Figure 3D is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 3D, the present disclosure relates to a model performance monitoring method, which can be executed by a terminal device. The method may include:

[0441] Step S3401: Receive the first instruction information.

[0442] The optional implementation of step S3401 can be found in the optional implementation of step S2101 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0443] Step S3402: Based on the first information, determine the third model that needs to be monitored for model performance from the first model and / or the second model.

[0444] The optional implementation of step S3402 can be found in the optional implementation of step S2102 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0445] Step S3403: Report the first CSI according to the third model.

[0446] The optional implementation of step S3403 can be found in the optional implementations of steps S2103 to S2104 in Figure 2A, steps S2303 to S2304 in Figure 2C, and other related parts in the embodiments involved in Figures 2A and 2C, which will not be repeated here.

[0447] In some embodiments, the first CSI may include a measured CSI and / or a compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI may be the CSI at M times output by the first part model and / or the third part model, where M is a positive integer.

[0448] In some embodiments, the first CSI includes measured CSI and / or compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI is the CSI at the M times output by the first part model and / or the third part model, where M is a positive integer.

[0449] In some embodiments, the first information includes at least one of the following: the value of M, a first numerical value, and the M time points, wherein the first numerical value is used to determine the M time points.

[0450] In some embodiments, determining the third model requiring model performance monitoring from the first model and / or the second model based on the first information includes at least one of the following:

[0451] If the value of M is greater than 1, the third model is determined to be the second model;

[0452] The value of M is equal to 1, the first value is less than 0, and the M times determined according to the first value are the same as one of the transmission times of the Channel State Information Reference Signal (CSI-RS) resource periodically transmitted by the network device, and the third model is determined to be the first model.

[0453] The value of M is equal to 1, the first value is less than 0, and the M times determined according to the first value are different from each transmission time of the network device periodically sending CSI-RS resources, so the third model is determined to be the second model;

[0454] Each of the M time points is the same as one of the time points when the network device periodically sends CSI-RS resources, thus determining the third model as the first model;

[0455] The M time points are different from each transmission time of the CSI-RS resources periodically sent by the network device, thus determining the third model as the second model.

[0456] In some embodiments, the value of M is determined by at least one of the following methods:

[0457] The terminal device determines the value of M;

[0458] The value of M is configured for the network device.

[0459] In some embodiments, the terminal device determines the value of M to include at least one of the following:

[0460] The model deployed on the terminal device does not include the fourth model, and the value of M is determined to be 1. The fourth model is a model used to predict CSI.

[0461] The model deployed on the terminal device includes the fourth model, and the number of times corresponding to CSI that the fourth model can predict is used as the value of M.

[0462] In some embodiments, the M times are determined by at least one of the following methods:

[0463] The M time points are determined based on the downlink time point corresponding to the CSI reference resource and the first value;

[0464] Based on the CSI reporting time and the first value, the M time points are determined;

[0465] The M times are determined based on the time corresponding to the first indication information and the first value.

[0466] In some embodiments, the time corresponding to the first indication information includes at least one of the following:

[0467] The moment when the network device sends the first indication information;

[0468] The time at which the terminal device receives the first instruction information.

[0469] In some embodiments, the determination of the first value includes at least one of the following:

[0470] The terminal device determines the first value;

[0471] The network device is configured with the first value;

[0472] The first value is determined based on an initial value and the time interval between two adjacent moments, wherein the initial value and the time interval are configured or predefined by the network device;

[0473] The first value is predefined.

[0474] In some embodiments, the first value is a predefined value including at least one of the following:

[0475] The first value is predefined as the downlink time corresponding to the CSI reference resource;

[0476] The first value is predefined as the time when the fourth model deployed on the terminal device predicts CSI, and the fourth model is a model used to predict CSI;

[0477] The first value is predefined as the time when the network device sends the downlink pilot.

[0478] In some embodiments, the first CSI includes the measurement CSI and the compressed CSI, the first CSI is used by the network device to determine model performance criteria, the model performance criteria are used to determine the performance of the third model; wherein, the model performance criteria include at least one of the following: squared cosine similarity (SGCS) and normalized mean square error (NMSE).

[0479] In some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0480] Receive the second CSI sent by the network device, wherein the second CSI is the CSI at the M time points recovered by the network device based on the compressed CSI, through the second part model and / or the fourth part model;

[0481] A model performance criterion is determined based on the measured CSI and the second CSI, the model performance criterion being used to indicate the performance of the third model; wherein the model performance criterion includes at least one of the following: SGCS, NMSE.

[0482] In some embodiments, receiving the second CSI sent by the network device includes at least one of the following:

[0483] Receive all second CSIs sent by the network device at the same time;

[0484] Receive the CSIs at each time point in the second CSI, which are sent sequentially by the network device in a first order.

[0485] In some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0486] Receive pilot signals sent by the network device through precoding and shaping, wherein the precoding is a second CSI, and the second CSI is the CSI at the M time points recovered by the network device based on the compressed CSI through the second part model and / or the fourth part model;

[0487] The performance of the third model is determined based on the precoded and shaped pilot signal.

[0488] In some embodiments, receiving the precoded and shaped pilot signal sent by the network device includes at least one of the following:

[0489] Receive M precoded and shaped pilot signals sent by the network device at the same time;

[0490] Receive M precoded and shaped pilot signals sent sequentially by the network device in a first order.

[0491] In some embodiments, reporting the first CSI to the network device according to the third model includes:

[0492] The compressed CSI is determined according to the third model;

[0493] The first CSI is reported to the network device.

[0494] In some embodiments, determining the compressed CSI according to the third model includes at least one of the following:

[0495] The third model is the first model, and the compressed CSI is determined to be the CSI at the M times output by the first part of the model;

[0496] The third model is the second model, and the compressed CSI is determined to be the CSI at the M times output by the third part of the model.

[0497] In some embodiments, reporting the first CSI to the network device includes at least one of the following:

[0498] Report all first CSIs to the network device at the same time;

[0499] The network device shall report the CSI at each time point of the first CSI in the second order.

[0500] In some embodiments, the time for reporting the first CSI is determined by at least one of the following methods:

[0501] The network device provides instructions via downlink signaling;

[0502] Predefined;

[0503] The time when the terminal device compresses and reports the CSI is taken as the time when the first CSI is reported.

[0504] Figure 4A is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 4A, the present disclosure relates to a model performance monitoring method, which can be executed by a network device. The method may include:

[0505] Step S4101: Send the first instruction information.

[0506] The optional implementation of step S4101 can be found in the optional implementation of step S2101 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0507] Step S4102: Receive compressed CSI.

[0508] The optional implementation of step S4101 can be found in the optional implementation of step S2104 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0509] Step S4103: Input the compressed CSI into the second part of the model and / or the fourth part of the model to obtain the second CSI.

[0510] The optional implementation of step S4103 can be found in the optional implementation of step S2105 in Figure 2A, as well as other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0511] Step S4104: Send the second CSI.

[0512] The optional implementation of step S4104 can be found in the optional implementation of step S2106a in Figure 2A, and other related parts in the embodiment involved in Figure 2A, which will not be repeated here.

[0513] The methods involved in the embodiments of this disclosure may include at least one of the steps S4101 to S4104 described above. For example, step S4101 may be implemented as an independent embodiment, step S4102 may be implemented as an independent embodiment, step S4103 may be implemented as an independent embodiment, step S4104 may be implemented as an independent embodiment, step S4102 + step S4103 may be implemented as an independent embodiment, step S4103 + step S4104 may be implemented as an independent embodiment, and step S4102 + step S4103 + step S4104 may be implemented as an independent embodiment, but are not limited thereto.

[0514] In some embodiments, the order of any two steps in steps S4101 to S4104 can be interchanged or they can be performed simultaneously.

[0515] In some embodiments, steps S4101 to S4104 are optional, and one or more of these steps may be omitted or substituted in different embodiments. For example, step S4101.

[0516] Figure 4B is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 4B, the present disclosure relates to a model performance monitoring method, which can be executed by a network device. The method may include:

[0517] Step S4201: Send the first instruction information.

[0518] The optional implementation of step S4101 can be found in the optional implementation of step S2101 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0519] Step S4201: Receive compressed CSI.

[0520] The optional implementation of step S4201 can be found in the optional implementation of step S2104 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0521] Step S4203: Input the compressed CSI into the second part of the model and / or the fourth part of the model to obtain the second CSI.

[0522] The optional implementation of step S4203 can be found in the optional implementation of step S2105 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0523] Step S4204: Send the pilot signal that has been precoded and shaped.

[0524] The optional implementation of step S4204 can be found in the optional implementation of step S2106b in Figure 2A, and other related parts in the embodiment involved in Figure 2A, which will not be repeated here.

[0525] The methods involved in the embodiments of this disclosure may include at least one of the steps S4201 to S4204 described above. For example, step S4201 may be implemented as an independent embodiment, step S4202 may be implemented as an independent embodiment, step S4203 may be implemented as an independent embodiment, step S4204 may be implemented as an independent embodiment, step S4202 + step S4203 may be implemented as an independent embodiment, step S4203 + step S4204 may be implemented as an independent embodiment, but are not limited thereto.

[0526] In some embodiments, the order of any two steps in steps S4201 to S4204 can be interchanged or they can be performed simultaneously.

[0527] In some embodiments, steps S4201 to S4204 are optional, and one or more of these steps may be omitted or substituted in different embodiments. For example, step S4201.

[0528] Figure 4C is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 4C, this disclosure relates to a model performance monitoring method, which can be executed by a network device. The method may include:

[0529] Step S4301: Send the first instruction information.

[0530] The optional implementation of step S4301 can be found in the optional implementation of step S2301 in Figure 2C, and other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0531] Step S4302: Receive the measurement CSI and compressed CSI.

[0532] The optional implementation of step S4302 can be found in the optional implementation of step S2304 in Figure 2C, and other related parts in the embodiment involved in Figure 2C, which will not be repeated here.

[0533] Step S4303: Input the compressed CSI into the second part of the model and / or the fourth part of the model to obtain the second CSI.

[0534] The optional implementation of step S4303 can be found in the optional implementation of step S2305 in Figure 2C, and other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0535] Step S4304: Determine the model performance criteria based on the measured CSI and the second CSI.

[0536] The optional implementation of step S4304 can be found in the optional implementation of step S2306 in Figure 2C, as well as other related parts in the embodiments involved in Figure 2C, which will not be repeated here.

[0537] The methods involved in the embodiments of this disclosure may include at least one of the steps S4301 to S4304 described above. For example, step S4301 may be implemented as a standalone embodiment, step S4302 may be implemented as a standalone embodiment, step S4303 may be implemented as a standalone embodiment, step S4302 + step S4303 may be implemented as a standalone embodiment, and step S4302 + step S4303 + step S4304 may be implemented as a standalone embodiment, but are not limited thereto.

[0538] In some embodiments, the order of any two steps in steps S4301 to S4304 can be interchanged or they can be performed simultaneously.

[0539] In some embodiments, steps S4301 to S4304 are all optional steps. For example, step S4301 is optional, and one or more of these steps may be omitted or substituted in different embodiments.

[0540] Figure 4D is a flowchart illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 4D, the present disclosure relates to a model performance monitoring method, which can be executed by a network device. The method may include:

[0541] Step S4401: Send the first instruction information.

[0542] The optional implementation of step S4101 can be found in the optional implementation of step S2101 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0543] Step S4402: Receive the first CSI.

[0544] In some embodiments, the first CSI may include a measured CSI and / or a compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI may be the CSI at M times output by the first part model and / or the third part model, where M is a positive integer.

[0545] In some embodiments, the first CSI includes measured CSI and / or compressed CSI, wherein the measured CSI is the CSI at M times measured or predicted by the terminal device, and the compressed CSI is the CSI at the M times output by the first part model and / or the third part model, where M is a positive integer.

[0546] In some embodiments, the first information includes at least one of the following: the value of M, a first numerical value, and the M time points, wherein the first numerical value is used to determine the M time points.

[0547] In some embodiments, the value of M is determined by at least one of the following methods:

[0548] The terminal device determines the value of M;

[0549] The value of M is configured for the network device.

[0550] In some embodiments, the terminal device determines the value of M to include at least one of the following:

[0551] The model deployed on the terminal device does not include the fourth model, and the value of M is determined to be 1. The fourth model is a model used to predict CSI.

[0552] The model deployed on the terminal device includes the fourth model, and the number of times corresponding to CSI that the fourth model can predict is used as the value of M.

[0553] In some embodiments, the M times are determined by at least one of the following methods:

[0554] The M time points are determined based on the downlink time point corresponding to the CSI reference resource and the first value;

[0555] Based on the CSI reporting time and the first value, the M time points are determined;

[0556] The M times are determined based on the time corresponding to the first indication information and the first value.

[0557] In some embodiments, the time corresponding to the first indication information includes at least one of the following:

[0558] The moment when the network device sends the first indication information;

[0559] The time at which the terminal device receives the first instruction information.

[0560] In some embodiments, the determination of the first value includes at least one of the following:

[0561] The terminal device determines the first value;

[0562] The network device is configured with the first value;

[0563] The first value is determined based on an initial value and the time interval between two adjacent moments, wherein the initial value and the time interval are configured or predefined by the network device;

[0564] The first value is predefined.

[0565] In some embodiments, the first value is a predefined value including at least one of the following:

[0566] The first value is predefined as the downlink time corresponding to the CSI reference resource;

[0567] The first value is predefined as the time when the fourth model deployed on the terminal device predicts CSI, and the fourth model is a model used to predict CSI;

[0568] The first value is predefined as the time when the network device sends the downlink pilot.

[0569] In some embodiments, the first CSI includes the measured CSI and the compressed CSI, and the method further includes:

[0570] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0571] The model performance criteria are determined based on the measured CSI and the second CSI, and the model performance criteria are used to determine the performance of the third model; wherein the model performance criteria include at least one of the following: squared cosine similarity (SGCS) and normalized mean square error (NMSE).

[0572] In some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0573] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0574] The second CSI is sent to the terminal device. The second CSI is used by the terminal device to determine the model performance criteria, which are used to determine the performance of the third model. The model performance criteria include at least one of the following: SGCS and NMSE.

[0575] In some embodiments, sending the second CSI to the terminal device includes at least one of the following:

[0576] Send all second CSIs to the terminal device at the same time;

[0577] The CSIs for each moment in the second CSI are sent to the terminal device in the first order.

[0578] In some embodiments, the first CSI includes the compressed CSI, and the method further includes:

[0579] The compressed CSI is input into the second part of the model and / or the fourth part of the model to obtain the second CSI;

[0580] A precoded and shaped pilot signal is sent to the terminal device, wherein the precoding is the second CSI, and the precoded and shaped pilot signal is used by the terminal device to determine the performance of the third model.

[0581] In some embodiments, sending the precoded and shaped pilot signal to the terminal device includes at least one of the following:

[0582] At the same time, M precoded and shaped pilot signals are sent to the terminal device;

[0583] M precoded and shaped pilot signals are sent to the terminal device in the first order.

[0584] In some embodiments, receiving the first CSI reported by the terminal device includes at least one of the following:

[0585] Receive all first CSIs reported by the terminal device at the same time;

[0586] The terminal device receives the CSIs at each time point in the first CSI, reported sequentially in a second order.

[0587] In some embodiments, the time when the terminal device reports the first CSI is determined by at least one of the following methods:

[0588] The network device provides instructions via downlink signaling;

[0589] Predefined;

[0590] The time when the terminal device compresses and reports the CSI is taken as the time when the first CSI is reported.

[0591] Figure 5 is an interactive schematic diagram illustrating a model performance monitoring method according to an embodiment of the present disclosure. As shown in Figure 5, the present disclosure relates to a model performance monitoring method, which can be executed by a communication system. The method may include:

[0592] Step S5101: The network device sends the first instruction information to the terminal device.

[0593] The optional implementation of step S5101 can be found in the optional implementation of step S2101 in Figure 2A, and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0594] Step S5102: The terminal device determines the third model that needs to be monitored for model performance from the first model and / or the second model based on the first information.

[0595] The optional implementation of step S5102 can be found in the optional implementation of step S2102 in Figure 2A and other related parts in the embodiments involved in Figure 2A, which will not be repeated here.

[0596] Step S5103: The terminal device reports the first CSI according to the third model.

[0597] The optional implementation of step S5103 can be found in the optional implementations of steps S2103 to S2104 in Figure 2A, steps S2303 to S2304 in Figure 2C, and other related parts in the embodiments involved in Figures 2A and 2C, which will not be repeated here.

[0598] In some embodiments, the above methods may include the methods described in the embodiments of the communication system, terminal device, network device, etc., which will not be repeated here.

[0599] In some embodiments, this disclosure uses a unified parameter configuration to monitor the performance of two bilateral AI / ML models, one of which includes prediction and the other does not.

[0600] In some embodiments, for NW-side model performance monitoring, the NW instructs the UE to include M (M is an integer greater than or equal to 1) time-based channel state information (CSI) in a single report via downlink signaling configuration.

[0601] Optionally, the NW instructs the UE to include the CSI at M times in multiple reports via downlink signaling.

[0602] In some embodiments, the reporting times of the CSI at M time points reported by the UE can be determined in the following ways:

[0603] The network provides instructions via downlink signaling;

[0604] Determined through predefined criteria;

[0605] Determined based on the timing of the UE sending the compressed CSI;

[0606] The above three methods are determined by a combination of multiple methods.

[0607] In some embodiments, for monitoring model performance on the UE side, NW transmits recovery target CSIs corresponding to M time points once or multiple times, or transmits precoded CSI-RS / DMRS once or multiple times, wherein the precoding is the recovery target CSI corresponding to M time points.

[0608] In some embodiments, the M time points can be determined in the following way:

[0609] Method 1: nref+l i , i = 0, ..., M-1, where nref represents the downlink time corresponding to the CSI reference resource;

[0610] Method 2: n+l i Where n represents the CSI reporting time;

[0611] Method 3: n'+l i , where n' represents the time when the bilateral AI / ML model performance monitoring signaling indicated by NW is received.

[0612] In some embodiments, the value of M can be determined in the following ways:

[0613] Method 1: The value of M is configured to the UE by the network;

[0614] Method 2: UE-side AI / ML model determination. If the UE side does not have an AI / ML model for predicting CSI, M = 1; if the UE side has deployed an AI / ML model for predicting CSI, M is determined based on the AI / ML output for predicting CSI.

[0615] In some embodiments, l i The value is determined by the following:

[0616] Method 1: Configure M units on the NW side i For UE;

[0617] Method 2: The value of M, the interval d between two adjacent time points, and l0 are determined by the NW side configuration. l0 represents the starting value of a time range, which can be determined by network configuration or predefined.

[0618] Method 3: l i Determined through predefinition, such as predefining l i The timing for sending downlink pilot signals is based on the most recent time referenced by the resource or the time predicted by the terminal-side prediction model.

[0619] It should be noted that the UE side will adjust the settings according to the above configuration. iThe value of M, or the method for determining the M time points mentioned above, determines whether the compressed reporting is based on the result of the current measurement time or on the result of CSI prediction. In other words, the UE can also select a suitable bilateral AI / ML model based on the network configuration parameters.

[0620] It should also be noted that the aforementioned downlink signaling includes one or more of RRC / MAC-CE / DCI signaling.

[0621] Example 1 (NW-side model performance monitoring):

[0622] Assume that a non-predictive CSI compressed feedback bilateral AI / ML model and a compressed bilateral AI / ML model containing predictive CSI are deployed on the UE and gNB sides. The gNB sends CSI-RS with a period of T to the UE. The UE estimates the downlink channel information based on the received CSI-RS, and then performs eigenvalue decomposition to obtain the eigenvector corresponding to the largest eigenvalue. This eigenvector is used as the input to the UE-side CSI generation part model as the target CSI. The output of this part model is quantized and reported to the gNB. The information obtained by the gNB through inverse quantization is used as the input to the gNB-side CSI recovery (recovering either predictive or non-predictive CSI) part model, thereby obtaining the recovered target CSI output by the CSI recovery part model.

[0623] Option 1: Monitor the performance of compressed bilateral AI / ML models that include CSI prediction:

[0624] The gNB configures parameters M=1 or M=2 for the UE via RRC signaling, and also configures parameters l0, l1=l0+T. Let n represent the time when the CSI is reported. The value of l0+n is an integer multiple of T and coincides with a certain time when the CSI-RS is periodically transmitted. When the UE receives an instruction from the gNB to report the target CSI at time n, the UE will obtain the target CSI at times l0+n and l1+n based on the received CSI-RS, and then report it at the respective reporting times, as shown in Figure 2B. The reported n1 and n2 can be indicated by the gNB via downlink signaling or determined predefinedly, such as based on the nth time, then offset by δ1 and δ2 times. The values ​​of δ1 and δ2 can be predefined or configured by the gNB. Optionally, the UE reports the target CSI and the compressed CSI together to the gNB.

[0625] gNB receives the target CSI (defined as v) i ) and compressed CSI, and the target CSI (defined as e) of the gNB-side partial model recovery. iThe SGCS is calculated using formula (1). The NW side determines the performance of the two-sided model based on the average value of the SGCS calculated at these two times.

[0626] Option 1: Monitor the performance of compressed bilateral AI / ML models that do not predict CSI:

[0627] Assuming the gNB needs to monitor the performance of a compressed bilateral AI / ML model with unpredictable CSI, the gNB configures parameter M=1 to the UE via RRC signaling, and also configures parameter l0, but l0<0. Based on parameter l0, the UE will report the target CSI corresponding to time l0+n at time n, or by default, the target CSI corresponding to the closest time to the time nref where the CSI reference resource is located. At time n, the UE reports the obtained target CSI and compressed CSI together or separately to the gNB. The gNB can monitor the bilateral model using formula (1) by receiving the target CSI and compressed CSI, as well as the recovered target CSI output after recovering part of the model.

[0628] As can be seen from the above, even if the gNB configures parameters M=1 and l0 for both the compressed bilateral AI / ML model containing predicted and unpredictable CSI, the UE can determine whether the gNB is monitoring the compressed bilateral AI / ML model containing unpredictable CSI or the compressed bilateral AI / ML model containing predicted CSI based on parameter l0.

[0629] It should be noted that CSI reporting can be sent only after receiving a gNB signal instructing the monitoring model performance; otherwise, target CSI will not be issued.

[0630] In some embodiments, the target CSI can be transmitted via PUSCH.

[0631] Example 2 (UE-side model performance monitoring):

[0632] For UE-side model performance monitoring, the process is similar to that of gNB-side model performance monitoring described above. It is assumed that the gNB sends the recovery target CSI at time M=1 to the UE. This time can also be determined using the method described in Example 1. Optionally, the gNB sends beamforming CSI-RS to the UE at time M=1.

[0633] The UE determines whether to monitor a compressed bilateral AI / ML model with non-predictive CSI or a compressed bilateral AI / ML model with predictive CSI based on the time corresponding to M=1. When M>1, the gNB will send data to the UE at multiple times. The transmission time can be determined based on the time of transmitting CSI-RS measurement resources, such as coinciding with the time of periodic CSI-RS measurement resources, or determined by the gNB's configuration of the interval d and start position l0 for transmitting multiple beamformed CSI-RS. The UE monitors the model performance based on the received beamformed CSI-RS.

[0634] In some embodiments, the target CSI for transmission recovery can be transmitted via PDSCH.

[0635] In some embodiments of this disclosure, a communication system is provided, which may include a terminal device and a network device, wherein the terminal device may execute the model performance monitoring method executed by the terminal device in the foregoing embodiments of this disclosure; and the network device may execute the model performance monitoring method executed by the network device in the foregoing embodiments of this disclosure.

[0636] This disclosure also provides an apparatus for implementing any of the above methods. For example, an apparatus is provided that includes units or modules for implementing the steps performed by the terminal in any of the above methods. Alternatively, another apparatus is provided that includes units or modules for implementing the steps performed by a network device (e.g., an access network device, a core network functional node, a core network device, etc.) in any of the above methods.

[0637] It should be understood that the division of units or modules in the above device is only a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, the units or modules in the device can be implemented by a processor calling software: for example, the device includes a processor connected to a memory containing instructions. The processor calls the instructions stored in the memory to implement any of the above methods or to implement the functions of the units or modules in the above device. The processor can be, for example, a general-purpose processor, such as a Central Processing Unit (CPU) or a microprocessor, and the memory can be internal or external to the device. Alternatively, the units or modules in the device can be implemented in the form of hardware circuits. The functionality of some or all of the units or modules can be achieved through the design of these hardware circuits, which can be understood as one or more processors. For example, in one implementation, the hardware circuit is an Application-Specific Integrated Circuit (ASIC), and the functionality of some or all of the units or modules is achieved through the design of the logical relationships between the components within the circuit. In another implementation, the hardware circuit can be implemented using a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), which can include a large number of logic gates. The connection relationships between the logic gates are configured through configuration files, thereby achieving the functionality of some or all of the units or modules. All units or modules of the above device can be implemented entirely through processor-called software, entirely through hardware circuits, or partially through processor-called software with the remaining parts implemented through hardware circuits.

[0638] In this embodiment, the processor is a circuit with signal processing capabilities. In one implementation, the processor can be a circuit with instruction read and execute capabilities, such as a Central Processing Unit (CPU), a microprocessor, a Graphics Processing Unit (GPU) (which can be understood as a microprocessor), or a Digital Signal Processor (DSP). In another implementation, the processor can implement certain functions through the logical relationships of hardware circuits. The logical relationships of the aforementioned hardware circuits are fixed or reconfigurable. For example, the processor is a hardware circuit implemented using an Application-Specific Integrated Circuit (ASIC) or a Programmable Logic Device (PLD), such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document and configuring the hardware circuit can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units or modules. In addition, it can also be hardware circuits designed for artificial intelligence, which can be understood as ASICs, such as Neural Network Processing Units (NPUs), Tensor Processing Units (TPUs), and Deep Learning Processing Units (DPUs).

[0639] Figure 6A is a schematic diagram of the structure of a terminal device according to an embodiment of this disclosure. As shown in Figure 6A, the terminal device 101 may include at least one of a transceiver module 6101, a processing module 6102, etc. In some embodiments, the transceiver module 6101 is configured to receive first indication information sent by a network device, the first indication information being used to indicate model performance monitoring; the processing module 6102 is configured to determine a third model requiring model performance monitoring from a first model and / or a second model based on first information, the first information including information corresponding to the first channel state information (CSI) that the terminal device needs to report, the first CSI being used to perform performance monitoring on the third model; the transceiver module 6101 is further configured to report the first CSI to the network device based on the third model; wherein, the first model includes a first part model deployed on the terminal device and a second part model deployed on the network device, the second model includes a third part model deployed on the terminal device and a fourth part model deployed on the network device, the first part model is used to compress the CSI, the second part model is used to recover the CSI after compression of the first part model, the third part model is used to compress and / or predict the CSI, and the fourth part model is used to recover and / or predict the CSI after compression of the third part model. Optionally, the transceiver module 6101 can be used to perform at least one of the communication steps (such as step S2101, step S2104, but not limited thereto) performed by the terminal device 101 in any of the above methods, which will not be elaborated here. Optionally, the processing module 6102 can be used to perform at least one of the other steps (such as step S2102, step S2103, but not limited thereto) performed by the terminal device 101 in any of the above methods, which will not be elaborated here.

[0640] In some embodiments, the transceiver module may include a transmitting module and / or a receiving module, which may be separate or integrated. Optionally, the transceiver module may be interchangeable with a transceiver.

[0641] In some embodiments, the processing module may be a single module or may include multiple sub-modules. Optionally, the multiple sub-modules may each perform all or part of the steps required by the processing module. Optionally, the processing module may be interchangeable with a processor.

[0642] Figure 6B is a schematic diagram of the structure of a network device according to an embodiment of this disclosure. As shown in Figure 6B, the network device 102 may include at least one of a transceiver module 6201, a processing module 6202, etc. In some embodiments, the transceiver module 6201 is configured to send first indication information to a terminal device, the first indication information being used to instruct model performance monitoring; the transceiver module 6201 is further configured to receive first channel state information (CSI) reported by the terminal device according to a third model, the first CSI being used to perform performance monitoring on the third model, the third model being a model that the terminal device determines from a first model and / or a second model that requires model performance monitoring based on first information, the first information including information corresponding to the first CSI; wherein, the first model includes a first part of the model deployed on the terminal device and a second part of the model deployed on the network device, the second model includes a third part of the model deployed on the terminal device and a fourth part of the model deployed on the network device, the first part of the model is used to compress the CSI, the second part of the model is used to recover the CSI after compression of the first part of the model, the third part of the model is used to compress and / or predict the CSI, and the fourth part of the model is used to recover and / or predict the CSI after compression of the third part of the model. Optionally, the transceiver module 6201 can be used to perform at least one of the communication steps (such as step S2101, but not limited thereto) performed by the network device 102 in any of the above methods, which will not be elaborated here. Optionally, the processing module 6202 can be used to perform at least one of the other steps (such as step S2105, but not limited thereto) performed by the network device 102 in any of the above methods, which will not be elaborated here.

[0643] In some embodiments, the transceiver module may include a transmitting module and / or a receiving module, which may be separate or integrated. Optionally, the transceiver module may be interchangeable with a transceiver.

[0644] In some embodiments, the processing module may be a single module or may include multiple sub-modules. Optionally, the multiple sub-modules may each perform all or part of the steps required by the processing module. Optionally, the processing module may be interchangeable with a processor.

[0645] Figure 7A is a schematic diagram of the structure of the communication device 7100 proposed in an embodiment of this disclosure. The communication device 7100 can be a network device (e.g., access network device, core network device, etc.), a terminal (e.g., user equipment, etc.), a chip, chip system, or processor that supports the first device in implementing any of the above methods, or a chip, chip system, or processor that supports the terminal in implementing any of the above methods. The communication device 7100 can be used to implement the methods described in the above method embodiments; for details, please refer to the descriptions in the above method embodiments.

[0646] As shown in Figure 7A, the communication device 7100 includes one or more processors 7101. The processor 7101 can be a general-purpose processor or a dedicated processor, such as a baseband processor or a central processing unit (CPU). The baseband processor can be used to process communication protocols and communication data, while the CPU can be used to control communication devices (e.g., base stations, baseband chips, IoT devices, IoT device chips, DUs or CUs, etc.), execute programs, and process program data. The communication device 7100 is used to execute any of the above methods.

[0647] In some embodiments, the communication device 7100 further includes one or more memories 7102 for storing instructions. Optionally, all or part of the memories 7102 may also be located outside the communication device 7100.

[0648] In some embodiments, the communication device 7100 further includes one or more transceivers 7103. When the communication device 7100 includes one or more transceivers 7103, the transceivers 7103 perform at least one of the communication steps such as sending and / or receiving in the above method (e.g., steps S2101, S2104, but not limited thereto), and the processor 7101 performs at least one of other steps (e.g., step S2102, but not limited thereto).

[0649] In some embodiments, a transceiver may include a receiver and / or a transmitter, which may be separate or integrated. Optionally, the terms transceiver, transceiver unit, transceiver, transceiver circuit, etc., may be used interchangeably; the terms transmitter, transmitting unit, transmitter, transmitting circuit, etc., may be used interchangeably; and the terms receiver, receiving unit, receiver, receiving circuit, etc., may be used interchangeably.

[0650] In some embodiments, the communication device 7100 may include one or more interface circuits. Optionally, the interface circuit is connected to the memory 7102, and the interface circuit can be used to receive signals from the memory 7102 or other devices, and can be used to send signals to the memory 7102 or other devices. For example, the interface circuit can read instructions stored in the memory 7102 and send the instructions to the processor 7101.

[0651] The communication device 7100 described in the above embodiments may be a first device or an Internet of Things (IoT) device, but the scope of the communication device 7100 described in this disclosure is not limited thereto, and the structure of the communication device 7100 may not be limited by FIG. 7A. The communication device may be a standalone device or part of a larger device. For example, the communication device may be: (1) a standalone integrated circuit IC, or chip, or chip system or subsystem; (2) a collection of one or more ICs, optionally, the IC collection may also include storage components for storing data and programs; (3) an ASIC, such as a modem; (4) a module that can be embedded in other devices; (5) a receiver, IoT device, smart IoT device, cellular phone, wireless device, handheld device, mobile unit, vehicle device, first device, cloud device, artificial intelligence device, etc.; (6) others, etc.

[0652] Figure 7B is a schematic diagram of the structure of the chip 7200 according to an embodiment of this disclosure. For cases where the communication device 7100 can be a chip or a chip system, the schematic diagram of the chip 7200 shown in Figure 7B can be referenced, but is not limited thereto.

[0653] Chip 7200 includes one or more processors 7201, which are used to perform any of the above methods.

[0654] In some embodiments, chip 7200 further includes one or more interface circuits 7203. Optionally, interface circuit 7203 is connected to memory 7202, and interface circuit 7203 can be used to receive signals from memory 7202 or other devices, and interface circuit 7203 can be used to send signals to memory 7202 or other devices. For example, interface circuit 7203 can read instructions stored in memory 7202 and send the instructions to processor 7201.

[0655] In some embodiments, the interface circuit 7203 performs at least one of the communication steps such as sending and / or receiving in the above method (e.g., steps S2101, S2104, but not limited thereto), and the processor 7201 performs at least one of the other steps (e.g., step S2102, but not limited thereto).

[0656] In some embodiments, the terms interface circuit, interface, transceiver pin, transceiver, etc., can be used interchangeably.

[0657] In some embodiments, chip 7200 further includes one or more memories 7202 for storing instructions. Optionally, all or part of the memories 7202 may be located outside of chip 7200.

[0658] This disclosure also proposes a storage medium storing instructions that, when executed on a communication device 7100, cause the communication device 7100 to perform any of the above methods. Optionally, the storage medium is an electronic storage medium. Optionally, the storage medium is a computer-readable storage medium, but not limited thereto; it may also be a storage medium readable by other devices. Optionally, the storage medium may be a non-transitory storage medium, but not limited thereto; it may also be a temporary storage medium.

[0659] This disclosure also provides a program product that, when executed by the communication device 7100, causes the communication device 7100 to perform any of the above methods. Optionally, the program product may be a computer program product.

[0660] This disclosure also proposes a computer program that, when run on a computer, causes the computer to perform any of the above methods.

Claims

1. A model performance monitoring method characterized by, The method is performed by a terminal device, and the method comprises: receiving first indication information sent by a network device, the first indication information being used for indicating that model performance monitoring is performed; determining a third model that needs to perform model performance monitoring from a first model and / or a second model according to first information, the first information comprising information corresponding to first channel state information (CSI) that needs to be reported by the terminal device, the first CSI being used for performance monitoring of the third model; reporting the first CSI to the network device according to the third model; wherein the first model comprises a first partial model deployed at the terminal device and a second partial model deployed at the network device, the second model comprises a third partial model deployed at the terminal device and a fourth partial model deployed at the network device, the first partial model is used for compression processing of CSI, the second partial model is used for CSI recovery of the CSI processed by the first partial model, the third partial model is used for compression and / or prediction processing of CSI, and the fourth partial model is used for CSI recovery and / or CSI prediction of the CSI processed by the third partial model.

2. The method of claim 1, wherein, The first CSI comprises measurement CSI and / or compressed CSI, the measurement CSI is CSI at M time instants measured or predicted by the terminal device, the compressed CSI is the CSI at the M time instants output by the first partial model and / or the third partial model, and M is a positive integer.

3. The method of claim 2, wherein, The first information comprises at least one of the following: a value of M, a first numerical value, and the M time instants, the first numerical value being used for determining the M time instants.

4. The method of claim 3, wherein, The determination of the third model that needs to perform model performance monitoring from the first model and / or the second model according to the first information comprises at least one of the following: the value of M is greater than 1, and the third model is determined as the second model; the value of M is equal to 1, the first numerical value is less than 0, the M time instants determined according to the first numerical value are the same as one of time instants at which the network device periodically sends channel state information reference signal (CSI-RS) resources, and the third model is determined as the first model; the value of M is equal to 1, the first numerical value is less than 0, and the M time instants determined according to the first numerical value are different from each of time instants at which the network device periodically sends CSI-RS resources, and the third model is determined as the second model; each of the M time instants is the same as one of time instants at which the network device periodically sends CSI-RS resources, and the third model is determined as the first model; the M time instants are different from each of time instants at which the network device periodically sends CSI-RS resources, and the third model is determined as the second model.

5. The method according to claim 3 or 4, characterized in that, The value of M is determined in at least one of the following ways: the terminal device determines the value of M; the network device configures the value of M.

6. The method of claim 5, wherein, The determination of the value of M by the terminal device comprises at least one of the following: The model deployed by the terminal device does not include a fourth model, and the value of M is determined as 1, the fourth model being a model for predicting CSI; The model deployed by the terminal device includes the fourth model, and the number of time instants corresponding to CSI that can be predicted by the fourth model is taken as the value of M.

7. The method according to any one of claims 3-6, characterized in that, The M time instants are determined in at least one of the following ways: The M time instants are determined according to the downlink time instant corresponding to the CSI reference resource and the first value; The M time instants are determined according to the reporting time instant of the CSI and the first value; The M time instants are determined according to the time instant corresponding to the first indication information and the first value.

8. The method of claim 7, wherein, The time instant corresponding to the first indication information includes at least one of the following: The time instant at which the network device sends the first indication information; The time instant at which the terminal device receives the first indication information.

9. The method according to any one of claims 3-8, characterized in that, The determination of the first value includes at least one of the following: The terminal device determines the first value; The network device configures the first value; The first value is determined according to an initial value and a time interval between two adjacent time instants, the initial value and the time interval being configured by the network device or predefined; The first value is predefined.

10. The method of claim 9, wherein, The first value is predefined in at least one of the following ways: The first value is predefined as the downlink time instant corresponding to the CSI reference resource; The first value is predefined as the time instant at which the fourth model deployed by the terminal device predicts the CSI, the fourth model being a model for predicting CSI; The first value is predefined as the time instant at which the network device sends a downlink pilot.

11. The method according to any one of claims 2-10, characterized in that, The first CSI includes the measured CSI and the compressed CSI, and the first CSI is used by the network device to determine a model performance criterion, the model performance criterion being used to determine the performance of the third model; wherein the model performance criterion includes at least one of the following: square of cosine similarity SGCS, normalized mean square error NMSE.

12. The method according to any one of claims 2-10, characterized in that, The first CSI includes the compressed CSI, and the method further includes: receiving second CSI sent by the network device, the second CSI being CSI of the M time instants recovered by the network device according to the compressed CSI, through the second part model and / or the fourth part model; determining a model performance criterion according to the measured CSI and the second CSI, the model performance criterion being used to determine the performance of the third model; wherein the model performance criterion includes at least one of the following: SGCS, NMSE.

13. The method of claim 12, wherein, The receiving of the second CSI sent by the network device includes at least one of the following: Receiving all second CSI sent by the network device at the same time; Receiving the CSI of each time instant in the second CSI sent by the network device in a first order.

14. The method according to any one of claims 2-10, characterized in that, The first CSI includes the compressed CSI, and the method further includes: receiving a pilot signal precoded by the network device, the precoding being a second CSI, the second CSI being the CSI of the M time instants recovered by the network device from the compressed CSI according to the second part model and / or the fourth part model; determining the performance of the third model according to the pilot signal precoded by the network device.

15. The method of claim 14, wherein, The receiving the pilot signal precoded by the network device includes at least one of: receiving M pilot signals precoded by the network device at the same time instant; receiving M pilot signals precoded by the network device sent in a first order.

16. The method according to any one of claims 2-15, characterized in that, The reporting the first CSI to the network device according to the third model includes: determining the compressed CSI according to the third model; reporting the first CSI to the network device.

17. The method of claim 16, wherein, The determining the compressed CSI according to the third model includes at least one of: the third model being the first model, the compressed CSI being the CSI of the M time instants output by the first part model; the third model being the second model, the compressed CSI being the CSI of the M time instants output by the third part model.

18. The method according to any one of claims 1 to 17, characterized in that, The reporting the first CSI to the network device includes at least one of: reporting all the first CSI to the network device at the same time instant; reporting the CSI of each time instant in the first CSI to the network device in a second order.

19. The method according to any one of claims 1 to 18, characterized in that, The time instant of reporting the first CSI is determined by at least one of: indicated by the network device through downlink signaling; predefined; the time instant of the terminal device compressing and reporting the CSI being the time instant of reporting the first CSI.

20. A model performance monitoring method, characterized by, The method is performed by a network device, and the method includes: sending first indication information to a terminal device, the first indication information being used to indicate that model performance monitoring is performed; receiving first channel state information (CSI) reported by the terminal device according to a third model, the first CSI being used to monitor the performance of the third model, the third model being a model that needs to be monitored in performance and is determined by the terminal device from a first model and / or a second model according to first information, the first information including information corresponding to the first CSI; wherein the first model includes a first part model deployed at the terminal device and a second part model deployed at the network device, the second model includes a third part model deployed at the terminal device and a fourth part model deployed at the network device, the first part model is used to compress CSI, the second part model is used to recover CSI after the CSI is compressed by the first part model, the third part model is used to compress and / or predict CSI, and the fourth part model is used to recover and / or predict CSI after the CSI is compressed by the third part model.

21. The method of claim 20, wherein, The first CSI includes measured CSI and / or compressed CSI, the measured CSI is CSI of M time instants measured or predicted by the terminal device, the compressed CSI is the CSI of the M time instants output by the first part model and / or the third part model, and M is a positive integer.

22. The method of claim 21, wherein, The first information includes at least one of the following: a value of M, a first numerical value, and the M time instants, the first numerical value being used to determine the M time instants.

23. The method of claim 22, wherein, The value of M is determined in at least one of the following ways: The terminal device determines the value of M. The network device configures the value of M.

24. The method of claim 23, wherein, The terminal device determines the value of M in at least one of the following ways: The model deployed by the terminal device does not include a fourth model, and the value of M is determined to be 1, the fourth model being a model for predicting CSI. The model deployed by the terminal device includes the fourth model, and the number of time instants corresponding to the CSI that can be predicted by the fourth model is taken as the value of M.

25. The method of any one of claims 22-24, wherein, The M time instants are determined in at least one of the following ways: According to a downlink time instant corresponding to a CSI reference resource and the first numerical value, the M time instants are determined. According to a reporting time instant of the CSI and the first numerical value, the M time instants are determined. According to a time instant corresponding to the first indication information and the first numerical value, the M time instants are determined.

26. The method of claim 25, wherein, The time instant corresponding to the first indication information includes at least one of the following: A time instant at which the network device sends the first indication information. A time instant at which the terminal device receives the first indication information.

27. The method of any one of claims 22-26, wherein, The determination of the first numerical value includes at least one of the following: The terminal device determines the first numerical value. The network device configures the first numerical value. The first numerical value is determined according to an initial value and a time interval between two adjacent time instants, the initial value and the time interval being configured by the network device or predefined. The first numerical value is predefined.

28. The method of claim 27, wherein, The first numerical value is predefined in at least one of the following ways: The first numerical value is predefined as a downlink time instant corresponding to a CSI reference resource. The first numerical value is predefined as a time instant at which a fourth model deployed by the terminal device predicts CSI, the fourth model being a model for predicting CSI. The first numerical value is predefined as a time instant at which the network device sends a downlink pilot.

29. The method of any one of claims 21-28, wherein, The first CSI includes the measured CSI and the compressed CSI, and the method further includes: Inputting the compressed CSI into the second part model and / or the fourth part model to obtain second CSI. Determining a model performance criterion according to the measured CSI and the second CSI, the model performance criterion being used to determine the performance of the third model, and the model performance criterion includes at least one of the following: square of cosine similarity SGCS and normalized mean square error NMSE.

30. The method of any one of claims 21-28, wherein, The first CSI includes the compressed CSI, and the method further includes: Inputting the compressed CSI into the second part model and / or the fourth part model to obtain second CSI. sending the second CSI to the terminal device, the second CSI being used by the terminal device to determine a model performance criterion, the model performance criterion being used to determine a performance of the third model; wherein the model performance criterion comprises at least one of: an SGCS, an NMSE.

31. The method of claim 30, wherein, The sending the second CSI to the terminal device comprises at least one of: sending all second CSIs to the terminal device at the same time; sending the CSI of each time in the second CSIs to the terminal device in a first order one by one.

32. The method of any one of claims 21-28, wherein, The first CSI comprises the compressed CSI, and the method further comprises: inputting the compressed CSI into the second partial model and / or the fourth partial model to obtain the second CSI; sending a pilot signal precoded and shaped to the terminal device, the precoding being the second CSI, the pilot signal precoded and shaped being used by the terminal device to determine the performance of the third model.

33. The method of claim 32, wherein, The sending the pilot signal precoded and shaped to the terminal device comprises at least one of: sending M pilot signals precoded and shaped to the terminal device at the same time; sending the M pilot signals precoded and shaped to the terminal device in a first order one by one.

34. The method of any one of claims 20-33, wherein, The receiving the first CSI reported by the terminal device comprises at least one of: receiving all first CSIs reported by the terminal device at the same time; receiving the CSI of each time in the first CSIs reported by the terminal device in a second order one by one.

35. The method of any one of claims 20-34, wherein, The time at which the terminal device reports the first CSI is determined in at least one of the following ways: indicated by the network device through downlink signaling; predefined; the time at which the terminal device reports the CSI after compression as the time at which the terminal device reports the first CSI.

36. A terminal device, comprising: comprises: a transceiver module configured to receive first indication information sent by a network device, the first indication information being used to indicate that model performance monitoring is performed; a processing module configured to determine a third model for which model performance monitoring needs to be performed from a first model and / or a second model according to first information, the first information comprising information corresponding to first channel state information (CSI) that needs to be reported by the terminal device, the first CSI being used to perform performance monitoring on the third model; the transceiver module is further configured to report the first CSI to the network device according to the third model; wherein the first model comprises a first partial model deployed at the terminal device and a second partial model deployed at the network device, and the second model comprises a third partial model deployed at the terminal device and a fourth partial model deployed at the network device, the first partial model being used to perform compression processing on CSI, the second partial model being used to perform CSI recovery on the CSI processed by the first partial model, the third partial model being used to perform compression and / or prediction processing on CSI, and the fourth partial model being used to perform CSI recovery and / or CSI prediction on the CSI processed by the third partial model. comprises:

37. A network device, comprising: ​ The transceiver module is configured to send first indication information to the terminal device, the first indication information being used to indicate to perform model performance monitoring; The transceiver module is further configured to receive first channel state information (CSI) reported by the terminal device according to a third model, the first CSI being used to perform performance monitoring on the third model, the third model being a model determined by the terminal device to need to perform model performance monitoring from a first model and / or a second model according to first information, the first information including information corresponding to the first CSI; The first model includes a first partial model deployed at the terminal device and a second partial model deployed at the network device, and the second model includes a third partial model deployed at the terminal device and a fourth partial model deployed at the network device, the first partial model being used to perform compression processing on CSI, the second partial model being used to perform CSI recovery on the CSI compressed by the first partial model, the third partial model being used to perform compression and / or prediction processing on CSI, and the fourth partial model being used to perform CSI recovery and / or CSI prediction on the CSI compressed by the third partial model.

38. A communications device, characterized by It is characterized by comprising: One or more processors; The communication device is configured to perform the model performance monitoring method in any one of claims 1 to 19 or claims 20 to 35.

39. A storage medium, the storage medium storing instructions, wherein, When the instructions run on the communication device, the communication device performs the model performance monitoring method in any one of claims 1 to 19 or claims 20 to 35.

40. A communication system, characterized by The communication system includes a terminal device and a network device, wherein the terminal device is configured to implement the model performance monitoring method in any one of claims 1 to 19, and the network device is configured to implement the model performance monitoring method in any one of claims 20 to 35.

Images