WO2023198093A1 - 模型有效性的确定方法、装置及通信设备 - Google Patents

模型有效性的确定方法、装置及通信设备 Download PDF

Info

Publication number
WO2023198093A1
WO2023198093A1 PCT/CN2023/087736 CN2023087736W WO2023198093A1 WO 2023198093 A1 WO2023198093 A1 WO 2023198093A1 CN 2023087736 W CN2023087736 W CN 2023087736W WO 2023198093 A1 WO2023198093 A1 WO 2023198093A1
Authority
WO
WIPO (PCT)
Prior art keywords
related information
model
target
reference value
communication device
Prior art date
Application number
PCT/CN2023/087736
Other languages
English (en)
French (fr)
Inventor
施源
Original Assignee
维沃移动通信有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 维沃移动通信有限公司 filed Critical 维沃移动通信有限公司
Publication of WO2023198093A1 publication Critical patent/WO2023198093A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0408—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas using two or more beams, i.e. beam diversity

Definitions

  • the present application belongs to the field of communication technology, and specifically relates to a method, device and communication equipment for determining the validity of a model.
  • AI Artificial Intelligence
  • the validity of the AI model can be verified to avoid the occurrence of AI model failure due to changes in internal and external factors.
  • the AI model validity verification process there is still a problem that the AI model is not valid. The problem of low accuracy of sexual verification results.
  • Embodiments of the present application provide a method, device, and communication device for determining model validity, which can ensure the accuracy of AI model validity verification results while implementing AI model validity verification.
  • a method for determining model validity including: a first communication device acquiring at least one beam group, each of the beam groups including at least one beam; the first communication device acquiring the at least one beam The first beam-related information corresponding to the beam group is input into the target AI model to predict at least one second beam-related information, and the second beam-related information is related to the beam group.
  • the first communication device performs performance verification on the target AI model based on the at least one second beam correlation information, or the first communication device sends the at least one second beam correlation to the second communication device Information, the at least one second beam related information is used by the second communication device to perform performance verification on the target AI model.
  • a method for determining model validity including any of the following: the second communication device receives at least one second beam correlation information sent by the first communication device, and based on the at least one second beam correlation The information performs performance verification on the target AI model; the second communication device receives the second indication information sent by the first communication device, and the second indication information is used to indicate whether the target AI model is valid or invalid.
  • a device for determining model validity including: an acquisition module, configured to acquire at least one beam group, each of the beam groups including at least one beam; and a prediction module, configured to obtain the at least one beam
  • the first beam-related information corresponding to the beam group is input into the target AI model respectively to predict at least one second beam-related information, and the second beam-related information corresponds to the beam group;
  • a verification module is used to predict according to the at least one
  • the second beam related information performs performance verification on the target AI model, or a sending module is configured to send the at least one second beam related information to the second communication device, where the at least one second beam related information is used for the second beam related information.
  • the second communication device performs performance verification on the target AI model.
  • a device for determining model validity including a receiving module for any of the following: receiving at least one second beam related information sent by the first communication device, and according to the at least one second beam The relevant information performs performance verification on the target AI model; and receives second indication information sent by the first communication device, where the second indication information is used to indicate whether the target AI model is valid or invalid.
  • a communication device in a fifth aspect, includes a processor and a memory.
  • the memory stores a program or instructions that can be run on the processor.
  • the program or instructions are implemented when executed by the processor. The steps of the method as described in the first aspect or the second aspect.
  • a communication device including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the first The steps of the method described in the second aspect, or the steps of implementing the method described in the second aspect.
  • a communication system including: a first communication device and a second communication device.
  • the first communication device can be used to perform the steps of the method described in the first aspect.
  • the second communication device can be used to perform the steps of the method as described in the first aspect. To perform the steps of the method described in the second aspect.
  • a readable storage medium is provided. Programs or instructions are stored on the readable storage medium. When the programs or instructions are executed by a processor, the steps of the method described in the first aspect are implemented, or the steps of the method are implemented as described in the first aspect. The steps of the method described in the second aspect.
  • a chip in a ninth aspect, includes a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement the method described in the first aspect. steps, or steps to implement the method described in the second aspect.
  • a computer program product/program product is provided, the computer program/program product is stored in a storage medium, and the computer program/program product is executed by at least one processor to implement as described in the first aspect
  • At least one beam group is obtained, and the first beam related information corresponding to the at least one beam group is used as the input of the target AI model to predict and obtain at least one second beam related information, and then according to the at least one second beam related information
  • the beam-related information verifies the performance of the target AI model. Therefore, on the one hand, it is possible to verify the effectiveness of the target AI model.
  • this embodiment uses at least one different beam group to verify the performance of the target AI model. , can avoid the problem of different model performance verification results due to different beam groups, and improve the accuracy of model performance verification results.
  • Figure 1 is a schematic structural diagram of a wireless communication system provided by an exemplary embodiment of the present application.
  • Figure 2 is a schematic flowchart of a method for determining model validity provided by an exemplary embodiment of the present application.
  • Figure 3 is a schematic flowchart of a method for determining model validity provided by another exemplary embodiment of the present application.
  • Figure 4 is a schematic flowchart of a method for determining model validity provided by yet another exemplary embodiment of the present application.
  • Figure 5 is a schematic structural diagram of a device for determining model validity provided by an exemplary embodiment of the present application.
  • Figure 6 is a schematic structural diagram of a device for determining model validity provided by another exemplary embodiment of the present application.
  • Figure 7 is a schematic structural diagram of a communication device provided by an exemplary embodiment of the present application.
  • Figure 8 is a schematic structural diagram of a terminal provided by an exemplary embodiment of the present application.
  • Figure 9 is a schematic structural diagram of a network-side device provided by an exemplary embodiment of the present application.
  • first, second, etc. in the description and claims of this application are used to distinguish similar objects and are not used to describe a specific order or sequence. It is to be understood that the terms so used are interchangeable under appropriate circumstances so that the embodiments of the present application can be practiced in sequences other than those illustrated or described herein, and that "first" and “second” are distinguished objects It is usually one type, and the number of objects is not limited.
  • the first object can be one or multiple.
  • “and/or” in the description and claims indicates at least one of the connected objects, and the character “/" generally indicates that the related objects are in an "or” relationship.
  • LTE Long Term Evolution
  • LTE-Advanced, LTE-A Long Term Evolution
  • LTE-A Long Term Evolution
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • SC-FDMA Single-carrier Frequency-Division Multiple Access
  • NR New Radio
  • FIG. 1 shows a block diagram of a wireless communication system to which embodiments of the present application are applicable.
  • the wireless communication system includes a terminal 11 and a network side device 12.
  • the terminal 11 may be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), or a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a palmtop computer, a netbook, or a super mobile personal computer.
  • Tablet Personal Computer Tablet Personal Computer
  • laptop computer laptop computer
  • PDA Personal Digital Assistant
  • PDA Personal Digital Assistant
  • UMPC ultra-mobile personal computer
  • UMPC mobile Internet device
  • MID mobile Internet Device
  • AR augmented reality
  • VR virtual reality
  • robots wearable devices
  • WUE Vehicle User Equipment
  • PUE Pedestrian User Equipment
  • smart home home equipment with wireless communication functions, such as refrigerators, TVs, washing machines or furniture, etc.
  • game consoles personal computers (personal computer, PC), teller machine or self-service machine and other terminal-side devices.
  • Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets) bracelets, smart anklets, etc.), smart wristbands, smart clothing, etc.
  • the network side device 12 may include an access network device or a core network device, where the access network device 12 may also be called a radio access network device, a radio access network (Radio Access Network, RAN), a radio access network function or Wireless access network unit.
  • the access network device 12 may include a base station, a Wireless Local Area Network (WLAN) access point or a Wireless Fidelity (WiFi) node, etc.
  • WLAN Wireless Local Area Network
  • WiFi Wireless Fidelity
  • the base station may be called a Node B, an Evolved Node B (eNB), Access point, Base Transceiver Station (BTS), Radio Base Station, Radio Transceiver, Basic Service Set (BSS), Extended Service Set (Extended Service) Set, ESS), home B-node, home evolved B-node, transmitting receiving point (Transmitting Receiving Point, TRP) or some other suitable term in the field.
  • eNB Evolved Node B
  • BTS Base Transceiver Station
  • BSS Basic Service Set
  • Extended Service Set Extended Service Set
  • TRP Transmitting Receiving Point
  • the base station is not limited to specific Regarding technical vocabulary, it should be noted that in the embodiment of this application, only the base station in the NR system is used as an example for introduction, and the specific type of the base station is not limited.
  • the method 200 may be, but is not limited to, executed by a first communication device. Specifically, it may be installed in the first communication device. hardware and/or software implementation. In this embodiment, the method 200 may include at least the following steps.
  • the first communication device acquires at least one beam group.
  • the first communication device and the second communication device mentioned later may be network side devices, such as base stations, etc., or they may be terminals or auxiliary network center units, etc., and the auxiliary network center unit is used for information interaction. unit.
  • each beam group acquired by the first communication device may include at least one beam.
  • the first communication device can select beams with beam IDs 1, 2, 3, and 4 as beam groups 1, 3, and 4.
  • the beams with beam IDs 4, 3, 2, and 1 are regarded as beam group 2
  • the beams with beam IDs 1, 4, 6, and 9 are regarded as beam group 3
  • the beams with beam IDs 1, 9, 18, and 30 are regarded as beam group 4. wait.
  • the first communication device obtains or selects the beam group, it can select it arbitrarily, or it can obtain it according to a pre-agreed or configured beam group acquisition method, which is not limited here.
  • the beam group may be the transmitting beam of the transmitting end, the receiving beam of the receiving end, the transmitting beam desired by the transmitting end, the transmitting beam desired by the receiving end, the receiving beam desired by the receiving end, the receiving beam desired by the receiving end.
  • the sending end may be a first communication device
  • the receiving end may be a second communication device, etc.
  • the beam group may also be called a beam pair, a beam combination, etc., which is not limited here.
  • the first communication device converts the first beam related information corresponding to the at least one beam group The information is input into the target AI model respectively to predict at least one second beam related information.
  • the target AI model may be a neural network, a decision tree, a support vector machine, a Bayesian classifier, etc., and is not limited here.
  • the first beam related information is used as input information of the target AI model, which may include but is not limited to at least one of the following (01)-(05).
  • the beam quality-related information mentioned in this application is information characterizing the beam quality, which may include but is not limited to Layer 1 (Layer 1, L1)-Signal-to-Noise and Interference Ratio, SINR), L1-Reference Signal Received Power (RSRP), L1-Reference Signal Received Quality (RSRQ), L3-SINR, L3-RSRP, L3-RSRQ, etc.
  • Layer 1 Layer 1, L1-Signal-to-Noise and Interference Ratio
  • SINR L1-Reference Signal Received Power
  • RSRQ L1-Reference Signal Received Quality
  • L3-SINR L3-RSRP
  • L3-RSRQ L3-SINR
  • L3-RSRP L3-RSRQ
  • Beam identification (ID) related information wherein the beam identification related information is used for the beam identity identification related information, including but not limited to at least one of the following: transmitting beam ID, receiving beam ID, the beam The corresponding reference signal set (set) ID, the reference signal resource (resource) ID corresponding to the beam, the uniquely identified random ID, the coded value processed by the additional AI model, etc.
  • Beam angle related information is used to represent the angle related information corresponding to the beam, including but not limited to at least one of the following: angle related information, transmitting angle related information, receiving angle related information, etc.
  • the aforementioned angle-related information is related information used to represent angles, such as angle, radian, index code value, code value processed by an additional AI model, etc.
  • Beam gain related information is used to characterize the gain related information of the beam and/or antenna, including but not limited to at least one of the following: antenna relative gain (unit dBi), omnidirectional radiation power, beam power spectrum (Effective Isotropic Radiated Power, EIRP), beam angle gain, beam angle gain spectrum (that is, the gain of a beam relative to different angles, including complete or partial gain spectrum information), EIRP corresponding to each beam angle, main lobe angle, side beam angle Lobe angle, number of side lobes, side lobe distribution, number of antennas, beam scanning horizontal coverage, beam scanning vertical coverage, 3dB width, 6dB width, etc.
  • the second beam related information is used as the output information of the target AI model, which corresponds to the beam group. For example, if the first communication device uses three beam groups to perform beam prediction, then the predicted second beam related information is three.
  • the second beam-related information may also include beam quality-related information, beam ID-related information, beam angle-related information, beam gain-related information, beam Width-related information, etc., will not be repeated here to avoid duplication.
  • the first beam related information and the second beam related information in addition to the aforementioned beam quality related information, beam ID related information, beam angle related information, beam gain related information, beam width related information, etc., you may also It is the time information related to it.
  • the first communication device performs performance verification on the target AI model based on the at least one second beam correlation information, or the first communication device sends the at least one second beam correlation to the second communication device.
  • Information, the at least one second beam related information is used by the second communication device to perform performance verification on the target AI model.
  • validation verification of the target AI model can also be understood as model testing, model adjustment, model fine-tuning, model updating, etc.
  • the first communication device predicts the relevant information of the second beam, it can independently conduct performance verification of the target AI model based on its own capabilities or communication needs, or it can also use the predicted second beam
  • the relevant information is sent to the second communication device for performance verification of the target AI model.
  • the first communication device obtains at least one beam group and uses the first beam-related information corresponding to the at least one beam group as the input of the target AI model to predict at least one second beam-related information, and then based on at least one
  • the second beam related information verifies the performance of the target AI model. Therefore, on the one hand, it is possible to verify the effectiveness of the target AI model; on the other hand, considering the second beam related information obtained by using different beam groups are different (that is, the performance of the target AI model obtained is different), but the tag beam related information corresponding to the different beam groups is the same, then this embodiment uses multiple different beam groups to perform performance verification of the target AI model, which can avoid the problem caused by the beam group.
  • the problem of different model performance verification results caused by different models improves the accuracy of model performance verification results. accuracy.
  • a schematic flowchart is provided for a method 300 for determining model validity according to an exemplary embodiment of the present application.
  • the method 300 may be, but is not limited to, executed by a first communication device. Specifically, it may be installed in the first communication device. hardware and/or software implementation. In this embodiment, the method 300 may include at least the following steps.
  • the first communication device acquires at least one beam group.
  • each beam group includes at least one beam.
  • the first communication device may acquire at least one beam group arbitrarily, or may acquire it as follows: At least one of (11)-(13) is obtained.
  • the first communication device acquires the at least one beam group in a manner agreed upon in the protocol.
  • the at least one beam group can be obtained according to the specific beam ID and other information stipulated in the protocol, or it can be obtained according to the beam group stipulated in the protocol.
  • the method or beam group acquisition condition (such as the first condition mentioned in the following (13)) is used to acquire the at least one beam group, which is not limited here.
  • the first communication device acquires the at least one beam group according to the received first indication information.
  • the first indication information may be an implicit indication or an explicit indication.
  • the reception timing of the first instruction information may also be determined by the aforementioned protocol agreement.
  • the first communication device acquires at least one beam group according to a first condition, wherein the first condition includes at least one of the following (131)-(135).
  • At least part of the beam ID related information corresponding to different beam groups is different. For example, assuming that the first communication device acquires beam group 1 and beam group 2, then beam group 1 includes beams identified as 1, 2, 3, and 5, and beam group 2 may include beams identified as 1, 2, and 5. 2, 3, 7 beams, etc.
  • At least part of the amount of beam ID related information corresponding to different beam groups is different. For example, assuming that the first communication device acquires beam group 1, beam group 2, and beam group 3, Then, the beam group 1 includes 5 beams, the beam group 2 includes 4 beams, and the beam group 3 includes 3 beams.
  • At least part of the order of the beam ID related information corresponding to different beam groups is different. For example, assuming that the first communication device acquires beam group 1 and beam group 2, then the beams included in the beam group 1 are in the order of 1, 2, 3, and 5, and the beams included in the beam group 2 are The order is 1, 2, 5, 3, etc.
  • the beam ID related information corresponding to different beam groups is obtained at different time periods. For example, assuming that the first communication device acquires beam group 1 and beam group 2, then the beam-related information corresponding to beam group 1 is acquired in the time period T1, and the beam-related information corresponding to beam group 2 is Obtained on time period T2.
  • beam groups with different beam combination conditions can belong to different beam groups.
  • the at least one beam group includes the beam group used when performing the first model verification on the target AI model, and/or the beam group used when using the target AI model.
  • Beam group the aforementioned "first model verification” can be understood as: assuming that the first communication device performs the 2nd, 3rd,...n times of performance verification on the target AI model, then at least one method used for performance verification
  • the beam group may include the beam group used in the first model verification of the target AI model.
  • the aforementioned "using the target AI model” can be understood as: the first communication device is using the target AI model to perform operations such as beam prediction (different from model performance verification).
  • the first communication device inputs the first beam related information corresponding to the at least one beam group into the target AI model to predict at least one second beam related information.
  • the second beam related information corresponds to the beam group.
  • the first communication device performs performance verification on the target AI model based on the at least one second beam related information.
  • the first communication device described in S330 performs the processing of the at least one second beam related information on the
  • the steps for performance verification of the target AI model may include S331 and S332 shown in Figure 3, as follows.
  • the first communication device matches the at least one second beam related information with the third beam related information respectively.
  • the third beam related information is tag beam related information corresponding to the at least one beam group (which can also be understood as real beam related information corresponding to the beam or actual beam related information, etc.). That is to say, when performing target AI model verification in this embodiment, the second beam related information obtained by using different beam groups is different (that is, the performance of the target AI model obtained is different), but the tag beams corresponding to different beam groups are related. The information has the same characteristics, and the predicted beam-related information corresponding to multiple different beam groups is matched with the real beam-related information, and then multiple different matching results are combined to verify the effectiveness of the target AI model and ensure The accuracy of verification results for the effectiveness of the target AI model.
  • the step of the first communication device matching the at least one second beam related information with the third beam related information may include any of the following (21)-(22) One item.
  • the first communication device For each second beam related information in the at least one second beam related information, the first communication device separately combines at least part of each second beam related information with the third beam related information. Make a match. That is to say, when the first communication device uses the second beam related information to verify the validity of the target AI model, it can match all the information in each of the second related information with the third beam related information, respectively. It is also possible to first select part of the second beam related information (such as beam quality related information, etc.), and then match the selected part of the information with the third beam related information respectively. This embodiment is not limited here.
  • the methods for selecting part of the information may be the same or different, and are not limited here.
  • the beam quality related information can be selected;
  • the beam identification related information can be selected, . . .
  • the first communication device For each second beam related information in the at least one second beam related information, the first communication device performs first processing on at least part of each of the second beam related information, and respectively processes The result is matched with the third beam related information. That is to say, when the first communication device uses the second beam related information to perform performance verification of the target AI model, it can first select at least part of the information from each of the second related information, and then perform at least part of the selected information. The first process is to finally match at least part of the information after the first process with the third beam related information respectively.
  • the first communication device performs a first processing step on at least part of each of the second beam related information, which may include At least one of the following (221)-(225).
  • the combination may be based on the corresponding beam group acquisition time, the combination may be based on the size of the beam quality-related information included in the second beam-related information, or the combination may be based on the acquisition time of the second beam-related information, etc., in There is no restriction on this.
  • the target beam quality related information includes the beam quality related information with the largest value, the at least part of the beam quality related information, the value greater than the th Any item of beam quality related information at a threshold.
  • the methods for selecting at least part of the information may be different or the same, and for at least part of the selected information, the first processing used may also be the same or different.
  • the specific selection method or first processing method can be implemented by protocol agreement or high-level configuration, and is not limited here.
  • S332 Determine whether the target AI model is valid or invalid according to the matching result.
  • the step of the first communication device determining whether the target AI model is valid or invalid based on the matching result may include at least one of the following (31)-(32).
  • the first reference value interval may be accuracy, probability of finding the optimal beam set, mean absolute error (Mean Absolute Error, MAE), mean squared error (MSE), normalized mean square error ( Normalized Mean Squared Error, NMSE), etc.
  • mean Absolute Error MAE
  • mean squared error MSE
  • Normalized mean square error Normalized Mean Squared Error, NMSE
  • the matching result is the difference between the second beam related information and the third beam related information
  • the target AI model is valid.
  • the matching result is the accuracy calculated based on the second beam related information and the third beam related information
  • the accuracy rate is within the first reference value interval (greater than or equal to the fourth threshold) , determine that the target AI model is effective.
  • the target can be determined when the probability of finding the optimal beam is located in the first reference value interval.
  • the AI model works.
  • Method 1 When there are multiple matching results and the multiple matching results are all located in the first reference value interval, determine that the target AI model is valid.
  • a matching result corresponds to a second beam related information, that is, a matching result is a result of matching the second beam related information and the third beam related information.
  • the second beam related information is a plurality of In the case of , if among the corresponding multiple matching results, they are all located in the first reference value interval, it is determined that the target AI model is valid.
  • Method 2 When there are multiple matching results and the number of matching results located in the first reference value interval reaches the second threshold, determine that the target AI model is valid.
  • a matching result corresponds to a second beam related information, that is, a matching result is a result of matching the second beam related information and the third beam related information.
  • the second beam related information is a plurality of In the case of , if 80% (ie, the second threshold) of the corresponding matching results are located in the first reference value interval, the target AI model is determined to be valid.
  • Method 3 When there are multiple matching results and the average of the multiple matching results is located in the first reference value interval, determine that the target AI model is valid.
  • the first communication device obtains the data according to the P time periods and different time periods correspond to The beam group of N beam groups obtains at least one beam group obtained by conditions. Then, when the matching result is located in the first reference value interval, the first communication device determines that the target AI model is valid, Including the following method 4 or method 5.
  • Method 4 Determine that the target AI model is valid when the target AI model is valid for L time periods, wherein the L time periods belong to the P time periods, and the L is greater than Or an integer equal to 0, L is less than or equal to P. For example, assuming that the at least one beam group is acquired over 5 (i.e., P) time periods, and the target AI model is effective at 3 (i.e., L) time periods, then it can be determined that the target AI model is effective .
  • the method for determining whether the target AI model is effective over the L time periods may include any of the following.
  • the matching results of the M1 second beam related information corresponding to the at least M1 beam groups in the time period and the third beam related information are located in the first
  • the M1 is an integer greater than or equal to 0, and M1 is less than or equal to N. That is to say, taking time period A among L time periods as an example, if 10 (i.e. N) beam groups are obtained in the time period, then if there are 8 (i.e. M1) beam groups corresponding to the 8th
  • the matching results of the second beam related information and the third beam related information are both located in the first reference value interval, then it is determined that the target AI model is valid in this time period A.
  • the mean value of the matching results of the M2 second beam related information corresponding to the at least M2 beam groups in the time period and the third beam related information is located in the In the case of the first reference value interval, it is determined that the target AI model is valid during this time period; wherein, the M2 is an integer greater than or equal to 0, and M2 is less than or equal to N. That is to say, taking time period B among L time periods as an example, if 10 (i.e. N) beam groups are obtained in the time period, then if there are 8 (i.e. M2) beam groups corresponding to the 8th If the mean value of the matching results of the two-beam related information and the third beam-related information is located in the first reference value interval, then it is determined that the target AI model is valid in this time period B.
  • Method 5 When the average of the matching results of the second beam related information and the third beam related information corresponding to S time periods is located in the first reference value interval, determine that the target AI model is valid, where, The S time periods belong to the P time periods, the S is an integer greater than or equal to 0, and S is less than or equal to P.
  • the method of determining or obtaining the first reference value may be different depending on the aforementioned matching results.
  • the method of obtaining the first reference value interval may include at least one of the following (41)-(44).
  • the first reference value interval can be directly configured through network side configuration or other methods, or a threshold or point value can be directly configured through protocol agreement, high-level configuration or network side configuration, and then the rules pre-configured by the first communication device And the threshold or point value determines the first reference value interval, which is not limited in this embodiment.
  • the first reference value interval is determined to be [80%, 100%], [85%, 100%] or [75%, 100%], etc., which is not limited here.
  • the first communication device can additionally interact with a set of beam data, which includes the input of the target AI model, the output of the AI model, the calculation method of loss, the tolerance range, etc., through this set of data, the target AI model can be verified. And/or obtain the first reference value interval for target AI model verification.
  • the model performance of the target AI model corresponding to the first beam group where the first beam group may be a beam group specified by protocol agreement or network side configuration.
  • the first communication device When obtaining the first reference value interval, the model performance of the target AI model can first be obtained based on the beam related information corresponding to the first beam group. For example, if the accuracy rate of the target AI model is 80%, then the first The communication device can determine the first reference value interval based on the "accuracy rate of 80%", such as [80%, 100%], [85%, 100%] or [75%, 100%], etc., which are not limited here.
  • the method of obtaining the third reference value interval includes at least one of the following (a)-(c).
  • the acquisition method of the third reference value interval defined in the aforementioned (a)-(c) is similar to the acquisition method of the first reference value interval defined in the aforementioned (41)-(43). In order to avoid duplication, I won’t go into details here.
  • the first auxiliary reference value can be obtained through interaction, such as protocol agreement, high-level configuration or network side configuration.
  • the method of the first communication device determining that the target AI model is invalid may also include the following methods 6 to 8, the contents are as follows.
  • Method 6 When there are multiple matching results and the multiple matching results are all located in the second reference value interval, determine that the target AI model is invalid.
  • Method 7 When there are multiple matching results and the number of matching results located in the second reference value interval reaches the fifth threshold, determine that the target AI model is invalid.
  • Method 8 When there are multiple matching results and the average of the multiple matching results is located in the second reference value interval, determine that the target AI model is invalid.
  • the first communication device obtains data based on the P time periods and different time periods correspond to The beam group of N beam groups obtains at least one beam group obtained by conditions. Then, when the matching result is located in the second reference value interval, the first communication device determines that the target AI model is invalid, Including the following method 9 or method 10.
  • Method 9 In the case where the target AI model fails in all L time periods, determine that the target AI model fails, wherein the L time periods belong to the P time periods, and the L is greater than Or an integer equal to 0, L is less than or equal to P.
  • the method for determining the failure of the target AI model in the L time periods may include any of the following.
  • the matching results of the M1 second beam related information corresponding to the at least M1 beam groups in the time period and the third beam related information are located in the first
  • the M1 is an integer greater than or equal to 0, and M1 is less than or equal to N.
  • the mean value of the matching results of the M2 second beam related information corresponding to the at least M2 beam groups in the time period and the third beam related information is located in the In the case of the second reference value interval, it is determined that the target AI model fails in this time period; wherein, the M2 is an integer greater than or equal to 0, and M2 is less than or equal to N.
  • Method 10 When the average of the matching results of the second beam related information corresponding to the S time periods and the third beam related information is located in the second reference value interval, it is determined that the target AI model is invalid, wherein, The S time periods belong to the P time periods, the S is an integer greater than or equal to 0, and S is less than or equal to P.
  • the beam group IDs range from 1-32
  • the target AI model input is related to 4 beam IDs
  • the output is the predicted beam quality of the 32 beam pairs (i.e., the second beam related information ).
  • the IDs of beam group 1 are 1, 2, 3, 4, the IDs of beam group 2 are 4, 3, 2, 1, the IDs of beam group 3 are 1, 4, 6, 9, and the IDs of beam group 4 are 1, 9, 18, 30.
  • the prediction result related information corresponding to the designated beam group 4 can be compared with the first reference value interval. If the prediction result related information is located in the first reference value interval, If the prediction result-related information is in the second reference value interval, it is determined that the performance of the target AI model is poor.
  • the prediction result-related information corresponding to the four beam groups can be compared with the first reference value interval respectively. If there are 2 (or 3/ 4) If the information related to the prediction results is located in the first reference value interval, it is determined that the performance of the target AI model is good. If there is 1 information related to the prediction results located in the first reference value interval, the performance of the target AI model is determined to be poor, or If 2 (or 3/4) prediction result-related information is located in the second reference value interval, it is determined that the performance of the target AI model is poor.
  • the acquisition method of the second reference value interval is also It may include at least one of the following: determined through signaling interaction; determined based on the verification results obtained during the first model verification of the target AI model; determined based on the model performance of the target AI model corresponding to the first beam group; Determined based on the fourth reference value interval and the second auxiliary reference value.
  • the acquisition method of the fourth reference value interval includes at least one of the following: through signaling interaction Obtained by the method; determined based on the verification results obtained during the first model verification of the target AI model; determined based on the model performance of the target AI model corresponding to the second beam group.
  • the method of obtaining the aforementioned second reference value is similar to the method of obtaining the aforementioned first reference value
  • the method of obtaining the aforementioned fourth reference value is similar to the method of obtaining the aforementioned third reference value. To avoid duplication, here No longer.
  • first threshold, second threshold, third threshold, fourth threshold, fifth threshold, P, S, L, M1, M2, etc. mentioned in this application can be determined interactively, as agreed in the agreement , high-level configuration or network-side configuration, etc., are not limited here.
  • the first communication device may send a second instruction to the second communication device after determining that the target AI model is valid or invalid.
  • the second indication information is used to indicate whether the target AI model is valid or invalid, thereby ensuring the consistency of the first communication device and the second communication device's understanding of the performance of the target AI model, Then determine the communication system performance.
  • a schematic flow chart of a method 400 for determining model validity is provided in an exemplary embodiment of the present application.
  • the method 400 can be, but is not limited to, executed by a second communication device. Specifically, it can be installed in the second communication device. hardware and/or software implementation. In this embodiment, the method 400 may include at least the following steps.
  • the second communication device receives at least one second beam related information sent by the first communication device, and performs performance verification on the target AI model based on the at least one second beam related information; or, the second communication device receives the first communication The second instruction information sent by the device.
  • the second indication information is used to indicate whether the target AI model is valid or invalid.
  • the second communication device can perform performance verification of the target AI model in the same manner as the method used to verify the performance of the target AI model in method embodiments 200-300, or it can also The use of performance verification methods different from the target AI model described in method embodiments 200-300 is not limited here.
  • the second beam related information includes at least one of the following: beam quality related information; beam ID related information; beam angle related information; beam gain related information; beam width related information.
  • each implementation manner mentioned in method embodiment 400 has the same or corresponding technical features as each implementation manner mentioned in the aforementioned method embodiment 200 and/or 300, therefore, the implementation manners mentioned in method embodiment 400
  • the implementation process of each implementation mode may refer to the relevant descriptions in method embodiments 200 and/or 300, and achieve the same or corresponding technical effects. To avoid duplication, they will not be described again here.
  • the execution subject may be a device for determining model validity.
  • the device for determining model validity is used as an example to illustrate the method for determining model validity performed by the device for determining model validity.
  • the device 500 includes: an acquisition module 510, configured to acquire at least one beam group. Each of the beam groups includes at least one beam; the prediction module 520 is used to respectively input the first beam related information corresponding to the at least one beam group into the target AI model to predict at least one second beam related information, the second beam related information Corresponding to the beam group; the verification module 530 is used to perform performance verification on the target AI model according to the at least one second beam related information, or the sending module 540 is used to send the at least one to the second communication device.
  • a second beam related information, the at least one second beam related information is used by the second communication device to perform performance verification on the target AI model.
  • the verification module 530 performs performance verification on the target AI model based on the at least one second beam related information, including: the verification module 530 separately compares the at least one second beam related information with The third beam related information is matched; and the target AI model is determined to be valid or invalid according to the matching result; wherein the third beam related information is the at least one Tag beam related information corresponding to each beam group.
  • the verification module 530 matches the at least one second beam related information with the third beam related information, including any of the following: for each third beam in the at least one second beam related information. For two beam related information, the verification module 530 respectively matches at least part of each second beam related information with the third beam related information; for each second beam related information in the at least one second beam related information, For beam-related information, the verification module 530 performs first processing on at least part of each of the second beam-related information, and matches the processing results with the third beam-related information respectively.
  • the verification module 530 performs a first processing step on at least part of each of the second beam related information, including at least one of the following: for each third of the at least one second beam related information.
  • Two beam related information select the beam quality related information with the largest value from a plurality of beam quality related information included in the second beam related information; for each second beam related information in the at least one second beam related information Information, select at least part of the beam quality-related information from a plurality of beam quality-related information included in the second beam-related information; for each second beam-related information in the at least one second beam-related information, select from the Select the beam quality-related information with a value greater than the first threshold from the plurality of beam quality-related information included in the second beam-related information; combine the at least one second beam-related information; and determine the beam ID corresponding to the target beam quality-related information.
  • the target beam quality related information includes any one of the beam quality related information with the largest value, the at least part of the beam quality related information, and the beam quality related information with the
  • the verification module 530 determines whether the target AI model is valid or invalid based on the matching result, including at least one of the following: when the matching result is within the first reference value interval, the verification module 530 It is determined that the target AI model is valid; when the matching result is located in the second reference value interval, the verification module 530 determines that the target AI model is invalid.
  • the verification module 530 determines that the target AI model is valid, including any of the following: when the matching results are multiple, and When multiple matching results are all located in the first reference value interval, determine the target The AI model is valid; when there are multiple matching results and the number of matching results located in the first reference value interval reaches the second threshold, it is determined that the target AI model is valid; when the matching result is If there are multiple matching results and the mean value of the multiple matching results is located in the first reference value interval, it is determined that the target AI model is valid.
  • the method for obtaining the first reference value interval includes at least one of the following: determining through signaling interaction; determining based on the verification results obtained during the first model verification of the target AI model; The model performance of the target AI model corresponding to a beam group is determined; determined based on the third reference value interval and the first auxiliary reference value; and/or, the second reference value interval is obtained in at least one of the following ways: through signaling The method of interaction is determined; determined based on the verification results obtained during the first model verification of the target AI model; determined based on the model performance of the target AI model corresponding to the first beam group; determined based on the fourth reference value interval and the second auxiliary The base value is determined.
  • the acquisition method of the third reference value interval includes at least one of the following: through signaling Obtained in an interactive manner; determined based on the verification results obtained during the first model verification of the target AI model; determined based on the model performance of the target AI model corresponding to the second beam group; and/or the second reference value
  • the acquisition method of the fourth reference value interval includes at least one of the following: acquisition through signaling interaction; according to the target The verification results obtained during the first model verification of the AI model are determined; it is determined based on the model performance of the target AI model corresponding to the second beam group.
  • the step for the acquisition module 510 to acquire at least one beam group includes at least one of the following: the acquisition module 510 acquires the at least one beam group according to the protocol stipulated; the acquisition module 510 acquires the at least one beam group according to the received
  • the first instruction information acquires the at least one beam group; the acquisition module 510 acquires at least one beam group according to a first condition, wherein the first condition includes at least one of the following: beam IDs corresponding to different beam groups At least part of the relevant information is different; at least part of the quantity of the beam ID related information corresponding to the different beam groups is different; at least part of the order of the beam ID related information corresponding to the different beam groups is different; different
  • the beam ID related information corresponding to the beam group is obtained in different time periods; in P time periods obtained on the above, and each of the time periods corresponds to N beam groups, where the P time periods are the P time periods closest to the performance verification of the target AI model by the verification module 530, and P and N are greater than 0 integer.
  • the time period is a distance from the first communication device.
  • the performance verification time of the target AI model is at least one recent time period.
  • the matching result is in the first reference value interval.
  • the verification module 530 determines that the target AI model is valid, including any of the following: when the target AI model is valid for L time periods, the verification module 530 determines that the target AI model is valid.
  • the AI model is valid, wherein the L time periods belong to the P time periods, the L is an integer greater than or equal to 0, and L is less than or equal to P; the second beam related information corresponding to the S time periods is the same as
  • the verification module 530 determines that the target AI model is valid, wherein the S time periods belong to the P time period, the S is an integer greater than or equal to 0, and S is less than or equal to P.
  • the target AI model is valid for the L time periods, including any of the following: for each of the L time periods, at least M1 beam groups in the time period When the matching results of the corresponding M1 second beam-related information and the third beam-related information are all located in the first reference value interval, it is determined that the target AI model is valid in this time period; for the L In each of the time periods, the mean value of the matching results of the M2 second beam-related information corresponding to the at least M2 beam groups in the time period and the third beam-related information is located at the first reference value. In the case of an interval, it is determined that the target AI model is valid during this time period; wherein, M1 and M2 are integers greater than or equal to 0, and M1 and M2 are less than or equal to N.
  • the first beam related information and/or the second beam related information includes at least one of the following: beam quality related information; beam ID related information; beam angle related information; beam increase Gain related information; beam width related information.
  • the at least one beam group includes the beam group used when performing the first model verification on the target AI model, and/or the beam group used when using the target AI model.
  • the sending module 540 is also configured to send second indication information to the second communication device, where the second indication information is used to indicate whether the target AI model is valid or invalid.
  • FIG. 6 it is a schematic structural diagram of a device 600 for determining model validity provided in an exemplary embodiment of the present application.
  • the device 600 includes: a receiving module 610, configured for any of the following: receiving a message sent by a first communication device. At least one second beam related information, and performance verification of the target AI model based on the at least one second beam related information; receiving second indication information sent by the first communication device, the second indication information is used to indicate the The target AI model is valid or invalid.
  • the second beam related information includes at least one of the following: beam quality related information; beam ID related information; beam angle related information; beam gain related information; beam width related information.
  • the device 500-600 for determining model validity in the embodiment of the present application may be a communication device, such as a terminal or a network-side device, or may be a component in the communication device, such as an integrated circuit or chip.
  • the terminal when the device 500-600 for determining the validity of the model is a terminal, the terminal may include but is not limited to the type of terminal 11 listed above, and when the device 500-600 for determining the validity of the model is a network side
  • the network side device may include but is not limited to the types of network side device 12 listed above, which are not specifically limited in the embodiment of this application.
  • model validity determination devices 500-600 provided by the embodiments of the present application can implement each process implemented by the method embodiments of Figures 2 to 4, and achieve the same technical effect. To avoid duplication, they will not be described again here.
  • this embodiment of the present application also provides a communication device 700, including a processor 701 and a memory 702.
  • the memory 702 stores programs or instructions that can be run on the processor 701, for example,
  • the communication device 700 is a terminal, when the program or instruction is executed by the processor 701, each step in the above method embodiments 200-400 is implemented, and the same technical effect can be achieved.
  • the communication device 700 is a network-side device, the above program or instruction is implemented when executed by the processor 701
  • Each step in the above-mentioned method embodiments 200-400 can achieve the same technical effect. To avoid repetition, it will not be described again here.
  • An embodiment of the present application also provides a terminal, including a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement the methods described in method embodiments 200-400. Method steps.
  • This terminal embodiment corresponds to the above-mentioned terminal-side method embodiment.
  • Each implementation process and implementation manner of the above-mentioned method embodiment can be applied to this terminal embodiment, and can achieve the same technical effect.
  • FIG. 8 is a schematic diagram of the hardware structure of a terminal that implements an embodiment of the present application.
  • the terminal 800 includes but is not limited to: radio frequency unit 801, network module 802, audio output unit 803, input unit 804, sensor 805, display unit 806, user input unit 807, interface unit 808, memory 809, processor 810, etc. at least some parts of it.
  • the terminal 800 may also include a power supply (such as a battery) that supplies power to various components.
  • the power supply may be logically connected to the processor 810 through a power management system, thereby managing charging, discharging, and power consumption through the power management system. Management and other functions.
  • the terminal structure shown in FIG. 8 does not constitute a limitation on the terminal.
  • the terminal may include more or fewer components than shown in the figure, or some components may be combined or arranged differently, which will not be described again here.
  • the input unit 804 may include a graphics processor (Graphics Processing Unit, GPU) 8041 and a microphone 8042.
  • the graphics processor 8041 is responsible for the image capture device (GPU) in the video capture mode or the image capture mode. Process the image data of still pictures or videos obtained by cameras (such as cameras).
  • the display unit 806 may include a display panel 8061, which may be configured in the form of a liquid crystal display, an organic light emitting diode, or the like.
  • the user input unit 807 includes a touch panel 8071 and at least one of other input devices 8072 .
  • Touch panel 8071 also known as touch screen.
  • the touch panel 8071 may include two parts: a touch detection device and a touch controller.
  • Other input devices 8072 may include but are not limited to physical keyboards, function keys (such as volume control keys, switch keys, etc.), trackballs, mice, and joysticks, which will not be described again here.
  • the radio frequency unit 801 after receiving downlink data from the network side device, can transmit it to the processor 810 for processing; in addition, the radio frequency unit 801 can send data to the network side device. Upstream data.
  • the radio frequency unit 801 includes, but is not limited to, an antenna, amplifier, transceiver, coupler, low noise amplifier, duplexer, etc.
  • Memory 809 may be used to store software programs or instructions as well as various data.
  • the memory 809 may mainly include a first storage area for storing programs or instructions and a second storage area for storing data, wherein the first storage area may store an operating system, an application program or instructions required for at least one function (such as a sound playback function, Image playback function, etc.) etc.
  • memory 809 may include volatile memory or non-volatile memory, or memory 809 may include both volatile and non-volatile memory.
  • non-volatile memory can be read-only memory (Read-Only Memory, ROM), programmable read-only memory (Programmable ROM, PROM), erasable programmable read-only memory (Erasable PROM, EPROM), electrically removable memory. Erase programmable read-only memory (Electrically EPROM, EEPROM) or flash memory.
  • Volatile memory can be random access memory (Random Access Memory, RAM), static random access memory (Static RAM, SRAM), dynamic random access memory (Dynamic RAM, DRAM), synchronous dynamic random access memory (Synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (Double Data Rate SDRAM, DDRSDRAM), enhanced synchronous dynamic random access memory (Enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (Synch link DRAM) , SLDRAM) and direct memory bus random access memory (Direct Rambus RAM, DRRAM).
  • RAM Random Access Memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • DRAM synchronous dynamic random access memory
  • SDRAM double data rate synchronous dynamic random access memory
  • Double Data Rate SDRAM Double Data Rate SDRAM
  • DDRSDRAM double data rate synchronous dynamic random access memory
  • Enhanced SDRAM, ESDRAM enhanced synchronous dynamic random access memory
  • Synch link DRAM synchronous link dynamic random access memory
  • SLDRAM direct memory bus
  • the processor 810 may include one or more processing units; optionally, the processor 810 integrates an application processor and a modem processor, where the application processor mainly handles operations related to the operating system, user interface, application programs, etc., Modem processors mainly process wireless communication signals, such as baseband processors. It can be understood that the above modem processor may not be integrated into the processor 810.
  • the processor 810 is configured to obtain at least one beam group, each of which includes at least one beam; and input the first beam-related information corresponding to the at least one beam group into the target AI respectively. model to predict and obtain at least one second beam-related information, the second beam-related information corresponding to the beam group; and the at least one second beam-related information performs performance verification on the target AI model, or by
  • the radio frequency unit 801 transmits information to the second communication device The at least one second beam related information is sent, and the at least one second beam related information is used by the second communication device to perform performance verification on the target AI model.
  • the radio frequency unit 801 is configured to receive at least one second beam related information sent by the first communication device, and perform performance verification on the target AI model based on the at least one second beam related information; or, receive Second indication information sent by the first communication device, the second indication information is used to indicate whether the target AI model is valid or invalid.
  • the first beam related information corresponding to the at least one beam group is used as the input of the target AI model to predict and obtain at least one second beam related information, and then according to the at least one second beam related information Information to verify the performance of the target AI model, thus, on the one hand, it is possible to verify the effectiveness of the target AI model, and on the other hand, considering that the second beam related information obtained by using different beam groups is different (i.e., obtained The performance of the target AI model is different), but the real values corresponding to different beam groups are the same. Then, this embodiment uses multiple different beam groups to verify the performance of the target AI model, which can avoid errors caused by different beam groups. The problem of different model performance verification results improves the accuracy of model performance verification results.
  • Embodiments of the present application also provide a network side device, including a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement as described in method embodiments 200-400. steps of the method described.
  • This network-side device embodiment corresponds to the above-mentioned network-side device method embodiment.
  • Each implementation process and implementation manner of the above-mentioned method embodiment can be applied to this network-side device embodiment, and can achieve the same technical effect.
  • the embodiment of the present application also provides a network side device.
  • the network side device 900 includes: an antenna 901, a radio frequency device 902, a baseband device 903, a processor 904 and a memory 905.
  • Antenna 901 is connected to radio frequency device 902.
  • the radio frequency device 902 receives information through the antenna 901 and sends the received information to the baseband device 903 for processing.
  • the baseband device 903 processes the information to be sent and sends it to the radio frequency device 902.
  • the radio frequency device 902 processes the received information and then sends it out through the antenna 901.
  • the method performed by the network side device in the above embodiment can be implemented in the baseband device 903, which includes a baseband processor.
  • the baseband device 903 may include, for example, at least one baseband board on which multiple chips are disposed, as shown in FIG. 9 .
  • One of the chips is, for example, a baseband processor, which is connected to the memory 905 through a bus interface to call the Program to perform the network device operations shown in the above method embodiments.
  • the network side device may also include a network interface 906, which is, for example, a common public radio interface (CPRI).
  • a network interface 906 which is, for example, a common public radio interface (CPRI).
  • CPRI common public radio interface
  • the network side device 900 in this embodiment of the present invention also includes: instructions or programs stored in the memory 905 and executable on the processor 904.
  • the processor 904 calls the instructions or programs in the memory 905 to execute Figure 5 or Figure 6
  • the execution methods of each module are shown and achieve the same technical effect. To avoid repetition, they will not be described in detail here.
  • Embodiments of the present application also provide a readable storage medium.
  • Programs or instructions are stored on the readable storage medium.
  • the program or instructions are executed by a processor, each process of the above method embodiments 200-400 is implemented, and can achieve The same technical effects are not repeated here to avoid repetition.
  • the processor is the processor in the terminal described in the above embodiment.
  • the readable storage medium includes computer readable storage media, such as computer read-only memory ROM, random access memory RAM, magnetic disk or optical disk, etc.
  • An embodiment of the present application further provides a chip.
  • the chip includes a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run network-side device programs or instructions to implement the above method embodiments.
  • Each process of 200-400 can achieve the same technical effect. To avoid repetition, we will not go into details here.
  • chips mentioned in the embodiments of this application may also be called system-on-chip, system-on-a-chip, system-on-chip or system-on-chip, etc.
  • Embodiments of the present application also provide a computer program product.
  • the computer program product includes a processor, a memory, and a program or instructions stored on the memory and executable on the processor.
  • the program or instructions are used by the processor.
  • An embodiment of the present application also provides a communication system, including: a first communication device and a second communication device.
  • the first communication device may be used to perform various steps of the method described in the above method embodiments 200-300
  • the second communication device may be used to perform the steps described in the method embodiment 400 above. various steps of the method.
  • the methods of the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is better. implementation.
  • the technical solution of the present application can be embodied in the form of a computer software product that is essentially or contributes to the existing technology.
  • the computer software product is stored in a storage medium (such as ROM/RAM, disk , CD), including several instructions to cause a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Signal Processing (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请公开了一种模型有效性的确定方法、装置及通信设备，属于通信技术领域，本申请实施例的模型有效性的确定方法包括：第一通信设备获取至少一个波束组，每个所述波束组中包括至少一个波束；所述第一通信设备将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，所述第一通信设备向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。

Description

模型有效性的确定方法、装置及通信设备
交叉引用
本发明要求在2022年04月13日提交中国专利局、申请号为202210384902.4、发明名称为“模型有效性的确定方法、装置及通信设备”的中国专利申请的优先权，该申请的全部内容通过引用结合在本发明中。
技术领域
本申请属于通信技术领域，具体涉及一种模型有效性的确定方法、装置及通信设备。
背景技术
随着人工智能(Artificial Intelligence，AI)的快速发展，其已在各个领域得到了广泛的应用，例如，对于通信领域，AI模块(如AI模型等)可部署在终端侧或网络侧，以用于波束信息预测、信道质量预测等。
其中，在AI模型的使用过程中，可通过AI模型的有效性验证来避免由于内外部因素变化而导致的AI模型失效问题的发生，但是在AI模型有效性验证过程中，依旧存在AI模型有效性验证结果准确性低的问题。
发明内容
本申请实施例提供一种模型有效性的确定方法、装置及通信设备，能够解决在实现AI模型有效性验证的同时，确保AI模型有效性验证结果的准确性。
第一方面，提供了一种模型有效性的确定方法，包括：第一通信设备获取至少一个波束组，每个所述波束组中包括至少一个波束；所述第一通信设备将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组 对应；所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，所述第一通信设备向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
第二方面，提供了一种模型有效性的确定方法，包括以下任一项：第二通信设备接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；第二通信设备接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
第三方面，提供了一种模型有效性的确定装置，包括：获取模块，用于获取至少一个波束组，每个所述波束组中包括至少一个波束；预测模块，用于将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；验证模块，用于根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，发送模块，用于向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
第四方面，提供了一种模型有效性的确定装置，包括接收模块，用于以下任一项：接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
第五方面，提供了一种通信设备，该通信设备包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如第一方面或第二方面所述的方法的步骤。
第六方面，提供了一种通信设备，包括处理器及通信接口，其中，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如第一 方面所述的方法的步骤，或实现如第二方面所述的方法的步骤。
第七方面，提供了一种通信系统，包括：第一通信设备及第二通信设备，所述第一通信设备可用于执行如第一方面所述的方法的步骤，所述第二通信设备可用于执行如第二方面所述的方法的步骤。
第八方面，提供了一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如第一方面所述的方法的步骤，或者实现如第二方面所述的方法的步骤。
第九方面，提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如第一方面所述的方法的步骤，或实现如第二方面所述的方法的步骤。
第十方面，提供了一种计算机程序产品/程序产品，所述计算机程序/程序产品被存储在存储介质中，所述计算机程序/程序产品被至少一个处理器执行以实现如第一方面所述的方法的步骤，或实现如第二方面所述的方法的步骤。
在本申请实施例中，通过获取至少一个波束组，将至少一个波束组对应的第一波束相关信息作为目标AI模型的输入，以预测得到至少一个第二波束相关信息，进而根据至少一个第二波束相关信息对目标AI模型的性能进行验证，由此，一方面，能够实现对目标AI模型有效性的验证，另一方面，本实施例利用至少一个不同的波束组进行目标AI模型的性能验证，能够避免由于波束组不同而导致的模型性能验证结果不同的问题，提高了模型性能验证结果的准确性。
附图说明
图1是本申请一示例性实施例提供的无线通信系统的结构示意图。
图2是本申请一示例性实施例提供的模型有效性的确定方法的流程示意图。
图3是本申请另一示例性实施例提供的模型有效性的确定方法的流程示意图。
图4是本申请又一示例性实施例提供的模型有效性的确定方法的流程示意图。
图5是本申请一示例性实施例提供的模型有效性的确定装置的结构示意图。
图6是本申请另一示例性实施例提供的模型有效性的确定装置的结构示意图。
图7是本申请一示例性实施例提供的通信设备的结构示意图。
图8是本申请一示例性实施例提供的终端的结构示意图。
图9是本申请一示例性实施例提供的网络侧设备的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员所获得的所有其他实施例，都属于本申请保护的范围。
本申请的说明书和权利要求书中的术语“第一”、“第二”等是用于区别类似的对象，而不用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换，以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施，且“第一”、“第二”所区别的对象通常为一类，并不限定对象的个数，例如第一对象可以是一个，也可以是多个。此外，说明书以及权利要求中“和/或”表示所连接对象的至少其中之一，字符“/”一般表示前后关联对象是一种“或”的关系。
值得指出的是，本申请实施例所描述的技术不限于长期演进型(Long Term Evolution，LTE)/LTE的演进(LTE-Advanced，LTE-A)系统，还可用于其他无线通信系统，诸如码分多址(Code Division Multiple Access，CDMA)、时分多址(Time Division Multiple Access，TDMA)、频分多址(Frequency Division Multiple Access，FDMA)、正交频分多址(Orthogonal Frequency  Division Multiple Access，OFDMA)、单载波频分多址(Single-carrier Frequency-Division Multiple Access，SC-FDMA)和其他系统。本申请实施例中的术语“系统”和“网络”常被可互换地使用，所描述的技术既可用于以上提及的系统和无线电技术，也可用于其他系统和无线电技术。以下描述出于示例目的描述了新空口(New Radio，NR)系统，并且在以下大部分描述中使用NR术语，但是这些技术也可应用于NR系统应用以外的应用，如第6代(6th Generation，6G)通信系统。
图1示出本申请实施例可应用的一种无线通信系统的框图。无线通信系统包括终端11和网络侧设备12。其中，终端11可以是手机、平板电脑(Tablet Personal Computer)、膝上型电脑(Laptop Computer)或称为笔记本电脑、个人数字助理(Personal Digital Assistant，PDA)、掌上电脑、上网本、超级移动个人计算机(ultra-mobile personal computer，UMPC)、移动上网装置(Mobile Internet Device，MID)、增强现实(augmented reality，AR)/虚拟现实(virtual reality，VR)设备、机器人、可穿戴式设备(Wearable Device)、车载设备(Vehicle User Equipment，VUE)、行人终端(Pedestrian User Equipment，PUE)、智能家居(具有无线通信功能的家居设备，如冰箱、电视、洗衣机或者家具等)、游戏机、个人计算机(personal computer，PC)、柜员机或者自助机等终端侧设备，可穿戴式设备包括：智能手表、智能手环、智能耳机、智能眼镜、智能首饰(智能手镯、智能手链、智能戒指、智能项链、智能脚镯、智能脚链等)、智能腕带、智能服装等。需要说明的是，在本申请实施例并不限定终端11的具体类型。网络侧设备12可以包括接入网设备或核心网设备，其中，接入网设备12也可以称为无线接入网设备、无线接入网(Radio Access Network,RAN)、无线接入网功能或无线接入网单元。接入网设备12可以包括基站、无线局域网(Wireless Local Area Network，WLAN)接入点或无线保真(Wireless Fidelity，WiFi)节点等，基站可被称为节点B、演进节点B(eNB)、接入点、基收发机站(Base Transceiver Station，BTS)、无线电基站、无线电收发机、基本服务集(Basic Service Set，BSS)、扩展服务集(Extended Service  Set，ESS)、家用B节点、家用演进型B节点、发送接收点(Transmitting Receiving Point，TRP)或所述领域中其他某个合适的术语，只要达到相同的技术效果，所述基站不限于特定技术词汇，需要说明的是，在本申请实施例中仅以NR系统中的基站为例进行介绍，并不限定基站的具体类型。
下面结合附图，通过一些实施例及其应用场景对本申请实施例提供的技术方案进行详细地说明。
如图2所示，为本申请一示例性实施例提供的模型有效性的确定方法200的流程示意图，该方法200可以但不限于由第一通信设备执行，具体可由安装于第一通信设备中的硬件和/或软件执行。本实施例中，所述方法200至少可以包括如下步骤。
S210，第一通信设备获取至少一个波束组。
其中，所述第一通信设备以及后续提及的第二通信设备可以是网络侧设备，如基站等，也可以是终端或辅助网络中心单元等，所述辅助网络中心单元是用于信息交互的单元。
基于此，在本实施例中，所述第一通信设备获取的每个所述波束组中可以包括至少一个波束。例如，假设一共有32个波束，各波束的标识(Identifier，ID)为1-32，那么，所述第一通信设备可以选取波束ID为1、2、3、4的波束作为波束组1、波束ID为4、3、2、1的波束作为波束组2、波束ID为1、4、6、9的波束作为波束组3、波束ID为1、9、18、30的波束作为波束组4等。当然，所述第一通信设备在获取或选取所述波束组时，可以任意选取，也可以按照预先约定或配置波束组获取方式进行获取，在此不做限制。
值得注意的是，所述波束组可以是所述发送端的发送波束、接收端的接收波束、发送端所期望的发送波束、接收端所期望的发送波束、接收端所期望的接收波束、接收端所期望的发送波束等，所述发送端可以是第一通信设备、所述接收端可以是第二通信设备等。另外，所述波束组也可以称作波束对、波束组合等，在此不做限制。
S220，所述第一通信设备将所述至少一个波束组对应的第一波束相关信 息分别输入目标AI模型，以预测得到至少一个第二波束相关信息。
其中，所述目标AI模型可以是神经网络、决策树、支持向量机、贝叶斯分类器等，在此不做限制。
可选的，在本实施例中，所述第一波束相关信息作为所述目标AI模型的输入信息，其可以包括但不限于以下(01)-(05)中的至少一项。
(01)波束质量相关信息。其中，本申请中提及的所述波束质量相关信息是表征波束质量的信息，其可以包括但不限于层1(Layer 1，L1)-信干噪比(Signal-to-Noise and Interference Ratio，SINR)、L1-参考信号接收功率(Reference Signal Received Power，RSRP)、L1-参考信号接收质量(Reference Signal Received Quality，RSRQ)、L3-SINR、L3-RSRP、L3-RSRQ等。
(02)波束标识(ID)相关信息，其中，所述波束标识相关信息用于所述波束身份识别的相关信息，包含但不限于以下至少之一：发送波束ID、接收波束ID、所述波束对应的参考信号集(set)ID、所述波束对应的参考信号资源(resource)ID、唯一标识的随机ID、额外AI模型处理后的编码值等。
(03)波束角度相关信息。其中，所述波束角度相关信息用于表征所述波束对应的角度相关信息，包含但不限于以下至少之一：角度相关信息、发送角度相关信息、接收角度相关信息等。前述的角度相关信息是用于表征角度的相关信息，例如，角度、弧度、索引编码值、额外AI模型处理后的编码值等。
(04)波束增益相关信息。其中，所述波束增益相关信息用于表征所述波束和/或天线的增益相关信息，包括但不限于以下至少之一：天线相对增益(单位dBi)、全向辐射功率波束功率谱(Effective Isotropic Radiated Power，EIRP)、波束角度增益、波束角度增益谱(也就是一个波束相对于不同角度上的增益，包括完整的或部分增益谱信息)、每个波束角度对应的EIRP、主瓣角度、副瓣角度、副瓣数量、副瓣分布、天线数量、波束扫描水平覆盖范围、波束扫描垂直覆盖范围、3dB宽度、6dB宽度等。
(05)波束宽度相关信息。
所述第二波束相关信息作为所述目标AI模型的输出信息，其与所述波束组对应。例如，所述第一通信设备利用3个波束组进行波束预测，那么预测得到的所述第二波束相关信息为3个。
基于此，在本实施例中，与所述第一波束相关信息对应，所述第二波束相关信息也可以包括波束质量相关信息、波束ID相关信息、波束角度相关信息、波束增益相关信息、波束宽度相关信息等，为避免重复，在此不再赘述。此外，对于所述第一波束相关信息和所述第二波束相关信息，除了前述的波束质量相关信息、波束ID相关信息、波束角度相关信息、波束增益相关信息、波束宽度相关信息等，还可以是与其相关的时间信息等。
S230，所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，所述第一通信设备向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
可以理解，前述的“目标AI模型的有效性验证”也可以理解为模型检验、模型调整、模型微调、模型更新等。
基于此，所述第一通信设备在预测得到所述第二波束相关信息后，可以根据自身能力或通信需求等，自主进行目标AI模型的性能验证，也可以将预测得到的所述第二波束相关信息发送给第二通信设备进行目标AI模型的性能验证。
本实施例中，第一通信设备通过获取至少一个波束组，将至少一个波束组对应的第一波束相关信息作为目标AI模型的输入，以预测得到至少一个第二波束相关信息，进而根据至少一个第二波束相关信息对目标AI模型的性能进行验证，由此，一方面，能够实现对目标AI模型有效性的验证；另一方面，考虑到利用不同的波束组获取到的第二波束相关信息不同(即获得的目标AI模型的性能不同)，但不同波束组对应的标签波束相关信息相同，那么，本实施例利用多个不同的波束组进行目标AI模型的性能验证，能够避免由于波束组不同而导致的模型性能验证结果不同的问题，提高了模型性能验证结果的 准确性。
如图3所示，为本申请一示例性实施例提供的模型有效性的确定方法300的流程示意图，该方法300可以但不限于由第一通信设备执行，具体可由安装于第一通信设备中的硬件和/或软件执行。本实施例中，所述方法300至少可以包括如下步骤。
S310，第一通信设备获取至少一个波束组。
其中，每个所述波束组中包括至少一个波束。
可以理解，S310的实现过程除了可参照方法实施例200中的相关描述之外，作为一种可能的实现方式，所述第一通信设备获取至少一个波束组可以是任意获取，也可以是按照以下(11)-(13)中的至少一项获取。
(11)所述第一通信设备根据协议约定的方式获取所述至少一个波束组。
其中，所述第一通信设备根据协议约定的方式获取所述至少一个波束组时，可以根据协议约定的具体的波束ID等信息获取所述至少一个波束组，也可以根据协议约定的波束组获取方式或波束组获取条件(如后续(13)中提及的第一条件)获取所述至少一个波束组，在此不做限制。
(12)所述第一通信设备根据接收到的第一指示信息获取所述至少一个波束组。
其中，所述第一指示信息可以是隐式指示也可以显式指示。另外，所述第一指示信息的接收时机等也可以通过前述的协议约定的方式确定。
(13)所述第一通信设备根据第一条件获取至少一个波束组，其中，所述第一条件包括以下(131)-(135)中的至少一项。
(131)不同的所述波束组对应的波束ID相关信息中的至少部分不同。例如，假设所述第一通信设备获取有波束组1和波束组2，那么，所述波束组1包括标识为1、2、3、5的波束，所述波束组2可以包括标识为1、2、3、7的波束等。
(132)不同的所述波束组对应的波束ID相关信息的数量中的至少部分不同。例如，假设所述第一通信设备获取有波束组1、波束组2、波束组3， 那么，所述波束组1包括5个波束，所述波束组2包括标识为4个波束，所述波束组3包括标识为3个波束。
(133)不同的所述波束组对应的波束ID相关信息的顺序中的至少部分不同。例如，假设所述第一通信设备获取有波束组1和波束组2，那么，所述波束组1包括的波束的顺序为1、2、3、5的波束，所述波束组2包括的波束的顺序为1、2、5、3等。
(134)不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得。例如，假设所述第一通信设备获取有波束组1和波束组2，那么，所述波束组1对应的波束相关信息是在时间周期T1上获取，所述波束组2对应的波束相关信息是在时间周期T2上获取。
(135)在P个时间周期上获取、且每个所述时间周期对应N个波束组，其中，所述P个时间周期是距离所述第一通信设备进行目标AI模型的性能验证最近的P个时间周期，P、N为大于0的整数。
值得注意的是，对于前述的波束组获取条件，拥有不同波束组合条件的波束组可以属于不同的波束组。
当然，在一种实现方式中，所述至少一个波束组中包括对所述目标AI模型进行第一次模型验证时所采用的波束组，和/或，使用所述目标AI模型时所采用的波束组。其中，对于前述的“第一次模型验证”可以理解为：假设第一通信设备在对目标AI模型进行第2、3，……n次性能验证时，那么，进行性能验证所采用的至少一个波束组中可以包括对所述目标AI模型进行第一次模型验证时所采用的波束组。相应的，前述的“使用所述目标AI模型”可以理解为：所述第一通信设备在利用所述目标AI模型进行波束预测等操作(不同于模型性能验证)。
S320，所述第一通信设备将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息。
其中，所述第二波束相关信息与所述波束组对应。
可以理解，S320的实现过程可以参照方法实施例200中的相关描述，为 避免重复，在此不再赘述。
S330，所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证。
可以理解，S330的实现过程除了参照方法实施例200中的相关描述之外，作为一种可能的实现方式，S330中所述的第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证的步骤可以包括图3中所示的S331和S332，内容如下。
S331，所述第一通信设备将所述至少一个第二波束相关信息分别与第三波束相关信息进行匹配。
其中，所述第三波束相关信息是所述至少一个波束组对应的标签波束相关信息(也可以理解为波束对应的真实波束相关信息或实际波束相关信息等)。也就是说，本实施例在进行目标AI模型验证时，利用不同的波束组获取到的第二波束相关信息不同(即获得的目标AI模型的性能不同)，但不同波束组对应的标签波束相关信息相同的特性，将预测得到的多个不同波束组对应的波束相关信息分别与真实的波束相关信息进行匹配，进而综合多个不同的匹配结果，实现对目标AI模型的有效性的验证，确保对目标AI模型的有效性的验证结果的准确性。
基于此，在一种实现方式中，所述第一通信设备将所述至少一个第二波束相关信息与第三波束相关信息进行匹配的步骤，可以包括以下(21)-(22)中的任一项。
(21)针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述第一通信设备分别将各所述第二波束相关信息中的至少部分与所述第三波束相关信息进行匹配。也就是说，所述第一通信设备在利用第二波束相关信息进行目标AI模型的有效性验证时，可以将各所述第二相关信息中的全部信息分别与第三波束相关信息进行匹配，也可以先从各所述第二波束相关信息中选取部分(如波束质量相关信息等)，再将选取的部分信息分别与第三波束相关信息进行匹配，本实施例在此不做限制。
当然针对不同的第二波束相关信息，进行部分信息选取的方式可以相同也可以不同，在此不做限制。例如，对于第二波束相关信息A，可以选取其中的波束质量相关信息；对于第二波束相关信息B，可以选取其中的波束标识相关信息，……。
(22)针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述第一通信设备对各所述第二波束相关信息中的至少部分进行第一处理，以及分别将处理结果与所述第三波束相关信息进行匹配。也就是说，所述第一通信设备在利用第二波束相关信息进行目标AI模型的性能验证时，可以先从各所述第二相关信息中选取至少部分信息，在对选取的至少部分信息进行第一处理，最后将经过第一处理后的至少部分信息分别与第三波束相关信息进行匹配。
一种实现方式中，假设所述第二波束相关信息中包括波束质量相关信息，那么所述第一通信设备对各所述第二波束相关信息中的至少部分进行第一处理的步骤，可以包括以下(221)-(225)中的至少一项。
(221)对于所述至少一个第二波束相关信息中的每个第二波束相关信息，分别从所述第二波束相关信息包括的多个波束质量相关信息中选取取值最大的波束质量相关信息。
(222)对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取至少部分波束质量相关信息。
(223)对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值大于第一阈值的波束质量相关信息。
(224)对所述至少一个第二波束相关信息进行组合。例如，可以按照对应的波束组获取时间进行组合，也可以按照第二波束相关信息中包括的波束质量相关信息的大小进行组合，还可以按照第二波束相关信息的获取时间及进行组合等，在此不做限制。
(225)确定目标波束质量相关信息对应的波束ID相关信息，所述目标波束质量相关信息包括所述取值最大的波束质量相关信息、所述至少部分波束质量相关信息、所述取值大于第一阈值的波束质量相关信息中的任一项。
值得注意的是，针对不同的第二波束相关信息，进行至少部分信息选取的方式可以不同或相同，以及针对选取的至少部分信息，所采用的第一处理也可以相同或不同。当然，具体采用哪种选取方式或哪种第一处理方式，可以由协议约定或高层配置等方式实现，在此不做限制。
S332，根据匹配结果确定所述目标AI模型有效或无效。
一种实现方式中，所述第一通信设备根据匹配结果确定所述目标AI模型有效或无效的步骤，可以包括以下(31)-(32)中的至少一项。
(31)在所述匹配结果位于第一基准值区间的情况下，确定所述目标AI模型有效。
其中，所述第一基准值区间可以是正确率、找到最优波束集合的概率、平均绝对误差(Mean Absolute Error，MAE)、均方差(Mean Squared Error，MSE)、规一化均方误差(Normalized Mean Squared Error，NMSE)等。
例如，假设匹配结果为所述第二波束相关信息与第三波束相关信息之间的差值，那么，可在所述差值位于第一基准值区间(小于或等于第三阈值)时，确定所述目标AI模型有效。
又例如，假设匹配结果为根据所述第二波束相关信息与第三波束相关信息计算得到的准确率，那么，可在所述准确率位于第一基准值区间(大于或等第四阈值)时，确定所述目标AI模型有效。
又例如，假设匹配结果为根据所述第二波束相关信息与第三波束相关信息找到最优波束的概率，那么，可在找到最优波束的概率位于第一基准值区间时，确定所述目标AI模型有效。
基于此，作为一种实现方式，所述在所述匹配结果位于第一基准值区间的情况下，确定所述目标AI模型有效的方式可以有多种，下面结合方式1-方式3对其实现过程进行说明。
方式1：在所述匹配结果为多个、且多个所述匹配结果均位于第一基准值区间的情况下，确定所述目标AI模型有效。
可以理解，一个匹配结果对应的一个第二波束相关信息，即一个匹配结果是一个第二波束相关信息与第三波束相关信息进行匹配的结果，那么，在所述第二波束相关信息为多个的情况下，对应的多个匹配结果中，如果其均位于所述第一基准值区间的情况下，确定所述目标AI模型有效。
方式2：在所述匹配结果为多个、且位于所述第一基准值区间的匹配结果的数量达到第二阈值的情况下，确定所述目标AI模型有效。
可以理解，一个匹配结果对应的一个第二波束相关信息，即一个匹配结果是一个第二波束相关信息与第三波束相关信息进行匹配的结果，那么，在所述第二波束相关信息为多个的情况下，对应的多个匹配结果中，如果存在80％(即第二阈值)的匹配结果位于所述第一基准值区间，确定所述目标AI模型有效。
方式3：在所述的匹配结果为多个、且多个所述匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效。
值得注意的是，除了前述的方式1-方式3之外，在一种可能的实现方式中，如果所述第一通信设备是根据所述在P个时间周期上获取、且不同的时间周期对应N个波束组这一波束组获取条件获取的至少一个波束组，那么，在所述匹配结果位于第一基准值区间的情况下，所述第一通信设备确定所述目标AI模型有效的步骤，包括以下方式4或方式5。
方式4：在所述目标AI模型在L个时间周期上均有效的情况下，确定所述目标AI模型有效，其中，所述L个时间周期属于所述P个时间周期，所述L为大于或等于0的整数，L小于或等于P。例如，假设所述至少一个波束组是在5(即P个)时间周期上获取，且所诉目标AI模型在3(即L)个时间周期上有效，那么，可以确定所述目标AI模型有效。
可选的，所述目标AI模型在所述L个时间周期上有效的确定方式可以包括以下任一项。
针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M1个波束组对应的M1个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效，其中，所述M1为大于或等于0的整数，M1小于或等于N。也就是说，以L个时间周期中的时间周期A为例，如果在时间周期上获取到10(即N)个波束组，那么，如果存在8(即M1)个波束组对应的8个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第一基准值区间，那么，确定所述目标AI模型在该时间周期A上有效。
针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M2个波束组对应的M2个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；其中，所述M2为大于或等于0的整数，M2小于或等于N。也就是说，以L个时间周期中的时间周期B为例，如果在时间周期上获取到10(即N)个波束组，那么，如果存在8(即M2)个波束组对应的8个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间，那么，确定所述目标AI模型在该时间周期B上有效。
方式5：在S个时间周期对应的第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效，其中，所述S个时间周期属于所述P个时间周期，所述S为大于或等于0的整数，S小于或等于P。
值得注意的是，所述第一通信设备采用前述方式1-方式5中的哪一个进行模型性能验证，可以由协议约定、高层配置或网络侧配置实现，在此不做限制。
一种实现方式中，根据前述的匹配结果的不同，所述第一基准值的确定方式或获取方式可以不同。在本实施例中，所述第一基准值区间的获取方式可以包括以下(41)-(44)中的至少一项。
(41)通过信令交互的方式确定。例如，可以通过协议约定、高层配置 或网络侧配置等方式直接配置所述第一基准值区间，也可以通过协议约定、高层配置或网络侧配置等方式直接配置一个阈值或点值，再由所述第一通信设备预先配置的规则以及所述阈值或点值确定所述第一基准值区间，本实施例在此不做限制。
(42)根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定。其中，假设所述第一通信设设备进行第一次模型验证时得到的验证结果为所述目标AI模型的准确率为80％，那么，所述第一通信设备可根据“准确率80％”确定第一基准值区间为[80％，100％]、[85％，100％]或[75％，100％]等，在此不做限制。
又或者，所述第一通信设备通过额外交互一组波束数据，其包括目标AI模型的输入、AI模型的输出、loss的计算方法，容差范围等，通过该组数据，可以验证目标AI模型和/或获得目标AI模型验证时的第一基准值区间。
(43)根据第一波束组对应的目标AI模型的模型性能确定，其中，所述第一波束组可以是协议约定或网络侧配置等方式指定的波束组，基于此，所述第一通信设备在获取第一基准值区间时，可先根据所述第一波束组对应的波束相关信息获取目标AI模型的模型性能，如所述目标AI模型的准确率为80％，那么，所述第一通信设备可根据“准确率80％”确定第一基准值区间，如[80％，100％]、[85％，100％]或[75％，100％]等，在此不做限制。
(44)根据第三基准值区间与第一辅助基准值确定。其中，所述第三基准值区间的获取方式包括以下(a)-(c)中的至少一项。
(a)通过信令交互的方式获取。
(b)根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定。
(c)根据第二波束组对应的目标AI模型的模型性能确定。
可以理解，前述(a)-(c)中所限定的第三基准值区间的获取方式与前述(41)-(43)中所限定的第一基准值区间的获取方式类似，为避免重复，在此不再赘述。
此外，所述第一辅助基准值可以通过交互的方式获取，如协议约定、高层配置或网络侧配置等方式。
(32)在所述匹配结果位于第二基准值区间的情况下，确定所述目标AI模型失效。
其中，与前述(31)中给出的验证目标AI模型有效的方式类似，所述第一通信设备确定所述目标AI模型失效的方式也可以包括以下方式6-方式8，内容如下。
方式6：在所述匹配结果为多个、且多个所述匹配结果均位于第二基准值区间的情况下，确定所述目标AI模型失效。
方式7：在所述匹配结果为多个、且位于所述第二基准值区间的匹配结果的数量达到第五阈值的情况下，确定所述目标AI模型失效。
方式8：在所述的匹配结果为多个、且多个所述匹配结果的均值位于所述第二基准值区间的情况下，确定所述目标AI模型失效。
值得注意的是，除了前述的方式6-方式8之外，在一种可能的实现方式中，如果所述第一通信设备是根据所述在P个时间周期上获取、且不同的时间周期对应N个波束组这一波束组获取条件获取的至少一个波束组，那么，在所述匹配结果位于第二基准值区间的情况下，所述第一通信设备确定所述目标AI模型失效的步骤，包括以下方式9或方式10。
方式9：在所述目标AI模型在L个时间周期上均失效的情况下，确定所述目标AI模型失效，其中，所述L个时间周期属于所述P个时间周期，所述L为大于或等于0的整数，L小于或等于P。
可选的，所述目标AI模型在所述L个时间周期上失效的确定方式可以包括以下任一项。
针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M1个波束组对应的M1个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第二基准值区间的情况下，确定所述目标AI模型在该时间周期上失效，其中，所述M1为大于或等于0的整数，M1小于或等于N。
针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M2个波束组对应的M2个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第二基准值区间的情况下，确定所述目标AI模型在该时间周期上失效；其中，所述M2为大于或等于0的整数，M2小于或等于N。
方式10：在S个时间周期对应的第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第二基准值区间的情况下，确定所述目标AI模型失效，其中，所述S个时间周期属于所述P个时间周期，所述S为大于或等于0的整数，S小于或等于P。
值得注意的是，所述第一通信设备采用前述方式6-方式10中的哪一个进行模型性能验证，可以由协议约定、高层配置或网络侧配置实现，在此不做限制。另外，关于前述方式6-方式10的实现过程课参照前述方式1-方式5中的相关描述，在此不再赘述。
示例性的，假设一共有32个波束对，波束组的ID从1-32，目标AI模型输入与4个波束ID相关，输出为预测的32个波束对的波束质量(即第二波束相关信息)。其中，波束组1的ID为1、2、3、4，波束组2的ID为4、3、2、1，波束组3的ID为1、4、6、9，波束组4的ID为1、9、18、30。那么，可以将指定的波束组4对应的预测结果相关信息(即第二波束相关信息与第三波束相关信息的匹配结果)与第一基准值区间进行对比，如果预测结果相关信息位于第一基准值区间，则确定目标AI模型的性能较好，如果预测结果相关信息位于第二基准值区间，则确定目标AI模型的性能较差。
或者，也可以将4个波束组对应的预测结果相关信息(即第二波束相关信息与第三波束相关信息的匹配结果)分别与第一基准值区间进行对比，如果其中存在2(或3/4)个预测结果相关信息位于第一基准值区间，则确定目标AI模型的性能较好，如果存在1个预测结果相关信息位于第一基准值区间，则确定目标AI模型的性能较差，或者如果2(或3/4)个预测结果相关信息位于第二基准值区间，则确定目标AI模型的性能较差。
此外，与前述第一基准值区间对应，所述第二基准值区间的获取方式也 可以包括以下至少一项：通过信令交互的方式确定；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第一波束组对应的目标AI模型的模型性能确定；根据第四基准值区间与第二辅助基准值确定。其中，所述第二基准值区间在根据所述第四基准值区间与第二辅助基准值确定的情况下，所述第四基准值区间的获取方式包括以下至少一项：通过信令交互的方式获取；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第二波束组对应的目标AI模型的模型性能确定。可以理解，前述的第二基准值的获取方式与前述第一基准值的获取方式类似，前述的第四基准值的获取方式与前述的第三基准值的获取方式类似，为避免重复，在此不再赘述。
值得注意的是，本申请中提及的第一阈值、第二阈值、第三阈值、第四阈值、第五阈值、P、S、L、M1、M2等可以通过交互方式确定，如协议约定、高层配置或网络侧配置等，在此不做限制。
当然，基于前述对模型有效性的确定过程的描述，作为一种可能的实现方式，所述第一通信设备在确定所述目标AI模型有效或无效后，可以向第二通信设备发送第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效，由此确保所述第一通信设备和所述第二通信设备对所述目标AI模型的性能的理解的一致性，进而确定通信系统性能。
如图4所示，为本申请一示例性实施例提供的模型有效性的确定方法400的流程示意图，该方法400可以但不限于由第二通信设备执行，具体可由安装于第二通信设备中的硬件和/或软件执行。本实施例中，所述方法400至少可以包括如下步骤。
S410，第二通信设备接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；或，第二通信设备接收第一通信设备发送的第二指示信息。
其中，所述第二指示信息用于指示所述目标AI模型有效或无效。需要说 明的是，在本实施例中，所述第二通信设备可以采用与方法实施例200-300中对目标AI模型的性能进行验证的方式相同的方式，进行目标AI模型的性能验证，也可以采用不同于方法实施例200-300中所述的目标AI模型的性能验证方式，在此不做限制。
可选的，所述第二波束相关信息包括以下至少一项：波束质量相关信息；波束ID相关信息；波束角度相关信息；波束增益相关信息；波束宽度相关信息。
可以理解，由于方法实施例400中提及的各实现方式具有与前述方法实施例200和/或300中提及的各实现方式相同或相应的技术特征，因此，方法实施例400中提及的各实现方式的实现过程可参照方法实施例200和/或300中的相关描述，并达到相同或相应的技术效果，为避免重复，在此不再赘述。
本申请实施例提供的模型有效性的确定方法200-400，执行主体可以为模型有效性的确定装置。本申请实施例中以模型有效性的确定装置执模型有效性的确定方法为例，说明本申请实施例提供的模型有效性的确定装置。
如图5所示，为本申请一示例性实施例提供的模型有效性的确定装置500的结构示意图，该装置500包括：获取模块510，用于获取至少一个波束组，每个所述波束组中包括至少一个波束；预测模块520，用于将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；验证模块530，用于根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，发送模块540，用于向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
可选的，所述验证模块530根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证的步骤，包括：所述验证模块530将所述至少一个第二波束相关信息分别与第三波束相关信息进行匹配；以及根据匹配结果确定所述目标AI模型有效或无效；其中，所述第三波束相关信息是所述至少一 个波束组对应的标签波束相关信息。
可选的，所述验证模块530将所述至少一个第二波束相关信息与第三波束相关信息进行匹配的步骤，包括以下任一项：针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述验证模块530分别将各所述第二波束相关信息中的至少部分与所述第三波束相关信息进行匹配；针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述验证模块530对各所述第二波束相关信息中的至少部分进行第一处理，以及分别将处理结果与所述第三波束相关信息进行匹配。
可选的，所述验证模块530对各所述第二波束相关信息中的至少部分进行第一处理的步骤，包括以下至少一项：对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值最大的波束质量相关信息；对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取至少部分波束质量相关信息；对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值大于第一阈值的波束质量相关信息；对所述至少一个第二波束相关信息进行组合；确定目标波束质量相关信息对应的波束ID相关信息，所述目标波束质量相关信息包括所述取值最大的波束质量相关信息、所述至少部分波束质量相关信息、所述取值大于第一阈值的波束质量相关信息中的任一项。
可选的，所述验证模块530根据匹配结果确定所述目标AI模型有效或无效的步骤，包括以下至少一项：在所述匹配结果位于第一基准值区间的情况下，所述验证模块530确定所述目标AI模型有效；在所述匹配结果位于第二基准值区间的情况下，所述验证模块530确定所述目标AI模型失效。
可选的，在所述匹配结果位于第一基准值区间的情况下，所述验证模块530确定所述目标AI模型有效的步骤，包括以下任一项：在所述匹配结果为多个、且多个所述匹配结果均位于第一基准值区间的情况下，确定所述目标 AI模型有效；在所述匹配结果为多个、且位于所述第一基准值区间的匹配结果的数量达到第二阈值的情况下，确定所述目标AI模型有效；在所述的匹配结果为多个、且多个所述匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效。
可选的，所述第一基准值区间的获取方式包括以下至少一项：通过信令交互的方式确定；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第一波束组对应的目标AI模型的模型性能确定；根据第三基准值区间与第一辅助基准值确定；和/或，所述第二基准值区间的获取方式包括以下至少一项：通过信令交互的方式确定；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第一波束组对应的目标AI模型的模型性能确定；根据第四基准值区间与第二辅助基准值确定。
可选的，所述第一基准值区间在根据所述第三基准值区间与第一辅助基准值确定的情况下，所述第三基准值区间的获取方式包括以下至少一项：通过信令交互的方式获取；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第二波束组对应的目标AI模型的模型性能确定；和/或，所述第二基准值区间在根据所述第四基准值区间与第二辅助基准值确定的情况下，所述第四基准值区间的获取方式包括以下至少一项：通过信令交互的方式获取；根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；根据第二波束组对应的目标AI模型的模型性能确定。
可选的，所述获取模块510获取至少一个波束组的步骤，包括以下至少一项：所述获取模块510根据协议约定的方式获取所述至少一个波束组；所述获取模块510根据接收到的第一指示信息获取所述至少一个波束组；所述获取模块510根据第一条件获取至少一个波束组，其中，所述第一条件包括以下至少一项：不同的所述波束组对应的波束ID相关信息中的至少部分不同；不同的所述波束组对应的波束ID相关信息的数量中的至少部分不同；不同的所述波束组对应的波束ID相关信息的顺序中的至少部分不同；不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得；在P个时间周期 上获取、且每个所述时间周期对应N个波束组，其中，所述P个时间周期是距离所述验证模块530进行目标AI模型的性能验证最近的P个时间周期，P、N为大于0的整数。
可选的，在所述第一条件包括所述不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得的情况下，所述时间周期是距离所述第一通信设备进行目标AI模型的性能验证的时间最近的至少一个时间周期。
可选的，在所述第一条件包括所述在P个时间周期上获取、且不同的时间周期对应N个波束组的情况下，所述在所述匹配结果位于第一基准值区间的情况下，所述验证模块530确定所述目标AI模型有效的步骤，包括以下任一项：在所述目标AI模型在L个时间周期上均有效的情况下，所述验证模块530确定所述目标AI模型有效，其中，所述L个时间周期属于所述P个时间周期，所述L为大于或等于0的整数，L小于或等于P；在S个时间周期对应的第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，所述验证模块530确定所述目标AI模型有效，其中，所述S个时间周期属于所述P个时间周期，所述S为大于或等于0的整数，S小于或等于P。
可选的，所述目标AI模型在所述L个时间周期上有效，包括以下任一项：针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M1个波束组对应的M1个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M2个波束组对应的M2个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；其中，所述M1、M2为大于或等于0的整数，M1、M2小于或等于N。
可选的，所述第一波束相关信息和/或所述第二波束相关信息包括以下至少一项：波束质量相关信息；波束ID相关信息；波束角度相关信息；波束增 益相关信息；波束宽度相关信息。
可选的，所述至少一个波束组中包括对所述目标AI模型进行第一次模型验证时所采用的波束组，和/或，使用所述目标AI模型时所采用的波束组。
可选的，所述发送模块540还用于向第二通信设备发送第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
如图6所示，为本申请一示例性实施例提供的模型有效性的确定装置600的结构示意图，该装置600包括：接收模块610，用于以下任一项：接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
可选的，所述第二波束相关信息包括以下至少一项：波束质量相关信息；波束ID相关信息；波束角度相关信息；波束增益相关信息；波束宽度相关信息。
本申请实施例中的模型有效性的确定装置500-600可以是通信设备，如终端或网络侧设备，也可以是通信设备中的部件，例如集成电路或芯片。示例性的，在所述模型有效性的确定装置500-600为终端时，终端可以包括但不限于上述所列举的终端11的类型，在所述模型有效性的确定装置500-600为网络侧设备时，网络侧设备可以包括但不限于上述所列举的网络侧设备12的类型，本申请实施例不作具体限定。
本申请实施例提供的模型有效性的确定装置500-600能够实现图2至图4的方法实施例实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
可选的，如图7所示，本申请实施例还提供一种通信设备700，包括处理器701和存储器702，存储器702存储有可在所述处理器701上运行的程序或指令，例如，该通信设备700为终端时，该程序或指令被处理器701执行时实现上述方法实施例200-400中的各个步骤，且能达到相同的技术效果。该通信设备700为网络侧设备时，该程序或指令被处理器701执行时实现上 述方法实施例200-400中的各个步骤，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供一种终端，包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如方法实施例200-400中所述的方法的步骤。该终端实施例是与上述终端侧方法实施例对应的，上述方法实施例的各个实施过程和实现方式均可适用于该终端实施例中，且能达到相同的技术效果。具体地，图8为实现本申请实施例的一种终端的硬件结构示意图。
该终端800包括但不限于：射频单元801、网络模块802、音频输出单元803、输入单元804、传感器805、显示单元806、用户输入单元807、接口单元808、存储器809、以及处理器810等中的至少部分部件。
本领域技术人员可以理解，终端800还可以包括给各个部件供电的电源(比如电池)，电源可以通过电源管理系统与处理器810逻辑相连，从而通过电源管理系统实现管理充电、放电、以及功耗管理等功能。图8中示出的终端结构并不构成对终端的限定，终端可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置，在此不再赘述。
应理解的是，本申请实施例中，输入单元804可以包括图形处理器(Graphics Processing Unit，GPU)8041和麦克风8042，图形处理器8041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元806可包括显示面板8061，可以采用液晶显示器、有机发光二极管等形式来配置显示面板8061。用户输入单元807包括触控面板8071以及其他输入设备8072中的至少一种。触控面板8071，也称为触摸屏。触控面板8071可包括触摸检测装置和触摸控制器两个部分。其他输入设备8072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆，在此不再赘述。
本申请实施例中，射频单元801接收来自网络侧设备的下行数据后，可以传输给处理器810进行处理；另外，射频单元801可以向网络侧设备发送 上行数据。通常，射频单元801包括但不限于天线、放大器、收发信机、耦合器、低噪声放大器、双工器等。
存储器809可用于存储软件程序或指令以及各种数据。存储器809可主要包括存储程序或指令的第一存储区和存储数据的第二存储区，其中，第一存储区可存储操作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外，存储器809可以包括易失性存储器或非易失性存储器，或者，存储器809可以包括易失性和非易失性存储器两者。其中，非易失性存储器可以是只读存储器(Read-Only Memory，ROM)、可编程只读存储器(Programmable ROM，PROM)、可擦除可编程只读存储器(Erasable PROM，EPROM)、电可擦除可编程只读存储器(Electrically EPROM，EEPROM)或闪存。易失性存储器可以是随机存取存储器(Random Access Memory，RAM)，静态随机存取存储器(Static RAM，SRAM)、动态随机存取存储器(Dynamic RAM，DRAM)、同步动态随机存取存储器(Synchronous DRAM，SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM，DDRSDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM，ESDRAM)、同步连接动态随机存取存储器(Synch link DRAM，SLDRAM)和直接内存总线随机存取存储器(Direct Rambus RAM，DRRAM)。本申请实施例中的存储器809包括但不限于这些和任意其它适合类型的存储器。
处理器810可包括一个或多个处理单元；可选的，处理器810集成应用处理器和调制解调处理器，其中，应用处理器主要处理涉及操作系统、用户界面和应用程序等的操作，调制解调处理器主要处理无线通信信号，如基带处理器。可以理解的是，上述调制解调处理器也可以不集成到处理器810中。
其中，在一种实现方式中，处理器810用于获取至少一个波束组，每个所述波束组中包括至少一个波束；将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；以及所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，通过射频单元801向第二通信设备 发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
在另一种实现方式中，射频单元801用于接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；或，接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
本实施例中，通过获取至少一个波束组，将至少一个波束组对应的第一波束相关信息作为目标AI模型的输入，以预测得到至少一个第二波束相关信息，进而根据至少一个第二波束相关信息对目标AI模型的性能进行验证，由此，一方面，能够实现对目标AI模型有效性的验证，另一方面，考虑到利用不同的波束组获取到的第二波束相关信息不同(即获得的目标AI模型的性能不同)，但不同波束组对应的真实值是相同的，那么，本实施例利用多个不同的波束组进行目标AI模型的性能验证，能够避免由于波束组不同而导致的模型性能验证结果不同的问题，提高了模型性能验证结果的准确性。
本申请实施例还提供一种网络侧设备，包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如方法实施例200-400中所述的方法的步骤。该网络侧设备实施例是与上述网络侧设备方法实施例对应的，上述方法实施例的各个实施过程和实现方式均可适用于该网络侧设备实施例中，且能达到相同的技术效果。
具体地，本申请实施例还提供了一种网络侧设备。如图9所示，该网络侧设备900包括：天线901、射频装置902、基带装置903、处理器904和存储器905。天线901与射频装置902连接。在上行方向上，射频装置902通过天线901接收信息，将接收的信息发送给基带装置903进行处理。在下行方向上，基带装置903对要发送的信息进行处理，并发送给射频装置902，射频装置902对收到的信息进行处理后经过天线901发送出去。
以上实施例中网络侧设备执行的方法可以在基带装置903中实现，该基带装置903包基带处理器。
基带装置903例如可以包括至少一个基带板，该基带板上设置有多个芯片，如图9所示，其中一个芯片例如为基带处理器，通过总线接口与存储器905连接，以调用存储器905中的程序，执行以上方法实施例中所示的网络设备操作。
该网络侧设备还可以包括网络接口906，该接口例如为通用公共无线接口(common public radio interface，CPRI)。
具体地，本发明实施例的网络侧设备900还包括：存储在存储器905上并可在处理器904上运行的指令或程序，处理器904调用存储器905中的指令或程序执行图5或图6所示各模块执行的方法，并达到相同的技术效果，为避免重复，故不在此赘述。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有程序或指令，该程序或指令被处理器执行时实现上述方法实施例200-400的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
其中，所述处理器为上述实施例中所述的终端中的处理器。所述可读存储介质，包括计算机可读存储介质，如计算机只读存储器ROM、随机存取存储器RAM、磁碟或者光盘等。
本申请实施例另提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行网络侧设备程序或指令，实现上述方法实施例200-400的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
应理解，本申请实施例提到的芯片还可以称为系统级芯片，系统芯片，芯片系统或片上系统芯片等。
本申请实施例还提供了一种计算机程序产品，该计算机程序产品包括处理器、存储器及存储在所述存储器上并可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时，实现上述方法实施例200-400的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供了一种通信系统，包括：第一通信设备及第二通信 设备，所述第一通信设备可用于执行如上所述的方法实施例200-300中所述的方法的各个步骤，所述第二通信设备可用于执行如上所述的方法实施例400中所述的方法的各个步骤。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外，需要指出的是，本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能，还可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能，例如，可以按不同于所描述的次序来执行所描述的方法，并且还可以添加、省去、或组合各种步骤。另外，参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以计算机软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述，但是本申请并不局限于上述的具体实施方式，上述的具体实施方式仅仅是示意性的，而不是限制性的，本领域的普通技术人员在本申请的启示下，在不脱离本申请宗旨和权利要求所保护的范围情况下，还可做出很多形式，均属于本申请的保护之内。

Claims (36)

  1. 一种模型有效性的确定方法，包括：
    第一通信设备获取至少一个波束组，每个所述波束组中包括至少一个波束；
    所述第一通信设备将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；
    所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，所述第一通信设备向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
  2. 如权利要求1所述的方法，其中，所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证的步骤，包括：
    所述第一通信设备将所述至少一个第二波束相关信息分别与第三波束相关信息进行匹配；
    根据匹配结果确定所述目标AI模型有效或无效；
    其中，所述第三波束相关信息是所述至少一个波束组对应的标签波束相关信息。
  3. 如权利要求2所述的方法，其中，所述第一通信设备将所述至少一个第二波束相关信息分别与第三波束相关信息进行匹配的步骤，包括以下任一项：
    针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述第一通信设备分别将各所述第二波束相关信息中的至少部分与所述第三波束相关信息进行匹配；
    针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述第一通信设备对各所述第二波束相关信息中的至少部分进行第一处理，以及分别将处理结果与所述第三波束相关信息进行匹配。
  4. 如权利要求3所述的方法，其中，所述第一通信设备对各所述第二波束相关信息中的至少部分进行第一处理的步骤，包括以下至少一项：
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值最大的波束质量相关信息；
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取至少部分波束质量相关信息；
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值大于第一阈值的波束质量相关信息；
    对所述至少一个第二波束相关信息进行组合；
    确定目标波束质量相关信息对应的波束标识ID相关信息，所述目标波束质量相关信息包括所述取值最大的波束质量相关信息、所述至少部分波束质量相关信息、所述取值大于第一阈值的波束质量相关信息中的任一项。
  5. 如权利要求2-4中任一项所述的方法，其中，根据匹配结果确定所述目标AI模型有效或无效的步骤，包括以下至少一项：
    在所述匹配结果位于第一基准值区间的情况下，确定所述目标AI模型有效；
    在所述匹配结果位于第二基准值区间的情况下，确定所述目标AI模型失效。
  6. 如权利要求5所述的方法，其中，在所述匹配结果位于第一基准值区间的情况下，确定所述目标AI模型有效的步骤，包括以下任一项：
    在所述匹配结果为多个、且多个所述匹配结果均位于第一基准值区间的情况下，确定所述目标AI模型有效；
    在所述匹配结果为多个、且位于所述第一基准值区间的匹配结果的数量达到第二阈值的情况下，确定所述目标AI模型有效；
    在所述的匹配结果为多个、且多个所述匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效。
  7. 如权利要求5所述的方法，其中，所述第一基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式确定；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第一波束组对应的目标AI模型的模型性能确定；
    根据第三基准值区间与第一辅助基准值确定；
    和/或，
    所述第二基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式确定；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第一波束组对应的目标AI模型的模型性能确定；
    根据第四基准值区间与第二辅助基准值确定。
  8. 如权利要求7所述的方法，其中，所述第一基准值区间在根据所述第三基准值区间与第一辅助基准值确定的情况下，所述第三基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式获取；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第二波束组对应的目标AI模型的模型性能确定；
    和/或，
    所述第二基准值区间在根据所述第四基准值区间与第二辅助基准值确定的情况下，所述第四基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式获取；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第二波束组对应的目标AI模型的模型性能确定。
  9. 如权利要求1-5中任一项所述的方法，其中，所述第一通信设备获取至 少一个波束组的步骤，包括以下至少一项：
    所述第一通信设备根据协议约定的方式获取所述至少一个波束组；
    所述第一通信设备根据接收到的第一指示信息获取所述至少一个波束组；
    所述第一通信设备根据第一条件获取至少一个波束组，其中，所述第一条件包括以下至少一项：
    不同的所述波束组对应的波束ID相关信息中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息的数量中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息的顺序中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得；
    在P个时间周期上获取、且每个所述时间周期对应N个波束组，其中，所述P个时间周期是距离所述第一通信设备进行目标AI模型的性能验证最近的P个时间周期，P、N为大于0的整数。
  10. 如权利要求9所述的方法，其中，在所述第一条件包括所述不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得的情况下，所述时间周期是距离所述第一通信设备进行目标AI模型的性能验证的时间最近的至少一个时间周期。
  11. 如权利要求9所述的方法，其中，在所述第一条件包括所述在P个时间周期上获取、且不同的时间周期对应N个波束组的情况下，所述在所述匹配结果位于第一基准值区间的情况下，确定所述目标AI模型有效的步骤，包括以下任一项：
    在所述目标AI模型在L个时间周期上均有效的情况下，确定所述目标AI模型有效，其中，所述L个时间周期属于所述P个时间周期，所述L为大于或等于0的整数，L小于或等于P；
    在S个时间周期对应的第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效，其中，所述S个时间周期属于所述P个时间周期，所述S为大于或等于 0的整数，S小于或等于P。
  12. 如权利要求11所述的方法，其中，所述目标AI模型在所述L个时间周期上有效，包括以下任一项：
    针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M1个波束组对应的M1个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；
    针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M2个波束组对应的M2个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；
    其中，所述M1、M2为大于或等于0的整数，M1、M2小于或等于N。
  13. 如权利要求1-12中任一项所述的方法，其中，所述第一波束相关信息和/或所述第二波束相关信息包括以下至少一项：
    波束质量相关信息；
    波束ID相关信息；
    波束角度相关信息；
    波束增益相关信息；
    波束宽度相关信息。
  14. 如权利要求1-13中任一项所述的方法，其中，所述至少一个波束组中包括对所述目标AI模型进行第一次模型验证时所采用的波束组，和/或，使用所述目标AI模型时所采用的波束组。
  15. 如权利要求1-14中任一项所述的方法，其中，所述第一通信设备根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证的步骤之后，所述方法还包括：
    所述第一通信设备向第二通信设备发送第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
  16. 一种模型有效性的确定方法，包括以下任一项：
    第二通信设备接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；
    第二通信设备接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
  17. 如权利要求16所述的方法，其中，所述第二波束相关信息包括以下至少一项：
    波束质量相关信息；
    波束ID相关信息；
    波束角度相关信息；
    波束增益相关信息；
    波束宽度相关信息。
  18. 一种模型有效性的确定装置，包括：
    获取模块，用于获取至少一个波束组，每个所述波束组中包括至少一个波束；
    预测模块，用于将所述至少一个波束组对应的第一波束相关信息分别输入目标AI模型，以预测得到至少一个第二波束相关信息，所述第二波束相关信息与所述波束组对应；
    验证模块，用于根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证，或者，发送模块，用于向第二通信设备发送所述至少一个第二波束相关信息，所述至少一个第二波束相关信息用于所述第二通信设备对目标AI模型进行性能验证。
  19. 如权利要求18所述的装置，其中，所述验证模块根据所述至少一个第二波束相关信息对所述目标AI模型进行性能验证的步骤，包括：
    所述验证模块将所述至少一个第二波束相关信息分别与第三波束相关信息进行匹配；以及根据匹配结果确定所述目标AI模型有效或无效；其中，所述第三波束相关信息是所述至少一个波束组对应的标签波束相关信息。
  20. 如权利要求19所述的装置，其中，所述验证模块将所述至少一个第二波束相关信息与第三波束相关信息进行匹配的步骤，包括以下任一项：
    针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述验证模块分别将各所述第二波束相关信息中的至少部分与所述第三波束相关信息进行匹配；
    针对所述至少一个第二波束相关信息中每个第二波束相关信息，所述验证模块对各所述第二波束相关信息中的至少部分进行第一处理，以及分别将处理结果与所述第三波束相关信息进行匹配。
  21. 如权利要求20所述的装置，其中，所述验证模块对各所述第二波束相关信息中的至少部分进行第一处理的步骤，包括以下至少一项：
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值最大的波束质量相关信息；
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取至少部分波束质量相关信息；
    对于所述至少一个第二波束相关信息中的每个第二波束相关信息，从所述第二波束相关信息包括的多个波束质量相关信息中选取取值大于第一阈值的波束质量相关信息；
    对所述至少一个第二波束相关信息进行组合；
    确定目标波束质量相关信息对应的波束ID相关信息，所述目标波束质量相关信息包括所述取值最大的波束质量相关信息、所述至少部分波束质量相关信息、所述取值大于第一阈值的波束质量相关信息中的任一项。
  22. 如权利要求19-21中任一项所述的装置，其中，所述验证模块根据匹配结果确定所述目标AI模型有效或无效的步骤，包括以下至少一项：
    在所述匹配结果位于第一基准值区间的情况下，所述验证模块确定所述目标AI模型有效；
    在所述匹配结果位于第二基准值区间的情况下，所述验证模块确定所述目标AI模型失效。
  23. 如权利要求22所述的装置，其中，在所述匹配结果位于第一基准值区间的情况下，所述验证模块确定所述目标AI模型有效的步骤，包括以下任一项：
    在所述匹配结果为多个、且多个所述匹配结果均位于第一基准值区间的情况下，确定所述目标AI模型有效；
    在所述匹配结果为多个、且位于所述第一基准值区间的匹配结果的数量达到第二阈值的情况下，确定所述目标AI模型有效；
    在所述的匹配结果为多个、且多个所述匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型有效。
  24. 如权利要求22所述的装置，其中，所述第一基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式确定；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第一波束组对应的目标AI模型的模型性能确定；
    根据第三基准值区间与第一辅助基准值确定；
    和/或，
    所述第二基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式确定；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第一波束组对应的目标AI模型的模型性能确定；
    根据第四基准值区间与第二辅助基准值确定。
  25. 如权利要求24所述的装置，其中，所述第一基准值区间在根据所述第三基准值区间与第一辅助基准值确定的情况下，所述第三基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式获取；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第二波束组对应的目标AI模型的模型性能确定；
    和/或，
    所述第二基准值区间在根据所述第四基准值区间与第二辅助基准值确定的情况下，所述第四基准值区间的获取方式包括以下至少一项：
    通过信令交互的方式获取；
    根据对所述目标AI模型进行第一次模型验证时得到的验证结果确定；
    根据第二波束组对应的目标AI模型的模型性能确定。
  26. 如权利要求18-22中任一项所述的装置，其中，所述获取模块获取至少一个波束组的步骤，包括以下至少一项：
    所述获取模块根据协议约定的方式获取所述至少一个波束组；
    所述获取模块根据接收到的第一指示信息获取所述至少一个波束组；
    所述获取模块根据第一条件获取至少一个波束组，其中，所述第一条件包括以下至少一项：
    不同的所述波束组对应的波束ID相关信息中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息的数量中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息的顺序中的至少部分不同；
    不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得；
    在P个时间周期上获取、且每个所述时间周期对应N个波束组，其中，所述P个时间周期是距离所述验证模块进行目标AI模型的性能验证最近的P个时间周期，P、N为大于0的整数。
  27. 如权利要求26所述的装置，其中，在所述第一条件包括所述不同的所述波束组对应的波束ID相关信息是在不同的时间周期上获得的情况下，所述时间周期是距离所述第一通信设备进行目标AI模型的性能验证的时间最近的至少一个时间周期。
  28. 如权利要求26所述的装置，其中，在所述第一条件包括所述在P个时间周期上获取、且不同的时间周期对应N个波束组的情况下，所述在所述 匹配结果位于第一基准值区间的情况下，所述验证模块确定所述目标AI模型有效的步骤，包括以下任一项：
    在所述目标AI模型在L个时间周期上均有效的情况下，所述验证模块确定所述目标AI模型有效，其中，所述L个时间周期属于所述P个时间周期，所述L为大于或等于0的整数，L小于或等于P；
    在S个时间周期对应的第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，所述验证模块确定所述目标AI模型有效，其中，所述S个时间周期属于所述P个时间周期，所述S为大于或等于0的整数，S小于或等于P。
  29. 如权利要求28所述的装置，其中，所述目标AI模型在所述L个时间周期上有效，包括以下任一项：
    针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M1个波束组对应的M1个第二波束相关信息与所述第三波束相关信息的匹配结果均位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；
    针对所述L个时间周期中的每个时间周期，在该时间周期上的至少M2个波束组对应的M2个第二波束相关信息与所述第三波束相关信息的匹配结果的均值位于所述第一基准值区间的情况下，确定所述目标AI模型在该时间周期上有效；
    其中，所述M1、M2为大于或等于0的整数，M1、M2小于或等于N。
  30. 如权利要求18-29中任一项所述的装置，其中，所述第一波束相关信息和/或所述第二波束相关信息包括以下至少一项：
    波束质量相关信息；
    波束ID相关信息；
    波束角度相关信息；
    波束增益相关信息；
    波束宽度相关信息。
  31. 如权利要求18-30中任一项所述的装置，其中，所述至少一个波束组中包括对所述目标AI模型进行第一次模型验证时所采用的波束组，和/或，使用所述目标AI模型时所采用的波束组。
  32. 如权利要求18-31中任一项所述的装置，其中，所述发送模块还用于向第二通信设备发送第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
  33. 一种模型有效性的确定装置，包括接收模块，用于以下任一项：
    接收第一通信设备发送的至少一个第二波束相关信息，以及根据所述至少一个第二波束相关信息对目标AI模型进行性能验证；
    接收第一通信设备发送的第二指示信息，所述第二指示信息用于指示所述目标AI模型有效或无效。
  34. 如权利要求33所述的装置，其中，所述第二波束相关信息包括以下至少一项：
    波束质量相关信息；
    波束ID相关信息；
    波束角度相关信息；
    波束增益相关信息；
    波束宽度相关信息。
  35. 一种通信设备，包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求1至15中任一项所述的模型有效性的确定方法的步骤，或实现如权利要求16至17中任一项所述的模型有效性的确定方法的步骤。
  36. 一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如权利要求1至15中任一项所述的模型有效性的确定方法的步骤，或实现如权利要求16至17中任一项所述的模型有效性的确定方法的步骤。
PCT/CN2023/087736 2022-04-13 2023-04-12 模型有效性的确定方法、装置及通信设备 WO2023198093A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210384902.4 2022-04-13
CN202210384902.4A CN116992250A (zh) 2022-04-13 2022-04-13 模型有效性的确定方法、装置及通信设备

Publications (1)

Publication Number Publication Date
WO2023198093A1 true WO2023198093A1 (zh) 2023-10-19

Family

ID=88328931

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/087736 WO2023198093A1 (zh) 2022-04-13 2023-04-12 模型有效性的确定方法、装置及通信设备

Country Status (2)

Country Link
CN (1) CN116992250A (zh)
WO (1) WO2023198093A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200266873A1 (en) * 2019-02-15 2020-08-20 Lg Electronics Inc. Method for supporting beam correspondence and apparatus thereof
CN113012057A (zh) * 2019-12-20 2021-06-22 通用电气精准医疗有限责任公司 对超声扫描仪中ai网络的连续训练
CN113994598A (zh) * 2019-04-17 2022-01-28 诺基亚技术有限公司 无线网络的波束预测

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200266873A1 (en) * 2019-02-15 2020-08-20 Lg Electronics Inc. Method for supporting beam correspondence and apparatus thereof
CN113994598A (zh) * 2019-04-17 2022-01-28 诺基亚技术有限公司 无线网络的波束预测
CN113012057A (zh) * 2019-12-20 2021-06-22 通用电气精准医疗有限责任公司 对超声扫描仪中ai网络的连续训练

Also Published As

Publication number Publication date
CN116992250A (zh) 2023-11-03

Similar Documents

Publication Publication Date Title
US11910418B2 (en) Electronic device and method for determining uplink operation in wireless communication system
CN113873651B (zh) 通信方法和通信装置
CN116266216A (zh) 客户端筛选方法及装置、客户端及中心设备
WO2023134650A1 (zh) 信息交互方法、装置及通信设备
WO2023198093A1 (zh) 模型有效性的确定方法、装置及通信设备
WO2023186014A1 (zh) 信号发送方法、信号接收方法及通信设备
WO2023179651A1 (zh) 波束处理方法、装置及设备
CN116347464A (zh) 感知测量方法、装置、通信设备及可读存储介质
CN114205201A (zh) 信号补偿方法、装置、中继设备、存储介质和程序产品
WO2023179649A1 (zh) 人工智能模型的输入处理方法、装置及设备
WO2023179653A1 (zh) 波束处理方法、装置及设备
WO2024120444A1 (zh) 模型监督方法、装置、终端、网络侧设备及可读存储介质
WO2024027576A1 (zh) 一种ai网络模型的性能监督方法、装置和通信设备
WO2023051529A1 (zh) 参考信号处理方法、装置、终端及介质
WO2023109828A1 (zh) 数据收集方法及装置、第一设备、第二设备
WO2024021805A1 (zh) 一种垂直定位方法和服务器
WO2024083004A1 (zh) Ai模型配置方法、终端及网络侧设备
WO2024017049A1 (zh) Bsc设备的识别方法、装置及通信设备
WO2023160571A1 (zh) 干扰或自激处理方法及装置、中继节点及宿主基站
WO2024007950A1 (zh) Sl定位方法、设备及可读存储介质
WO2024022166A1 (zh) Ra方法、装置、ue、网络侧设备、通信系统及可读存储介质
US20240236911A1 (en) Sensing Device Registration Method
WO2024140445A1 (zh) 模型信息传输方法、装置及设备
WO2024067665A1 (zh) Csi预测处理方法、装置、通信设备及可读存储介质
WO2023029320A1 (zh) 通信方法及装置、计算机可读存储介质、通信设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23787736

Country of ref document: EP

Kind code of ref document: A1