Terminal device for monitoring performance of beam selection of radio communication using artificial intelligence (AI) / machine learning (ML), base station device, control method, and program

Abstract

To properly configure a feedback from a terminal device to a base station device for a selection method, by an AI / ML model, of a beam that the base station device uses.SOLUTION: A terminal device capable of measuring radio qualities of reference signals which are transmitted from a base station device using a plurality of first beams formed by the base station device and reference signals which are transmitted using a plurality of second beams less than the plurality of first beams performs a first measurement of the radio qualities of the reference signals transmitted from the base station device using the second beams, inputs at least a part of the radio qualities measured for the second beams to a model which is learned by machine learning, thereby acquiring predictive radio qualities regarding the plurality of first beams, performs a measurement of the radio qualities of the reference signals transmitted from the base station device using at least a part of the first beams and reports to the base station device information on a differential value between the actually measured radio quality obtained by the measurement and the predictive radio quality regarding at least a part of the plurality of first beams.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present invention relates to a technology for monitoring the performance of beam selection in wireless communications using artificial intelligence (AI) / machine learning (ML). [Background technology]

[0002] In the standardization work of the 3rd Generation Partnership Project (3GPP), the use of artificial intelligence (AI) / machine learning (ML) is being considered for a communication device capable of forming multiple beams to perform wireless communication, to determine which of the multiple beams to use to communicate with a partner device. Non-Patent Document 1 describes an AI / ML model in which a terminal device measures the reference signal received power (RSRP) of some beams included in multiple beams with narrow beam widths that can be formed in a network (base station device), and selects a beam to use for communication from all of the multiple beams based on the RSRP. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] 3GPP (registered trademark) Contribution, R1-2203142 Summary of the Invention [Problem to be solved by the invention]

[0004] The AI / ML model used to select a beam may become unable to select an appropriate beam, for example, due to changes in the circumstances in which the terminal device is located, so it is important to periodically check that the selected beam is appropriate and ensure that the appropriate beam continues to be used. [Means for solving the problem]

[0005] The present invention provides a technology for efficiently monitoring the performance of a beam used by a selected base station device using an AI / ML model.

[0006] A terminal device according to one embodiment of the present invention includes a measurement means capable of measuring the wireless quality of a reference signal transmitted from a base station device using each of a plurality of first beams formed by the base station device and a reference signal transmitted using each of a number of second beams that is fewer than the plurality of first beams; an acquisition means for acquiring a predicted wireless quality for each of the plurality of first beams by inputting at least a portion of the wireless quality of the reference signal transmitted using each of the second beams into a trained model using machine learning; and a reporting means for reporting to the base station device information regarding a difference value between the actual wireless quality obtained by the measurement and the predicted wireless quality for at least a portion of the plurality of first beams.

[0007] A base station device according to one embodiment of the present invention includes a transmitting means for transmitting a reference signal using a plurality of first beams and a number of second beams that is smaller than the plurality of first beams; a receiving means for receiving from the terminal device information regarding a difference value between a predicted wireless quality in the terminal device for at least some of the plurality of first beams, which is obtained by inputting at least some of the wireless qualities of the reference signals transmitted using each of the second beams into a trained model using machine learning, and an actual measured wireless quality obtained by measuring the reference signal in the terminal device for at least some of the plurality of first beams; and a determining means for determining the performance of the trained model based on the information regarding the difference value. [Effects of the Invention]

[0008] According to the present invention, it is possible to efficiently monitor the performance of a beam used by a base station device selected using an AI / ML model. [Brief explanation of the drawings]

[0009] [Figure 1]FIG. 1 is a diagram illustrating an example of the configuration of a wireless communication system. [Figure 2] FIG. 10 is a diagram illustrating an example of machine learning for beam determination. [Figure 3] 10A and 10B are diagrams illustrating examples of information reported by a terminal device to a base station device. [Figure 4] 10A and 10B are diagrams illustrating examples of information reported by a terminal device to a base station device. [Figure 5] FIG. 1 is a diagram illustrating an example of a flow of processing executed in a wireless communication system. [Figure 6] FIG. 2 illustrates an example of a hardware configuration of the apparatus. [Figure 7] FIG. 2 is a diagram illustrating an example of a functional configuration of a terminal device. [Figure 8] FIG. 2 is a diagram illustrating an example of a functional configuration of a base station device. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, the embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the scope of the invention as claimed, and not all combinations of features described in the embodiments are necessarily essential to the invention. Two or more of the features described in the embodiments may be combined in any desired manner. Furthermore, the same reference numerals are used to designate identical or similar components, and redundant descriptions will be omitted.

[0011] 1(A) and 1(B) show an example of the configuration of a wireless communication system according to this embodiment. The wireless communication system is, for example, a cellular communication system conforming to the cellular communication standard of the Third Generation Partnership Project (3GPP (registered trademark)), and is configured to include a base station device 101 and a terminal device 102. The base station device 101 can form multiple first beams 111 and selectively uses one of the beams to communicate with the terminal device 102. The selection of the beam to be used is performed, for example, based on the result of measuring the reception quality of reference signals transmitted by all beams in the beams at the terminal device 102. This selection is particularly effective in the downlink where signals are transmitted from the base station device 101 to the terminal device 102. The reference signal here is, for example, a synchronization signal / physical broadcast channel (SS / PBCH) block (SSB) or a channel state information reference signal (CSI-RS). Furthermore, the reception quality is, for example, reference signal received power (RSRP), reference signal received quality (RSRQ), signal-to-interference-and-noise ratio (SINR), etc. For example, the terminal device 102 measures the reception quality of the reference signal transmitted in each of the multiple first beams 111 and notifies the base station device 101 of the measurement results. Based on the measurement results, the base station device 101 determines, for example, a beam with high reception quality as the beam to be used for communication with the terminal device 102.

[0012] However, if the terminal device 102 measures the reference signals for all of the first beams 111 to determine the reception quality and notify the base station device 101 of the reception quality, the terminal device 102 needs to perform a large number of measurements, which lengthens the time required to determine the beam. Furthermore, multiple notifications for multiple beams may waste radio resources. In response to this, for example, by selecting some beams that are expected to provide good quality in the terminal device 102 and having the terminal device 102 measure and report only those beams, it is possible to simplify the measurements and reduce the amount of information to be reported. However, this process does not allow the terminal device 102 to select a beam to be used from all of the first beams 111. To address this issue, the use of artificial intelligence (AI) / machine learning (ML) to select a beam to be used is being considered. For example, at least a portion of the measurement results by the terminal device 102 of reference signals transmitted in a second beam that is smaller than the first beam 111 is input into a trained model obtained by machine learning. Then, as an output of the trained model, information capable of identifying a small number of candidate beams to be used by the base station device 101 for communication (e.g., downlink, and in some cases, also uplink) with the terminal device 102 may be output. For example, the trained model may output the predicted wireless quality of each of the first beams 111, and a predetermined number of beams with the highest predicted wireless quality, or beams whose predicted wireless quality exceeds a predetermined value, may be identified as candidate beams to be used for communication. Note that the predicted wireless quality may not be output, and the candidate beams to be used for communication may be output directly. Thereafter, the terminal device 102 may measure the reception quality of a reference signal transmitted from the base station device 101 using each of the small number of candidate beams, and the beam to be used for communication may be determined based on the measurement result. Alternatively, the beam to be used for communication may be determined all at once, rather than the candidate beams, based on the output of the trained model.

[0013] The second beam, the reception quality of which is input to the trained model, may be, for example, a part of the first beam 111, as shown in beam 112 in FIG. 1(A). That is, a part of the many first beams 111 that the base station device 101 can form (for example, one beam 112 is set for each predetermined number of first beams 111) is set as the second beam, and the reception quality of the reference signal transmitted by the second beam is measured in the terminal device 102. Then, at least a part of the measurement result is input to the trained model, thereby determining the beam to be used in the base station device 101 for communication with the terminal device 102. Also, a beam having a wider beam width than the first beam 111, for example, as shown in beam 113 in FIG. 1(B), may be used as the second beam. For example, if the SSB is transmitted via beam 113 and the CSI-RS is transmitted via the first beam 111, the radio quality of the SSB is measured by the terminal device 102 and the result is input into the trained model, thereby determining from the first beam 111 the beam that the base station device 101 will use to communicate with the terminal device 102.

[0014] In this way, beam selection based on AI / ML makes it possible to determine a beam suitable for communication between the base station device 101 and the terminal device 102 from among all of the first beams 111, without the terminal device 102 measuring the wireless quality of all of the first beams 111. Note that in this embodiment, beam selection based on AI / ML is performed in the terminal device 102. In this case, for example, the terminal device 102 observes a reference signal for each of the second beams to acquire wireless quality, inputs the wireless quality into a trained model, and identifies candidate beams to be used for communication. Then, the terminal device 102 notifies the base station device 101 of information about the candidate beams. The base station device 101 transmits a reference signal using the candidate beams, and the terminal device 102 acquires the wireless quality for each of the candidate beams. Then, the terminal device 102 notifies the base station device 101 of an optimal beam (for example, one with the best wireless quality). As a result, the beam notified from the terminal device 102 to the base station device 101 is used in subsequent communications.

[0015] Here, an example of machine learning for beam determination will be described with reference to Figures 2(A) and 2(B). Figure 2(A) shows an example of the learning phase in machine learning, and Figure 2(B) shows an example after the learning phase is completed and a trained model for beam determination processing has been obtained. Note that these are merely examples, and the beams to be used may be determined by other configurations.

[0016] In the learning phase shown in the example of FIG. 2(A), machine learning is performed using information on the reception quality in the terminal device 102 of reference signals transmitted from each of the first beams 111 as training data 203, and information on the reception quality in the terminal device 102 of reference signals transmitted from each of the second beams as input 202. Here, the training data may be, for example, data on the reception quality itself corresponding to each of the first beams 111, or information indicating the probability that each of the first beams 111 will be selected as the best beam (e.g., the beam with the best wireless quality) when a certain input 202 is given. Furthermore, although the reception quality of the second beam is represented as RSRP in FIG. 2(A), it may also be other index values ​​such as RSRQ, SINR, signal-to-noise ratio (SNR), etc. The reception quality of the second beam is input to a learning model 201, and from that information, information in a format corresponding to the training data 203 is output. For example, when the training data 203 is the reception quality for each of the first beams 111, the output of the learning model 201 is the predicted reception quality for each of the first beams 111. Then, the difference value between the output and the training data 203 is fed back to the learning model 201, and the learning model 201 is updated. This process is repeatedly performed, and when the difference value is maintained at a sufficiently small value, or when the number of times the process is repeated reaches a predetermined number, the machine learning ends, and the learning model at that time is output as a trained model.

[0017] In FIG. 2(B), an input 212 related to the reception quality for the second beam, which is the same as the input 202 in the learning phase, is supplied to the trained model 211. Based on the input 212, the trained model 211 infers and outputs a beam 213 (or a candidate beam in some cases) to be used for communication with the terminal device 102 in the base station device 101. In this way, based on measurement results of reference signals transmitted by fewer beams than the first beam 111, a beam to be used for communication with the terminal device 102 can be determined from all of the first beams 111. Note that the examples of FIGS. 2(A) and 2(B) are examples of machine learning, and an AI / ML model may be defined by other methods such that a value of the measurement result for the second beam is input to identify a beam (or a candidate beam) to be used for communication from the first beam 111.

[0018] In this embodiment, the learning model 201 and the trained model 211 are assumed to output predicted wireless quality values ​​for each of the first beams 111.

[0019] When an AI / ML model is used, multiple models are prepared that can be used depending on the environment in which the terminal device 102 is placed, for example. In other words, if the same model is always used for the terminal device 102, an inappropriate beam may be selected as the beam to be used. For this reason, in order to select a beam with high accuracy, it is important to monitor whether the selected beam is appropriate and to reselect an appropriate model or update the model.

[0020] In this embodiment, the terminal device 102 selects a beam (candidate) to be used using the trained model 211. In this case, for example, the terminal device 102 may monitor the wireless quality of the selected beam and report information indicating the wireless quality to the base station device 101. Then, if the base station device 101 determines that the wireless quality is insufficient (for example, below a predetermined threshold), it may instruct the terminal device 102 to retrain the trained model 211 or use another model. However, such wireless quality information cannot determine whether the result of the AI / ML inference is normal. In other words, even if the reported wireless quality is insufficient, there may be cases where the wireless quality does not deviate significantly from the wireless quality predicted by the AI / ML and is the best beam among the first beams 111. In this case, even if retraining or changing the model does not produce better results, and in some cases, a model with lower performance may be used, resulting in the selection of an inappropriate beam.

[0021] In view of these circumstances, this embodiment provides a mechanism for allowing information to be fed back from the terminal device 102 to the base station device 101, enabling the base station device 101 to evaluate the performance of an AI / ML model.

[0022] In one example, the terminal device 102 feeds back to the base station device 101 a difference value between the predicted wireless quality obtained by the trained model 211 and the actually measured wireless quality. If the magnitude (absolute value) of this difference value exceeds a predetermined value, the base station device 101 can instruct the terminal device 102 to re-learn or use another model. In this way, by notifying the base station device 101 of the discrepancy between the value inferred by AI / ML and the actually measured value, the base station device 101 can determine whether or not poor wireless quality observed in the terminal device 102 is due to the performance of the inference by AI / ML, for example.

[0023] The terminal device 102 may, for example, periodically measure the wireless quality for each of the second beams, input the wireless quality to the trained model 211, obtain the predicted wireless quality for each of the first beams 111, identify a difference value for one of the beams being used, and report the difference value to the base station device 101. Furthermore, as described above, when selecting a beam, the terminal device 102 may identify a predetermined number of beams having the highest predicted wireless quality as candidate beams to be used, notify the base station device 101 of information about the candidate beams, and then measure a reference signal transmitted from the base station device 101 using each of the candidate beams. In this case, the terminal device 102 may report to the base station device 101 a difference value between the predicted wireless quality and the actually measured wireless quality for each of the candidate beams. That is, only one difference value for the beam that is actually being used may be reported, or multiple difference values ​​for each of multiple beams (part or all of the first beams 111) may be reported. When the base station device 101 receives multiple difference values, for example, if the proportion (number) of the multiple difference values ​​whose magnitude exceeds a predetermined value exceeds a certain value, the base station device 101 may determine that the performance of the AI / ML inference is insufficient. In one example, the base station device 101 may determine that the performance of the AI / ML inference is insufficient if any one of the multiple difference values ​​exceeds a predetermined value. The base station device 101 may also determine that the performance of the AI / ML inference is insufficient if the average value of the difference values ​​exceeds a predetermined value. The terminal device 102 may identify a difference value for each of multiple beam candidates obtained by AI / ML and notify the base station device 101 of the average value of the difference values. The terminal device 102 may input the acquired difference values ​​into a predetermined function (e.g., a function for performing statistical processing) and notify the base station device 101 of the value obtained. As an example, a statistical value such as the variance of the acquired difference values ​​may be notified to the base station device 101.

[0024] Furthermore, for example, when the difference value exceeds a predetermined value, the terminal device 102 may notify the base station device 101 of this information. That is, for example, the terminal device 102 may notify a flag such as "1" when the difference value between the predicted wireless quality obtained by AI / ML and the actually measured wireless quality for the beam in use exceeds a predetermined value, or "0" when the difference value does not exceed the predetermined value. The terminal device 102 may periodically notify the base station device 101 of this information. Also, the terminal device 102 may make a predetermined report when the difference value between the predicted wireless quality obtained by AI / ML and the actually measured wireless quality for the beam in use exceeds a predetermined value, and may not make the predetermined report when the difference value does not exceed the predetermined value. That is, the terminal device 102 may make this report aperiodically.

[0025] FIG. 3 shows an example of information reported from the terminal device 102 to the base station device 101. The terminal device 102 may generate a report including, for example, identification information 301 indicating which beam the report relates to and a differential value 302 between the predicted wireless quality and the actually measured wireless quality, and report the report to the base station device 101. Here, for example, the base station device 101 may notify the terminal device 102 of the number of differential values ​​to be notified in one report, the number of messages for which differential values ​​should be reported, etc. When differential values ​​are reported by multiple messages, a report number 303 for identifying the report message may be assigned. Note that the differential value 302 is expected to be sufficient if it can express a narrower range of values ​​than when indicating the value of wireless quality (e.g., RSRP) itself, and therefore may be expressed using fewer bits than, for example, a conventional report of wireless quality measurement results. Furthermore, the differential value 302 may be expressed in detail using, for example, the same number of bits as or more than, a conventional report of wireless quality measurement results. Furthermore, as shown in FIG. 4, a report including identification information 401 about a specific beam and information 402 indicating that the magnitude of the difference value between the predicted wireless quality and the actually measured wireless quality has exceeded a predetermined threshold may be transmitted from the terminal device 102 to the base station device 101. This information 402 may be expressed as 1-bit information. Note that FIG. 3 illustrates an example in which information about multiple beams is reported, and FIG. 4 illustrates an example in which information about a single beam is reported, but this is not limiting. That is, in FIG. 3, only information about one beam may be transmitted, and in FIG. 4, information 402 about multiple beams may be transmitted. In one example, the information as shown in FIG. 4 may be transmitted as a bitmap in which 1-bit information 402 is arranged for each of multiple beams. Furthermore, when it is sufficient to transmit information that can determine the difference value between the predicted wireless quality and the actually measured wireless quality, and it is not necessary to identify which beam has a large difference between the predicted and actually measured wireless quality, the beam identification information 301 / 401 may be omitted. Furthermore, when a report is made only about one beam in use, the base station device 101 can naturally recognize which beam the report relates to, so the beam identification information 301 / 401 may be omitted.

[0026] The base station device 101 may instruct the terminal device 102 to perform the above-described monitoring. That is, the terminal device 102 may identify and report a difference value between the predicted wireless quality and the actually measured wireless quality in response to receiving an instruction from the base station device 101. For example, when receiving a predetermined instruction, the terminal device 102 measures reference signals transmitted using each of the second beams and inputs the measurement result values ​​to the trained model 211. Then, the terminal device 102 acquires the value of the predicted wireless quality of the first beam 111 output from the trained model 211. Thereafter, for example, based on the predicted wireless quality, the terminal device 102 identifies a predetermined number of beams, which are a part of the first beam 111, as candidate beams to be used for communication, and notifies the base station device 101 of the identified information. The base station device 101 transmits reference signals using each of the identified candidate beams and causes the terminal device 102 to measure the wireless quality of the reference signals. The terminal device 102 measures wireless quality by observing reference signals transmitted using each of the beam candidates among the first beam 111. Then, the terminal device 102 calculates a difference value between the measurement result and the above-mentioned predicted wireless quality, and notifies the base station device 101 of the calculation result. Note that during this process, the terminal device 102 may (anew) identify a beam to be used for communication, and notify the base station device 101 of information indicating the identified beam together with information on the difference value. That is, while performing a process of selecting a beam to be used for communication using AI / ML, the terminal device 102 may identify a difference value for determining the performance of the beam selection process using AI / ML and notify the base station device 101 of the difference value.

[0027] Furthermore, the base station device 101 may notify the terminal device 102 of configuration information regarding the configuration of information to be reported. The configuration information may include, for example, information indicating which of the above-mentioned report formats to use for the report. The configuration information may include, for example, information specifying that difference values ​​should be reported only for beams in use, or that difference values ​​should be reported for a predetermined number of beams (some or all of the candidate beams to be used, or some or all of the first beams 111). The configuration information may also include information specifying that an average value of difference values ​​for a predetermined number of beams (some or all of the candidate beams to be used, or some or all of the first beams 111) should be reported. Furthermore, when statistical processing other than averaging is to be performed, information indicating a method for processing the difference values, such as information indicating the statistical processing, may be notified as the configuration information. Furthermore, the configuration information may include information specifying that a report should be sent regarding whether or not the magnitude of the difference value for a beam in use exceeds a predetermined value, or that a report should be sent only when the magnitude of the difference value for a beam in use exceeds a predetermined value. The setting information may also include information specifying whether or not the magnitude of the difference value for a predetermined number (one or more) of beams among candidate beams to be used or a predetermined number (one or more) of beams among the first beams 111 exceeds a predetermined value should be reported. The setting information may also include information specifying that a report should be transmitted only when the magnitude of the difference value for a predetermined number (one or more) of beams among candidate beams to be used or a predetermined number (one or more) of beams among the first beams 111 exceeds a predetermined value. The setting information may also include information indicating the number of beams to be reported when a difference value should be reported, and information on the number of bits used to indicate the difference value. Note that multiple tables may be prepared in advance, in which, for each number of bits used to indicate a difference value, a value indicated by the number of bits is associated with the magnitude of the difference value, and information indicating which of the multiple tables should be used may be specified by the setting information. The setting information may also include information indicating the number of report messages to be used to report a difference value.Furthermore, the setting information may include information indicating whether or not the above-mentioned monitoring should be performed, and if monitoring should be performed, the period during which the monitoring should be performed.

[0028] 5 shows an example of the flow of processing executed in the wireless communication system of this embodiment. Details of each step have been described above, so they will not be repeated here, and only an overview of the processing will be described. In the following, an example will be described in which the terminal device 102 reports the above-mentioned difference values ​​and the like for evaluating AI / ML performance, along with the procedure for selecting a beam to be used.

[0029] First, the base station device 101 transmits configuration information for AI / ML performance evaluation to the terminal device 102 (S501). This configuration information may include, for example, information indicating to the terminal device 102 that monitoring should be performed and information specifying a format for reporting the monitoring results. The terminal device 102 starts the monitoring process based on the configuration information. Note that here, it is assumed that the terminal device 102 also selects a beam to be used using an AI / ML model. The terminal device 102 observes reference signals transmitted from the base station device 101 using each of the second beams and acquires measurement results of wireless quality (S502). Then, the terminal device 102 inputs the measurement results into the AI / ML model to acquire predicted wireless quality for each of the first beams 111 (S503). Note that the AI / ML model may be, for example, a model provided in advance by the base station device 101, or may be acquired by executing machine learning or the like in the terminal device 102. The terminal device 102 selects one or more candidate beams to be used from the first beams 111 based on the predicted wireless quality obtained in S503 (S504). It is assumed here that multiple candidate beams to be used are selected. The terminal device 102 selects a predetermined number of beams with the best predicted wireless quality as candidate beams to be used. The terminal device 102 notifies the base station device 101 of information on the selected candidate beams (S505). The terminal device 102 then observes reference signals transmitted using each of the candidate beams (part of the first beams 111) and measures their wireless quality (S506). Based on the wireless quality measurement results, the terminal device 102 can select, for example, a beam that can provide the best wireless quality as a beam to be used in subsequent communications (S507). In addition, the terminal device 102 calculates the difference value between the measurement result of S506 and the predicted wireless quality obtained in S503 for the beam decided to be used or for each candidate beam to be used, and identifies the value to be reported to the base station device 101 (S508).For example, the terminal device 102 may calculate a value in a format specified in S501 or determined in advance by a standard, for example, such as the wireless quality of the beam decided to be used, the wireless quality for each candidate beam, or the average value of multiple wireless qualities corresponding to multiple candidate beams.The terminal device 102 then notifies the base station device 101 of the value identified in S508 (S509).At this time, the terminal device 102 may also notify the base station device 101 of information specifying the beam decided to be used together with the value identified in S508.

[0030] FIG. 6 shows an example of the hardware configuration of the base station device 101 and the terminal device 102 according to this embodiment. In one example, the base station device 101 and the terminal device 102 are configured to include a processor 601, a ROM 602, a RAM 603, a storage device 604, and a communication circuit 605. The processor 601 is a computer configured to include one or more processing circuits, such as a general-purpose CPU (Central Processing Unit) or an ASIC (Application Specific Integrated Circuit), and executes the overall processing of the device and each of the above-mentioned processes by reading and executing programs stored in the ROM 602 or the storage device 604. The ROM 602 is a read-only memory that stores information such as programs and various parameters related to the processing executed by the base station device 101 and the terminal device 102. The RAM 603 functions as a workspace when the processor 601 executes a program and is a random access memory that stores temporary information. The storage device 604 is configured, for example, by a removable external storage device. The communication circuit 605 is configured, for example, by a circuit for wireless communication of 5G or its successor standards. Although FIG. 6 illustrates one communication circuit 605, the base station device 101 and the terminal device 102 may have multiple communication circuits. For example, the base station device 101 and the terminal device 102 may have wireless communication circuits for 5G and its successor standard, and a common antenna for these circuits. The base station device 101 and the terminal device 102 may have separate antennas suitable for each standard. The base station device 101 may also have a wired communication circuit used when communicating with other base station devices or nodes in the core network. The terminal device 102 may also have a communication circuit conforming to a wireless communication standard other than the cellular communication standard, such as a wireless local area network (LAN) or Bluetooth (registered trademark). The base station device 101 and the terminal device 102 may have separate communication circuits 605 for each of multiple available frequency bands, or may have a common communication circuit 605 for at least some of these frequency bands.

[0031] FIG. 7 shows an example of the functional configuration of the terminal device 102. The terminal device 102 includes, for example, a measurement unit 701, an inference unit 702, a reporting unit 703, and a setting information receiving unit 704. Note that FIG. 7 only shows functions particularly related to this embodiment, and does not illustrate various other functions that the terminal device 102 may have. For example, the terminal device 102 naturally has other functions that terminal devices compliant with 5G or subsequent standards generally have. The functional blocks in FIG. 7 are shown schematically, and the respective functional blocks may be integrated or further subdivided. Each function in FIG. 7 may be realized, for example, by the processor 601 executing a program stored in the ROM 602 or the storage device 604, or may be realized, for example, by a processor within the communication circuit 605 executing predetermined software. Since the details of the processing performed by each functional unit are as described above, only the general functions of the terminal device 102 will be outlined here.

[0032] The measurement unit 701 observes a reference signal transmitted from the base station device 101 and measures wireless quality. The measurement unit 701 has the ability to measure wireless quality based on reference signals transmitted from each of the first beams 111 and each of the second beams formed by the base station device 101, for example. The inference unit 702 inputs the wireless quality of each of the second beams measured by the measurement unit 701 into an AI / ML model to perform inference processing and acquire predicted wireless quality for each of the first beams 111. Then, the inference unit 702 selects, for example, a predetermined number of beams with the best predicted wireless quality as candidates for beams to be used in subsequent communications. The reporting unit 703 reports, for example, information specifying the predetermined number of beams as a result of the inference by the inference unit 702 to the base station device 101. The measurement unit 701 observes reference signals for beams reported by the reporting unit 703 among the first beams 111 formed by the base station device 101 to acquire wireless quality. The reporting unit 703 then reports to the base station device 101 information specifying a beam corresponding to the best value of the acquired wireless qualities. This enables the base station device 101 to use the specified beam in communication with the terminal device 102. The reporting unit 703 may notify the base station device 101 of measurement results of wireless qualities for each of a plurality of candidate beams, and the base station device 101 may select a beam to use based on the measurement results. The reporting unit 703 may also identify a difference value between the predicted wireless quality acquired by the inference unit 702 and the actual wireless quality measured by the measurement unit 701 for at least a portion of the first beam 111, and report information about the difference value to the base station device 101. As described above, the information about the difference value may include only information about the beam determined to be used, or may include information about at least a portion of the candidate beams, or may include information about all of the first beams 111 in some cases. The information about the difference value may also be information indicating whether or not the magnitude of the difference value exceeds a predetermined threshold.In this case, information indicating whether the magnitude of the difference value exceeds a predetermined threshold may be transmitted, or the information may be transmitted only when the magnitude exceeds the predetermined threshold, thereby implicitly indicating that the magnitude has exceeded the predetermined threshold. Information such as the average value and variance of the difference value may also be notified. The setting information receiving unit 704 receives various setting information from the base station device 101, such as setting information for reporting on the difference value.

[0033] FIG. 8 shows an example of the functional configuration of the base station device 101. The base station device 101 includes a reference signal transmitter 801, a report receiver 802, a performance determiner 803, and a controller 804. Note that FIG. 8 only illustrates functions particularly related to this embodiment, and does not illustrate various other functions that the base station device 101 may have. For example, the base station device 101 naturally has other functions that are generally included in base station devices compliant with 5G and subsequent standards. The functional blocks in FIG. 8 are shown schematically, and the functional blocks may be integrated or further subdivided. Each function in FIG. 8 may be implemented, for example, by the processor 601 executing a program stored in the ROM 602 or the storage device 604, or by a processor within the communication circuit 605 executing predetermined software. Since the details of the processing performed by each functional unit are as described above, only the general functions of the base station device 101 will be outlined here.

[0034] The reference signal transmitting unit 801 transmits reference signals (e.g., CSI-RS or SSB) using each of the first beam 111 and the second beam. The report receiving unit 802 receives the above-mentioned report related to at least the differential value from the terminal device 102. The performance determining unit 803 determines whether performance degradation of beam selection by the AI / ML model has occurred based on the information related to the received differential value. For example, the performance determining unit 803 may determine that performance degradation has occurred when the magnitude of the reported value exceeds a predetermined value. Furthermore, when multiple measurement values ​​corresponding to multiple beams are reported, the performance determining unit 803 may determine that performance degradation has occurred when the number of measurement values ​​whose magnitudes exceed a predetermined value among the multiple measurement values ​​exceeds a predetermined number. Furthermore, the performance determining unit 803 may calculate the average value and variance of the multiple measurement values, and determine that performance degradation has occurred when the magnitude of the value exceeds a predetermined value. Furthermore, when a predetermined notification is sent from the terminal device 102 in a situation where it is determined that performance degradation has occurred, such as when the magnitude of the difference value exceeds a predetermined threshold, the performance determination unit 803 may determine that performance degradation has occurred in response to receiving the predetermined notification. When the performance determination unit 803 determines that performance degradation has occurred, the control unit 804 executes predetermined control to suppress or improve performance degradation, such as instructing the terminal device 102 to update the AI / ML model or use a different model.

[0035] As described above, in this embodiment, in an environment where the beam to be used by the base station device 101 is determined using an AI / ML model in the terminal device 102, information regarding the difference between the predicted wireless quality and the actually measured wireless quality is notified from the terminal device 102 to the base station device 101. This enables the base station device 101 to determine the performance of the AI / ML model used in the terminal device 102, and as a result, enables communication using an appropriate beam. This makes it possible to contribute to Goal 9 of the United Nations-led Sustainable Development Goals (SDGs), which is to "Develop resilient infrastructure, promote sustainable industrialization and foster innovation."

[0036] The invention is not limited to the above-described embodiment, and various modifications and variations are possible within the scope of the gist of the invention.

Claims

1. A terminal device, a measurement means for measuring radio quality of a reference signal transmitted from a base station device using each of a plurality of first beams formed by the base station device and a reference signal transmitted using each of a number of second beams that is less than the number of the plurality of first beams; an acquisition means for acquiring a predicted wireless quality for each of the plurality of first beams by inputting at least a part of wireless qualities of reference signals transmitted using each of the second beams into a trained model by machine learning; a reporting unit configured to report, to the base station device, information on a difference value between the actual radio quality obtained by the measurement and the predicted radio quality for at least some of the plurality of first beams; A terminal device comprising:

2. the reporting means reports to the base station device information on a candidate beam to be used for communication between the base station device and the terminal device, the candidate beam being selected from the plurality of first beams based on the predicted wireless quality, before transmitting information on the difference value; the measuring means measures radio quality for each of the beam candidates; the reporting means reports the difference value between the predicted wireless quality and the actually measured wireless quality for at least some of the beam candidates to the base station device as information on the difference value.

2. The terminal device according to claim 1, wherein:

3. the reporting means reports to the base station device information on a candidate beam to be used for communication between the base station device and the terminal device, the candidate beam being selected from the plurality of first beams based on the predicted wireless quality, before transmitting information on the difference value; the measuring means measures radio quality for each of the beam candidates; the reporting means reports to the base station device information specifying a beam to be used for communication between the base station device and the terminal device based on the wireless quality obtained for each of the candidate beams, and the difference value between the predicted wireless quality and the actually measured wireless quality for the beam, as information on the difference value.

2. The terminal device according to claim 1, wherein:

4. 2. The terminal device according to claim 1, wherein the reporting means reports information indicating whether the magnitude of the difference value exceeds a predetermined threshold to the base station device as information about the difference value.

5. 2. The terminal device according to claim 1, wherein the reporting means transmits a predetermined report to the base station device as information relating to the difference value when the magnitude of the difference value exceeds a predetermined threshold.

6. The terminal device described in Claim 1, characterized in that the reporting means reports information regarding the average value or variance of the difference values ​​corresponding to at least a portion of the plurality of first beams to the base station device as information regarding the difference values.

7. 2. The terminal device according to claim 1, further comprising: a receiving unit configured to receive, from said base station device, setting information for reporting information relating to said difference value.

8. A base station device, a transmitting means for transmitting a reference signal using each of a plurality of first beams and a number of second beams that is less than the plurality of first beams; a receiving means for receiving, from the terminal device, information regarding a difference value between a predicted wireless quality in a terminal device for at least a part of the plurality of first beams, which is obtained by inputting at least a part of wireless qualities of reference signals transmitted using each of the plurality of second beams into a trained model by machine learning, and an actual measured wireless quality obtained by measuring a reference signal in the terminal device for the at least a part of the plurality of first beams; A determination means for determining the performance of the trained model based on information about the difference value; A base station device comprising:

9. 9. The base station apparatus according to claim 8, further comprising a notification unit that notifies the terminal apparatus of setting information for reporting information related to the difference value.

10. A control method executed by a terminal device capable of measuring radio quality of a reference signal transmitted from a base station device using each of a plurality of first beams formed by the base station device and a reference signal transmitted using each of a number of second beams that is less than the plurality of first beams, the method comprising: performing a first measurement of radio quality of a reference signal transmitted from the base station device using each of the second beams; inputting at least a portion of the wireless qualities measured for each of the second beams into a trained model by machine learning to obtain a predicted wireless quality for each of the plurality of first beams; performing a second measurement of radio quality of a reference signal transmitted from the base station device using at least a part of the first beam; reporting, to the base station device, information on a difference value between the actual radio quality obtained by the second measurement and the predicted radio quality for at least the part of the plurality of first beams; A control method comprising:

11. A control method executed by a base station device, Transmitting a reference signal using each of a plurality of first beams and a number of second beams that is less than the plurality of first beams; receiving, from the terminal device, information regarding a difference value between a predicted wireless quality in a terminal device for at least a part of the plurality of first beams, which is obtained by inputting at least a part of wireless qualities of reference signals transmitted using each of the second beams into a trained model by machine learning, and an actual measured wireless quality obtained by measuring a reference signal in the terminal device for the at least a part of the plurality of first beams; Determining the performance of the trained model based on information about the difference value; A control method comprising:

12. A program for causing a computer provided in a terminal device to execute the control method according to claim 10.

13. A program for causing a computer provided in a base station device to execute the control method according to claim 11.

Images