Prediction device, prediction method, and program

Abstract

The present invention comprises: an acquisition unit (51) for acquiring first channel information, generated by a base station (21) on the basis of a first reference signal transmitted from a terminal device (11) at a first time point, and second channel information, generated by the base station on the basis of a second reference signal transmitted from the terminal device at a second time point later than the first time point; and a prediction unit for predicting third channel information at a third time point later than the second time point on the basis of channel information generated by the base station (21) on the basis of the first channel information and the second channel information acquired by the acquisition unit (51). The first reference signal is transmitted at predetermined transmission intervals, and the second reference signal is transmitted at least once during the predetermined transmission intervals.

Description

Prediction device, prediction method, and program

[0001] The present invention relates to a prediction device, a prediction method, and a program, and more particularly to a prediction device, a prediction method, and a program that enable a base station to perform processing using more accurate channel information in communication between a base station and a terminal device.

[0002] In mobile communications such as 5G, a sounding reference signal (SRS) transmitted from a UE is used as a reference signal for measuring uplink channel quality, reception timing, and the like in a base station. Resource elements for the UE to transmit the SRS in the uplink radio resources are allocated by the base station. The base station refers to the SRS to measure the channel quality, reception timing, and the like used in communication with each UE, and generates channel information (e.g., a channel matrix) related to the uplink channel.

[0003] In addition, in the TDD (Time Division Duplex) system adopted in 5G, the downlink and uplink use the same frequency to more effectively use the wireless section. Therefore, channel information related to the uplink channel can also be used in the downlink. For example, it is also possible to generate weights for precoding processing required for MIMO transmission based on channel information related to the uplink channel.

[0004] It is known to generate channel information for a UE by performing channel estimation based on a Demodulation Reference Signal (DMRS) or an SRS signal transmitted by the UE (Prior Art Document 1).It is also known to obtain uplink channel information based on the DMRS transmitted from each UE and to obtain a downlink channel matrix based on channel reciprocity (Prior Art Document 2).

[0005] International Publication No. WO2017 / 038529 Japanese Patent Application Laid-Open No. 2011-130438

[0006] However, when a base station generates current channel information based on SRS received in the past, there is a problem that accuracy decreases due to, for example, an increase in the number of terminals communicating with the base station or an increase in the number of channels used by each terminal, which results in a gap between the most recent time SRS was received and the current time.

[0007] On the other hand, since DMRS has lower accuracy than SRS, when channel information is generated based on DMRS, the accuracy of the generated channel information decreases.

[0008] An object of one aspect of the present invention is to realize a technology that enables a base station to perform processing using more accurate channel information in communications between a base station and a terminal device.

[0009] A prediction device according to one aspect of the present invention is a prediction device that predicts channel information relating to a channel between a base station and a terminal device in a mobile communication network, and includes: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time; and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit, wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0010] A prediction method according to one aspect of the present invention is a prediction method of a prediction device that predicts channel information related to a channel between a base station and a terminal device in a mobile communication network, wherein an acquisition unit acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time, and a prediction unit predicts third channel information at a third time that is temporally later than the second time based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit, wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0011] Each aspect of the present invention may be realized by a computer. In this case, a program that causes a computer to execute each step of the above method, and a computer-readable recording medium on which the program is recorded, also fall within the scope of the present invention.

[0012] In communications between a base station and a terminal device, the base station can perform processing using more accurate channel information.

[0013] 1 is a diagram illustrating an example of a channel information prediction system according to an embodiment. FIG. 2 is a block diagram illustrating an example of the functional configuration of a prediction device. FIG. 3 is a diagram illustrating an example of a resource grid of radio resources used for communication between a UE and a base station. FIG. 4 is a diagram illustrating an SRS transmission period and a DMRS transmission period. FIG. 5 is an arrow chart illustrating an example of processing executed between a base station, a UE, and a prediction device according to an embodiment. FIG. 6 is a diagram illustrating an example of the configuration of a computer that executes instructions of a program that is software that realizes each function.

[0014] First Embodiment Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a diagram showing an example of a channel information prediction system 10 according to this embodiment.

[0015] (Channel Information Prediction System) UE11 and UE12 shown in Fig. 1 are, for example, terminal devices carried by users, and may be constituted by smartphones, personal computers, etc. Alternatively, UE11 and UE12 may be devices such as routers, cameras, sensors, etc. UE11 communicates with base station 21 wirelessly. Although only two UEs are shown in the figure, in reality, many more UEs exist.

[0016] The base station 21 is part of a network that constitutes a 5G mobile communication system, and communicates wirelessly with the UE 11 and also communicates with a core network 22 of the 5G mobile communication system. That is, the UE 11 is connected to the Internet or the like via the base station 21 and the core network 22.

[0017] UE11 and UE12 each transmit a Sounding Reference Signal (SRS), which is a reference signal for measuring uplink channel conditions, etc. The SRS is a signal transmitted from each UE corresponding to a predetermined subcarrier. Symbols for transmitting the SRS from each UE in the uplink radio resources are assigned by base station 21.

[0018] Furthermore, UE11 and UE12 transmit a demodulation reference signal (DMRS) as a reference signal for demodulating a physical uplink channel such as a physical uplink shared channel (PUSCH). Symbols for transmitting the DMRS from each UE in the uplink radio resources are allocated by the base station 21.

[0019] The base station 21 measures uplink channel quality, reception timing, etc. by using the SRS transmitted from the UE 11 as a reference signal. The base station 21 measures channel quality, reception timing, etc. related to communication with each UE by referring to the SRS, and generates uplink channel information. Here, the channel information may be, for example, a channel matrix.

[0020] Furthermore, the base station 21 demodulates a physical uplink channel such as a PUSCH by using the DMRS transmitted from the UE 11 as a reference signal. At this time, the base station 21 generates uplink channel information. Here, the channel information may be, for example, a channel matrix.

[0021] The prediction device 41 is connected to, for example, a Distributed Unit (DU) of the base station 21, and predicts more accurate channel information based on channel information generated by the base station 21. The prediction device 41 predicts a more accurate channel matrix based on, for example, a channel matrix generated by the base station 21 based on a first reference signal (e.g., SRS) transmitted from the UE 11 and a channel matrix generated by the base station 21 based on a second reference signal (e.g., DMRS) transmitted from the UE 11. The prediction device 41 also supplies the predicted channel matrix to the base station 21.

[0022] In the TDD (Time Division Duplex) system adopted in 5G, the uplink and downlink use the same frequency to more effectively use the wireless section. Therefore, when the TDD system is adopted, channel information related to the uplink can also be used in the downlink. For example, it is also possible to generate weights for precoding processing required for downlink MIMO transmission based on channel information related to the uplink.

[0023] The base station 21 executes processing related to communication with the UE 11 and the UE 12 using the channel matrix supplied from the prediction device 41. For example, the base station 21 executes processing to generate weights for a precoding process using the channel matrix supplied from the prediction device 41.

[0024] 2 is a block diagram showing an example of the functional configuration of the prediction device 41. In the example shown in the figure, the prediction device 41 includes an acquisition unit 51 and a prediction unit 52.

[0025] The prediction device 41 is a prediction device that predicts channel information related to a channel between a base station (e.g., base station 21) and a terminal device (e.g., UE 11) in a mobile communication network, and includes an acquisition unit 51 that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time, and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time, and a prediction unit 52 that predicts third channel information at a third time that is temporally later than the second time, based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit 51, wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0026] The first reference signal may be a Sounding Reference Signal (SRS). The second reference signal may be a Demodulation Reference Signal (DMRS). The prediction unit 52 may predict a channel matrix related to an uplink between the terminal device and the base station as the third channel information.

[0027] (Acquisition Unit) The acquisition unit 51 acquires, for example, a first channel matrix generated by the base station 21 based on the SRS transmitted by the UE 11 and a second channel matrix generated based on the DMRS transmitted by the UE 11. The first channel matrix and the second channel matrix acquired by the acquisition unit 51 are used to predict a third channel matrix in the prediction unit 52.

[0028] Similarly, the acquisition unit 51 acquires, for example, a first channel matrix generated by the base station 21 based on the SRS transmitted by the UE 12 and a second channel matrix generated based on the DMRS transmitted by the UE 12. The first channel matrix and the second channel matrix acquired by the acquisition unit 51 are used by the prediction unit 52 to predict a third channel matrix.

[0029] (Prediction Unit) The prediction unit 52 is a functional block that executes prediction using a machine learning model. The prediction unit 52 generates a third channel matrix, which is a more accurate channel matrix, based on the first channel matrix and the second channel matrix acquired by the acquisition unit 51.

[0030] As an example, the machine learning model employed by the prediction unit 52 may be configured as a model employing a self-attention mechanism. A model employing a self-attention mechanism is suitable for prediction using time-series data and is considered to be suitable for predicting the most recent channel information. Furthermore, compared to, for example, a convolutional neural network (CNN) model, a model employing a self-attention mechanism can flexibly change the number of dimensions of input data and is considered to be suitable for predicting a channel matrix.

[0031] The learning of the machine learning model included in the prediction unit 52 will be described later.

[0032] (SRS Transmission Interval) Fig. 3 is a diagram showing an example of a resource grid of radio resources used for communication between UE11 and UE12 and base station 21. The horizontal axis of the diagram represents time, the vertical axis represents frequency, and each rectangular grid represents a resource element.

[0033] 3, for convenience, a resource grid configured by the first to sixth subcarriers and the first to tenth symbols is shown. In this example, it is assumed that six UEs, UE-a to UE-f, are connected to the base station 21.

[0034] In Figure 3, the hatched grid indicates the resource elements from which each UE transmits the SRS. In this example, UE-a transmits the SRS in the resource element corresponding to the second symbol on the first subcarrier. UE-b transmits the SRS in the resource element corresponding to the third symbol on the second subcarrier. UE-c transmits the SRS in the resource element corresponding to the fourth symbol on the third subcarrier.

[0035] In this way, the resource elements used for transmitting the SRS from each UE are allocated in a round-robin manner, which means that it takes a time Ts from the time UE-a transmits the first SRS until the time UE-a transmits the second SRS.

[0036] 4 is a diagram illustrating the transmission period of an SRS and a DMRS. In this diagram, the horizontal axis represents time and the vertical axis represents frequency, and a portion of a two-slot resource block is shown. It should be noted that all 14 resource elements shown in this diagram are assumed to be allocated to uplink communications of the same UE (e.g., UE11).

[0037] In the figure, blank rectangles indicate resource elements where data is transmitted, vertically hatched rectangles indicate resource elements where DMRS is transmitted, and horizontally hatched rectangles indicate resource elements where SRS is transmitted.

[0038] In the example of Figure 4, SRS 102-1 is transmitted in the resource element corresponding to the 14th symbol. Meanwhile, DMRS 101-1 is transmitted in the resource element corresponding to the 4th symbol, and DMRS 101-2 is transmitted in the resource element corresponding to the 11th symbol. In this way, the transmission frequency of DMRS is higher than the transmission frequency of SRS. That is, after one SRS is transmitted, DMRS is transmitted multiple times (at least one) within the transmission period until the next SRS is transmitted.

[0039] When the base station 21 receives an SRS, it generates a channel matrix based on the SRS, and when it receives a DMRS, it generates a channel matrix based on the DMRS, and the respective channel matrices are acquired as a first channel matrix and a second channel matrix by the acquisition unit 51 of the prediction device 41. That is, within a predetermined transmission period, a second reference signal is transmitted multiple times from a terminal device (e.g., UE 11), the base station 21 generates second channel information corresponding to each of the second reference signals transmitted multiple times, and the acquisition unit 51 acquires the first channel information and the multiple pieces of second channel information.

[0040] Furthermore, in the TDD (Time Division Duplex) system adopted in 5G, the downlink and uplink use the same frequency in order to use the wireless section more effectively. Therefore, when the TDD system is adopted, channel information related to the uplink channel can also be used in the downlink.

[0041] (Precoding Weights) The base station 21 can also generate precoding weights required for downlink MIMO transmission, for example, by referring to channel information based on the SRS transmitted from each UE. When generating the precoding weights, it is desirable to use channel information (channel matrix) that is as accurate as possible.

[0042] Here, an example of a method for calculating weights for precoding processing will be described.

[0043] If the transmitted symbol is represented by x and the noise is represented by n, the autocorrelation matrix R of x is ｘｘ and the autocorrelation matrix R of n ｎｎ can be expressed by equations (1) and (2), respectively, where E represents an expected value.

[0044] ...(1)

[0045] ...(2) Also, the cross-correlation matrix R between x and n ｘｎ and the cross-correlation matrix R between n and x ｎｘ can be expressed by equations (3) and (4), respectively.

[0046] ...(3)

[0047] (4) In the formulas (3) and (4), Nt represents the number of transmitting antennas, and Nr represents the number of receiving antennas.

[0048] Furthermore, when the received symbol is represented by y, the channel matrix is ​​represented by H, and the equivalent weight is represented by W, the estimated transmitted symbol x^ can be expressed by equation (5). Note that the notation of x^ is represented by a ^ (hat) placed above x.

[0049] ...(5) Here, the equivalent weight can be calculated by equation (6). Here, the equivalent weight W ＭＭＳＥ The following will be calculated.

[0050] ...(6) The autocorrelation matrix R in equation (6) ｘｘ and the autocorrelation matrix R ｎｎ can be approximated using the SNR. Therefore, by specifying the channel matrix H and the SNR, W ＭＭＳＥ It is possible to calculate

[0051] By performing an operation similar to that shown in Equation (6) before transmitting data, the weight of the precoding process is obtained as W ＭＭＳＥ can be calculated.

[0052] However, as described with reference to Fig. 3, for the same UE (e.g., UE-a), it takes a certain time (e.g., time Ts in Fig. 3) from receiving an SRS once until receiving the next SRS. For example, if the bandwidth of radio resources expands and the number of UEs connected to base station 21 increases, the time required from receiving an SRS once until receiving the next SRS becomes even longer.

[0053] For example, it may take several hundred milliseconds since UE11 last transmitted an SRS. For example, when UE11 is moving at high speed, even if the channel matrix H is generated based on the last transmitted SRS, the channel quality may have changed significantly. In such a case, for example, when generating weights for precoding processing related to communication with UE11, it may not be possible to generate appropriate weights by referring to the channel matrix generated based on the last transmitted SRS.

[0054] Therefore, in this embodiment, a channel matrix is ​​predicted by the prediction device 41. For example, based on a first channel matrix generated based on the SRS last transmitted by UE 11 and a second channel matrix generated based on the DRMS ​​transmitted by UE 11 after transmitting the SRS, the prediction device 41 generates a third channel matrix, which is a more accurate channel matrix.

[0055] The base station 21 generates weights for the precoding process related to communication with the UE 11 by using the third channel matrix supplied from the prediction device 41 as the channel matrix H in equation (6). In this way, it becomes possible to generate weights for the precoding process more appropriately.

[0056] (Machine Learning Model of Prediction Unit) Next, an example of a learning method for the machine learning model included in the prediction unit 52 of the prediction device 41 will be described.

[0057] As an example, a first channel matrix generated based on an SRS previously transmitted by a certain UE and a channel matrix generated based on a DMRS transmitted by the UE after transmitting the SRS are used in learning of the machine learning model included in the prediction unit 52. Furthermore, the channel matrix generated based on the SRS transmitted by the UE after transmitting the DMRS is used as a correct label.

[0058] 4, assume that SRS 102-1 is the SRS transmitted for the second time from UE 11. Three channel matrices are identified: a channel matrix generated based on SRS 102-0 (not shown) transmitted for the first time by UE 11, a channel matrix generated based on DMRS 101-1, and a channel matrix generated based on DMRS 101-2. Then, training data is generated in which the channel matrix generated based on SRS 102-1 is associated with these three channel matrices as a correct label.

[0059] A plurality of such teacher data are generated for UE 11. Similarly, a plurality of such teacher data are generated for UE 12 and each of the other plurality of UEs. Then, model parameters of the machine learning model are calculated based on the teacher data, and the machine learning model included in the prediction unit 52 is trained.

[0060] Although an example has been described here in which a set of training data consisting of one channel matrix generated based on SRS and three channel matrices generated based on DMRS is used, other combinations may be used. For example, a set of training data consisting of one channel matrix generated based on SRS and two (or one) channel matrices generated based on DMRS may be used. Alternatively, a set of training data consisting of one channel matrix generated based on SRS and four (or more) channel matrices generated based on DMRS may be used.

[0061] When the learning of the machine learning model included in the prediction unit 52 is completed, the prediction unit 52 can predict channel information (channel matrix) using the machine learning model. That is, the prediction unit 52 can receive the channel matrix generated based on the SRS and the channel matrix generated based on the DMRS from the acquisition unit 51, and output a more accurate channel matrix as a prediction result.

[0062] In this way, the prediction unit 52 has a machine learning model trained using training data consisting of a set of first channel information generated based on a first reference signal transmitted at a first time, second channel information generated based on a second reference signal transmitted at a second time, and another first channel information generated based on another first reference signal transmitted from a terminal device (e.g., UE11) after the second time, and predicts the third channel information using the machine learning model.

[0063] Next, an example of processing executed between the base station 21, the UE 11, and the prediction device 41 according to this embodiment will be described with reference to Fig. 5. Fig. 5 is an arrow chart illustrating an example of processing executed between the base station 21, the UE 11, and the prediction device 41 according to this embodiment.

[0064] By performing this process, for example, the base station 21 can perform processing using a more accurate channel matrix predicted by the prediction device 41. For example, even if it takes a long time for the base station 21 to receive the next SRS after receiving one SRS, by using the channel matrix predicted by the prediction device 41, weights for the precoding process can be generated more appropriately.

[0065] In step S101, the base station 21 transmits an RRC setup command to the UE 11, which is received by the UE 11 in step S121. The transmitted RRC setup command includes a description related to SRS-config.

[0066] 5, the description related to SRS-config is "SRS config: periodicity=20 slots, symbol 8." This indicates that the SRS transmission period is 20 slot intervals, and the eighth symbol is the SRS.

[0067] In step S122, UE 11 transmits the SRS in the 20th slot of the uplink, which is received by base station 21 in step S102. That is, UE 11 periodically transmits the SRS to base station 21 in accordance with the description related to SRS-config included in the RRC setup command received in step S121.

[0068] In step S103, the base station 21 generates channel information based on the SRS received in step S102. At this time, the base station 21 refers to the SRS to measure the channel quality and reception timing related to communication with the UE 11, and generates, for example, a channel matrix as uplink channel information.

[0069] In step S104, the base station 21 performs calculations related to the precoding process based on the channel information generated in step S103. At this time, for example, weights for the precoding process may be calculated.

[0070] In step S105, the base station 21 supplies the channel information generated in step S103 to the prediction device 41. The prediction device 41 acquires the supplied channel information in step S141.

[0071] At this time, the base station 21 supplies the channel information (e.g., channel matrix) generated in step S103 to the prediction device 41, together with, for example, the identification number of UE11, information indicating that the channel information is channel information generated based on the SRS, and the identification number of the slot in which the SRS referenced when generating the channel information was transmitted.

[0072] In step S123, UE11 transmits PUSCH and DMRS using resource elements in a predetermined slot of the uplink, which are received by base station 21 in step S106.

[0073] In step S107, the base station 21 generates channel information based on the DMRS received in step S106. At this time, the base station 21 demodulates the PUSCH by using the DMRS transmitted from the UE 11 as a reference signal, for example, and generates a channel matrix as the channel information.

[0074] In step S108, the base station 21 supplies the channel information based on the DMRS received in step S106 (the channel information generated in step S107) to the prediction device 41. The prediction device 41 acquires the supplied channel information in step S142.

[0075] At this time, the base station 21 supplies the channel information (e.g., a channel matrix) generated in step S107 to the prediction device 41, together with, for example, the identification number of UE11, information indicating that the channel information is channel information generated based on the DMRS, and the identification number of the slot in which the DMRS referenced when generating the channel information was transmitted.

[0076] In step S143, the prediction device 41 predicts more accurate channel information based on the channel information acquired in step S141 and the channel information acquired in step S142. At this time, the prediction unit 52 receives the channel matrix generated based on the SRS and the channel matrix generated based on the DMRS from the acquisition unit 51, and predicts the more accurate channel matrix related to the UE 11.

[0077] That is, the prediction device 41 predicts a third channel matrix in step S143 based on the first channel matrix acquired in step S141 and the second channel matrix acquired in step S142.

[0078] The processes of steps S123, S106, S107, S108, S142, and S143 are repeatedly executed each time a DMRS is transmitted from UE 11 during a predetermined period. That is, these processes are repeatedly executed each time a DMRS is transmitted from UE 11 during the period from when an SRS is transmitted from UE 11 to base station 21 in step S122 until the next SRS is transmitted. For example, if the SRS transmission period is 20 slot intervals in the description related to SRS-config in the RRC setup command transmitted in step S101, these processes are repeatedly executed during the period from the 21st slot to the 40th slot.

[0079] In step S143, elements of the previously predicted channel matrix (third channel matrix) may be updated every time a channel matrix (second channel matrix) based on a new DMRS is supplied. That is, the prediction unit 52 may update the prediction result of the third channel information every time the acquisition unit 51 acquires each of the plurality of pieces of second channel information.

[0080] Now, for example, assume that a large number of UEs are connected to the base station 21 during the period from the 21st slot to the 40th slot. For example, when a large number of UEs are connected to the base station 21, the predetermined time (e.g., time Ts in FIG. 3) between receiving an SRS once and receiving the next SRS for the same UE becomes long. In this case, the base station 21 determines that it must set the SRS transmission period longer.

[0081] In step S108, the base station 21 transmits an RRC Reconfig command, which is received by the UE 11 in step S124. The transmitted RRC setup command includes a description relating to update SRS-config.

[0082] 5, "update SRS config: periodicity = 40 slots" is written as a description related to update SRS-config. This indicates that the SRS transmission period is updated to 40 slot intervals, and base station 21 transmits information specifying a longer SRS transmission period to UE 11. In other words, when the number of UEs connected to base station 21 increases, for example, the SRS transmission period of 20 slot intervals specified by SRS-config included in the RRC setup command transmitted in step S101 is updated to 40 slot intervals. As a result, UE 11 will transmit SRS at 40 slot intervals in the uplink from now on.

[0083] In step S109, the base station 21 transmits a request for the prediction result, which is received by the prediction device 41 in step S144. That is, the base station 21 requests the prediction device 41 to transmit the channel information (channel matrix) predicted by the prediction device 41. The request for the prediction result transmitted here includes an identifier of the UE related to the requested channel information (e.g., the identifier of UE 11).

[0084] In step S145, the prediction device 41 transmits a response corresponding to the request received in step S144, which is received by the base station 21 in step S110. The transmitted response includes channel information related to the UE corresponding to the identification information included in the request received in step S144. As a result, the prediction device 41 transmits, for example, channel information (channel matrix) related to the UE 11 predicted by the processing in step S143 to the base station 21.

[0085] In step S111, the base station 21 executes a calculation related to the precoding process using the channel information (channel matrix) included in the response received in step S110. At this time, for example, weights for the precoding process may be calculated.

[0086] In this manner, the processing is executed between the base station 21, the UE 11, and the prediction device 41 according to this embodiment.

[0087] (Effects of this embodiment) According to this embodiment, the prediction unit 52 predicts third channel information at a third time that is temporally later than the second time, based on channel information generated by the base station based on the first channel information and second channel information acquired by the acquisition unit 51. Furthermore, the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0088] 5 , the weights for the precoding process calculated in step S104 are obtained based on the channel matrix generated based on the SRS received in step S102. That is, in step S104, it is possible to calculate the weights for the precoding process using the channel matrix corresponding to the SRS received from UE 11 immediately before the calculation of the weights for the precoding process.

[0089] On the other hand, when calculating the weights for the precoding process in step S111, the SRS most recently received from UE 11 is still the SRS received in step S102. If a channel matrix corresponding to the SRS most recently received from UE 11 is used in step S111 as well, the weights for the precoding process will be calculated several tens of slots after the most recently received SRS from UE 11. That is, if the weights for the precoding process are calculated in step S111 using a channel matrix corresponding to the SRS most recently received from UE 11, as in the case of step S104, there is a possibility that an appropriate weight cannot be calculated.

[0090] That is, several tens of slots after receiving the SRS in step S102, the channel quality, etc. related to UE 11 may have changed over time. In particular, when UE 11 is moving at high speed, the change in the channel quality, etc. related to UE 11 over time becomes significant. Therefore, at the time step S111 is executed, the accuracy of the channel matrix generated in step S103 is considered to be low. In this way, if a channel matrix with low accuracy is used, it may not be possible to generate appropriate weights as weights for the precoding process related to communication with UE 11.

[0091] In this embodiment, when the base station 21 executes a calculation related to a precoding process, it is possible to acquire a prediction result from the prediction device 41. In this way, even when the calculation related to a precoding process is executed at a point in time when a time has elapsed since the base station 21 most recently received an SRS from the UE 11, it is possible to execute a calculation using a highly accurate channel matrix.

[0092] In other words, by using the channel matrix predicted by the prediction device 41, it becomes possible to perform calculations using the channel matrix corresponding to the SRS most recently received by UE11, even if some time has passed since the most recent SRS was received from UE11.

[0093] Second Embodiment In the first embodiment, an example in which the prediction device 41 and the base station 21 are configured separately has been described. However, the prediction device 41 may be configured as an integrated unit with the base station 21 .

[0094] The functions of the base station 21 may also be realized as a virtualized RAN (vRAN). For example, RUs (radio units) may be installed at multiple sites as devices that perform processing of the antenna portion of the base station and the lowest layer portion of the PHY layer, and signal modulation / demodulation and retransmission of lost signals may be controlled by a distributed unit (DU) located in a data center or the like. Furthermore, a RAN intelligent controller (RIC) that controls the operation of the RU and DU may be provided in the data center or the like.

[0095] When the function of the base station 21 is realized as a vRAN, the prediction device 41 may be integrated with, for example, a computer constituting an RIC. In this case, for example, the RIC may predict third channel information, and calculation of weights for precoding processing may be performed using the third channel information.

[0096] (Software Implementation Example) The prediction device 41 described above is a program for causing a computer to function, and can be realized by a program for causing a computer to function as the prediction device 41. In this case, the prediction device 41 includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. An example of such a computer is shown in FIG. 6.

[0097] The computer 500 includes at least one processor 501 and at least one memory 502. The memory 502 stores a program 520 for causing the computer 500 to operate as the prediction device 41. In the computer 500, the processor 501 reads and executes the program 520 from the memory 502, thereby realizing each function of the prediction device 41.

[0098] The processor 501 may be, for example, a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), a PPU (Physics Processing Unit), a microcontroller, or a combination of these.

[0099] The memory 502 may be, for example, a flash memory, a hard disk drive (HDD), a solid state drive (SSD), or a combination thereof.

[0100] The computer 500 may further include a RAM (Random Access Memory) for expanding the program 520 during execution and for temporarily storing various data. The computer 500 may also include a communication interface for transmitting and receiving data to and from other devices. The computer 500 may also include an input / output interface for connecting input / output devices such as a keyboard, a mouse, a display, and a printer.

[0101] Furthermore, the program 520 for causing the computer 500 to operate as the prediction device 41 can be recorded on a non-transitory, tangible recording medium 530 that is readable by the computer 500. Such a recording medium 530 can be, for example, a tape, a disk, a card, a semiconductor memory, or a programmable logic circuit. The computer 500 can acquire the program 520 via such a recording medium 530.

[0102] Furthermore, the program 520 for causing the computer 500 to operate as the prediction device 41 can be transmitted via a transmission medium. Examples of such a transmission medium include a communication network and broadcast waves. The computer 500 can also acquire the program 520 via such a transmission medium.

[0103] Furthermore, some or all of the functions of the prediction device 41 can be realized by logic circuits. For example, an integrated circuit in which a logic circuit that functions as each of the above control blocks is formed is also included in the scope of the present invention. In addition, the functions of each of the above control blocks can also be realized by, for example, a quantum computer.

[0104] Furthermore, in each of the above-described embodiments, examples have been described in which the present invention is applied to a 5G communication system, but the present invention can also be applied to communication systems from 6G onwards.

[0105] According to each aspect of the present invention described above, the above-mentioned effects can be achieved, thereby contributing to the achievement of Goal 9 of the Sustainable Development Goals (SDGs), "Build resilient infrastructure, promote inclusive and sustainable industrialization, and build resilient infrastructure."

[0106] The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0107] [Summary] A prediction device according to a first aspect of the present invention is a prediction device that predicts channel information relating to a channel between a base station and a terminal device in a mobile communication network, and includes: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time; and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time, based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit; wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0108] A prediction device according to aspect 2 of the present invention is, in the above-mentioned aspect 1, characterized in that the prediction unit has a machine learning model trained using training data consisting of a set of the first channel information generated based on a first reference signal transmitted at the first time, the second channel information generated based on a second reference signal transmitted at the second time, and another first channel information generated based on another first reference signal transmitted from the terminal device after the second time, and predicts the third channel information using the machine learning model.

[0109] A prediction device according to Aspect 3 of the present invention is the prediction device of Aspect 2 above, wherein the machine learning model possessed by the prediction unit is configured by a model employing a self-attention mechanism.

[0110] A prediction device according to aspect 4 of the present invention is the same as that according to aspect 2 above, in which the second reference signal is transmitted from the terminal device multiple times within the specified transmission period, the base station generates second channel information corresponding to each of the second reference signals transmitted multiple times, and the acquisition unit acquires the first channel information and the multiple pieces of second channel information.

[0111] A prediction device according to aspect 4 of the present invention is, in any of aspects 1 to 3 above, such that the second reference signal is transmitted from the terminal device multiple times within the specified transmission period, the base station generates second channel information corresponding to each of the second reference signals transmitted multiple times, and the acquisition unit acquires the first channel information and the multiple pieces of second channel information.

[0112] A prediction device according to aspect 5 of the present invention is the same as in aspect 2 above, wherein the prediction unit updates the prediction result of the third channel information each time the acquisition unit acquires each of the plurality of second channel information.

[0113] A prediction device according to a sixth aspect of the present invention is the prediction device of any one of the first to fifth aspects, wherein the first reference signal is a Sounding Reference Signal (SRS), the second reference signal is a Demodulation Reference Signal (DMRS), and the prediction unit predicts, as the third channel information, a channel matrix related to an uplink between the terminal device and the base station.

[0114] A prediction device according to aspect 7 of the present invention is the same as in aspect 6 above, in that the base station transmits information to the terminal device specifying the transmission period of the SRS, and when the number of terminal devices connected to the base station increases, the base station transmits information to the terminal device specifying a longer transmission period of the SRS.

[0115] A prediction device according to aspect 8 of the present invention is configured as an integrated part of the base station in aspect 7 above, and the base station calculates weights for the precoding process using the third channel information predicted by the prediction unit.

[0116] A prediction method according to aspect 9 of the present invention is a prediction method of a prediction device that predicts channel information relating to a channel between a base station and a terminal device in a mobile communication network, the prediction method including: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time; and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit; the first reference signal is transmitted at a predetermined transmission period; and the second reference signal is transmitted at least once within the predetermined transmission period.

[0117] A program according to aspect 10 of the present invention causes a computer to function as a prediction device that predicts channel information relating to a channel between a base station and a terminal device in a mobile communication network, the prediction device comprising: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time; and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit, wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

[0118] REFERENCE SIGNS LIST 10 Channel information prediction system 11 UE 12 UE 21 Base station 22 Core network 41 Prediction device 51 Acquisition unit 52 Prediction unit

Claims

1. A prediction device for predicting channel information relating to a channel between a base station and a terminal device in a mobile communication network, comprising: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time, and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time, based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit; wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

2. The prediction device described in claim 1, wherein the prediction unit has a machine learning model trained using training data consisting of a set of the first channel information generated based on a first reference signal transmitted at the first time, the second channel information generated based on a second reference signal transmitted at the second time, and another first channel information generated based on another first reference signal transmitted from the terminal device after the second time, and predicts the third channel information using the machine learning model.

3. The prediction device according to claim 2, wherein the machine learning model possessed by the prediction unit is configured as a model employing a self-attention mechanism.

4. The prediction device according to claim 1, wherein the second reference signal is transmitted from the terminal device multiple times within the predetermined transmission period, the base station generates second channel information corresponding to each of the second reference signals transmitted multiple times, and the acquisition unit acquires the first channel information and the multiple pieces of second channel information.

5. The prediction device according to claim 4, wherein the prediction unit updates the prediction result of the third channel information each time the acquisition unit acquires each of the plurality of second channel information.

6. The prediction device according to claim 1, wherein the first reference signal is an SRS (Sounding Reference Signal), the second reference signal is a DMRS (Demodulation Reference Signal), and the prediction unit predicts, as the third channel information, a channel matrix related to an uplink between the terminal device and the base station.

7. The prediction device according to claim 6, wherein the base station transmits information specifying a transmission period of the SRS to the terminal device, and when the number of terminal devices connected to the base station increases, the base station transmits information specifying a longer transmission period of the SRS to the terminal device.

8. The prediction device according to claim 7, which is configured integrally with the base station, and the base station calculates weights for precoding processing using the third channel information predicted by the prediction unit.

9. A prediction method for a prediction device that predicts channel information related to a channel between a base station and a terminal device in a mobile communication network, comprising: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time; and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; a prediction unit that predicts third channel information at a third time that is temporally later than the second time, based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit; wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

10. A program that causes a computer to function as a prediction device that predicts channel information related to a channel between a base station and a terminal device in a mobile communication network, comprising: an acquisition unit that acquires first channel information generated by the base station based on a first reference signal transmitted from the terminal device at a first time, and second channel information generated by the base station based on a second reference signal transmitted from the terminal device at a second time that is temporally later than the first time; and a prediction unit that predicts third channel information at a third time that is temporally later than the second time, based on the channel information generated by the base station based on the first channel information and the second channel information acquired by the acquisition unit; wherein the first reference signal is transmitted at a predetermined transmission period, and the second reference signal is transmitted at least once within the predetermined transmission period.

Images