Data generation system and data generation method

Abstract

Using known data and shared information, a management device (10) generates teacher data used when a base station (30) uses a wireless signal received from a user terminal (UE) as an explanatory variable and the user terminal generates a model in which the wireless signal transmitted to the base station serves as an objective variable.

Description

Data generation system and data generation method

[0001] The present invention relates to a data generation system and a data generation method.

[0002] An invention is known that uses machine learning with training data to improve matching accuracy in a process of matching an unknown transmitting device based on a signal wirelessly transmitted from the device (Patent Document 1). The invention described in Patent Document 1 improves matching accuracy by using a machine-learned model that obtains a correct sample feature for a known signal or known radio wave feature and outputs a correct label indicating the sample feature.

[0003] Japanese Patent Application Publication No. 2023-080852

[0004] When a base station demodulates a radio signal received from a UE (User Equipment), it performs two-dimensional interpolation on the channel information estimated at each pilot position and demodulates the radio signal using the channel estimate at the data position obtained by this. While such conventional two-dimensional linear interpolation is easy to implement, its estimation accuracy is not necessarily sufficient.

[0005] Therefore, the inventors focused on demodulating wireless signals using artificial intelligence (AI) that utilizes machine learning, which has been actively developed in recent years. It is difficult for a base station to grasp what information a UE is carrying in its wireless resources. More specifically, it is difficult for a base station to grasp the "original format of the wireless signal transmitted from the UE" regarding the wireless signal received from the UE. Therefore, unlike Patent Document 1, when attempting to generate a model that infers "the wireless signal transmitted from the UE" from "the wireless signal received by the base station," there is no training data.

[0006] In this case, learning is performed without teacher data, but the optimal parameter set is found based on the results of numerous tests performed to determine what data is obtained when a specific parameter set is used, a process known as reinforcement learning. However, with this type of reinforcement learning, there is a risk that the wireless quality processed by the base station will become unstable until the optimal parameter set is found. Therefore, it is necessary to generate teacher data that indicates the "wireless signals transmitted by the UE."

[0007] In order to solve the above problem, a data generation system according to one embodiment of the present invention is a data generation system including a user terminal, a base station, and a management device, wherein when the user terminal accesses a predetermined network node, the user terminal transmits predetermined, uniquely determined known data to the network node, the base station transmits shared information shared with the user terminal to the management device, and the management device uses the known data and the shared information to generate training data to be used when generating a model in which the radio signal received by the base station from the user terminal is an explanatory variable and the radio signal transmitted by the user terminal to the base station is an objective variable.

[0008] In order to solve the above problem, a data generation method according to one aspect of the present invention is a data generation method used in a data generation system including a user terminal, a base station, and a management device, and includes: a first transmission step in which, when the user terminal accesses a predetermined network node, the user terminal transmits uniquely determined known data to the network node; a second transmission step in which the base station transmits shared information shared with the user terminal to the management device; and a generation step in which the management device uses the known data and the shared information to generate training data to be used when generating a model in which the radio signal received by the base station from the user terminal is an explanatory variable and the radio signal transmitted by the user terminal to the base station is an objective variable.

[0009] By generating training data that indicates the "wireless signals transmitted by the UE," training with training data becomes possible. This makes the above-mentioned reinforcement learning unnecessary, and can resolve the instability of the wireless quality of the base station that occurs during reinforcement learning.

[0010] 1 is a diagram illustrating an example of the configuration of a wireless communication system according to an embodiment of the present invention; FIG. 2 is a diagram illustrating an example of functional blocks of a management device according to an embodiment of the present invention; FIG. 3 is a diagram illustrating an example of channel estimation; FIG. 4 is a sequence chart illustrating an example of processing executed by a wireless communication system according to an embodiment of the present invention; and FIG. 5 is a block diagram illustrating an example of the configuration of a computer that can be used as a management device, a base station, etc.

[0011] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the description of the drawings, the same parts are designated by the same reference numerals and the description thereof will be omitted.

[0012] (Configuration of Wireless Communication System) FIG. 1 is a diagram illustrating an example of the configuration of a wireless communication system 1 according to this embodiment.

[0013] 1, the wireless communication system 1 includes a management device 10, a web server 20, an application server 21, a base station 30, and a user terminal UE (User Equipment). The wireless communication system 1 is intended to be applied to a 5G network, but is not limited to this. For example, the wireless communication system 1 may be applied to a future network following the 5G network, a 4G network, or a legacy network prior to the 4G network.

[0014] The base station 30 communicates with user terminals UE located inside a cell 31, which is a wireless communication area.

[0015] 1 shows one base station 30 and one user terminal UE for ease of viewing, but the number of base stations connected to the core network 40 is not limited to one and may be multiple. Also, the number of user terminals UE connected to the base station 30 may be two or more. Also, the number of web servers 20 and application servers 21 is not limited to one and may be multiple.

[0016] Although the user terminal UE is shown in Fig. 1 as a smartphone that can be carried by a user, this is an example and is not limited to a smartphone. For example, the user terminal UE may be a tablet terminal, a smart watch, a mobile phone, etc. Furthermore, the user terminal UE may be an electronic device attached to a vehicle, etc. The user terminal UE may be connected to the base station 30 by wire or wirelessly.

[0017] The management device 10, the web server 20, the application server 21, and the base station 30 are connected to each other via a core network 40 so as to be able to communicate with each other.

[0018] The core network 40 is a network having functions such as a UPF (User Plane Function), a NEF (Network Exposure Function), a PCF (Policy Control Function), an AMF (Access and Mobility Management Function), and a SMF (Session Management Function), and is used to establish user sessions and transfer user data.

[0019] The management device 10, web server 20, and application server 21 may be composed of one or more general-purpose computers, and may have a processor for executing programs, memory for storing programs and data, auxiliary storage devices (HDD, SSD, etc.), and a network interface.

[0020] The web server 20 processes, for example, a request sent from a user terminal UE, and sends a web page or the like as a response to the user terminal UE.

[0021] The application server 21 processes requests sent from the user terminal UE, for example, and executes necessary business logic such as database access, data calculation, and external API calls.

[0022] Such web server 20 and application server 21 are examples of network nodes. A network node refers to a device or software for communicating within a network, and provides functions such as sending and receiving, processing, and relaying data within the network. In addition to the web server 20 and application server 21, network nodes may also include database servers, mail servers, file servers, etc.

[0023] (Functional Blocks of Management Device) FIG. 2 is a diagram showing an example of functional blocks of the management device 10 according to this embodiment.

[0024] The management device 10 may be configured as a hyper-converged infrastructure (HCI) including multiple servers. The management device 10 includes a communication unit 110, a teacher data generation server 120, a screening server 130, and a learning server 140. In the present embodiment, the teacher data generation server 120, the screening server 130, and the learning server 140 are integrated into a single management device 10. However, this is not limited to this example, and each server may be configured as a separate device. The management device 10 may also be configured as a RAN Intelligent Controller (RIC). The RIC is a controller that manages and controls nodes such as RUs (radio units), DUs (distributed units), and CUs (aggregate units) that constitute a 5G RAN, as well as software on the devices. The management device 10 may also be a device managed and operated by a telecommunications carrier (carrier).

[0025] The communication unit 110 communicates with the web server 20 , the application server 21 , and the base station 30 via the core network 40 .

[0026] (Configuration of Teacher Data Generation Server) The teacher data generation server 120 includes a control unit 121 , a memory 124 , and a storage unit 125 .

[0027] The memory 124 temporarily stores various programs executed by the control unit 121 and various data referenced by these programs.

[0028] The storage unit 125 stores information to be read, written, and referenced by the control unit 121.

[0029] The control unit 121 includes a data acquisition unit 122 and a teacher data generation unit 123 .

[0030] The data acquisition unit 122 acquires a Cell ID and a Cell-Radio Network Temporary Identifier (C-RNTI) from the base station 30. The Cell ID is an identifier for identifying the cell 31 of the base station 30. The C-RNTI is an identifier that the base station 30 assigns to the user terminal UE for communication with the user terminal UE. The data acquisition unit 122 also acquires an MCS (Modulation and Coding Scheme), a codebook, a MIMO specification (Multiple Input Multiple Output), and an RB specification (Resource Block) from the base station 30.

[0031] The MCS indicates the modulation and coding scheme for the transmission data. The Codebook indicates predetermined precoding weight matrix candidates. The MIMO specification specifies the number of transmit antennas and the transmission scheme to be used by the user terminal UE. The RB specification specifies resources in both the frequency and time dimensions.

[0032] The Cell ID and C-RNTI are shared information shared between the base station 30 and the teacher data generation server 120, and are used to identify the base station 30 and the user terminal UE. The Cell ID and C-RNTI are preferably used in a trained model, which will be described later, when reproducing a radio signal transmitted from a specific user terminal (for example, a user terminal UE) to a specific base station (for example, the base station 30).

[0033] The MCS, Codebook, MIMO designation, and RB designation are also shared information shared between the base station 30 and the teacher data generation server 120. The MCS, Codebook, MIMO designation, and RB designation are also shared information shared between the base station 30 and the user terminal UE. Such MCS, Codebook, MIMO designation, and RB designation are used when converting known data into a PUSCH signal (Physical Uplink Shared Channel), which is a radio signal. The PUSCH signal is a physical channel for transmitting data on the uplink from the user terminal UE to the base station 30.

[0034] Here, the "known data" according to this embodiment will be described. "Known data" is data that is uniquely determined in advance, and is data that the user terminal UE is arranged to send to the corresponding web server 20 or application server 21 when the user terminal UE accesses an arbitrary web (web server 20) or an arbitrary application (application server 21). The format of the "known data" is not particularly limited, and may be any format.

[0035] The "known data" is processed in the user terminal UE and converted into a PUSCH signal through an inverse Fourier transform (IFFT: Inverse Fast Fourier Transform) and D / A conversion. The MCS, Codebook, MIMO designation, and RB designation used in the process of converting the known data into a PUSCH signal are shared between the base station 30 and the user terminal UE. Furthermore, the C-RNTI is also used in the process of converting the known data into a PUSCH signal, but since the C-RNTI is an identifier assigned by the base station 30 to the user terminal UE, the C-RNTI can also be said to be shared information shared between the base station 30 and the user terminal UE.

[0036] The C-RNTI is a static value, and the MCS, Codebook, MIMO designation, and RB designation are dynamic values. Note that the shared information listed here is an example of information required to convert known data into a PUSCH signal, and the shared information shared between the base station 30 and the user terminal UE may include information other than the above.

[0037] The "known data" may be stored in advance in the storage unit 125.

[0038] The data acquisition unit 122 outputs shared information shared between the base station 30 and the teacher data generation server 120, i.e., Cell ID, C-RNTI, MCS, Codebook, MIMO specification, and RB specification, to the teacher data generation unit 123.

[0039] The teacher data generating unit 123 generates teacher data using the known data stored in the storage unit 125 and the shared information acquired from the data acquiring unit 122 .

[0040] Here, we will explain "training data." Generally, "training data" is correct data used in machine learning. Machine learning is a method of learning the features contained in input data and generating a "model" that infers results corresponding to newly input data.

[0041] The "teacher data" according to this embodiment is a PUSCH signal transmitted by the user terminal UE to the base station 30. It is difficult to directly acquire the PUSCH signal transmitted by the user terminal UE to the base station 30. Therefore, in this embodiment, a configuration is adopted in which the PUSCH signal transmitted by the user terminal UE to the base station 30 is reproduced. Specifically, the teacher data generation unit 123 reproduces the PUSCH signal transmitted by the user terminal UE to the base station 30 using known data stored in the storage unit 125 and shared information acquired from the data acquisition unit 122. As described above, the PUSCH signal is obtained by converting known data using shared information, so the PUSCH signal can be reproduced if the known data and shared information are available. The "PUSCH signal transmitted by the user terminal UE to the base station 30" reproduced in this way becomes the "teacher data." In other words, the "teacher data" according to this embodiment can be said to be correct data for the "PUSCH signal received by the base station 30 from the user terminal UE." When the management device 10 including the teacher data generation server 120 is managed and operated by a telecommunications carrier as described above, the management device 10 can ascertain which web server the user terminal UE has accessed. Therefore, when the user terminal UE accesses a specific web server, the management device 10 can also ascertain the known data that the user terminal UE has transmitted to the relevant web server. Therefore, the teacher data generation unit 123 can acquire the relevant known data from the known data stored in the storage unit 125 and reproduce the PUSCH signal that the user terminal UE transmitted to the base station 30.

[0042] The teacher data generation server 120 transmits the generated teacher data to the learning server 140 .

[0043] It has been stated above that it is difficult to directly acquire the radio signal (PUSCH signal) transmitted by the user terminal UE to the base station 30, but in other words, it is difficult for the base station 30 to ascertain "in what format the radio signal received from the user terminal UE was originally transmitted from the user terminal UE." This point will be explained with reference to Fig. 3.

[0044] 3 is a diagram illustrating an example of channel estimation. Reference numeral 60 denotes a radio resource that receives a pilot signal, as defined by the 3GPP protocol, that is known to both the base station 30 and the user terminal UE. Reference numeral 61 adjacent to reference numeral 60 denotes a radio resource that receives traffic such as internet data, i.e., a signal whose type of data the base station 30 cannot determine.

[0045] When the base station 30 demodulates a radio signal received from a user terminal UE, it performs two-dimensional interpolation of the channel information estimated at each pilot position, and demodulates the radio signal using the channel estimation value at the data position obtained thereby. However, as shown on the left side of Figure 3, the pilot signal is intermittent on the frequency and time axes. Therefore, this method does not necessarily provide sufficient estimation accuracy.

[0046] Therefore, the inventors focused on demodulating wireless signals using artificial intelligence (AI) that utilizes machine learning, which has been actively developed in recent years. Machine learning with training data makes it possible to complement channel estimation using a process similar to the process used to refine a coarse-pixelated image, as shown on the right side of Figure 3, and thus demodulate wireless signals with high accuracy. Training data is required to achieve this.

[0047] In this embodiment, when the user terminal UE accesses the web server 20 or the application server 21, the user terminal UE is configured to transmit known data to the web server 20 or the application server 21. With this configuration, teacher data can be generated with a simple arrangement.

[0048] (Configuration of Screening Server) The screening server 130 includes a control unit 131 , a memory 134 , and a storage unit 135 .

[0049] The functions of the memory 134 and the storage unit 135 are similar to those of the memory 124 and the storage unit 125, and therefore a description thereof will be omitted.

[0050] The control unit 131 includes a data acquisition unit 132 and a data deletion unit 133 .

[0051] When a predetermined condition is satisfied, the base station 30 transmits the radio signal (PUSCH signal) received from the user terminal UE to the screening server 130. In other words, when a predetermined condition is satisfied, the data acquisition unit 132 acquires the radio signal from the base station 30.

[0052] The known data may include a MAC level identifier that can be determined by the base station 30. The MAC level identifier may be a 5QI (5G QoS Identifier). The above-mentioned "predetermined condition" may be a condition such as "5QI = xx." Therefore, the base station 30 may transmit to the screening server 130 a radio signal received from the user terminal UE when the 5QI flowing in the network is xx.

[0053] The data acquisition unit 132 outputs the radio signal acquired from the base station 30 to the data deletion unit 133 .

[0054] The data deletion unit 133 deletes data related to communication secrecy from the wireless signal and transmits the deleted data to the learning server 140. This enables machine learning in a state where data related to communication secrecy has been removed.

[0055] (Configuration of Learning Server) The learning server 140 includes a control unit 141 , a memory 144 , and a storage unit 145 .

[0056] The functions of the memory 144 and the storage unit 145 are similar to those of the memory 124 and the storage unit 125, and therefore a description thereof will be omitted.

[0057] The control unit 141 includes a data acquisition unit 142 and a learning unit 143 .

[0058] The data acquisition unit 142 acquires the teacher data from the teacher data generation server 120. The data acquisition unit 142 acquires the deleted data from the screening server 130.

[0059] The "teacher data (radio signals transmitted by the user terminal UE to the base station 30)" is correct answer data for the "deleted data (radio signals received by the base station 30 from the user terminal UE)." The data acquisition unit 142 associates the acquired "deleted data" with the "teacher data" and stores them in the storage unit 145. By repeating the data acquisition process and the data storage process, the data acquisition unit 142 generates multiple training data sets, each of which is a set of "deleted data" and "teacher data," and stores these in the storage unit 145.

[0060] The learning unit 143 acquires a training data set by referring to the storage unit 145. The learning unit 143 generates a trained model for inferring "the radio signal (PUSCH signal) transmitted by the user terminal UE to the base station 30" from "the data after deletion" by machine learning using the training data set. The trained model is a computation model in which "the data after deletion" is an explanatory variable and "the radio signal transmitted by the user terminal UE to the base station 30" is an objective variable. There are no particular limitations on the machine learning algorithm. For example, the learning unit 143 may generate the trained model using a neural network, or may generate the trained model using regression analysis, random forest, or the like. The trained model generated by such machine learning is stored in the storage unit 135.

[0061] The learning server 140 may provide the generated trained model to the base station 30. By using the trained model, the base station 30 can accurately demodulate the radio signal received from the user terminal UE (see FIG. 3 ).

[0062] As described above, according to this embodiment, when the user terminal UE accesses the web server 20 or the application server 21, the user terminal UE transmits known data to the web server 20 or the application server 21, and this simple arrangement makes it possible to generate teacher data indicating the radio signals transmitted by the user terminal UE to the base station 30. This makes learning with teacher data possible and eliminates the need for the reinforcement learning described above, which can resolve the instability of the radio quality of the base station 30 that occurs during reinforcement learning. Furthermore, with learning with teacher data, the time required to find the optimal parameter set is shorter than with reinforcement learning. Furthermore, with learning with teacher data, improved accuracy of the parameter set is also expected compared to reinforcement learning.

[0063] (Processing Flow) The processing flow executed by the wireless communication system 1 will be described with reference to Fig. 4. Fig. 4 is a sequence chart showing an example of processing executed by the wireless communication system 1.

[0064] In step S1, the base station 30 transmits the Cell ID and C-RNTI to the teacher data generation server 120. This allows the base station 30 to share the Cell ID and C-RNTI with the teacher data generation server 120.

[0065] In step S2, the base station 30 assigns a C-RNTI to the user terminal UE, thereby enabling the base station 30 to share the C-RNTI with the user terminal UE.

[0066] In step S3, the base station 30 transmits the MCS, Codebook, MIMO specification, and RB specification to the user terminal UE. This allows the base station 30 to share the MCS, Codebook, MIMO specification, and RB specification with the user terminal UE.

[0067] In step S4, when the user terminal UE accesses the web server 20, it transmits known data to the web server 20. At this time, if the known data includes 5QI, information that "5QI = xx" is transmitted to the network. Note that, although FIG. 4 shows an example in which the user terminal UE accesses the web server 20, this is not limiting. The user terminal UE may also access the application server 21. In this case, the user terminal UE transmits known data to the application server 21.

[0068] Although not shown in FIG. 4 , the user terminal UE converts known data into a radio signal (PUSCH signal) using the shared information, and transmits the radio signal to the base station 30.

[0069] In step S5, the base station 30 transmits to the teacher data generation server 120 the MCS, Codebook, MIMO designation, and RB designation that were transmitted to the user terminal UE in the processing of step S3. This allows the base station 30 to share the MCS, Codebook, MIMO designation, and RB designation with the teacher data generation server 120. As described above, the shared information shared between the base station 30 and the user terminal UE is not limited to the Cell ID, C-RNTI, MCS, Codebook, MIMO designation, and RB designation, and may include other information. For example, the shared information shared between the base station 30 and the user terminal UE may include information related to overall physical layer control that the base station 30 instructs the user terminal UE to perform in order to cause the user terminal UE to transmit a radio signal, such as RB designation. In this case, the base station 30 may transmit to the teacher data generation server 120, in addition to the above information, information related to overall physical layer control.

[0070] In step S6, since the 5QI flowing through the network is xx, the base station 30 transmits the radio signal received from the user terminal UE to the screening server 130. Here, "the 5QI flowing through the network is xx" means "a predetermined condition is met."

[0071] In step S7, the teacher data generation server 120 converts the known data stored in the storage unit 125 using the shared information received from the base station 30 in the processes of steps S1 and S5, and reproduces the "radio signal (PUSCH signal) transmitted by the user terminal UE to the base station 30." The reproduced radio signal becomes the teacher data. In other words, the teacher data generation server 120 generates the teacher data using the known data and the shared information. Note that the shared information used by the teacher data generation server 120 to generate the teacher data may further include information related to the overall control of the physical layer described above. The shared information may also include other information necessary for generating the teacher data.

[0072] In step S8, the teacher data generation server 120 transmits the teacher data generated in the processing of step S7 to the learning server 140.

[0073] In step S9, the screening server 130 deletes data relating to the confidentiality of communications from the radio signal received in the process of step S6.

[0074] In step S10, the screening server 130 transmits the deleted data to the learning server 140.

[0075] In step S11, the learning server 140 associates the teacher data received from the teacher data generation server 120 with the deleted data received from the screening server 130 and stores them in the memory unit 145. The learning server 140 repeats the data acquisition process and the data storage process to generate multiple training datasets, each consisting of "deleted data" and "teacher data," and stores them in the memory unit 145. The learning server 140 generates a trained model for inferring "the radio signal transmitted from the user terminal UE to the base station 30" from the "deleted data" through machine learning using the training datasets.

[0076] 4, the learning server 140 may provide the generated trained model to the base station 30. By using the trained model, the base station 30 can accurately demodulate the radio signal received from the user terminal UE.

[0077] The processes of steps S5 to S11 do not need to be performed in real time, and may be performed, for example, at night when the communication load is lighter. Also, the process flow in the sequence chart shown in Figure 4 is an example, and steps may be deleted or new steps may be added within the scope of the gist of the invention.

[0078] As described above, the wireless communication system 1 (data generation system) according to this embodiment includes a user terminal UE, a base station 30, and a management device 10. When the user terminal UE accesses a predetermined network node, it transmits uniquely determined known data to the network node. The base station 30 transmits shared information shared with the user terminal UE to the management device 10. The management device 10 uses the known data and shared information to generate training data that is used when generating a model in which the radio signal received by the base station 30 from the user terminal UE is used as an explanatory variable and the radio signal transmitted by the user terminal UE to the base station 30 is used as a target variable.

[0079] The network nodes include at least a web server 20 and an application server 21. In addition to the web server 20 and the application server 21, the network nodes may also include a database server, a mail server, a file server, and the like.

[0080] In this embodiment, the management device 10 has been described as an HCI including multiple servers (teacher data generation server 120, screening server 130, and learning server 140). However, these servers may be configured as a single device or as separate devices. For example, the teacher data generation server 120 may be configured as a first management device. The screening server 130 may be configured as a second management device. The learning server 140 may be configured as a third management device. The first management device, second management device, and third management device may all be configured as a single device. The known data may include an identifier that can be identified by the base station 30. When the identifier satisfies a predetermined condition, the base station 30 may transmit a wireless signal to the second management device (screening server 130). The second management device may delete data related to communication secrecy from the wireless signal and transmit the deleted data to the third management device (learning server 140). The first management device (teacher data generation server 120) may transmit the teacher data to the third management device. The third management device may generate a model by machine learning using the deleted data and training data.

[0081] Note that "when the identifier satisfies a predetermined condition" may also mean "when the 5QI flowing through the network is xx."

[0082] Furthermore, the data generation method according to this embodiment may be a data generation method used in a wireless communication system 1 (data generation system) that includes a user terminal UE, a base station 30, and a management device 10. The data generation method includes a first transmission step (S4) in which, when the user terminal UE accesses a predetermined network node, the user terminal UE transmits uniquely determined known data to the network node, a second transmission step (S5) in which the base station 30 transmits shared information that is shared with the user terminal UE to the management device 10, and a generation step (S7) in which the management device 10 generates, using the known data and the shared information, teacher data that is used when generating a model in which the radio signal received by the base station 30 from the user terminal UE is an explanatory variable and the radio signal transmitted by the user terminal UE to the base station 30 is a target variable.

[0083] In the above description, it has been explained that the known data may be stored in advance in the storage unit 125 of the teacher data generation server 120, but this is not limited to this. For example, when the user terminal UE accesses the web server 20 or the application server 21, the corresponding web server 20 or application server 21 may transmit the known data received from the user terminal UE to the teacher data generation server 120.

[0084] (Examples of hardware configuration and software implementation) The control blocks of the management device 10, web server 20, application server 21, base station 30, and user terminal UE (particularly the units included in control unit 121, control unit 131, and control unit 141) may be implemented by logic circuits (hardware) formed on an integrated circuit (IC chip) or the like, or may be implemented by software using a CPU (Central Processing Unit). In the latter case, the management device 10, web server 20, application server 21, base station 30, and user terminal UE may be configured using a computer. For ease of explanation, the management device 10, web server 20, application server 21, base station 30, and user terminal UE will be collectively referred to as the "management device 10, etc."

[0085] FIG. 5 is a block diagram illustrating the configuration of a computer 910 that can be used as the management device 10, etc. The computer 910 includes an arithmetic unit 912, a main memory device 913, an auxiliary memory device 914, and an input / output interface 915, all connected to one another via a bus 911. The arithmetic unit 912, the main memory device 913, and the auxiliary memory device 914 may each be, for example, a CPU, a RAM (random access memory), a solid-state drive, or a hard disk drive. The input / output interface 915 is connected to an input device 920 through which a user inputs various information to the computer 910, and an output device 930 through which the computer 910 outputs various information to the user. The input device 920 and the output device 930 may be built into the computer 910 or may be connected (externally) to the computer 910. For example, the input device 920 may be a button, keyboard, mouse, touch sensor, etc., and the output device 930 may be a lamp, display, printer, speaker, etc. It is also possible to apply a device having the functions of both the input device 920 and the output device 930, such as a touch panel in which a touch sensor and a display are integrated. The communication interface 916 is an interface that allows the computer 910 to communicate with external devices.

[0086] The auxiliary storage device 914 stores an information processing program for causing the computer 910 to operate as the management device 10, etc. The arithmetic device 912 then deploys the information processing program stored in the auxiliary storage device 914 onto the main storage device 913 and executes instructions contained in the information processing program, thereby causing the computer 910 to function as each unit of the management device 10, etc. Note that the recording medium used by the auxiliary storage device 914 to record information such as the information processing program may be any computer-readable "non-transitory tangible medium," and may be, for example, a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, etc.

[0087] Alternatively, the computer 910 may be configured to function using a program stored on a recording medium external to the computer 910 or a program supplied to the computer 910 via any transmission medium (such as a communication network or broadcast waves).The present invention may also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission.

[0088] The present invention is not limited to the above-described embodiment, and various modifications are possible within the scope of the claims.

[0089] REFERENCE SIGNS LIST 1 wireless communication system 10 management device 20 web server 21 application server 30 base station 120 training data generation server 130 screening server 140 learning server UE user terminal

Claims

1. A data generation system including a user terminal, a base station, and a management device, wherein when the user terminal accesses a predetermined network node, the user terminal transmits uniquely determined known data to the network node, the base station transmits shared information shared with the user terminal to the management device, and the management device uses the known data and the shared information to generate training data to be used in generating a model in which the radio signal received by the base station from the user terminal is an explanatory variable and the radio signal transmitted by the user terminal to the base station is an objective variable.

2. The data generation system of claim 1, wherein the management device is a first management device, the known data includes an identifier that can be identified by the base station, the base station transmits the wireless signal to a second management device when the identifier satisfies a predetermined condition, the second management device deletes data related to communication secrecy from the wireless signal and transmits the deleted data to a third management device, the first management device transmits the training data to the third management device, and the third management device generates the model by machine learning using the deleted data and the training data.

3. The data generation system according to claim 1 or 2, wherein the known data is stored in the management device.

4. A data generation method used in a data generation system including a user terminal, a base station, and a management device, comprising: a first transmission step of transmitting uniquely predetermined known data to a predetermined network node when the user terminal accesses the network node; a second transmission step of transmitting shared information shared with the user terminal by the base station to the management device; and a generation step of generating, by the management device, training data using the known data and the shared information, to be used when generating a model in which the radio signal received by the base station from the user terminal is an explanatory variable and the radio signal transmitted by the user terminal to the base station is a target variable.

Images