Estimation device

Abstract

This estimation device comprises: a first distance information generation unit that, for each combination of a first transmission point and a first reception point, generates information about a first distance between the first transmission point and the first reception point on the basis of coordinate information for the first transmission point and coordinate information for the first reception point; a first propagation information generation unit that, for each combination, generates first diffraction information and first reflection information for a first radio wave propagating from the first transmission point to the first reception point, on the basis of first terrain information for a first area where the first transmission point and the first reception point are arranged, the coordinate information for the first transmission point, and the coordinate information for the first reception point; a training unit that generates a trained model using the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination, and using the measured value of the propagation loss of the first radio wave as a first objective variable for each combination; and an estimation unit that inputs second distance information, second diffraction information, and second reflection information as second explanatory variables to the trained model, and acquires from the trained model an estimated value of the propagation loss of a second radio wave as a second objective variable.

Description

estimation device

[0001] The present invention relates to an estimation device.

[0002] In order to estimate the propagation loss of radio waves between a transmission point and a reception point on flat ground, predetermined features may be extracted from building information and topographical information. An estimation device generates a trained model using a machine learning technique based on the extracted features. The estimation device estimates the propagation loss of radio waves between a transmission point and a reception point on flat ground using the generated trained model (see Patent Document 1). Patent Document 2 also discloses an estimation device that uses a deep learning technique to extract such features.

[0003] JP 2023-004867 A JP 2019-122008 A

[0004] Buildings (land features) and mountains act as shields that block propagating radio waves. Some of the propagating radio waves are reflected by the sides of the shield, and some of the propagating radio waves are diffracted around the shield.

[0005] In flat areas, there are many buildings, so many feature quantities related to radio wave propagation loss need to be extracted from building information. In contrast, in mountainous areas, there are few buildings, so many feature quantities related to radio wave propagation loss do not need to be extracted from building information. However, there is a problem in that it is not possible to improve the efficiency of estimating radio wave propagation loss between a transmitting point and a receiving point in mountainous areas.

[0006] In view of the above circumstances, an object of the present invention is to provide an estimation device that can improve the efficiency of estimating the propagation loss of radio waves between a transmission point and a reception point in mountainous areas.

[0007] According to one aspect of the present invention, there is provided a first distance information generation unit that generates, for each combination of a first transmission point and a first reception point, first distance information between the first transmission point and the first reception point based on coordinate information of the first transmission point and coordinate information of the first reception point; a first propagation information generation unit that generates, for each combination, first diffraction information and first reflection information of a first radio wave propagating between the first transmission point and the first reception point based on first topographical information of a first region in which the first transmission point and the first reception point are located, coordinate information of the first transmission point, and coordinate information of the first reception point; and a trained model that generates a trained model using the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination and a measurement value of the propagation loss of the first radio wave as a first objective variable for each combination. a second distance information generation unit that generates second distance information between the second transmission point and the second reception point based on coordinate information of the second transmission point and coordinate information of the second reception point; a second propagation information generation unit that generates second diffraction information and second reflection information of a second radio wave propagating between the second transmission point and the second reception point based on second topographical information of a second region in which the second transmission point and the second reception point are located, coordinate information of the second transmission point, and coordinate information of the second reception point; an estimation unit that inputs the second distance information, the second diffraction information, and the second reflection information into the trained model as second explanatory variables and obtains an estimated value of the propagation loss of the second radio wave from the trained model as a second objective variable; and an output unit that outputs the estimated value of the propagation loss of the second radio wave.

[0008] According to the present invention, it is possible to improve the efficiency of estimating the propagation loss of radio waves between a transmitting point and a receiving point in mountainous regions.

[0009] FIG. 1 is a diagram showing an example of the configuration of an estimation device in an embodiment. FIG. 2 is a diagram showing an example of a cross section of a terrain between a transmission point and a reception point, and each diffraction point between the transmission point and the reception point in an embodiment. FIG. 3 is a diagram showing an example of a straight-line distance between a transmission point and a reception point, a straight-line distance between a transmission point and a reflection point, and a straight-line distance between a reception point and a reflection point in an embodiment. FIG. 4 is a diagram showing an example of a plurality of points defined in a region in an embodiment. FIG. 5 is a flowchart showing an example of the operation of a learning process of an estimation device in an embodiment. FIG. 6 is a flowchart showing an example of the operation of an estimation process of an estimation device in an embodiment.

[0010] An embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram showing an example configuration of an estimation device 1 according to the embodiment. The estimation device 1 is a device (learning device) that generates a trained model for estimating propagation loss of radio waves between a transmission point and a reception point using a machine learning technique. The estimation device 1 also estimates propagation loss of radio waves between a transmission point and a reception point using the generated trained model. The estimation device 1 can improve the efficiency of estimating propagation loss of radio waves between a transmission point and a reception point in mountainous areas.

[0011] The estimation device 1 includes a storage device 11, a memory 12, an acquisition unit 13, a signal processing unit 14, and an output unit 15. The signal processing unit 14 includes a first distance information generation unit 141, a first propagation information generation unit 142, a learning unit 143, a second distance information generation unit 144, a second propagation information generation unit 145, and an estimation unit 146.

[0012] The storage device 11 includes a non-volatile recording medium (non-transitory recording medium). The storage device 11 stores a program in advance. When the estimation device 1 is started up, the program is loaded from the storage device 11 into the memory 12.

[0013] The estimation device 1 can also be realized by a computer and a program, and the program can be recorded on a recording medium or provided via a network.

[0014] The estimation device 1 is realized as software by a processor such as a CPU (Central Processing Unit) executing a program loaded from the storage device 11 to the memory 12. The program may be recorded on a computer-readable recording medium. Examples of the computer-readable recording medium include portable media such as a flexible disk, a magneto-optical disk, a ROM (Read Only Memory), and a CD-ROM (Compact Disc Read Only Memory), and non-transitory recording media such as a hard disk or a solid state drive (SSD) built into a computer system.

[0015] The estimation device 1 may be realized using hardware (accelerator) including an electronic circuit (electronic circuit or circuitry) using, for example, an LSI (Large Scale Integrated circuit), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array).

[0016] Next, the estimation device 1 (learning device) in the learning stage of machine learning will be described. The acquisition unit 13 acquires topographical information of a first region having mountainous terrain as topographical information of a space through which radio waves (hereinafter referred to as "first radio waves") for which propagation loss is to be learned propagate. That is, the acquisition unit 13 acquires the topographical information of the first region having mountainous terrain as one of the learning data (explanatory variables) in the teacher data. The first region (horizontal plane) is pre-divided into a matrix of squares (mesh). The length of one side of the square is, for example, 10 meters. A point is defined at the center of each square. The topographical information includes three-dimensional coordinates (e.g., longitude, latitude, and altitude) of each point. Hereinafter, the transmission point (first transmitting antenna) of the first radio waves will be referred to as the "first transmitting point." The reception point (first receiving antenna) of the first radio waves will be referred to as the "first receiving point."

[0017] The acquisition unit 13 acquires multiple combinations of coordinate information (three-dimensional coordinates) of the first transmission point and coordinate information (three-dimensional coordinates) of the first reception point as one of the learning data (explanatory variables) in the teacher data. The altitude information in the coordinate information of the first transmission point represents the altitude of the first transmission antenna. The altitude information in the coordinate information of the first reception point represents the altitude of the first reception antenna. The acquisition unit 13 acquires a measurement value of the propagation loss of the radio wave for each combination of the first transmission point and the first reception point as a correct label (objective variable) in the teacher data. The measurement value of the propagation loss is a measurement value of the difference between the power of the first radio wave received by the first reception point and the power of the first radio wave transmitted by the first transmission point.

[0018] The first distance information generator 141 generates distance information between the first transmission point and the first reception point (hereinafter referred to as "first distance information") for each combination of the first transmission point and the first reception point based on the coordinate information of the first transmission point and the coordinate information of the first reception point. Hereinafter, the distance information may be, for example, information on the straight-line distance between the transmission point and the reception point, or information on the free space propagation loss between the transmission point and the reception point.

[0019] The first propagation information generation unit 142 generates, for each combination, first diffraction information and first reflection information of the first radio wave propagating between the first transmission point and the first reception point, based on first topographical information of the first region in which the first transmission point and the first reception point are located, coordinate information of the first transmission point, and coordinate information of the first reception point.

[0020] The first diffraction information includes the number of diffraction points (number of diffractions) between the first transmitting point and the first receiving point, the maximum diffraction angle between the first transmitting point and the first receiving point, and the sum of all diffraction angles between the first transmitting point and the first receiving point (cumulative diffraction angle).

[0021] Hereinafter, a point in an area that satisfies a predetermined condition and that has no obstructions (diffraction points) between it and either the transmitting point or the receiving point will be referred to as a "reflection point." The predetermined condition is, for example, a condition that a predetermined inequality is satisfied. Hereinafter, the ratio of the number of reflection points in an area that satisfies the predetermined condition to the total number of points in the area that satisfies the predetermined condition will be referred to as the "reflection point number ratio."

[0022] Hereinafter, in an area that satisfies predetermined conditions, a point where a radio wave shield (diffraction point) exists between the transmitting point and one of the receiving points is referred to as a "quasi-reflection point." A shield is an object that shields the propagating radio waves, such as a mountain or a building (land feature). Hereinafter, the ratio of the number of quasi-reflection points in an area that satisfies predetermined conditions to the total number of points in the area that satisfies the predetermined conditions is referred to as the "quasi-reflection point number ratio."

[0023] The first reflection information includes the ratio (hereinafter referred to as "first reflection point number ratio") of the number of points (hereinafter referred to as "first reflection points") in a first region that satisfies a specified condition and between which there are no obstructions (diffraction points) and the first transmitting point and the first receiving point, to the total number of points in the first region that satisfy the specified condition.

[0024] The first reflection information may also include the ratio (hereinafter referred to as "first quasi-reflection point number ratio") of the number of points (hereinafter referred to as "first quasi-reflection points") where a first radio wave blocker (diffraction point) exists between one of the first transmitting point and the first receiving point in the first region that satisfies a specified condition to the total number of points in the first region that satisfy the specified condition.

[0025] The learning unit 143 uses the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination of the first transmission point and the first reception point, and the measured value of the propagation loss of the first radio wave as the first objective variable for each combination of the first transmission point and the first reception point, and generates a learned model using a machine learning technique.

[0026] The machine learning technique is, for example, supervised learning. The supervised learning is, for example, gradient boosting. The gradient boosting is, for example, XGBoost (eXtreme Gradient Boosting). The learning unit 143 records the trained model in the storage device 11. The learning unit 143 may record the trained model in an external storage device (not shown). The output unit 15 may output a measurement value of the propagation loss of the first radio wave.

[0027] Next, the estimation device 1 in the estimation stage after the learning stage will be described. The acquisition unit 13 acquires topographical information of a second region having mountainous terrain as one of the pieces of topographical information (explanatory variables) of information on the topography of the space through which radio waves (hereinafter referred to as "second radio waves") for which propagation loss is to be estimated propagate. The second region (horizontal plane) is pre-divided into a matrix of squares (mesh), similar to the first region of the topographical information acquired in the learning stage. The length of one side of the square is, for example, 10 meters, similar to the squares pre-defined for the first region of the topographical information acquired in the learning stage. A point is defined at the center of each square. Similar to the topographical information acquired in the learning stage, the topographical information includes three-dimensional coordinates (e.g., longitude, latitude, and altitude) of each point. Hereinafter, the transmission point of the second radio waves (second transmitting antenna) will be referred to as the "second transmitting point." The reception point of the second radio waves (second receiving antenna) will be referred to as the "second receiving point." Note that it is desirable that the carrier frequency of the first radio waves and the carrier frequency of the second radio waves are the same.

[0028] The acquisition unit 13 acquires coordinate information (three-dimensional coordinates) of the second transmission point and coordinate information (three-dimensional coordinates) of the second reception point as one of the explanatory variables. The altitude information in the coordinate information of the second transmission point represents the altitude of the second transmission antenna. The altitude information in the coordinate information of the second reception point represents the altitude of the second reception antenna.

[0029] The second distance information generator 144 generates distance information between the second transmission point and the second reception point (hereinafter referred to as "second distance information") based on the coordinate information of the second transmission point and the coordinate information of the second reception point. Here, the second distance information generator 144 generates, for example, the straight-line distance or the free space propagation loss between the second transmission point and the second reception point as the second distance information.

[0030] The second propagation information generation unit 145 generates second diffraction information and second reflection information of the second radio wave propagating between the second transmission point and the second reception point based on second topographical information of the second region in which the second transmission point and the second reception point are located, coordinate information of the second transmission point, and coordinate information of the second reception point.

[0031] The second diffraction information includes the number of diffraction points (number of diffractions) between the second transmitting point and the second receiving point, the maximum diffraction angle between the second transmitting point and the second receiving point, and the sum of all diffraction angles between the second transmitting point and the second receiving point (cumulative diffraction angle).

[0032] The second reflection information includes the ratio (hereinafter referred to as the "second reflection point number ratio") of the number of points (hereinafter referred to as the "second reflection points") in the second region that satisfy a specified condition where there are no obstructions (diffraction points) between the second transmission point and the second reception point and the total number of points in the second region that satisfy the specified condition.

[0033] The second reflection information may also include the ratio (hereinafter referred to as "second quasi-reflection point number ratio") of the number of points (hereinafter referred to as "second quasi-reflection points") where a second radio wave blocker (diffraction point) exists between one of the second transmitting point and the second receiving point in the second region that satisfies a specified condition to the total number of points in the second region that satisfy the specified condition.

[0034] The estimation unit 146 acquires a trained model from the storage device 11 or an external storage device (not shown). The estimation unit 146 inputs the second distance information, the second diffraction information, and the second reflection information into the trained model as second explanatory variables, and acquires an estimated value (estimation result) of the propagation loss of the second radio wave from the trained model as a second objective variable. The estimated value of the propagation loss is an estimated value of the difference between the power of the second radio wave received by the second receiving point and the power of the second radio wave transmitted by the second transmitting point. The output unit 15 is, for example, a display device. The output unit 15 outputs the estimated value of the propagation loss of the second radio wave.

[0035] Next, the explanatory variables will be described in detail. Fig. 2 is a diagram showing an example of a cross section of the terrain between the transmission point 2 and the reception point 3, and each diffraction point 4 between the transmission point 2 and the reception point 3, in an embodiment. The obstructing object 21 is a mountain. The obstructing object 21 may also be a building (land feature). Hereinafter, the xy plane represents the ground plane. The positive direction of the z axis represents the vertically upward direction.

[0036] Transmission point 2 is the first transmission point in the learning stage and the second transmission point in the estimation stage. Reception point 3 is the first reception point in the learning stage and the second reception point in the estimation stage.

[0037] If there is no shielding object 21 on the line connecting the transmitting point 2 and the receiving point 3, the radio wave transmitted from the transmitting point 2 is received by the receiving point 3 as a direct wave.

[0038] 2 , on the other hand, a radio wave transmitted from a transmission point 2 is diffracted at diffraction points 4 by a shielding body 21 and received at a reception point 3. The number of diffraction points (number of diffractions) in the diffraction information is, for example, 2. The maximum diffraction angle in the diffraction information is, for example, "θ2." The sum of the diffraction angles in the diffraction information (cumulative diffraction angle) is, for example, "θ1 + θ2."

[0039] 3 is a diagram showing an example of the linear distance between the transmission point 2 and the reception point 3, the linear distance between the transmission point 2 and the reflection point 5, and the linear distance between the reception point 3 and the reflection point 5 in an embodiment. Hereinafter, distance "a" represents the linear distance between the transmission point 2 and the reflection point 5. Distance "b" represents the linear distance between the reception point 3 and the reflection point 5. Distance "c" represents the linear distance between the transmission point 2 and the reception point 3.

[0040] Fig. 4 is a diagram showing an example of a plurality of points defined in an area in an embodiment. In Fig. 4, coordinate information of transmission points 2 and coordinate information of reception points 3 are predetermined for an area divided into 25 squares in a matrix. The reflection information includes a ratio of the number of reflection points and a ratio of the number of quasi-reflection points for each coefficient "r". Hereinafter, the coefficient "r" is a real number equal to or greater than 1 and is predetermined. Hereinafter, the symbol "n" represents the number of reflection points. The symbol "m" represents the number of quasi-reflection points.

[0041] 4, when the coefficient "r" is, for example, 1.5, the number "N" of points in the area that satisfy the inequality "a+b<r×c" is 21. These 21 points include one transmission point 2, one reception point 3, three diffraction points 4, 11 reflection points 5, four flat points 6 (points on flat ground), and one quasi-reflection point 7.

[0042] Therefore, the ratio of the number of reflection points "n / N" is 11 / 21. The ratio of the number of quasi-reflection points "m / N" is 1 / 21. The number of points "N" may be a positive constant. When the number of points "N" is 1, the ratio of the number of reflection points is equal to the number of reflection points. Also, the ratio of the number of quasi-reflection points is equal to the number of quasi-reflection points.

[0043] Next, a description will be given of an example of the operation of the estimation device 1. Fig. 5 is a flowchart showing an example of the operation of the learning process of the estimation device 1 according to an embodiment. In the learning stage of machine learning, the acquisition unit 13 acquires topographical information of a learning target area, coordinate information of transmission points, and coordinate information of reception points (step S101).

[0044] The signal processing unit 14 performs predetermined preprocessing. For example, the first distance information generation unit 141 generates first distance information for each combination of a first transmission point and a first reception point. For example, the first propagation information generation unit 142 generates the number of diffraction points between the first transmission point and the first reception point for each combination of the first transmission point and the first reception point. For example, the first propagation information generation unit 142 generates the maximum diffraction angle between the first transmission point and the first reception point for each combination of the first transmission point and the first reception point. For example, the first propagation information generation unit 142 generates the sum of all diffraction angles between the first transmission point and the first reception point for each combination of the first transmission point and the first reception point. For example, the first propagation information generation unit 142 generates a first reflection point number ratio for each combination of the first transmission point and the first reception point. For example, the first propagation information generating unit 142 generates a first quasi-reflection point number ratio for each combination of a first transmission point and a first reception point (step S102).

[0045] The acquisition unit 13 acquires the measurement value of the propagation loss of the first radio wave. The learning unit 143 acquires the measurement value of the propagation loss of the first radio wave from the acquisition unit 13 (step S103).

[0046] The learning unit 143 uses the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination of the first transmission point and the first reception point, and the measured value of the propagation loss of the first radio wave as the first objective variable for each combination of the first transmission point and the first reception point, and generates a learned model using a machine learning technique (step S104).

[0047] The learning unit 143 determines whether or not there are any remaining measurement values ​​(hereinafter referred to as "other measurement values") that have not been used to generate the trained model (step S105). If it is determined that there are any remaining measurement values ​​(step S105: YES), the estimation device 1 returns the process to step S101. If it is determined that there are no remaining measurement values ​​(step S105: NO), the learning unit 143 records the trained model in the storage device 11 (step S106).

[0048] 6 is a flowchart showing an example of the operation of the estimation process of the estimation device 1 according to the embodiment. In the estimation stage, the acquisition unit 13 acquires topographical information of a learning target area, coordinate information of transmission points, and coordinate information of reception points (step S201).

[0049] The signal processing unit 14 performs predetermined preprocessing. For example, the second distance information generation unit 144 generates second distance information for each combination of a second transmission point and a second reception point. For example, the second propagation information generation unit 145 generates the number of diffraction points between the second transmission point and the second reception point for each combination of the second transmission point and the second reception point. For example, the second propagation information generation unit 145 generates the maximum diffraction angle between the second transmission point and the second reception point for each combination of the second transmission point and the second reception point. For example, the second propagation information generation unit 145 generates the sum of all diffraction angles between the second transmission point and the second reception point for each combination of the second transmission point and the second reception point. For example, the second propagation information generation unit 145 generates a second reflection point number ratio for each combination of the second transmission point and the second reception point. For example, the second propagation information generating unit 145 generates a second quasi-reflection point number ratio for each combination of a second transmission point and a second reception point (step S202).

[0050] The estimation unit 146 acquires a trained model from the storage device 11 or an external storage device (not shown) (step S203). The estimation unit 146 inputs the second distance information, the second diffraction information, and the second reflection information as second explanatory variables to the trained model, and acquires an estimated value of the propagation loss of the second radio wave as a second objective variable (estimation result) from the trained model (step S204). The output unit 15 outputs the estimated value of the propagation loss of the second radio wave (step S205).

[0051] As described above, in the learning stage of machine learning, the first distance information generation unit 141 generates first distance information for each combination of a first transmission point and a first reception point based on coordinate information of the first transmission point and coordinate information of the first reception point. The first propagation information generation unit 142 generates first diffraction information and first reflection information of the first radio wave propagating between the first transmission point and the first reception point for each combination based on first topographical information of the first region in which the first transmission point and the first reception point are located, coordinate information of the first transmission point, and coordinate information of the first reception point. The learning unit 143 generates a learned model using the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination of a first transmission point and a first reception point, and using the measured value of the propagation loss of the first radio wave as a first objective variable for each combination of a first transmission point and a first reception point.

[0052] In the estimation stage after the learning stage, the second distance information generation unit 144 generates second distance information based on coordinate information of the second transmission point and coordinate information of the second reception point. The second propagation information generation unit 145 generates second diffraction information and second reflection information of the second radio wave propagating between the second transmission point and the second reception point based on second topographical information of the second region where the second transmission point and the second reception point are located, and the coordinate information of the second transmission point and the second reception point. The estimation unit 146 inputs the second distance information, the second diffraction information, and the second reflection information into the trained model as second explanatory variables and obtains an estimated value of the propagation loss of the second radio wave from the trained model as a second objective variable. The output unit 15 outputs the estimated value of the propagation loss of the second radio wave.

[0053] This makes it possible to improve the efficiency of estimating the propagation loss of radio waves between a transmitting point and a receiving point in mountainous areas.

[0054] (Example of Effect) For comparison with the error (estimation efficiency) of the estimate by the estimation device 1, when propagation loss is estimated based on the straight-line distance between the transmitting point and the receiving point and the diffraction loss of the radio waves without using a trained model, the mean square error of the estimated value of the radio wave propagation loss with respect to the measured value of the radio wave propagation loss is, for example, 10.8 dB. In contrast, when propagation loss is estimated by the estimation device 1, the mean square error of the estimated value of the radio wave propagation loss with respect to the measured value of the radio wave propagation loss is, for example, 6.3 dB when the coefficient "r" is 2.

[0055] Although an embodiment of the present invention has been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention.

[0056] The present invention is applicable to a system that assists in determining the placement of base stations.

[0057] 1...estimation device, 2...transmission point, 3...reception point, 4...diffraction point, 5...reflection point, 6...flat point, 7...quasi-reflection point, 11...storage device, 12...memory, 13...acquisition unit, 14...signal processing unit, 15...output unit, 21...obstruction object, 141...first distance information generation unit, 142...first propagation information generation unit, 143...learning unit, 144...second distance information generation unit, 145...second propagation information generation unit, 146...estimation unit

Claims

1. A first distance information generation unit that generates first distance information between the first transmission point and the first reception point for each combination of the first transmission point and the first reception point based on the coordinate information of the first transmission point and the coordinate information of the first reception point; a first propagation information generation unit that generates, for each combination, first diffraction information and first reflection information of a first radio wave propagating between the first transmission point and the first reception point based on the first terrain information of a first area where the first transmission point and the first reception point are arranged, the coordinate information of the first transmission point, and the coordinate information of the first reception point; a learning unit that generates a learned model using the first distance information, the first diffraction information, and the first reflection information as first explanatory variables for each combination and using the measured value of the propagation loss of the first radio wave as a first target variable for each combination; a second distance information generation unit that generates second distance information between the second transmission point and the second reception point based on the coordinate information of the second transmission point and the coordinate information of the second reception point; a second propagation information generation unit that generates, for each combination, second diffraction information and second reflection information of a second radio wave propagating between the second transmission point and the second reception point based on the second terrain information of a second area where the second transmission point and the second reception point are arranged, the coordinate information of the second transmission point, and the coordinate information of the second reception point; an estimation unit that inputs the second distance information, the second diffraction information, and the second reflection information as second explanatory variables into the learned model and obtains an estimated value of the propagation loss of the second radio wave as a second target variable from the learned model; and an output unit that outputs the estimated value of the propagation loss of the second radio wave.

2. The estimation device according to claim 1, wherein the first diffraction information includes the number of diffraction points between the first transmission point and the first reception point, the maximum diffraction angle between the first transmission point and the first reception point, and the sum of all diffraction angles between the first transmission point and the first reception point, and the second diffraction information includes the number of diffraction points between the second transmission point and the second reception point, the maximum diffraction angle between the second transmission point and the second reception point, and the sum of all diffraction angles between the second transmission point and the second reception point.

3. The first reflection information includes the ratio of the number of first reflection points, which are points where no shielding body exists between any of the first transmission point and the first reception point in the first area satisfying the predetermined condition, to the total number of points in the first area satisfying the predetermined condition, and the ratio of the number of first quasi-reflection points, which are points where a diffraction point of the first radio wave exists between one of the first transmission point and the first reception point in the first area, to the total number of points in the first area satisfying the predetermined condition. The second reflection information includes the ratio of the number of second reflection points, which are points where no shielding body exists between any of the second transmission point and the second reception point in the second area satisfying the predetermined condition, to the total number of points in the second area satisfying the predetermined condition, and the ratio of the number of second quasi-reflection points, which are points where a diffraction point of the second radio wave exists between one of the second transmission point and the second reception point in the second area, to the total number of points in the second area satisfying the predetermined condition. The estimation device according to claim 1.

Images