Information processing device, method, and program

Abstract

To shorten the time taken until a terminal device is connected to a network.SOLUTION: An information processing device for estimating a propagation characteristic of an electromagnetic wave is configured to: acquire transmission / reception point information including positions and heights of a transmission point for transmitting an electromagnetic wave and a reception point for receiving the electromagnetic wave transmitted from the transmission; acquire structure information including the position and the height of a structure within a prescribed range including the positions of the transmission point and the reception point; generate reflection point information showing the structure that becomes a reflection point at which the electromagnetic wave transmitted from the transmission point reflects once from the acquired transmission / reception point information and structure information; and perform learning of a machine learning model for estimating a propagation characteristic of the electromagnetic wave between transmission / reception points by using the reflection point information.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an information processing device, method, and program for estimating propagation characteristics of electromagnetic waves. [Background technology]

[0002] Techniques for estimating the propagation characteristics of electromagnetic waves using machine learning models have been proposed. Cited Document 1 discloses a method for estimating spatiotemporal propagation characteristics using a machine learning model that uses a visible range image and information linked to the visible range image. According to Cited Document 1, by using line-of-sight information from the transmitting and receiving points, it becomes possible to estimate propagation characteristics that take into account buildings on the propagation path of radio waves that are reflected and diffracted only once, which has a significant impact on the propagation characteristics of electromagnetic waves. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-106445 [Non-patent literature]

[0004] [Non-Patent Document 1] T. Imai, K. Kitao and M. Inomata, “Radio Propagation Prediction Model Using Convolutional Neural Networks by Deep Learning,” 2019 13th European Conference on Antennas and Propagation (EuCAP), Krakow, Poland, 2019, pp. 1-5. Summary of the Invention [Problem to be solved by the invention]

[0005] However, when information on the line of sight from the transmitting and receiving points is treated as a visible range image, it also contains a lot of information about areas that are not related to the propagation path.As a result, there are cases where a large amount of data is required to train the machine learning model, or the amount of calculation required for training is large, which poses a problem of reduced convenience in training machine learning models that estimate the propagation characteristics of electromagnetic waves. [Means for solving the problem]

[0006] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a technology for improving the convenience of learning a machine learning model that estimates radio wave propagation characteristics between transmitting and receiving points.

[0007] An information processing device according to one aspect of the present invention includes: An information processing device that estimates propagation characteristics of electromagnetic waves, a first acquisition means for acquiring transmitting / receiving point information including the positions and heights of a transmitting point that transmits electromagnetic waves and a receiving point that receives the electromagnetic waves transmitted from the transmitting point; a second acquisition means for acquiring structure information including the positions and heights of structures within a predetermined range including the positions of the transmission point and the reception point; a generating means for generating reflection point information indicating the structure that becomes a reflection point at which the electromagnetic wave transmitted from the transmission point is reflected once, from the acquired transmission / reception point information and structure information; a learning means for learning a machine learning model that uses the reflection point information to estimate the propagation characteristics of electromagnetic waves between transmitting and receiving points; Equipped with 、 The generating means generates the reflection point information in which the relative height of a structure including the reflection point is associated with a plane passing through the transmission point and the reception point. It is characterized by: [Effects of the Invention]

[0008] According to the present invention, it is possible to provide a technology for improving the convenience of learning a machine learning model that estimates radio wave propagation characteristics between transmitting and receiving points. [Brief explanation of the drawings]

[0009] [Figure 1]Hardware configuration diagram of an information processing device according to this embodiment [Figure 2] Software configuration diagram of the information processing device according to the present embodiment [Figure 3] 1A and 1B are sequence diagrams illustrating an example of processing executed by an information processing device according to this embodiment. [Figure 4] (A) is a diagram showing elevation information around the transmitting and receiving points, and (B) is a diagram showing relative height. [Figure 5] Conceptual diagram showing how to identify reflection points [Figure 6] (A) and (B) are conceptual diagrams showing how to determine the angle of incidence and the angle of reflection at the reflection point. [Figure 7] FIG. 10 is a diagram showing an example of reflection point information. DETAILED DESCRIPTION OF THE INVENTION

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

[0011] <Hardware configuration> 1 is a diagram showing the configuration of an information processing device according to this embodiment. The information processing device 1 includes, as components, a CPU (Central Processing Unit) 101, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, and an HDD (Hard Disk Drive) 104. These components are communicatively connected via a bus 105. The information processing device 1 is, for example, a computer such as a server.

[0012] The CPU 101 is one or more processors, programmable logic circuits, or microprocessors that control the overall operation of the information processing device 1. The CPU 101 also reads out a control program stored in the ROM 102 and works together with the RAM 103 and the like to estimate the propagation characteristics of electromagnetic waves, which will be described later.

[0013] The ROM 102 is a storage area for the control program, such as a flexible disk, magnetic disk, optical disk, magneto-optical disk, CD-ROM, CD-R, DVD-ROM, magnetic tape, non-volatile memory card, EEPROM, or silicon disk.

[0014] The RAM 103 is a volatile memory used as a work area for executing various programs by the CPU 101. The HDD 104 is a large-capacity recording medium that stores image data and various programs.

[0015] <Software configuration> 2 shows functional blocks of the information processing device 1 according to this embodiment. The information processing device 1 functions as an altitude information management unit 201, a transmission / reception point information acquisition unit 202, a reflection point information generation unit 203, a learning unit 204, and a propagation characteristics estimation unit 205 by the CPU 101 loading into the RAM 103 and executing a program stored in the ROM 102 or HDD 104.

[0016] The altitude information management unit 201 manages altitude information that can identify the altitude of a given location. For example, the altitude information management unit 201 stores altitude in association with latitude and longitude. The altitude may include the elevation of the terrain, the height of artificial structures such as buildings, or the height of natural objects such as trees. In this embodiment, the altitude information is described as including structure information that indicates the elevation of the structure and the shape of the structure as viewed from above. However, the altitude information may also include three-dimensional shape data of the structure. This makes it possible to identify the angle of the side surface of the structure, as will be described later, and to identify the angle of incidence and angle of reflection with respect to the side surface of the structure. However, when estimating propagation characteristics based only on the undulations of the terrain, the shape data of the structure may not be included.

[0017] The transmission / reception point information acquisition unit 202 acquires transmission / reception point information indicating the positions and heights of the transmitter (transmission point) and receiver (reception point) of the electromagnetic waves. In one example, the transmission / reception point information includes the latitude and longitude as the positions of the transmission point and the reception point, and the altitude as the altitude. In one example, the altitude in the transmission / reception point information may include the antenna height of the transmission / reception antenna.

[0018] The reflection point information generation unit 203 identifies positions that become reflection points between the transmission and reception points based on the altitude information managed by the altitude information management unit 201 and the transmission and reception point information acquired by the transmission and reception point information acquisition unit 202. As will be described later with reference to Fig. 3, reflection points include points where electromagnetic waves sent from a transmission point reach a reception point by being reflected once, that is, points that function as a path of the electromagnetic waves between the transmission and reception points. Also, reflection points include points where electromagnetic waves sent from a transmission point are reflected in a direction different from the reception point, that is, points that block the electromagnetic waves between the transmission and reception points.

[0019] The learning unit 204 has a known machine learning model and learns the machine learning model based on the transmission and reception point information and the reflection point information. As described above, the estimation result of the propagation characteristics between predetermined transmission and reception points is estimated using the measurement results of the propagation characteristics or the output related to the propagation characteristics simulated by other simulation methods as training data. As the machine learning algorithm, a predetermined machine learning model such as a random forest or a neural network can be applied. Details of the learning unit 204 will be described later with reference to FIG. 3(A). The propagation characteristics estimation unit 205 estimates the propagation characteristics between predetermined transmission and reception points by inputting the transmission and reception point information and the reflection point information identified based on the transmission and reception point information to the trained machine learning model learned by the learning unit 204. Details of the propagation characteristics estimation unit 205 will be described later with reference to FIG. 3(B).

[0020] Previously, a technology for estimating an electromagnetic wave propagation model using a machine learning model has been proposed. The propagation characteristic estimation in Cited Document 1 uses a machine learning model that accepts, as input, a visible light image from a transmitting point or a receiving point and information associated with the visible light image. However, such machine learning models have a large number of input parameters, and therefore, unless a large amount of training data is prepared for learning, the estimation accuracy of the machine learning model does not improve, resulting in a long learning time. Therefore, the machine learning model of this embodiment identifies structures that exist around the transmitting and receiving points and that may become reflection points on the propagation path, generates reflection point information based on the shape of the structures, and uses this information as input to the machine learning model. This reduces the number of input parameters of the machine learning model and shortens the learning time. Below, a method for generating reflection point information and training the machine learning model is described.

[0021] <Example of learning process for machine learning model> An example of processing executed by the information processing device 1 will be described with reference to Fig. 3(A). In this embodiment, an example will be described in which a structure that can be a reflection point as seen from a transmission point is identified and a machine learning model is trained. The processing in Fig. 3(A) is executed when an instruction to execute the processing is received from a user.

[0022] The information processing device 1 acquires transmission and reception point information (S301). In S301, the information processing device 1 receives the transmission and reception point information including the positions and heights of the transmission and reception points from the user. Next, the information processing device 1 acquires altitude information around the transmission and reception points from the HDD 104 (S302). The range of positions from which the altitude information is acquired in S302 can be changed based on the frequency of the electromagnetic waves for estimating the propagation characteristics, the output strength of the electromagnetic waves expected to be transmitted from the transmission point, etc. Note that if the information processing device 1 has a communication unit (not shown), the altitude information may be acquired from another information processing device that manages external altitude information. Next, the information processing device 1 calculates the relative height in the transmission and reception path based on the altitude information and the transmission and reception point information (S303).

[0023] Here, a method for calculating the relative height on the route between transmitting and receiving points based on altitude information will be described with reference to Figures 4(A) and 4(B). Figure 4(A) shows the altitude of the transmitting and receiving points and their surroundings. For example, a structure 401 is shown to have a height of 18 m. The relative height H of a given structure to the transmitting and receiving point is R can be calculated by a method similar to the calculation method of the relative height according to the Fresnel criterion described in Non-Patent Document 1. In this embodiment, the relative height H R is calculated according to the following formula 1:

[0024] H R =H A -H T -{(H R -H T )D T} / (D T +D R ) (Equation 1) where H A is the height of the given structure, and H Tis the height of the transmitting point, and H R is the height of the receiving point, and D T is the horizontal distance from the transmitting point to the specified point, D R is the distance on the horizontal plane from a specified point to the receiving point. By calculating this for each structure or for a specified position, the relative height for each point can be calculated. Therefore, if there is a structure A with a first height and a structure B with a height higher than the first height, the relative height of structure A may be higher than the relative height of structure B depending on the heights of the transmitting and receiving points. In other words, the relative height may be the height relative to a plane that passes through the line of sight of the transmitting and receiving points and a line that intersects the line of sight perpendicularly and is parallel to the horizontal plane.

[0025] Figure 4(B) shows an example of the results of calculating the relative height of each structure according to Equation 1. Note that the relative height of the ground is not shown in Figure 4(B). The structure 411 shown in Figure 4(B) has a relative height of -5 m, which means that it is lower than the plane passing through the line of sight connecting the transmitting and receiving points and a line that intersects the line of sight perpendicularly and is parallel to the horizontal plane. As a result, there is no shielding of the electromagnetic waves sent from the transmitting point, and no strong reflection close to regular reflection occurs.

[0026] Next, the information processing device 1 identifies position information of reflection points for each predetermined angle from the transmission point based on the calculation result of the relative height (S304). In this embodiment, the information processing device 1 identifies the distance from the transmission point to the position of the reflection point for each predetermined angle from the transmission point.

[0027] A method for identifying the positions of reflection points is shown in Fig. 5. In the example of Fig. 5, the direction from the transmission point to the reception point is set to 0 degree, and the positions of reflection points are identified at 45-degree intervals, but the method is not limited to the example of Fig. 5 as long as the reflection points are determined at each predetermined angle.

[0028] In this embodiment, for each angle, structures whose relative height as seen from the transmission point is 0 or more are identified as reflection points. For this reason, structures 501 to 506 shown by solid lines in Fig. 5 are identified as reflection points. Note that when there is no structure, such as in the direction of an angle of 135° as seen from the transmission point, or when the relative height of an existing structure is less than 0, such as in the direction of an angle of 270° as seen from the transmission point, it can be determined that there is no reflection point, that is, no reflection occurs. Note that the height of the reflection point is determined as the position on a plane that passes through the line of sight connecting the transmitting and receiving points and a line that intersects the line of sight perpendicularly and is parallel to the horizontal plane, that is, the relative height H in Equation 1. R = 0. Therefore, the height of the reflection point is H A -H R It can be calculated as:

[0029] Next, the angle of incidence and the angle of reflection at the reflection point determined for each angle are determined (S305). R is greater than 0, that is, when there is no structure that can be a reflection point, the angle of incidence and the angle of reflection at the reflection point may be set to 0.

[0030] FIG. 6A shows a method for determining the angle of incidence and the angle of reflection at a reflection point. The information processing device 1 according to this embodiment acquires three-dimensional information about a structure as elevation information. Therefore, as shown in FIG. 6A, it is possible to determine the side surface 601 of the structure on which the electromagnetic waves transmitted from the transmission point are incident, based on the position of the transmission point, the direction of transmission of the electromagnetic waves from the transmission point, and the position and shape of the structure. In the example of FIG. 6A, it is possible to determine that reflection occurs at the side surface 601 of the structure 506. Therefore, it is possible to determine the angle of incidence from the transmission point and the direction from the reflection point to the reception point as the angle of reflection, based on a normal 602 perpendicular to the side surface 601 of the structure 506. By calculating the angle of incidence and the angle of reflection relative to the normal direction in this way, it is possible to determine whether the angle of incidence and the angle of reflection are close, i.e., whether a strong reflection close to regular reflection occurs at the reflection point, or whether the difference between the angle of incidence and the angle of reflection is large, i.e., whether a weak reflection occurs at the reflection point. In the example of FIG. 6(A), the angle of incidence and the angle of reflection are shown on a horizontal plane, but the angle of incidence and the angle of reflection may be determined on a plane including the transmitting / receiving points and the reflecting points.

[0031] Following S305, the information processing device 1 creates a list of reflection point information in which the positions of reflection points, angles of incidence, and angles of reflection are associated with each other (S306). An example of reflection point information is shown in Fig. 7. The reflection point information indicates the direction [°] of the reflection point as seen from the transmission point, with the direction from the transmission point to the reception point being 0 degrees, the relative height [m] of the structure where the reflection point is located, the distance [m] from the transmission point, the distance [m] from the reception point, the angle of incidence [°], and the angle of reflection [°]. In the example of Fig. 7, a list of reflection points at intervals of 1° from the transmission point is created as reflection point information.

[0032] Following S306, the information processing device 1 acquires training data indicating propagation characteristics between the transmitting and receiving points (S307), and uses reflection point information as input to train the machine learning model so as to minimize the error between the output and the training data (S308). In S308, the machine learning model is trained based on the error between the estimated value of the propagation characteristics of the output electromagnetic wave and the measurement results (training data) or the output of another simulator, for a plurality of different transmitting and receiving points, along with the positions of the transmitting and receiving points and the reflection points at those positions. The adjustment method can be performed using known techniques, and therefore will not be described here. The propagation characteristics may be at least one of the radio wave intensity detected at the receiving point when electromagnetic waves are transmitted from the transmitting point, the delay spread between the transmitting and receiving points, and the angular spread. The training data can be changed depending on the type of output of the machine learning model.

[0033] In this embodiment, the description will be given assuming that reception strength is estimated as the propagation characteristic at a specific frequency. However, when the information processing device 1 estimates propagation characteristics at multiple frequencies, information indicating a frequency band may be input to the machine learning model. In other words, the input parameters of the machine learning model may include information indicating a frequency.

[0034] This allows for training of a machine learning model that estimates propagation characteristics between specified transmitting and receiving points when the positions and heights of the transmitting and receiving points are input. Also, by identifying reflection points for each angle seen from the transmitting point and inputting only the reflection point information into the machine learning model, training of the machine learning model can be performed with a short learning time and computational load without degrading the estimation accuracy of the propagation characteristics.

[0035] According to the above-described process, as shown in Fig. 6(B), there may be a case where a structure is located between the transmission point and a predetermined reflection point, or between the predetermined reflection point and the reception point. Even in such a case, by calculating the angle of incidence and the angle of reflection with respect to the normal to the side surface of the structure, it is possible to include in the reflection point information that either the transmission point or the reception point is not on the line of sight due to being blocked by the structure at the reflection point. In other words, according to the above-described process, it is possible to identify the structure that is the blocking point.

[0036] <Propagation characteristics estimation process> The estimation of propagation characteristics performed by the information processing device 1 according to this embodiment will be described with reference to Fig. 3(B). The process of Fig. 3(B) is performed when an instruction to perform the process is received from the user.

[0037] In S351, the information processing device 1 acquires transmission and reception point information by accepting input of the positions and heights of the transmission and reception points from the user. In one example, the instruction to execute a process from the user may include information specifying the positions and heights of the transmission and reception points, and in this case, the information processing device 1 acquires information on the positions and heights of the transmission and reception points from the instruction to execute a process.

[0038] The processing of S352 to S356 is the same as that of S302 to S306 described above, and therefore description thereof will be omitted. Subsequently, the information processing device 1 inputs the transmission and reception point information and the reflection point information into the trained machine learning model, and outputs an estimated value of the propagation characteristics (S357).

[0039] As described above, according to the machine learning model of this embodiment, the reflection point information is input as an input parameter to the trained machine learning model to estimate the propagation characteristics of the electromagnetic wave. This reduces the number of input parameters of the machine learning model, thereby reducing the calculation load and calculation time involved in estimating the propagation characteristics.

[0040] <Other embodiments> In this embodiment, an example has been described in which reflection point information is generated as a reflection point of an electromagnetic wave transmitted from a transmission point. However, it is also possible to identify structures whose relative height is 0 or more in the line of sight direction for each predetermined angle on a horizontal plane using the reception point as a reference, and generate reflection point information for the reception point as described with reference to FIGS. 4(A) to 6(B). In addition, in one example, a machine learning model can be applied in which both the reflection points of the electromagnetic wave transmitted from the transmission point and the reflection points as the path of the electromagnetic wave incident on the reception point are input. In other words, it is sufficient to generate reflection point information from points that are reflection points for at least one of the transmission point and the reception point.

[0041] In addition, in the present embodiment, the directivity between the transmitting and receiving points has been described as omnidirectional. However, by performing learning including the directivity of the transmitting and receiving antennas as input parameters of the machine learning model, it is possible to estimate propagation characteristics with high accuracy. In this case, the transmitting and receiving point information may include the antenna directivity, such as the antenna gain in each direction, in addition to the positions and heights of the transmitting and receiving points.

[0042] The invention is not limited to the above-described embodiment, and various modifications and variations are possible within the scope of the gist of the invention. [Explanation of symbols]

[0043] 1: Information processing device, 101: CPU, 102: ROM, 103: RAM, 104: HDD, 201: Altitude information management unit, 202: Transmission and reception point information acquisition unit, 203: Reflection point information generation unit, 204: Learning unit, 205: Propagation characteristic estimation unit

Claims

1. An information processing device that estimates propagation characteristics of electromagnetic waves, a first acquisition means for acquiring transmitting / receiving point information including the positions and heights of a transmitting point that transmits electromagnetic waves and a receiving point that receives the electromagnetic waves transmitted from the transmitting point; a second acquisition means for acquiring structure information including the positions and heights of structures within a predetermined range including the positions of the transmission point and the reception point; a generating means for generating reflection point information indicating the structure that becomes a reflection point at which the electromagnetic wave transmitted from the transmission point is reflected once, from the acquired transmission / reception point information and structure information; a learning means for learning a machine learning model that uses the reflection point information to estimate the propagation characteristics of electromagnetic waves between transmitting and receiving points; Equipped with The information processing device is characterized in that the generation means generates the reflection point information in which the relative height of a structure including the reflection point is associated with a plane passing through the transmission point and the reception point.

2. 2. The information processing device according to claim 1, wherein the generating means includes identifying, as a reflection point, a structure that is closest to the transmitting point or the receiving point among structures whose relative height has a positive value at a predetermined angle seen from the transmitting point or the receiving point.

3. 2. The information processing device according to claim 1, wherein the generating means generates the reflection point information in which a first distance from the transmission point to the reflection point and a second distance from the reflection point to the reception point are associated with each other.

4. The structure information can identify an angle of a side surface of the structure, 2. The information processing apparatus according to claim 1, wherein the generating means generates the reflection point information in which an incident angle from the transmission point to the reflection point and a reflection angle from the reflection point to the reception point are associated with each other.

5. 5. The information processing apparatus according to claim 4, wherein the incident angle and the reflection angle are angles relative to a normal direction of the side surface of the structure.

6. 2. The information processing apparatus according to claim 1, wherein the generating means identifies reflection points at predetermined angular intervals from the transmission point or the reception point on a horizontal plane, with the transmission point or the reception point as a reference.

7. 2. The information processing device according to claim 1, wherein the learning means performs learning of the machine learning model further based on a measurement result at the reception point of the electromagnetic wave transmitted from the transmission point.

8. The information processing device according to claim 1 , further comprising an estimation means for estimating propagation characteristics of electromagnetic waves between transmitting and receiving points at predetermined positions and heights using the trained machine learning model.

9. An information processing method for estimating propagation characteristics of electromagnetic waves, comprising: acquiring transmission and reception point information including the positions and heights of a transmission point that transmits electromagnetic waves and a reception point that receives the electromagnetic waves transmitted from the transmission point; acquiring structure information including the positions and heights of structures within a predetermined range including the positions of the transmitting point and the receiving point; generating reflection point information indicating the structure that serves as a reflection point at which the electromagnetic wave transmitted from the transmission point is reflected once from the acquired transmission / reception point information and structure information; Learning a machine learning model using the reflection point information; Including, An information processing method characterized in that, in generating the reflection point information, the reflection point information is generated in which the relative height of a structure including the reflection point is associated with a plane passing through the transmission point and the reception point.

10. A program for causing a computer to function as the information processing device according to any one of claims 1 to 8.

Images