Data correction device, data correction method, and program

Abstract

This data correction device includes: a reference data set determination unit configured to determine a reference data set on the basis of the number of pieces of data from a set of pieces of data for each route measured at a plurality of points by a plurality of terminals that each have moved on a different route; a correction value determination unit which, when a difference between the pieces of data of each of two routes in an overlapping section of the routes is defined as a difference between the sets of the pieces of data related to the two routes, is configured to determine, for each set of the pieces of data, a correction value of the set on the basis of an accumulated value of the differences from the reference data set; and a data correction unit configured to correct each set of the pieces of data on the basis of the correction value. Thus, the data correction device, with respect to a set of pieces of data measured by a plurality of terminals, reduces the influence of the terminals.

Description

Data correction device, data correction method, and program

[0001] The present invention relates to a data correction device, a data correction method, and a program.

[0002] In a wireless communication system targeting mobile terminals such as LTE and 5G, fluctuations in wireless quality occur due to the movement of the terminals. To enable stable use of applications by predicting the wireless quality and controlling based on the prediction results, various methods for predicting the wireless quality at the destination have been proposed.

[0003] For example, Non-Patent Document 1 discloses a technique in which data measured by various terminals at various locations is directly registered in a DB used for calculating predicted values of wireless quality.

[0004] Japanese Patent Application Laid-Open No. 2020-71042

[0005] Wakao, Kawamura, Moriyama, "Quality Prediction Technology for Optimal Use of Multiple Wireless Accesses", NTT Technical Journal, vol. 32, no. 4, pp. 11-13, April 2020

[0006] However, in the prior art, since a DB is constructed based on data acquired by various terminals at various points, the data also includes differences in the device performance and measurement conditions of each terminal. For example, the data measured by terminal a at point A and the data measured by terminal b at point B include differences in the functional performance of terminal a and terminal b. Then, the predicted values calculated based on such a DB may have a large error.

[0007] The present invention has been made in view of the above points, and an object thereof is to reduce the influence of terminals on a set of data measured by a plurality of terminals.

[0008] To solve the above problems, the data correction device includes: a reference data set determination unit configured to determine a reference data set based on the number of data points from a set of data for each route measured at multiple points by multiple terminals that have traveled along different routes; a correction value determination unit configured to determine a correction value for each data set based on the cumulative value of the difference from the reference data set, using the difference in the data for each of the two routes in the overlapping section of the two routes as the difference in the sets of data related to the two routes; and a data correction unit configured to correct each of the sets of data based on the correction value.

[0009] This method can reduce the influence of individual devices on a set of data measured by multiple devices.

[0010] This figure shows an example of a system configuration in an embodiment of the present invention. This figure illustrates an overview of the data processing performed by the data correction device 10. This figure illustrates a method for calculating the difference in overlapping intervals. This figure shows an example of the hardware configuration of the data correction device 10 in an embodiment of the present invention. This figure shows an example of the functional configuration of the terminal 20 and the data correction device 10 in an embodiment of the present invention. This is a flowchart illustrating an example of a processing procedure performed by the data correction device 10.

[0011] Embodiments of the present invention will be described below with reference to the drawings. Figure 1 is a diagram showing an example of a system configuration in an embodiment of the present invention. In Figure 1, a plurality of terminals 20-1 to X (hereinafter referred to as "terminal 20" when not distinguishing between them) are wirelessly connected to the data correction device 10 via a network.

[0012] Terminal 20 is a device that measures data related to wireless quality at each location while moving. For example, a mobile device such as a smartphone or tablet, or an in-vehicle device, are examples of terminal 20. Data related to wireless quality (hereinafter referred to as "quality data") is, for example, the radio wave strength and throughput from a base station. Each terminal 20 moves along a different route and transmits data (hereinafter referred to as "measurement data"), which includes the quality data measured at multiple points along the route and the location information of those points, to the data correction device 10. If terminal 20-1 takes the measurement data at point i as d_1[i], and terminal 20-1 measures data at N points from point 1 to point N along the travel route, the set of measurement data D_1 transmitted from terminal 20-1 is N d_1s in {d_1[1], ..., d_1[N]}. The content of one d_1 is {location information, quality data}.

[0013] The data correction device 10 is one or more computers that receive and store measurement data from each terminal 20, and process the stored measurement data to generate a set of quality data in which the influence of each terminal 20, such as the performance and measurement conditions of each terminal 20, is minimized. The set of quality data generated by the data correction device 10 can be used, for example, to predict the wireless quality at each location. For example, this set may be used as training data for a machine learning model that predicts wireless quality.

[0014] Figure 2 is a diagram illustrating the overview of the data processing performed by the data correction device 10. Figure 2 shows data sets D_1 to D_4 from terminals 20-1 to 20-4, respectively. Data set D_1 = {d_1[1], ..., d_1[N]} contains N data d_1, data set D_2 = {d_2[1], ..., d_2[M]} contains M data d_2, data set D_3 = {d_3[1], ..., d_3[K]} contains K data d_3, and data set D_4 = {d_4[1], ..., d_4[L]} contains L data d_4.

[0015] The data set D_1 of terminal 20-1 is illustrated by lines indicating movement paths obtained by plotting each of the {d_1[1], ..., d_1[N]} belonging to the data set D_1 at the location information of the measured data d_1. The same applies to data sets D_2 to D_4. If each terminal 20 measures data at equal intervals, the length of the lines will increase in proportion to the number of data points in each data set D. Therefore, if the number of data points in data set D is |D|, then in Figure 2, |D_1| > |D_3| > |D_2| > |D_4|.

[0016] In a situation where data sets D_1 to D_4 are stored in the data correction device 10, the data correction device 10 designates the data set with the largest number of data points among D_1 to D_4 as the reference data set. In the example in Figure 2, D_1 is the reference data set.

[0017] Next, the data correction device 10 calculates the difference in quality data between data sets other than the reference data set and the reference data set. The difference in quality data between a given data set and the reference data set is calculated using the respective measurement data in the overlapping interval between that data set and the reference data set.

[0018] For example, if data set D_2 overlaps with the reference data set in interval s1. Therefore, the data correction device 10 uses the difference (d_1 - d_2) between the quality data of any d_1 in the reference data set whose position information is included in interval s1 and the quality data of any d_2 in data set D_2 whose position information is included in interval s1 as the difference for data set D_2 relative to the reference data set. The data correction device 10 uses this difference as the correction value to correct all measurement data d_2 belonging to data set D_2. If the difference is -α, the data correction device 10 corrects data set D_2 by adding -α to all d_2 quality data (subtracting α from all d_2 quality data) to generate data set D'_2.

[0019] The same correction can be applied to data set D_3 as to data set D_2. In this case, the difference (d_1 - d_3) between the quality data of any d_1 whose position information is included in interval s2 in the reference data set and the quality data of any d_3 whose position information is included in interval s2 in data set D_3 is taken as the difference for data set D_3 relative to the reference data set, and this difference is taken as the correction value. If the difference is -β, the data correction device 10 corrects data set D_3 by adding -β to all d_3 quality data (subtracting β from all d_3 quality data) and generates data set D'_3.

[0020] Data set D_4 has no overlapping interval with the reference data set. On the other hand, data set D_4 overlaps with data set D_3 (D'_3) in interval s3. Therefore, the data correction device 10 takes the difference (d'_3 - d_4) between the corrected quality data of any d'_3 whose position information is included in interval s3 in data set D'_3 and the quality data of any d_4 whose position information is included in interval s3 in data set D_4 as the difference for data set D_4 with respect to data set D'_3. If this difference is -γ, the data correction device 10 corrects data set D_4 by adding -γ to all d_4 quality data (subtracting γ from all d_4 quality data). The difference for data set D_4 with respect to set D_3 was -β - γ. Therefore, the correction value for data set D_4 with respect to the reference data set is -β - γ. Thus, the correction of the data set is performed in a chain-like (recursive) manner based on the reference data set. In other words, the corrected value of each data set is the cumulative value of the chain-like differences from the reference data set.

[0021] The data correction device 10 combines (integrates) the reference data set and the corrected data sets D'_2, D'_3, and D'_4 into a single data set. Combining means finding the union.

[0022] In the above, we described a method in which only one data set D_1 is used as the reference data set (hereinafter referred to as the "first reference data set determination method"). This method is effective when the performance differences between the respective terminals 20 for each data set D_1 to D_4 are large.

[0023] On the other hand, there may be cases where the performance difference between any two or more terminals 20 is small (for example, when the models or model numbers are the same). In such cases, the data sets relating to the two or more terminals 20 with small performance differences may be combined into a single reference data set. This method for determining such a reference data set will be referred to below as the "second reference data set determination method".

[0024] In the second method for determining the reference data set, the data correction device 10 designates a data set in which the difference in quality data is less than or equal to a preset threshold as one data set, and then designates the data set with the largest number of data points as the reference data set. In Figure 2, if the absolute value of the difference -α between data set D_1 and data set D_2, |-α|, is less than or equal to the threshold, the data correction device 10 combines data set D_1 and data set D_2. The combined data set is called data set D_12. Also, if the absolute value of the difference (-β-γ) between data set D_3 and data set D_4, |-β-γ|, is less than or equal to the threshold, the data correction device 10 combines data set D_3 and data set D_4. The combined data set is called data set D_34. If the number of data points in data set D_12 is greater than that in data set D_34, the data correction device 10 uses data set D_12 as the reference data set and performs correction on data set D_34 based on the difference with respect to data set D_12.

[0025] Furthermore, two or more terminals 20 with small performance differences may be identified based on the attribute information of the terminals 20 (such as model and model number). In this case, data sets that do not have overlapping intervals may be combined. In Figure 2, if terminals 20-1 and 20-4 have the same model number, the data correction device 10 combines data set D_1 and data set D_4 to form data set D_14. If the number of data points in data set D_14 is greater than that of other data sets, the data correction device 10 uses data set D_14 as the reference data set.

[0026] Next, we will explain how to calculate the difference between two data sets D that have overlapping intervals (for example, intervals s1 to s3). Figure 3 is a diagram illustrating how to calculate the difference in an overlapping interval. Figure 3 shows three calculation methods for the difference: (1) to (3). In (1) to (3), the measurement data that was measured at a point within interval s2 in Figure 2, among the respective measurement data belonging to data sets D_1 and D_3, is indicated by black circles.

[0027] The difference calculation method (1) is a method of calculating the difference based on a comparison of one measurement data point enclosed by a dashed line within the overlapping interval. For example, in data set D_1 and data set D_3, the difference in quality data of measurement data measured at the same location within the overlapping interval s1 is calculated. However, each terminal 20 does not necessarily perform measurements at the same location. Therefore, if there is no measurement data for the same location, it is sufficient to calculate the difference in quality data between any one measurement data point in one data set D and the measurement data point in the other data set D that contains the location information closest to the location information of that measurement data point.

[0028] The difference calculation method (2) is a method of calculating the difference based on a comparison of multiple consecutive measurement data points that are included in the same section enclosed by a dashed line within the overlapping section. For example, the difference between the median or mean of multiple quality data measured in the section enclosed by a dashed line in data set D_1 and the median or mean of multiple quality data measured in the section enclosed by a dashed line in data set D_3 is taken as the difference between data set D_1 and data set D_3. According to the difference calculation method (2), the influence of differences due to antenna patterns can be reduced, as can the influence of instantaneous fluctuations such as fading.

[0029] The difference calculation method (3) is a method of calculating the difference based on a comparison of multiple measurement data related to multiple points enclosed by dashed lines within an overlapping interval (for example, multiple points that are separated by a certain distance from each other). Measurement data related to a certain point refers to measurement data in which the location information indicates the position closest to that point. For example, the difference between the median or mean of three quality data enclosed by dashed lines in data set D_1 and the median or mean of three quality data enclosed by dashed lines in data set D_3 is considered the difference between data set D_1 and data set D_3. If the two data sets D have multiple overlapping intervals that are separated from each other, measurement data for calculating the difference may be sampled for each overlapping interval. The difference calculation method (3) makes it less susceptible to the effects of fading.

[0030] Furthermore, each of the difference calculation methods (1) to (3) can be combined with either the first reference data set determination method or the second reference data set determination method.

[0031] The data correction device 10 will be explained in more detail below.

[0032] Figure 4 shows an example of the hardware configuration of the data correction device 10 in an embodiment of the present invention. The data correction device 10 in Figure 4 includes a drive device 100, an auxiliary storage device 102, a memory device 103, a processor 104, and an interface device 105, etc., which are all interconnected by bus B.

[0033] The program that enables processing in the data correction device 10 is provided on a recording medium 101 such as a CD-ROM. When the recording medium 101 containing the program is set in the drive device 100, the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100. However, the program does not necessarily have to be installed from the recording medium 101; it may also be downloaded from another computer via a network. The auxiliary storage device 102 stores the installed program as well as necessary files and data.

[0034] The memory device 103 reads and stores a program from the auxiliary storage device 102 when a program startup command is received. The processor 104 is either a CPU or a GPU (Graphics Processing Unit), or both a CPU and a GPU, and executes the functions related to the data correction device 10 according to the program stored in the memory device 103. The interface device 105 is used as an interface for connecting to a network.

[0035] Figure 5 shows an example of the functional configuration of the terminal 20 and the data correction device 10 in an embodiment of the present invention. In Figure 5, the terminal 20 has a measurement unit 21. The measurement unit 21 is realized by a process that a program installed on the terminal 20 causes the processor of the terminal 20 to execute.

[0036] The measurement unit 21 measures quality data at multiple points along the path the terminal 20 travels, using an antenna or the like provided by the terminal 20, and generates measurement data consisting of the location information of the point where the quality data was measured and the quality data itself. The measurement unit 21 transmits the generated measurement data to the data correction device 10. The transmission of measurement data may be performed sequentially in response to the generation of new measurement data, or it may be performed in batches at regular intervals.

[0037] The data correction device 10 includes a receiving unit 11, a reference data set determination unit 12, a correction value determination unit 13, and a data correction unit 14. Each of these units is realized by processing that one or more programs installed in the data correction device 10 cause the CPU 104 to execute. The data correction device 10 also utilizes databases (storage units) such as a temporary storage DB 15 and an actual value DB 16. Each of these databases can be realized, for example, using an auxiliary storage device 102 or a storage device that can be connected to the data correction device 10 via a network.

[0038] The receiving unit 11 receives a set of measurement data transmitted from each terminal 20 (data sets D1 to D4, etc., as explained in Figure 2), and stores the received data set in the temporary storage DB 15 for each terminal 20.

[0039] The reference data set determination unit 12 determines a reference data set from among the data sets stored in the temporary storage DB 15 based on the number of data in each data set. Alternatively, the reference data set determination unit 12 determines a reference data set after combining data sets in which the difference in quality data in overlapping intervals is less than or equal to a threshold.

[0040] As explained in Figure 2, the correction value determination unit 13 determines a correction value for each data set. More specifically, the correction value determination unit 13 takes the difference in quality data for each of the two paths in the overlapping section of the two paths as the difference in the data sets related to the two paths, and for each data set, determines a correction value for that data set based on the cumulative value of the difference from the reference data.

[0041] The data correction unit 14 corrects each data set based on the correction value and combines the corrected data sets into a single data set. The data correction unit 14 stores the combined data set in the actual value DB 16.

[0042] The following describes the processing procedure performed by the data correction device 10. Figure 6 is a flowchart illustrating an example of the processing procedure performed by the data correction device 10.

[0043] In step S101, the receiving unit 11 stores the data set received from each terminal 20 in the temporary storage DB15. Note that the reception of the data set from each terminal 20 may be performed for each measurement data (a set of position information and quality data), or may be performed for each data set for a certain period of time.

[0044] For example, when the data sets from each terminal 20 are sufficiently stored in the temporary storage DB15, steps S102 and subsequent steps are executed.

[0045] Subsequently, the reference data set determination unit 12 determines whether to combine two or more data sets with a difference below a threshold as a data set of one terminal 20 (S102). This determination corresponds to the determination of which of the first reference data set determination method and the second reference data set determination method to use for determining the reference data set, and is made based on the setting information preset regarding which method to adopt. Also, the difference may be calculated by any one of the difference calculation methods (1) to (3) described in FIG. 3.

[0046] When combining two or more data sets with a difference below a threshold as a data set of one terminal 20 (when it is set to adopt the second reference data set determination method) (Yes in S102), the reference data set determination unit 12 updates the temporary storage DB15 by combining two or more data sets with the difference in quality data below a preset threshold as one data set (S103). The difference in quality data refers to the absolute value of the difference in the quality data of the measurement data in the overlapping interval in the combined data sets.

[0047] When not combining two or more data sets with a difference below a threshold as a data set of one terminal 20 (No in S102), or subsequent to step S103, the reference data set determination unit 12 sets the data set with the largest number of data (number of elements) as the reference data set (S104). Note that when step S103 is executed, the reference data set is determined based on the number of data in the combined data set.

[0048] Subsequently, the correction value determination unit 13 determines whether there is a data set having an overlapping section with the reference data set among the data sets not combined with the reference data set (S105). For example, if the distance between the position indicated by the position information included in any measurement data of a certain data set and the position indicated by the position information included in any measurement data of the reference data set is within a predetermined value (or if there are a predetermined number or more of such measurement data in a certain data set), it may be determined that the data set and the reference data set have an overlapping section.

[0049] When there is a corresponding data set (hereinafter referred to as "target data set") (Yes in S105), the correction value determination unit 13 determines whether the end condition is satisfied (S106). The end condition is a condition set in advance to prevent the recursive processing (loop processing) after step S105 from being repeated more than necessary (for example, to avoid performing processing related to a data set related to a route that exceeds the necessary geographical range). For example, it may be set that the end condition is that the number of executions or the execution time of the processing reaches a preset upper limit value.

[0050] When the end condition is not satisfied (No in S106), the correction value determination unit 13 determines a correction value based on the difference between the target data set and the reference data set for the target data set (S107). The method for determining the correction value is as described above. When there are a plurality of target data sets, correction values are determined for each of them.

[0051] Subsequently, the data correction unit 14 performs correction on each quality data belonging to the target data set for each target data set based on the determined correction value (S108).

[0052] Subsequently, the data correction unit 14 combines each target data set with the reference data set (S109). As a result, the reference data set becomes a set of more data (a data set covering more routes).

[0053] Next, steps S105 and onward are repeated. In this case, the "reference data set" is the reference data set into which the target data sets were combined in step S109. If, during the repetition of steps S105 and onward, there are no data sets that have overlapping intervals with the reference data set (No in S105), or if the termination condition is met (Yes in S106), the process proceeds to step S110.

[0054] In step S110, the data correction unit 14 saves the reference data set at that point in time as a data set for one terminal 20 in the actual value DB 16.

[0055] As described above, according to this embodiment, a reference data set is determined based on the number of data points, and other data sets are corrected by the cumulative difference from the reference data set. As a result, the influence of each terminal can be reduced on the data set measured by multiple terminals 20.

[0056] Furthermore, by using a data set with a relatively large number of data points as the reference data set, the quality data of other data sets can be corrected to quality data corresponding to terminal 20, which has a high market share. As a result, it is expected that the actual value DB 16 will store a data set with high generality.

[0057] Although embodiments of the present invention have been described in detail above, the present invention is not limited to these specific embodiments, and various modifications and changes are possible within the scope of the gist of the present invention as described in the claims.

[0058] 10 Data Correction Device 11 Receiving Unit 12 Reference Data Set Determination Unit 13 Correction Value Determination Unit 14 Data Correction Unit 15 Temporary Storage DB 16 Actual Value DB 20 Terminal 21 Measurement Unit 100 Drive Device 101 Recording Medium 102 Auxiliary Storage Device 103 Memory Device 104 Processor 105 Interface Device B Bus

Claims

1. A data correction device comprising: a reference data set determination unit configured to determine a reference data set based on the number of data points from a set of data for each route measured at multiple points by multiple terminals that have traveled along different routes; a correction value determination unit configured to determine a correction value for each data set based on the cumulative value of the difference from the reference data set, using the difference in the data for each of the two routes in the overlapping section of the two routes as the difference in the sets of data related to the two routes; and a data correction unit configured to correct each of the sets of data based on the correction value.

2. The data correction device according to claim 1, characterized in that the reference data set determination unit is configured to determine the reference data set after combining sets of data in which the difference of the data in the overlapping interval is less than or equal to a threshold.

3. A data correction method characterized in that a computer performs the following steps: a reference data set determination step, in which a reference data set is determined based on the number of data points from a set of data for each route measured at multiple points by multiple terminals that have traveled along different routes; a correction value determination step, in which the difference between the data for each of the two routes in the overlapping section of the two routes is taken as the difference between the sets of data related to the two routes, and for each set of data, a correction value is determined based on the cumulative value of the difference from the reference data set; and a data correction step, in which each set of data is corrected based on the correction value.

4. A program for causing a computer to execute the following: a reference data set determination procedure, which determines a reference data set based on the number of data points from a set of data for each route measured at multiple points by multiple terminals that traveled along different routes; a correction value determination procedure, which determines a correction value for each data set based on the cumulative value of the difference from the reference data set, using the difference in the data for each of the two routes in the overlapping section of the two routes as the difference in the sets of data related to the two routes; and a data correction procedure, which corrects each of the sets of data based on the correction value.

Images