Estimation device, estimation method, and recording medium
A sensor includes: a complex transfer function calculator that calculates a complex transfer function representing propagation characteristics between each N transmission antenna element and each M reception antenna element, using radio waves transmitted as a reception signal in a space where at least one living body is present from each N transmission antenna element and received by each M reception antenna element; a spectrum calculator that: calculates likelihood spectra, each indicating a likelihood of presence of each living body, by an estimation algorithm for estimating the presence from living body information that is a living body component in the complex transfer function, using different values as the number of the at least one living body; and calculates an integrated spectrum by integrating the likelihood spectra calculated; and an estimator that estimates living body information indicating at least the number of living bodies, and outputs the living body information estimated.
1. An estimation device, comprising:
a complex transfer function calculator that calculates a complex transfer function representing propagation characteristics between each of N transmission antenna elements and each of M reception antenna elements, using radio waves that are transmitted in a space as a reception signal from each of the N transmission antenna elements and received by each of the M reception antenna elements, where N and M are natural numbers greater than or equal to two, the space being a space in which at least one living body is present;
a spectrum calculator that:
(a) calculates likelihood spectra by use of an estimation algorithm for estimating presence of each of the at least one living body from living body information, using a total number of the at least one living body as a variable, the variable being changed when calculating the likelihood spectra, the likelihood spectra each indicating a likelihood of the presence, the living body information being a living body component included in the complex transfer function; and
(b) calculates an integrated spectrum by integrating the likelihood spectra calculated by changing the variable; and
an estimator that estimates, from the integrated spectrum, living body information indicating at least the total number of the at least one living body that is present in the space, and outputs the living body information estimated.
2. The estimation device according to claim 1 ,
wherein the estimator estimates, from the integrated spectrum, the living body information further indicating a position of each of the at least one living body that is present in the space, and outputs the living body information estimated.
3. The estimation device according to claim 1 ,
wherein the variable is less than or equal to (N×M−1), less than or equal to N, or less than or equal to M.
4. The estimation device according to claim 1 ,
wherein the variable is less than or equal to a value that is defined as a maximum number of living bodies that can be present in the space.
5. The estimation device according to claim 1 , further comprising:
a storage that stores the living body information estimated by the estimator in the past,
wherein the variable is in a range that includes a total number of living bodies indicated by the living body information stored in the storage.
6. The estimation device according to claim 1 ,
wherein the estimator:
obtains at least three local maximum values among a plurality of local maximum values in the likelihood spectra, each of which takes a maximum value in a predetermined range that includes a corresponding one of the at least three local maximum values;
determines a first local maximum value that is one of the at least three local maximum values obtained and whose differential from a second local maximum value is largest, the second local maximum value being second largest after the first local maximum value; and
estimates, as the total number of the at least one living body, a number indicating a place of the first local maximum value determined in descending order of the at least three local maximum values.
7. The estimation device according to claim 6 ,
wherein the estimator determines the first local maximum value, using, as the at least three local maximum values, only at least one third local maximum value that is one of the at least three local maximum values and whose differential from a value is greater than or equal to a threshold, the value being obtained by multiplying, by a predetermined ratio, a value included in a predetermined range that includes the at least one third local maximum value.
8. The estimation device according to claim 1 ,
wherein the estimator estimates, as the total number of the at least one living body, a total number of sections in which the likelihood in each of the likelihood spectra is greater than or equal to a threshold, the sections being sections of the likelihood spectra.
9. The estimation device according to claim 1 ,
wherein the estimator estimates, as the total number of the at least one living body, a total number of living bodies that is outputted by inputting the integrated spectrum calculated by the spectrum calculator to a model that is preliminarily created by machine learning that uses, as training data, the total number of the at least one living body and an image representing each of the likelihood spectra that indicates the likelihood of the presence of each of the at least one living body in the space.
10. The estimation device according to claim 9 ,
wherein the estimator outputs the living body information, using, as the model, a convolutional neural network model.
11. The estimation device according to claim 1 ,
wherein the spectrum calculator calculates the likelihood spectra, using, as the estimation algorithm, an estimation algorithm for estimating presence of each living body indicated by a total number of living bodies that are present in the space, when the total number of living bodies has been entered to the estimation algorithm.
12. The estimation device according to claim 1 ,
wherein the spectrum calculator calculates the likelihood spectra, using, as the estimation algorithm, MUltiple SIgnal Classification (MUSIC) method.
13. An estimation method, comprising:
calculating a complex transfer function representing propagation characteristics between each of N transmission antenna elements and each of M reception antenna elements, by use of radio waves that are transmitted in a space as a reception signal from each of the N transmission antenna elements and received by each of the M reception antenna elements, where N and M are natural numbers greater than or equal to two, the space being a space in which at least one living body is present;
calculating likelihood spectra by use of an estimation algorithm for estimating presence of each of the at least one living body from living body information, using a total number of the at least one living body as a variable, the variable being changed when calculating the likelihood spectra, the likelihood spectra each indicating a likelihood of the presence, the living body information being a living body component included in the complex transfer function;
calculating an integrated spectrum by integrating the likelihood spectra calculated in the calculating of the complex transfer function by changing the variable; and
estimating, from the integrated spectrum, living body information indicating at least the total number of the at least one living body that is present in the space, and outputting the living body information estimated in the estimating.
14. A non-transitory computer-readable recording medium having a program recorded thereon for causing a computer to execute the estimation method according to claim 13 .