JP7624178B2 - OBJECT DETECTION DEVICE, OBJECT DETECTION METHOD, AND OBJECT DETECTION SYSTEM - Google Patents

Description

This disclosure relates to an object detection device, an object detection method, and an object detection system, and in particular to an object detection device, an object detection method, and an object detection system that are suitable for detecting an object that does not carry a special device, based on propagation channel information measured by a wireless terminal station.

In the field of wireless communications, efforts are being made to improve communication speeds using a variety of technologies, such as beamforming, which controls the directionality of radio waves, in order to stably accommodate the ever-increasing amount of wireless traffic.

On the other hand, services are being considered that use signal strength information of wireless signals (e.g., RSS (Received Signal Strength)) to detect wireless terminal stations within a communication area and provide location information. In such services, for example, a wireless terminal station measures the RSS of beacon signals transmitted from multiple wireless base stations, and the location of the wireless terminal station is calculated from the multiple measured RSSs. This method is widely used as a positioning system in indoor environments where it is difficult to use GPS (Global Positioning System).

As disclosed in the following non-patent document 1, a device-free detection method is also known. With the device-free detection method, it is possible to detect objects that do not hold a special device such as an antenna, such as a person.

The device-free detection method is used in an environment where fixed wireless base stations and wireless terminal stations exist within a detection area. The wireless terminal stations measure propagation channel information (CSI) based on the radio signals transmitted from the wireless base station. The CSI reflects the presence or absence of an object in the detection area between the wireless base station and the wireless terminal stations, or the state of the object. Then, object detection is performed based on the fluctuation characteristics that appear in the CSI.

“WiFi Sensing with Channel State Information: A Survey”, YONGSEN MA, GANG ZHOU, and SHUANGQUAN WANG, Computer Science Department, College of William & Mary, USA, ACM Computing Surveys, Vol. 52, No. 3, Article 46. Publication date: June 2019.

However, device-free detection methods are implemented based on small fluctuations that occur in the propagation environment. For this reason, it is more difficult to ensure detection accuracy than when detecting an object that holds a device such as an antenna. One method to address this issue is to perform a large amount of calculations based on the large amount of CSI acquired by the wireless terminal station. However, this solution creates a new issue in that it creates an extremely high calculation load. Furthermore, the high computer load can have a negative impact on the demand for real-time object detection.

The present disclosure has been made in light of the above-mentioned problems, and has a first objective to provide an object detection device that can detect device-free objects within a detection area with high accuracy without excessive computational load.
A second object of the present disclosure is to provide an object detection method for detecting device-free objects in a detection area with high accuracy without excessive calculation load.
Furthermore, a third object of the present disclosure is to provide an object detection system that can detect device-free objects within a detection area with high accuracy without excessive computational load.

In order to achieve the above object, a first aspect of the present disclosure is an object detection device that detects an object in a detection area between a wireless base station and a wireless terminal station, comprising:
A process of receiving a feedback signal that carries information related to propagation channel information estimated for each subcarrier of a communication wave and is transmitted by the wireless terminal station;
extracting information related to the propagation channel information from the feedback signal;
a reduction process for reducing information related to the propagation channel information by applying a reduction condition selected as meeting a requirement for a detection rate of object detection;
A process of performing the object detection based on a reduced impulse response corresponding to the information after the reduction process has been performed;
Preferably, the system is configured to execute:

A second aspect of the present disclosure is an object detection method for detecting an object in a detection area between a wireless base station and a wireless terminal station, comprising:
a step of transmitting a radio signal to a radio terminal station by the radio base station on each of a plurality of subcarriers included in a frequency band used for communication;
The wireless terminal station estimates propagation channel information for each of the subcarriers;
The wireless terminal station transmits a feedback signal carrying information related to the propagation channel information;
extracting information related to said propagation channel profile from said feedback signal;
a reduction process for reducing the extracted information related to the propagation channel information by applying a reduction condition selected as satisfying a detection rate requirement for object detection;
performing the object detection based on a reduced impulse response corresponding to the information after the reduction process;
It is preferable that the present invention includes the following:

A third aspect of the present disclosure is an object detection system,
a wireless base station that transmits wireless signals on a plurality of subcarriers included in a frequency band used for communication;
a wireless terminal station that receives the wireless signal, estimates propagation channel information for each of the subcarriers, and transmits a feedback signal carrying information related to the propagation channel information;
an object detection device that detects an object in a detection area between the wireless base station and the wireless terminal station,
The object detection device includes:
receiving the feedback signal;
extracting information related to the propagation channel information from the feedback signal;
a reduction process for reducing information related to the propagation channel information by applying a reduction condition selected as meeting a requirement for a detection rate of object detection;
It is preferable that the apparatus is configured to execute a process of performing the object detection based on a reduced impulse response corresponding to the information after the reduction process has been performed.

According to the present disclosure, device-free objects within a detection area can be detected with high accuracy without excessive computational load.

FIG. 1 is a diagram for explaining a configuration of an object detection system according to a first embodiment of the present disclosure. FIG. 2 is a diagram for explaining an example of a detection area targeted by the first embodiment of the present disclosure. 1 is a diagram for explaining an overview of propagation channel information acquired by a wireless terminal station in the first embodiment of the present disclosure. FIG. FIG. 1 is a diagram illustrating an example of frequency characteristics appearing in CSI. FIG. 1 is a diagram illustrating an example of an impulse response obtained by performing a Fourier transform on the frequency characteristics appearing in the CSI. FIG. 6 is a diagram showing a state in which a window function is applied to the impulse response shown in FIG. 5 . FIG. 11 is a diagram illustrating an example of a relationship between an average object detection rate and a window function. 1 is a flowchart for explaining the flow of an object detection process performed in the first embodiment of the present disclosure. 13 is a flowchart for explaining the flow of an object detection process performed in a second embodiment of the present disclosure. FIG. 13 is a diagram for explaining features of the Fourier transform performed in the third embodiment of the present disclosure. FIG. 11 is a diagram for explaining a first modified example of a frequency band covered by the present disclosure. FIG. 13 is a diagram for explaining a second modified example of a frequency band targeted by the present disclosure.

Embodiment 1.
[Configuration of First Embodiment]
1 is a block diagram showing a configuration of an object detection system according to a first embodiment of the present disclosure. The system according to the present embodiment includes a wireless base station 10. The wireless base station 10 includes M antennas 12-1 to 12-M. Hereinafter, when there is no need to distinguish between the antennas 12-1 to 12-M, the subscripts of the reference symbols will be omitted and the antennas will be referred to as "antennas 12."

The wireless base station 10 transmits wireless signals belonging to a pre-assigned frequency band from each of the antennas 12. More specifically, this frequency band includes multiple subcarriers, and the antennas 12 transmit wireless signals for each of these subcarriers.

The system of this embodiment includes a wireless terminal station 20. The wireless terminal station 20 includes N antennas 22-1 to 22-N. Each of the antennas 22-1 to 22-N receives wireless signals transmitted on multiple subcarriers from each of the antennas 12-1 to 12-M. Hereinafter, when there is no need to distinguish between the antennas 22-1 to 22-N, the subscripts of the reference symbols will be omitted and the antennas will be referred to as "antennas 22."

Figure 2 is a diagram for explaining the environment in which the antennas 12 of the wireless base station 10 and the antennas 22 of the wireless terminal station 20 are placed. This shows an example in which the wireless base station 10 has four antennas 12 and the wireless terminal station 20 has one antenna 22.

In this embodiment, both antenna 12 and antenna 22 are fixedly disposed. A detection area is set between antenna 12 and antenna 22. FIG. 2 shows an example in which the detection area is divided into eight areas (1) to (8). This embodiment discloses a system that detects whether an object that does not hold a device that emits a wireless signal, such as a person, is present within the detection area, and further, in which area and at what density the object is present within the detection area.

Referring again to FIG. 1, radio units 24-1 to 24-N are connected to antennas 22-1 to 22-N of wireless terminal station 20, respectively. Hereinafter, when there is no need to distinguish between radio units 24-1 to 24-N, the subscripts of the reference numbers will be omitted and they will be referred to as "radio unit 24."

The radio unit 24 converts the radio signal received by the antenna 22 into a format that allows estimation of propagation channel information (CSI). The signal converted by the radio unit 24 is provided to the CSI estimation unit 26. Here, CSI refers to amplitude information, phase information, and information on relative values between antennas for each subcarrier of OFDM (Orthogonal Frequency Division Multiplexing) on the propagation path.

M types of radio signals emitted from antennas 12-1 to 12-M of radio base station 10 arrive at antenna 22-1 of radio terminal station 20. The same applies to the other antennas 22-2 to 22-N. As radio units 24-1 to 24-N each receive M types of radio signals, M×N types of signals are provided to CSI estimation unit 26. Each of these M×N types of signals contains a frequency element for each subcarrier.

Figure 3 is a diagram for explaining an overview of the CSI estimated by the CSI estimation unit 26. Figure 3 shows how an M x N CSI matrix is defined for each subcarrier. As shown in Figure 3, the CSI estimation unit 26 estimates the CSI in the form of an M x N matrix arranged in the frequency direction.

As shown in FIG. 1, the CSI estimated by the CSI estimation unit 26 is provided to the conversion unit 28. The conversion unit 28 converts the M×N CSI matrix into a right singular matrix by singular value decomposition according to standardized techniques for wireless communication. The converted right singular matrix is provided to the compression unit 30.

The compression unit 30 compresses the information of the right singular matrix using a known method defined in standardization technology. The compressed information is converted into a radio signal by the radio unit 32 and transmitted as a feedback signal from the antennas 22-1 to 22-N.

The system of this embodiment includes an object detection device 40 in addition to a wireless base station 10 and a wireless terminal station 20. The CSI feedback signal emitted from the wireless terminal station 20 is received by the wireless base station 10 and also by an antenna 42 of the object detection device 40. The signal received by the antenna 42 is provided to an extraction unit 44.

The extraction unit 44 extracts compressed CSI information from the signal provided by the antenna 42. The extracted compressed information is provided from the extraction unit 44 to the decompression unit 46. The decompression unit 46 reconstructs the right singular matrix from the extracted information using a decompression method defined by standardization. This reconstructs the right singular matrix for each subcarrier, that is, the right singular matrix arranged in the frequency direction.

Figure 4 shows the result of extracting specific elements of the right singular matrix and arranging them in the frequency direction. Each element of the right singular matrix generated for each subcarrier can be expressed as a frequency characteristic as shown in Figure 4.

The information of the right singular matrices restored by the decompression unit 46 is provided to the information reduction unit 48, as shown in FIG. 1. The information reduction unit 48 first performs an inverse Fourier transform on the right singular matrices arranged in the frequency direction. This converts the information in the frequency domain into information in the time domain. As a result, information on the impulse response represented in a time series is generated.

Figure 5 shows an example of an impulse response generated by the above process. The impulse response generated in this way contains information that is essentially meaningless from the perspective of object detection, such as periodic noise called aliases.

The information reduction unit 48 shown in FIG. 1 generates an impulse response and then applies an appropriate window function to the impulse response.

Figure 6 shows an example of an impulse response after applying a window function to reduce meaningless information. In this way, this embodiment aims to reduce the amount of information to be applied to the object detection process.

As shown in FIG. 1, the information processed by the information reduction unit 48 is provided to the determination unit 50. The determination unit 50 performs object detection processing using a known method based on the impulse response provided in this manner. More specifically, the determination unit 50 applies machine learning to the above impulse response to perform object detection using a known method.

Figure 7 shows the results of object detection simulations performed with different window functions. In the figure, the leftmost bar is the result of object detection performed using all information (64 points). The "T/4" bar is the result when a window function that limits the amount of information to 1/4 (16 points) is used. Similarly, the "T/8" and "T/16" bars are the results when window functions that limit the amount of information to 1/8 (8 points) and 1/16 (4 points), respectively. In this example, the average detection rate improves as the "window" is narrowed.

In this embodiment, a window function that achieves a high average detection rate is selected in advance based on the results shown in FIG. 7. Therefore, the information reduction unit 48 deletes only information that is meaningless or unimportant in terms of object detection, and only a small amount of information that is significant is provided to the determination unit 50. As a result, the system of this embodiment achieves highly accurate object detection without an excessive computational load.

[Processing flow in object detection device]
Fig. 8 is a flowchart for explaining the flow of processing executed by the object detection device 40 in this embodiment. In this embodiment, the object detection device 40 may include a calculation processing unit and a memory that stores a program. The processing shown in Fig. 8 may be realized by the calculation processing unit executing the program stored in the memory.

In the routine shown in Figure 8, first, a CSI feedback signal emitted from the wireless terminal station 20 is received (step 100).

Next, the compressed CSI information is extracted from the received signal (step 102). This realizes the function of the extraction unit 44 shown in FIG. 1.

Next, the right singular matrix is restored from the compressed information by a default decompression process (step 104). This realizes the function of the decompression unit 46 shown in FIG. 1.

The object detection device 40 then reduces information from the restored right singular matrix (step 106). Specifically, first, an inverse Fourier transform is performed on the right singular matrix arranged in the frequency direction. Next, an appropriate window function that has been set in advance is applied to the impulse response generated as a result to reduce information. By executing the processing of this step 106, the function of the information reduction unit 48 shown in FIG. 1 is realized.

Finally, object detection processing is performed based on the impulse response after the information has been reduced (step 108). The processing of this step 108 realizes the function of the determination unit 50 shown in FIG. 1.

As described above, the system of this embodiment can perform object detection calculations by deleting information that is meaningless or unimportant from the perspective of object detection. Therefore, this system can achieve highly accurate object detection without incurring an excessive computational load.

Embodiment 2.
Next, a second embodiment of the present disclosure will be described with reference to Fig. 9 together with Fig. 1. The object detection system of the present embodiment can be realized by the configuration shown in Fig. 1, similar to the first embodiment. More specifically, the object detection system of the present embodiment can be realized by causing the object detection device 40 to execute a routine shown in Fig. 9, which will be described later, instead of the routine shown in Fig. 8 above.

In the above-described first embodiment, the object detection device 40 executes the routine shown in FIG. 8 on the premise that the window function used to reduce information is set in advance. The routine shown in FIG. 9 executed in this embodiment includes a process for selecting an appropriate window function. The object detection system of this embodiment is characterized in that the routine shown in FIG. 9 is used to provide the object detection device 40 with the function of selecting an appropriate window function.

Figure 9 shows a flowchart of a routine executed by the object detection device 40 in this embodiment. In Figure 9, steps that are the same as those shown in Figure 8 are given the same reference numerals, and their explanations are omitted or simplified.

In the routine shown in FIG. 9, in step 106, as in the first embodiment, the information of the impulse response based on the right singular matrix is reduced using an appropriate window function. Then, in step 108, object detection is performed based on the reduced information.

In the routine shown in FIG. 9, following the above process, it is determined whether the search for window functions has ended (step 110). Candidates for window functions for information reduction are predefined in the object detection device 40. Here, it is determined whether the processes of steps 106 and 108 have been performed for all of the candidates. If it is determined that these processes have not been performed for all of the candidates, the determination of the end of the search is denied.

If the search is not completed in step 110, the window function is changed to an unused one (step 112). After that, the processes in steps 106 and 108 are performed again using the changed window function.

On the other hand, if the search is terminated in step 110, a threshold value for the detection rate of object detection is set (step 114). More specifically, the allowable amount of deterioration in the detection rate relative to the maximum detection rate is set as the threshold value.

Next, based on the object detection results obtained in step 108, a window function that clears the above threshold is extracted (step 116). Specifically, an average detection rate is calculated for each window function. Next, the window function whose average detection rate clears the above threshold is extracted as pass data.

Once the extraction of the acceptable data is completed, the window function that achieves the largest reduction rate among the extracted window functions is selected (step 118). This allows the selection of a window function that can achieve the desired detection rate and achieve a large reduction rate.

Next, it is determined whether the selected window function satisfies the processing time condition (step 120). More specifically, it is determined whether the object detection processing performed based on the information reduced using the window function fits within a specified processing time. The specified processing time is determined in advance according to, for example, the real-time requirement for object detection.

In this embodiment, priority is given to the requirement of real-time performance over the accuracy of object detection. Therefore, if the time condition is not met in step 120 above, the process returns to step 114 and the detection rate threshold is reviewed.

On the other hand, if the time condition is found to be met in step 120, the window function selected at that time is finally determined to be the window function for information reduction (step 122).

As described above, the routine shown in FIG. 9 makes it possible to appropriately select a window function that achieves both the desired detection rate and the desired real-time performance. Therefore, the object detector detection system of this embodiment can realize the function of detecting objects with high accuracy without excessive computational load.

In the above-described second embodiment, one of the requirements for selecting a window function is to meet a time condition, but the present disclosure is not limited to this. In an environment in which the time condition is not important, this condition may be removed and the window function may be selected with only the detection rate as a requirement.

Embodiment 3.
Next, a third embodiment of the present disclosure will be described with reference to Fig. 10 together with Fig. 1 and Fig. 8. The object detection system of the present embodiment can be realized by the configuration shown in Fig. 1, similarly to the first and second embodiments.

In the above-mentioned first and second embodiments, the frequency domain information is converted into time domain information, and then a window function is used to reduce the information. However, the method of reducing the information is not limited to this. Specifically, the information reduction of the right singular matrix may be performed at the frequency characteristics stage, before performing an inverse Fourier transform to convert to a time domain impulse response.

Figure 10 is a diagram for explaining the timing at which the object detection device 40 performs information reduction in this embodiment. In Figure 10, the upper left area shows how specific elements of the right singular matrix are arranged in the frequency direction. Here, each of the dots that make up the dashed line and each of the circles spaced at intervals of 4Δf represent data for a specific element.

The lower left area of Figure 10 shows the results of extracting data corresponding to circles from the data shown in the upper left area. Specifically, the amount of data has been reduced to 1/4 by extracting data at intervals of 4Δf.

The area on the right side of Figure 10 is the result of performing an inverse Fourier transform on the frequency domain information reduced by 1/4. This result is equivalent to the result shown in Figure 6, which was obtained by applying a window function to the impulse response shown in Figure 5. In this way, by using a technique for extracting data at regular intervals from frequency domain information, it is possible to achieve information reduction similar to that achieved by applying a window function to time domain information.

In this embodiment, the object detection device 40 performs the object detection process in accordance with the flowchart shown in Fig. 8, similarly to the first embodiment. However, in this embodiment, in step 106, information reduction is performed in the following procedure.
1. A frequency characteristic based on the restored right singular matrix (see the upper left region of FIG. 4 or FIG. 10) is generated.
2. Extract information from the above frequency characteristics at regular intervals to generate a reduced frequency characteristic.
3. An inverse Fourier transform is applied to the extracted reduced frequency characteristic to obtain an impulse response with reduced information.

After that, in step 108, object detection processing is performed based on the impulse response after the information has been reduced, as in the first embodiment. Therefore, the system of this embodiment can achieve highly accurate object detection without an excessive computational load, as in the first embodiment.

[Modification of the third embodiment]
Incidentally, in the above-described third embodiment, the object detection device 40 performs the object detection process in accordance with the flowchart shown in Fig. 8, as in the first embodiment. However, the present disclosure is not limited to this. That is, the method of reducing information at the frequency domain stage may be combined with the process in accordance with the flowchart shown in Fig. 9.

[Modifications common to the first to third embodiments]
In addition, in the above-described first to third embodiments, it is assumed that only one frequency band having multiple subcarriers is used. However, the present disclosure is not limited to this, and the object detection system may handle multiple frequency bands.

Figure 11 shows multiple frequency bands arranged adjacent to frequency band 52 used in embodiments 1 to 3. The object detection device of the present disclosure can also perform object detection after performing information reduction using multiple frequency bands as shown in Figure 11. In this case, a more detailed impulse response can be obtained, which is expected to improve the accuracy of the detection rate.

Figure 12 shows multiple frequency bands spaced apart to the right of frequency band 52 used in embodiments 1 to 3. The object detection system of the present disclosure can also handle multiple frequency bands spaced apart, as shown in Figure 12.

10 Wireless base station 20 Wireless terminal station 40 Object detection device 48 Information reduction unit

Claims (7)

An object detection device that detects an object in a detection area between a wireless base station and a wireless terminal station,
A process of receiving a feedback signal that carries information related to propagation channel information estimated for each subcarrier of a communication wave and is transmitted by the wireless terminal station;
extracting information related to the propagation channel information from the feedback signal;
a reduction process for reducing information related to the propagation channel information by applying a reduction condition selected as meeting a requirement for a detection rate of object detection;
A process of performing the object detection based on a reduced impulse response corresponding to the information after the reduction process is performed ;
The reduction condition is a condition that a detection rate of the object detection clears a pass threshold,
An object detection device in which the detection rate is an average detection rate of object detection performed by applying the results of machine learning that uses the reduced impulse response as input . the reduction condition is a condition for a window function to be applied to the impulse response,
The reduction process includes:
performing an inverse Fourier transform on frequency domain information relating to the propagation channel information to generate the impulse response;
The object detection apparatus according to claim 1 , further comprising: a process of applying the window function to the impulse response to generate the reduced impulse response. the reduction condition is a condition for an extraction process of extracting frequency domain information related to the propagation channel information at a certain frequency interval;
The reduction process includes:
performing the extraction process on frequency domain information related to the propagation channel information;
The object detection device according to claim 1 , further comprising a process of performing an inverse Fourier transform on the frequency domain information after the extraction process to generate the reduced impulse response. In order to select the reduction conditions,
calculating a detection rate of the object detection for each of the available reduction conditions;
A process of extracting a reduction condition that satisfies the pass condition as the pass condition,
selecting an optimal condition that achieves a maximum information reduction rate from among the pass conditions;
a selection process for selecting the optimal condition as the reduction condition;
4. The object detection device according to claim 1 , further configured to perform the following: In order to select the reduction conditions,
a process of determining whether a processing time for object detection based on the reduced impulse response generated by applying the optimal condition satisfies a time condition;
If it is determined that the processing time does not satisfy the time condition, a process of relaxing the judgment condition of the pass condition is further executed;
The object detection device according to claim 4 , wherein the selection process includes a process of selecting the optimum condition determined to satisfy the time condition as the reduction condition. An object detection method for detecting an object in a detection area between a wireless base station and a wireless terminal station, comprising:
a step of transmitting a radio signal to a radio terminal station by the radio base station on each of a plurality of subcarriers included in a frequency band used for communication;
The wireless terminal station estimates propagation channel information for each of the subcarriers;
The wireless terminal station transmits a feedback signal carrying information related to the propagation channel information;
extracting information related to said propagation channel profile from said feedback signal;
a reduction step of reducing the extracted information related to the propagation channel information by applying a reduction condition selected as meeting a requirement of a detection rate of object detection;
performing the object detection based on a reduced impulse response corresponding to information after the reduction step has been performed ;
The reduction condition is a condition that a detection rate of the object detection clears a pass threshold,
An object detection device in which the detection rate is an average detection rate of object detection performed by applying the results of machine learning that uses the reduced impulse response as input . a wireless base station that transmits wireless signals on a plurality of subcarriers included in a frequency band used for communication;
a wireless terminal station that receives the wireless signal, estimates propagation channel information for each of the subcarriers, and transmits a feedback signal carrying information related to the propagation channel information;
an object detection device that detects an object in a detection area between the wireless base station and the wireless terminal station,
The object detection device includes:
receiving the feedback signal;
extracting information related to the propagation channel information from the feedback signal;
a reduction process for reducing information related to the propagation channel information by applying a reduction condition selected as meeting a requirement for a detection rate of object detection;
A process of performing the object detection based on a reduced impulse response corresponding to the information after the reduction process is performed ;
The reduction condition is a condition that a detection rate of the object detection clears a pass threshold,
An object detection system in which the detection rate is an average detection rate of object detection performed by applying the results of machine learning using the reduced impulse response as input .

Images