Method and device for motion detection based on wireless LAN signal

Abstract

The motion detection device for performing motion detection by using information indicating a channel state measured using a wireless LAN signal comprises: a memory; and a processor. The processor determines whether the degree of change in the measurement time interval of the information indicating the channel state is within a predetermined value, and determines whether motion occurs around a device that has transmitted the wireless LAN signal on the basis of the determination and the information indicating the channel state.

Description

Wireless LAN signal-based motion detection method and device

[0001] The present invention relates to a method and device for detecting operation using wireless LAN signals.

[0002] With the recent expansion of mobile device adoption, Wireless LAN technology, capable of providing fast wireless internet services to these devices, is gaining significant attention. Based on short-range wireless communication technology, Wireless LAN enables mobile devices—such as smartphones, smart pads, laptop computers, portable multimedia players, and embedded devices—to connect to the internet wirelessly in homes, businesses, or specific service areas.

[0003] With the proliferation of wireless LANs, research is also underway to detect motion around access points (APs) and stations using signals exchanged between them. Motion detection using wireless LAN signals has the advantage of minimizing unnecessary privacy exposure, as it does not require directly capturing images of users. However, opinions have been raised that further research is needed regarding methods to improve the accuracy of motion detection and to avoid interfering with existing wireless LAN communications.

[0004] One embodiment of the present invention aims to provide a method and device for detecting operation using a wireless LAN signal.

[0005] According to one embodiment of the present invention, a motion detection device that performs motion detection using information indicating a channel state measured using a wireless LAN signal includes a memory; and a processor. The processor determines whether the degree of change in the information indicating the channel state is within a predetermined value during a measurement time interval, and determines whether motion occurs around the device that transmitted the wireless LAN signal based on the determination and the information indicating the channel state.

[0006] The processor can reduce the influence of the information representing the channel state on the motion detection when the maximum deviation of the information representing the channel state corresponding to the subcarrier is within a predetermined value.

[0007] The processor can reduce the influence of information indicating the channel state on motion detection based on whether the strength of the wireless LAN signal, which is the target of measurement of the channel state, is smaller than a predetermined size in the measurement time interval.

[0008] The processor can reduce the influence of the channel state information that measured the wireless LAN signal on the motion detection when the strength of the wireless LAN signal, which is the target of measurement of the channel state, is smaller than a predetermined size in the measurement time interval.

[0009] The processor can determine whether the movement occurs based on information indicating the channel state obtained from each of a plurality of measurement intervals having different durations.

[0010] The above processor can obtain information indicating the channel state by measuring the L-LTF (long training field) of the wireless LAN signal.

[0011] The above processor may not use a subcarrier among the subcarriers of the L-LTF that does not transmit a signal.

[0012] A method of operation of a motion detection device that performs motion detection using information indicating a channel state measured using a wireless LAN signal according to an embodiment of the present invention may include a step of determining whether the degree of change in the measurement time interval of the information indicating the channel state is within a predetermined value, and a step of determining whether motion occurs around the device that transmitted the wireless LAN signal based on the determination and the information indicating the channel state.

[0013] The step of determining whether movement occurs around the device that transmitted the wireless LAN signal may include a step of reducing the influence of the information representing the channel state on the movement detection when the maximum deviation of the information representing the channel state corresponding to the subcarrier is within a predetermined value.

[0014] The step of determining whether movement occurs around the device that transmitted the wireless LAN signal may include a step of reducing the influence of information indicating the channel state on the movement detection based on whether the strength of the wireless LAN signal, which is the measurement target of the channel state, is smaller than a predetermined size in the measurement time interval.

[0015] The step of reducing the influence of information indicating the channel state on the motion detection may include reducing the influence of the channel state information that measured the wireless LAN signal on the motion detection when the strength of the wireless LAN signal, which is the measurement target of the channel state, is smaller than a predetermined size in the measurement time interval.

[0016] The step of determining whether movement occurs around the device that transmitted the wireless LAN signal may include the step of determining whether movement occurs based on information indicating the channel state obtained from each of a plurality of measurement intervals having different durations.

[0017] The above operation method may include the step of measuring the L-LTF (long training field) of the wireless LAN signal to obtain information indicating the channel state.

[0018] The step of obtaining information indicating the channel state by measuring the L-LTF (long training field) of the wireless LAN signal may include a step of not using a subcarrier that does not transmit a signal among the subcarriers of the L-LTF.

[0019] One embodiment of the present invention provides a method and device for detecting operation using a wireless LAN signal.

[0020] FIG. 1 shows a system that detects user actions using wireless LAN signals according to an embodiment of the present invention.

[0021] FIG. 2 shows a station according to an embodiment of the present invention acquiring CSI.

[0022] FIG. 3 shows that in a wireless channel where a station and an anchor exchange signals according to an embodiment of the present invention, a change in channel state occurs due to various events.

[0023] FIG. 4 shows that a station according to an embodiment of the present invention maintains a fixed number of time series data through a plurality of measurement values ​​measured immediately before the most recent measurement value.

[0024] FIGS. 5 and 6 show a motion detection algorithm according to an embodiment of the present invention.

[0025] FIG. 7 shows a method for a station to collect CSI-related data according to an embodiment of the present invention.

[0026] FIGS. 8 and 9 show the operation of a noise and outlier removal filter applied to data regarding CSI by a motion detection device according to an embodiment of the present invention.

[0027] FIGS. 10 and 11 show the operation of a noise and outlier removal filter applied to data regarding RSSI by a motion detection device according to an embodiment of the present invention.

[0028] FIGS. 12 and 13 show the operation of a motion detection device according to an embodiment of the present invention acquiring features from a CSI after preprocessing the CSI.

[0029] Figure 14 shows a flat CSI sample and a standard sample together.

[0030] FIG. 15 shows the operation of a motion detection processing device according to an embodiment of the present invention quantifying the degree of flatness of a flat CSI.

[0031] FIG. 16 shows a motion detection processing device according to an embodiment of the present invention calculating the size range of CSI.

[0032] FIG. 17 shows a flat-CSI filter function used by a motion detection device according to an embodiment of the present invention.

[0033] FIGS. 18 and 19 show the operation of a motion detection device according to an embodiment of the present invention acquiring features from RSSI.

[0034] FIGS. 20 and 21 show the operation of a motion detection device according to an embodiment of the present invention outputting a probability of motion occurrence based on the previously described feature.

[0035] FIG. 22 shows the operation of a motion detection device according to an embodiment of the present invention, which monitors the probability of motion occurrence during a predetermined time interval and outputs a final detection result.

[0036] The terms used in this specification have been selected to be as widely used as possible, taking into account their functions in the present invention; however, these may vary depending on the intent, convention, or emergence of new technologies of those skilled in the art. In addition, in certain cases, terms have been arbitrarily selected by the applicant, and in such cases, their meanings will be described in the relevant description of the invention. Therefore, it should be noted that the terms used in this specification should be interpreted based on their actual meanings and the overall content of this specification, rather than merely their names.

[0037] Throughout the specification, when a configuration is described as being "connected" to another configuration, this includes not only cases where they are "directly connected," but also cases where they are "electrically connected" with other components interposed between them. Furthermore, when a configuration is described as "including" a specific component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. In addition, limitations such as "greater than or equal to" or "less than or equal to" based on a specific threshold value may be appropriately replaced with "greater than" or "less than," respectively, depending on the embodiment.

[0038]

[0039] FIG. 1 shows a system that detects user actions using wireless LAN signals according to an embodiment of the present invention.

[0040] A non-AP station can be associated with multiple APs. For convenience of explanation, a non-AP station is referred to as a station in this specification. A station can detect movement around the AP associated with the station by measuring RF signals transmitted by the AP. When movement occurs around the AP, the signal measured by the station changes. Channel state information (CSI), as defined in the 802.11 standard, is information indicating such changes in the signal. The station can detect movement occurring between the station and the AP by tracking changes in the signal. In this specification, the AP that measures the signal or transmits the signal is referred to as an anchor.

[0041] The station can determine the movement of an object based on the measured RF signal. The station can represent the detected movement as an activity level. The activity level can be transmitted by the station to a cloud server connected to the AP. The cloud server can store information regarding the activity level in a database and process it. By combining server logic, the cloud server can send a notification to an end-user application, such as a smartphone application.

[0042] FIG. 2 shows a station according to an embodiment of the present invention acquiring CSI.

[0043] Channel State Information (CSI) can indicate signal distortion during propagation from the transmitter to the receiver. Additionally, CSI can indicate effects such as scattering, fading, and power attenuation with distance. Transceivers can respond to changes in channel conditions based on CSI. This enables transceivers to maintain high reliability in communication using multiple antennas at high data rates. Since CSI can indicate distortion during signal propagation from the transmitter to the receiver, it needs to be measured at the receiver. The measurement procedure must be performed continuously at appropriate times. Therefore, it is recommended that the measurement procedure be performed by a power-supplied device rather than a battery.

[0044] In Figure 2, the station (STA) sends a null data packet to the AP, and the AP sends an ACK frame to the station. The AP is referred to as the anchor. The station (STA) measures the CSI for the signal containing the ACK frame. The CSI reflects how the signal propagates from the anchor to the station (STA) while the ACK is being transmitted.

[0045] FIG. 3 shows that in a wireless channel where a station and an anchor exchange signals according to an embodiment of the present invention, a change in channel state occurs due to various events.

[0046] When there is no obstruction between the anchor and the station, the AP and the station can exchange signals via the optimal path. However, the optimal path between the AP and the station can be disrupted by human movement. This can alter the path of the signal transmitted between the AP and the sensing node.

[0047] In Fig. 3, signal exchange between the AP and the station occurs via multipath. In this case, the received signal is expressed as the sum of all signals that reached the receiver through multipath. When there are no obstacles between the AP and the station, the signal propagated via the line-of-sight path (t0) generally has the greatest influence. If an obstacle, such as a person, passes through the optimal path (t0) between the AP and the station, the AP and the station will use a path other than the line-of-sight path (t n Signals propagated through ) can have the greatest impact on the received signal. Therefore, if there is no change in the surrounding environment of the AP and the station, the value of CSI can be stable. In addition, if there is a change in the surrounding environment of the AP and the station, the time deviation of the CSI for the received signal, which is composed of the sum of multipath signals, may increase.

[0048]

[0049] FIG. 4 shows that a station according to an embodiment of the present invention maintains a fixed number of time series data through a plurality of measurement values ​​measured immediately before the most recent measurement value.

[0050] The station may use a value representing changes such as reflection, absorption, and attenuation experienced as a signal propagates from the anchor to the station, such as CSI, as a measurement value. The station may obtain the sum of the most recent measurement value and N-1 measurement values ​​measured immediately before the most recent value. Based on the obtained values, the station may generate a data set used as input for an algorithm. Additionally, the station may obtain measurement values ​​by measuring the channel state at regular time intervals. For convenience of explanation, the time interval during which N measurement values ​​are obtained is referred to as the measurement window. Furthermore, N may be a pre-specified value or may be adjusted by a sensing device. For example, N may be a square number greater than 9. Also, N may change proportionally to the window and the measurement interval. Accordingly, if the length of the data set measured immediately before the most recent value is N, the station may generate a data set by deleting the oldest measured data among the measurements constituting the data set and adding the most recent measured value. In a specific embodiment, if the measurement window is 10 seconds and the measurement interval is 0.1 seconds, N may be 100. To maintain consistency of values ​​measured at different bandwidths, the data within the window may consist of CSI values ​​of a 20 MHz channel, e.g., a primary channel. The measurement values ​​measured by the station increase in proportion to the number of receiving antennas of the station. Therefore, the number of data sets may be a value proportional to the number of receiving antennas.

[0051] FIGS. 5 and 6 show a motion detection algorithm according to an embodiment of the present invention.

[0052] The motion detection process can be performed by a cloud server or a station in the system described through Fig. 1. For convenience of explanation, the device performing the motion detection process is referred to as a motion detection device. The motion detection process can be performed by a device that measures CSI and RSSI or by any device capable of receiving the measured CSI and RSSI.

[0053] The motion detection process of a motion detection device begins with acquiring new wireless LAN data from the physical layer. The wireless LAN data may include CSI and RSSI (received signal strength indicator). CSI provides specific and fine-grained measurement results of the communication channel. Specifically, CSI can be extracted from the LTF (long training field) of a wireless LAN packet. Additionally, CSI can be measured for each specific subcarrier of the wireless LAN channel. CSI sensitively reflects environmental changes, including movement, along the signal path. RSSI can be a scalar value representing the measured power level of the received signal. RSSI is a relatively coarse-grained indicator of signal strength that reflects movement.

[0054] The motion detection device stores newly acquired CSI and RSSI data in a data queue. The data queue can function as a buffer and collect continuous data for a predetermined window period. This enables more stable analysis of signal characteristics over a specific duration. The motion detection device can process CSI data and RSSI data through distinct preprocessing steps.

[0055] First, the preprocessing process of RSSI data is explained. During the preprocessing of RSSI data, the motion detection device can remove noise and outlier values ​​from the RSSI data. Noise and outlier values ​​can cause erroneous detection. By removing noise and outliers, the motion detection device can increase the reliability of the RSSI data used for feature extraction.

[0056] The CSI preprocessing process may include flat CSI filtering. Specifically, flat CSI filtering is a filtering method designed to identify subcarrier data frames with abnormally constant magnitudes and to mitigate the influence of such data frames. Through flat CSI filtering, motion detection devices can prevent false detections caused by subcarrier data frames with abnormally constant magnitudes. Following flat CSI filtering, noise and outlier values ​​can be removed, similar to the preprocessing of RSSI data.

[0057] The motion detection device determines whether motion has occurred by extracting features from preprocessed CSI and RSSI data. Specifically, the CSI and RSSI data can be passed to a feature extraction block. The motion detection device can output the probability that motion has occurred from the extracted blocks using a model classifier, which is a machine learning model trained on the CSI and RSSI data. The motion detection device can finally determine whether motion has occurred after post-processing. Specifically, the motion detection device can finally determine whether motion has occurred based on the probabilities of multiple measurement windows.

[0058] In the embodiments described above, when the motion detection device accumulates data regarding CSI and RSSI, the motion detection device may operate two distinct queues with different durations for data accumulation. Specifically, it may operate a short queue that accumulates data regarding CSI and RSSI for the shorter of the two queues, and a long queue that accumulates data regarding CSI and RSSI for the longer of the two queues. Features extracted from the data accumulated in the short queue can reflect sudden movements occurring around the anchor and station. Additionally, features extracted from the data accumulated in the long queue can reliably reflect movements occurring around the anchor and station.

[0059] FIG. 5 shows a motion detection algorithm according to an embodiment of the present invention in which short queue and long queue are not applied, and FIG. 6 shows a motion detection algorithm according to an embodiment of the present invention in which short queue and long queue are applied.

[0060] FIG. 7 shows a method for a station to collect CSI-related data according to an embodiment of the present invention.

[0061] Wireless LAN packets received by a station may contain multiple LTFs depending on the wireless LAN generation. The motion detection device can calculate the CSI for the L-LTF among the multiple LTFs. Since all wireless LAN packets contain the L-LTF, the motion detection device can calculate the CSI from the L-LTF to increase the reliability of CSI extraction.

[0062] The motion detection device calculates the magnitude of the CSI. Specifically, the value of the CSI of each subcarrier has a real number and an imaginary number, and the motion detection device can calculate the magnitude of the CSI as the square root of the square of the real number and the square of the imaginary number of the CSI value.

[0063] The motion detection device can remove noise and outliers from the magnitude values ​​of the CSI by applying noise and outlier removal filtering. Through this, the motion detection device can prevent false detections caused by random noise, interference from the surroundings, and signal fading.

[0064] Additionally, the motion detection device may exclude CSI values ​​corresponding to inactive subcarriers, such as pilot subcarriers, DC subcarriers, or guard subcarriers, which are subcarriers that do not transmit signals. A guard subcarrier is an inactive subcarrier that is not used to prevent interference with adjacent channels. Therefore, a guard subcarrier may not have a CSI value. Among the active subcarriers, the CSI values ​​of pilot subcarriers and DC subcarriers may be 0 (null). Therefore, if motion detection is performed by including CSI values ​​corresponding to pilot subcarriers and DC subcarriers of the motion detection device, the accuracy of motion detection may be reduced.

[0065] FIGS. 8 and 9 show the operation of a noise and outlier removal filter applied to data regarding CSI by a motion detection device according to an embodiment of the present invention.

[0066] The motion detection device can identify outlier values ​​by applying Z-Score normalization to CSI values, that is, an operation that sets the mean to 0 and the standard deviation to 1. Specifically, if the Z-Score normalized value exceeds a pre-specified threshold, the motion detection device can determine it as an outlier value. Additionally, the motion detection device can replace the identified outlier values ​​with the mean value. Through this, the motion detection device can enhance the stability of CSI-related data. Figure 8 illustrates the operation of a noise and outlier removal filter to which this outlier removal operation is applied. In the embodiment of Figure 8, the motion detection device excludes the signals of the DC subcarrier and the pilot subcarrier. Furthermore, the motion detection device performs Z-Score normalization on the subcarrier axis. Subsequently, the motion detection device replaces the values ​​identified as outliers on the time axis of the Z-Score normalized CSI with the mean value.

[0067] In the embodiment of FIG. 9, the motion detection device removes outlier values ​​using an isolation forest outlier filter as well as Z-Score normalization. In a specific embodiment, the motion detection device may use an isolation forest algorithm that determines outlier values ​​based on the number of lines required to separate the data. At this time, the motion detection device may use the isolation forest algorithm on the time axis. For example, the motion detection device may perform Z-Score normalization on the subcarrier axis and perform isolation forest on the time axis.

[0068] FIGS. 10 and 11 show the operation of a noise and outlier removal filter applied to data regarding RSSI by a motion detection device according to an embodiment of the present invention.

[0069] The motion detection device can apply the previously described Z-Score outlier removal filter to RSSI data as well. Through this, the motion detection device can remove data corresponding to outliers from the data within the measurement window. In addition, as shown in the embodiment of FIG. 11, the motion detection device can perform Min-Max scaling normalization after applying Z-Score outlier removal filtering. Min-Max scaling normalization maps the RSSI values ​​to a predetermined range, for example, between 0 and 1. Through this, the motion detection device can minimize the effects of signal distortion.

[0070] FIGS. 12 and 13 show the operation of a motion detection device according to an embodiment of the present invention acquiring features from a CSI after preprocessing the CSI.

[0071] The motion detection processing device can obtain a value representing the degree of change along the time axis among the data regarding CSI, and acquire features from the value representing the change. In a specific embodiment, first, the motion detection processing device can calculate a filter value for flat CSI filtering on the data regarding CSI that has not undergone the preprocessing described above. Examples regarding the filter value are explained in detail through FIGS. 15 to 17. The motion detection processing device can calculate the variance along the time axis for each subcarrier from the preprocessed data regarding CSI. This is because, since motion is detected based on the change in CSI values ​​over time, the variance value along the time axis is suitable for indicating whether or not motion is occurring. Additionally, the motion detection device can obtain a representative value from the variance values ​​for each subcarrier. In this case, the representative value may be the median value. The motion detection processing device can acquire features from the value obtained by applying flat CSI filtering to the data regarding CSI that has not undergone preprocessing and the acquired representative value. FIG. 12 illustrates this operation of the motion detection device. The motion detection device can perform robust and accurate motion detection through these embodiments.

[0072] Additionally, the motion detection device can obtain variance from each set of data regarding multiple CSIs, where the duration of the time interval for acquiring the CSIs is different, and obtain features from the obtained variance. Specifically, the motion detection device can obtain the variance on the time axis of the data regarding the CSIs acquired in the first time interval and the variance on the time axis of the data regarding the CSIs acquired in the second time interval. At this time, the duration of the first time interval may be shorter than the duration of the second time interval. In the process of calculating the variance for each time interval, the motion detection device can obtain the variance on the time axis for each subcarrier and obtain a representative value that represents the variance value for each subcarrier. At this time, the representative value may be the median value. The motion detection device can obtain features from the variance value representing multiple first time intervals and the variance value representing multiple second time intervals. FIG. 13 illustrates such operation of the motion detection device. Through these embodiments, the motion detection device can reliably detect both fast and slow motions.

[0073] Flat CSI is explained in detail through FIGS. 14 to 16.

[0074] Figure 14 shows a flat CSI sample and a standard sample together.

[0075] Flat CSI samples refer to CSI data that exhibits relatively small variations compared to typical CSI samples within the time interval in which the CSI was acquired. Specifically, motion detection devices can distinguish whether a sample is a flat CSI based on the difference between the maximum and minimum values ​​of the CSI. When a signal path is stably formed and very strong signals are transmitted and received, flat CSI samples with minimal variation in CSI values ​​may be collected. Since the collected CSI samples do not reflect changes in the surrounding environment with excessive sensitivity, this can interfere with motion detection. Specifically, if Z-score normalization is performed including flat CSI samples, the standard deviation becomes small, causing even very small changes to be amplified into excessively large values. Therefore, it is necessary to identify flat CSI and process it before performing motion detection.

[0076] FIG. 15 shows the operation of a motion detection processing device according to an embodiment of the present invention quantifying the degree of flatness of a flat CSI.

[0077] As previously explained, the motion detection processing unit can quantify the degree of change in CSI and obtain features from the numerical value. Through this, the motion detection unit can reduce the influence of flat CSI. The motion detection processing unit can first calculate the size range of the CSI. In addition, the motion detection processing unit can apply a flat CSI filter function based on the calculated size range.

[0078] FIG. 16 shows a motion detection processing device according to an embodiment of the present invention calculating the size range of CSI.

[0079] The motion detection device can acquire CSI values ​​for each of the N subcarriers at N time points. At this time, the motion detection device can acquire the difference (Rt) between the maximum and minimum CSI values ​​of the subcarriers at a specific time point (t). This can be expressed by the following formula.

[0080] R t = (max 1<=s<=S C t,s ) - (min 1<=s<=S C t,s )

[0081] The motion detection device is a vector containing the maximum deviation, which is the difference (Rt) between the maximum and minimum values ​​at each time point, R=[R1, R2, … R t ] can be obtained. The motion detection device can select a representative value, such as the median value, from the vector R as the representative value. In this case, the representative value can be expressed by the following formula.

[0082] M range =median([R1, R2, …, R t ])

[0083] This allows the influence of outlier values ​​in motion detection to be reduced.

[0084] FIG. 17 shows a flat-CSI filter function used by a motion detection device according to an embodiment of the present invention.

[0085] The motion detection device can input the acquired value into a flat-CSI filter function. The flat-CSI filter function can output 0 if the acquired CSI data corresponds to flat-CSI, and output 1 if the acquired CSI data does not correspond to flat-CSI. In this case, the flat-CSI filter function may be a sigmoid function. Specifically, the flat-CSI filter function can be represented by the following formula.

[0086]

[0087] In the above formula represents the first threshold and the inflection point distinguishing between flat and non-flat general CSI, and can indicate the slope of the sigmoid function. In Fig. 13, M rangeIf the value of is smaller than the threshold, the CSI is highly likely to be classified as a flat CSI. Also, M range If the value of is greater than the threshold, the probability that the CSI is classified as not being a flat CSI is high in Fig. 13. The value of is 12, and can be 1.5. The above formula outputs 1 if the CSI is determined to be non-flat CSI, and outputs 0 if the CSI is determined to be flat CSI. The motion detection device may reduce the significance or reliability of the variance measured in the CSI determined to be flat CSI.

[0088] FIGS. 18 and 19 show the operation of a motion detection device according to an embodiment of the present invention acquiring features from RSSI.

[0089] The motion detection device can acquire features from a determination regarding whether the RSSI of the wireless LAN signal, which is the target of channel state measurement, is weaker than a preset strength. Specifically, if the RSSI of the wireless LAN signal, which is the target of channel state measurement, is weaker than a preset strength, the motion detection device can reduce the influence of channel state information on the corresponding wireless LAN signal upon motion detection. For convenience of explanation, an RSSI weaker than a preset strength is referred to as a "weak RSSI." In this case, the motion detection device can acquire features from a determination regarding whether the preprocessed RSSI is a weak RSSI. The motion detection processing device can acquire the average value of the RSSI over the time axis. In this case, the motion processing device can acquire features based on the average value of the RSSI and a determination regarding whether it is a weak RSSI based on the average value. In this case, the motion detection device can determine whether the RSSI is a weak RSSI by using the average value of the RSSI as input to a preset classification function. The average value of the RSSI provides a stable trend of signal strength, and the determination regarding whether it is a weak RSSI can minimize the influence of noise. A motion detection device can determine whether an RSSI is a weak RSSI by inputting the average value of the RSSI into a pre-specified classification function. In this case, the classification function may be a sigmoid function. Specifically, the classification function may be the following formula.

[0090]

[0091] is the signal strength threshold that serves as the criterion for determining whether it is a weak signal, and can be a gain factor that reduces the slope of the filter response. In a specific embodiment It can be -67.5 dBm. Also, can be 1.69. The above formula outputs 1 if it is determined to be an RSSI that is not weak RSSI, and outputs 0 if it is determined to be an RSSI that is weak RSSI. The determination of whether the feature obtained from RSSI is weak RSSI can be used to scale or adjust the feature obtained from CSI. Specifically, the determination of whether the feature obtained from RSSI is weak RSSI can be used as a weight for CSI when performing motion detection from the feature obtained from CSI. If the RSSI value is too small, the magnitude of the CSI value may be excessively small. Also, if the RSSI value is too large, the magnitude of the CSI value may be excessively large. Therefore, the motion detection device can use the determination of whether the feature obtained from RSSI is weak RSSI to normalize the magnitude of the CSI. FIG. 18 illustrates such an embodiment.

[0092] The motion detection device can acquire features from the difference between the maximum and minimum RSSI values ​​measured in the measurement interval and the median RSSI. Specifically, the motion detection device can use the difference between the maximum and minimum RSSI values ​​measured in the measurement interval and the median RSSI to scale or adjust features acquired from the CSI. For example, the motion detection device can use the difference between the maximum and minimum RSSI values ​​measured in the measurement interval and the median RSSI to normalize the magnitude of the CSI. Additionally, the motion detection device can acquire features from the difference between the maximum and minimum RSSI values ​​and the median RSSI from each of the measurement intervals with different durations. Through this, it is possible to accurately detect motion occurring in a relatively short time and motion occurring slowly over a long period. FIG. 19 illustrates such an embodiment.

[0093] FIGS. 20 and 21 show the operation of a motion detection device according to an embodiment of the present invention outputting a probability of motion occurrence based on the previously described feature.

[0094] In the embodiment of FIG. 20, the machine learning model outputs a motion occurrence probability based on a judgment of whether it is a weak RSSI, features obtained from the RSSI, and features obtained from the CSI, according to the embodiments described through FIG. 18. Specifically, the motion detection device can compensate for features obtained from the CSI by multiplying the features obtained from the CSI by a value indicating a judgment of whether it is a weak RSSI. At this time, the closer the RSSI is to a weak RSSI, the closer the value indicating a judgment of a weak RSSI may be to 0. Through this, the probability of false detection due to noise can be reduced by excluding CSI obtained from very weak signals.

[0095] The machine learning model can output the probability that motion has occurred based on features obtained from the rewarded CSI and features obtained from the RSSI. In this case, the machine learning model may be a model trained to determine whether motion has occurred using features obtained from the rewarded CSI and features obtained from the RSSI. Additionally, the output value may be a value between 0 and 1.

[0096] In the embodiment of FIG. 21, the machine learning model outputs a probability of motion occurrence based on features obtained from the difference between the maximum and minimum values ​​of RSSI values ​​and the median value of RSSI in each measurement interval with different durations according to the embodiments described through FIG. 19.

[0097] FIG. 22 shows the operation of a motion detection device according to an embodiment of the present invention, which monitors the probability of motion occurrence during a predetermined time interval and outputs a final detection result.

[0098] The motion detection device can determine motion based on whether the probability of motion occurrence of the machine learning model described above is greater than a threshold value. In this case, the motion detection device can assign weight values ​​according to the magnitude of the probability of motion occurrence. For example, if the probability of motion occurrence has a value equal to or greater than a first threshold value, a first weight value may be assigned, and if it has a value smaller than the first threshold value but equal to or greater than a second threshold value, a second weight value may be assigned. Additionally, if it has a value smaller than a third threshold value, a first weight value may be assigned, and if it has a value smaller than the second threshold value but equal to or greater than a third threshold value, a second weight value may be assigned. In this case, the sensing motion detection device can adjust the sensitivity of motion detection by adjusting the magnitude of each threshold value. In the embodiment of FIG. 22, the first threshold value is BUSY confidence , the second threshold is 0.5, and the third threshold is IDLE confidence It is. Also, the first weight value is 1, and the second weight value is 0.5.

[0099] In addition, the motion detection device can monitor detection results for a certain period and output a final detection result. In this case, the certain period is referred to as a voting interval, and the length of the voting interval may be referred to as the voting size. Specifically, if the sum of the products of multiple detection results output during the certain period and their weight values ​​is equal to or greater than a pre-specified threshold, e.g., 0.5, the motion detection device may determine that motion has been finally detected. In this case, the detection result is a type of flag, which may have a value of 1 if the probability of motion occurring is high and a value of 0 if the probability of motion not occurring is high. Specifically, the motion detection device may obtain the value determining the final motion detection result through the following formula.

[0100]

[0101] N is the number of motion results included in the voting window, Motion Flag is the detection result, and voteWeight represents the weight value based on the previously explained motion occurrence probability.

[0102] Specifically, the motion detection device can determine whether final motion is detected using the following formula.

[0103]

[0104] Specifically, if the final detection result value obtained by adding the values ​​obtained by multiplying each detection result by a weight value is equal to or greater than a threshold value, for example 0.5, the motion detection device can determine that motion has finally occurred.

[0105] Through this operation, the motion detection device can increase the reliability of the motion detection results.

[0106]

[0107] Although the present invention has been described above using wireless LAN communication as an example, the invention is not limited thereto and can be applied in the same way to other communication systems, such as cellular communication. Furthermore, while the method, apparatus, and system of the present invention have been described in relation to specific embodiments, some or all of the components and operations of the present invention may be implemented using a computer system having a general-purpose hardware architecture.

[0108] The features, structures, effects, etc. described in the embodiments above are included in at least one embodiment of the present invention and are not necessarily limited to only one embodiment. Furthermore, the features, structures, effects, etc. exemplified in each embodiment may be combined or modified and implemented in other embodiments by a person skilled in the art to which the embodiments belong. Therefore, details regarding such combinations and modifications should be interpreted as being included within the scope of the present invention.

[0109] Although the above description has focused on exemplary embodiments, this is merely illustrative and does not limit the invention. Those skilled in the art will understand that various modifications and applications not exemplified above are possible within the scope of the essential characteristics of the embodiments. For example, each component specifically shown in the embodiments may be modified. Furthermore, differences related to such modifications and applications should be interpreted as being included within the scope of the invention as defined in the appended claims.

Claims

1. In a motion detection device that performs motion detection using information indicating the channel status measured using a wireless LAN signal, Memory; and Includes a processor, The above processor Determining whether the degree of change in the measurement time interval of the information indicating the channel status above is within a preset value, and Determining whether movement occurs around the device that transmitted the wireless LAN signal based on the above judgment and information indicating the above channel status Motion detection device.

2. In Paragraph 1, The above processor If the maximum deviation of the information indicating the channel state corresponding to the above subcarrier is within a preset value, the influence of the information indicating the channel state on the motion detection is reduced. Motion detection device.

3. In Paragraph 1, The above processor Reducing the influence of information indicating the channel state on motion detection based on whether the strength of the wireless LAN signal, which is the measurement target of the channel state, is smaller than a preset size in the measurement time interval. Motion detection device.

4. In Paragraph 3, The above processor If the strength of the wireless LAN signal, which is the measurement target of the above channel state, is smaller than a predetermined size during the above measurement time interval, the influence of the channel state information measuring the wireless LAN signal on the motion detection is reduced Motion detection device.

5. In Paragraph 1, The above processor Determining whether the movement occurs based on information indicating the channel state obtained from each of a plurality of measurement intervals having different durations. Motion detection device.

6. In Paragraph 1, The above processor Obtaining information indicating the channel state by measuring the L-LTF (long training field) of the above wireless LAN signal Motion detection device.

7. In Paragraph 6, The above processor Among the subcarriers of the above L-LTF, the subcarrier that does not transmit a signal is not used Motion detection device.

8. A method of operation of a motion detection device that performs motion detection using information indicating a channel state measured using a wireless LAN signal, A step of determining whether the degree of change in the measurement time interval of information indicating the channel status above is within a preset value and A step of determining whether movement occurs around the device that transmitted the wireless LAN signal based on the above determination and information indicating the channel status. Method of operation.

9. In Paragraph 8, The step of determining whether movement occurs around the device that transmitted the above wireless LAN signal is If the maximum deviation of the information indicating the channel state corresponding to the subcarrier is within a predetermined value, the method includes a step of reducing the influence of the information indicating the channel state on the motion detection. Method of operation.

10. In Paragraph 8, The step of determining whether movement occurs around the device that transmitted the above wireless LAN signal is A step comprising reducing the influence of information indicating the channel state on motion detection based on whether the strength of the wireless LAN signal, which is the measurement target of the channel state, is smaller than a predetermined size in the measurement time interval. Method of operation.

11. In Paragraph 10, The step of reducing the influence of information indicating the channel state on the motion detection If the strength of the wireless LAN signal, which is the target of measurement for the channel state, is smaller than a predetermined size in the measurement time interval, the method includes a step of reducing the influence of the channel state information that measured the wireless LAN signal on the motion detection. Method of operation.

12. In Paragraph 8, The step of determining whether movement occurs around the device that transmitted the above wireless LAN signal is A step of determining whether the movement occurs based on information indicating the channel state obtained from each of a plurality of measurement intervals having different durations. Method of operation.

13. In Paragraph 8, The above method of operation A method comprising the step of measuring the L-LTF (long training field) of the wireless LAN signal to obtain information indicating the channel state. Method of operation.

14. In Paragraph 13, The step of obtaining information indicating the channel state by measuring the L-LTF (long training field) of the wireless LAN signal A step comprising not using a subcarrier among the subcarriers of the above L-LTF that does not transmit a signal. Method of operation.

Images