Offline adjustment system for detecting new motion regions in motion detection systems

Abstract

In a general aspect, an offline system detects new motion regions for a motion detection system. In some examples, observed channel response data is obtained from the motion detection system. The observed channel response data is associated with a plurality of subcarriers for each wireless link over a period of time. Estimated channel response data is generated for each wireless link. Transitions between activity regions are identified in the estimated channel response data for each of the plurality of subcarriers for each wireless link. New motion regions for the motion detection system are detected based on the estimated channel response data and the identified transitions.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Patent Application No. 16 / 360,250, filed on March 21, 2019, entitled “OFFLINE ADJUSTMENT SYSTEM FOR DETECTING NEW MOTION ZONES IN MOTION DETECTION SYSTEMS,” the disclosure of which is incorporated herein by reference. Background Art

[0003] The following description relates to motion detection.

[0004] Motion detection systems have been used to detect the movement of objects within a room or outdoor area, for example. In some example motion detection systems, infrared or optical sensors are used to detect the movement of objects within the sensor's field of view. Motion detection systems have been used in security systems, automated control systems, and other types of systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Figure 1 is a block diagram illustrating an example offline adjustment system.

[0006] Figure 2 is a diagram showing an example time sequence of subcarriers.

[0007] Figure 3 is a diagram illustrating an example adaptive model for identifying new activity regions.

[0008] Figure 4 is a diagram showing an example of updating hyperparameters in an adaptive model.

[0009] Figure 5 is a diagram showing an example aggregation of link data.

[0010] Figure 6 is a diagram showing a flow chart for identifying a new motion area.

[0011] Figure 7 is a block diagram illustrating an example external sensor device.

[0012] Figure 8 Shown is the output of an actual experiment conducted using an example offline training system. DETAILED DESCRIPTION

[0013] In some aspects described, a new motion zone of a motion detection system is identified. A motion zone is an area in a space where motion can be detected and is associated with a certain position in the space. A zone can be a specific room in a space or a combination of rooms or zones. A space can be, for example, a house, a business, or other types of buildings. The motion detection system can initially be trained to associate detected motion with certain motion zones, which are sometimes determined by a user performing supervised training of the system. For example, when a motion detection system is installed in a space (e.g., a house or a business), the system can be trained for that specific space. For example, a user can be instructed to walk through certain areas (e.g., motion zones) in the space while performing certain movements. This type of training is called supervised training. During supervised training, the motion detection system records readings of motion from various motion zones. In some cases, the user provides corresponding tags for various motion zones (e.g., kitchen, living room, bedroom, etc.). However, supervised training can be time-consuming and tedious for the user, and the user may be reluctant to explicitly train the system. Requiring the user to walk in different zones and, for example, to mark each zone via a mobile application is a cumbersome process and is prone to user errors. Furthermore, the user may be required to supervise periodic retraining, for example when the space is reorganized, or to adjust to an updated motion algorithm. In some cases, it is desirable to periodically train the system to automatically recognize new or changed motion zones, thereby requiring minimal user involvement.

[0014] In some examples, the systems and techniques described herein can provide one or more advantages. For example, the accuracy of motion sensing and detection and the accuracy of localization of motion within a space served by a motion detection system can be improved. In addition, identification of new motion zones can be performed offline without affecting the operation of the motion detection system and without user supervision.

[0015] Figure 1 is a block diagram illustrating an example offline adjustment environment 100. In the example offline adjustment environment, an offline adjustment system 110 interfaces with a motion detection system 120, an application running a user device 140, and a machine learning system 160. In an implementation, the motion detection system 120 is operating in a home (or other type of space) and providing motion and positioning information to its user. The offline adjustment system 110 and the machine learning system 160 can be implemented on any suitable computer system (such as, for example, any type of computer, server or server cluster, distributed computing system or other computing resource, etc.). In some cases, the offline adjustment system 110 and the machine learning system 160 operate in a cloud environment (e.g., remote from the motion detection system); or in some cases, one or both of the offline adjustment system 110 and the machine learning system 160 can operate in a location or network environment associated with the motion detection system 120 or the user device 140.

[0016] The motion detection system 120 may use motion detection techniques to detect the motion (or lack of motion, presence, etc.) of an object in space. The motion detection system 120 may also use positioning techniques to pinpoint the location of an object in space, for example, to a specific node of the system. Figure 1 In the example shown, the motion detection system 120 uses wireless signals to detect the movement of objects in a home or other type of space. The space accessed by the wireless signal can be an indoor or outdoor space, which can include, for example, one or more fully or partially enclosed areas, open areas that are not enclosed, etc. The space can be or can include the interior of a room, multiple rooms or buildings, etc. In some instances, based on the wireless signal (for example, based on channel response information obtained from the wireless signal, or other types of channel state information), the motion detection system 120 can detect the movement, presence or position of objects in the space. The object can be any type of static or movable object, and can be animate or inanimate. For example, the object can be a person, an animal, an inorganic object, or other device, equipment or accessory, an object that defines all or part of the boundaries of the space, or other types of objects.

[0017] In some instances, wireless signals received at various wireless communication devices in the motion detection system 120 may be analyzed to determine channel response information for wireless links in the network (e.g., between corresponding pairs of wireless communication devices in the network). The channel response information may represent the application of a transfer function to a physical medium of a wireless signal traveling through space. In some instances, the channel response information includes channel characteristics of the wireless link and may describe how the wireless signal propagates from a transmitter to a receiver, thereby representing, for example, the combined effects of scattering, fading, and power attenuation within the space between the transmitter and the receiver. In some instances, the channel response information includes beamforming state information, which is used in a multi-antenna (multiple-input / multiple-output (MIMO)) radio system for directional signal transmission or reception.

[0018] Example motion detection and localization algorithms that may be used to detect motion based on wireless signals include the techniques described in the following, as well as other techniques: U.S. Patent 9,523,760, entitled “Detecting Motion Based on Repeated Wireless Transmissions,” U.S. Patent 9,584,974, entitled “Detecting Motion Based on Reference Signal Transmissions,” U.S. Patent 10,051,414, entitled “Detecting Motion Based On Decompositions Of Channel Response Variations,” U.S. Patent 10,048,350, entitled “Motion Detection Based on Groupings of Statistical Parameters of Wireless Signals,” U.S. Patent 10,048,350, entitled “Motion Detection Based on Machine Learning of Wireless Signal Characteristics,” Properties,” U.S. Patent 10,108,903, entitled “Motion Localization in a Wireless Mesh Network Based on Motion Indicator Values,” U.S. Patent 10,109,167, entitled “Motion Localization Based on Channel Response Characteristics,” and U.S. Patent 10,109,168, entitled “Motion Localization Based on Channel Response Characteristics,”

[0019] In some cases, the offline adjustment system 110 automatically retrains the motion detection system 120 periodically to capture environmental changes to the space (e.g., movement of furniture, installation / removal of ceiling fans, remodels, etc.) that result in changes to the multipath environment in the home. Retraining can also be used to account for improvements in the detection, localization, and machine learning algorithms utilized by the motion detection system 120. In these cases, the offline adjustment process can be implemented without requiring active participation by the user for training purposes.

[0020] In implementations, the offline adaptation process also automatically identifies new motion regions in the space.For example, the offline adaptation environment may identify and segment motion activity that may be associated with various regions in the space where the motion detection system 120 is operating.

[0021] In some examples, each time a calibration event is run (eg, every night), the offline adjustment system 110 obtains the observed channel data 130 from the motion detection system 120 and determines the delineation of the motion zone. Figure 1 As shown, the offline adjustment system 110 receives actual observation data 130 from the motion detection system 120. The offline adjustment system 110 processes the received data offline, that is, separately from the motion detection system 120 processing the data to detect motion. The channel data 130 may be received and processed regularly at certain intervals, such as every day, every few days, or only on weekends, etc. In some cases, the interval may be configurable by the user. For example, when the system is initially installed, offline adjustments of the system may be performed every day. After a period of time, the interval may be adjusted so that offline adjustments are performed less regularly, for example, weekly, biweekly, monthly, etc. In some instances, the offline adjustment system 110 receives data during times when minimal motion is occurring (e.g., between midnight and 3 a.m.). In some instances, the data 130 is processed when the offline adjustment system 110 identifies activity in a portion of the received data 130.

[0022] Initially, the motion detection system 120 may only be able to locate motion near the installation node. For example, each node for motion detection can be implemented as a wireless communication device (e.g., a radio frequency (RF) device), which can carry a software stack for motion detection. In some instances, the wireless communication device is part of the motion detection system 120. Initially, the number of zones available to the system and known to the user is limited by the number of nodes in the motion detection system 120. In an example, at the time of installation, two nodes can be installed, for example, one node in the kitchen and another node in the living room. The offline adjustment system 110 operates to extract (pull) channel data 130 captured by the motion detection system 120, identify multiple motion zones by analyzing the data, and identify additional motion zones based on the data. In some instances, the additional zone improves the accuracy of motion detection and positioning of motion.

[0023] In some instances, the offline adjustment system 110 operates in a semi-supervised mode to elicit human assistance and input from the user to accurately name newly identified zones. However, the user is not required to undergo a supervised training process to identify new areas. For example, in some instances, the offline adjustment system 110 identifies zones and their occurrence times offline, and only involves the user when a change in configuration or a new unlabeled zone is identified. By processing independently of the user, the system iterates over time, each time improving the quantitative performance (which refers to the accuracy of the derived information) and the qualitative performance (which refers to the relevance of the information provided to the user, such as the location of the motion) of the motion detection system 120.

[0024] In some instances, a message 150 is sent via a user interface of a motion detection application on a user device 140 (e.g., a smart phone, personal digital assistant, personal computer, laptop, or other device) to prompt the user for feedback. In some instances, the message alerts the user that motion was detected in an unlabeled zone at a certain time of day, and further asks the user if the user is interested in giving the unlabeled zone a name (such as "front corridor", "decoration", etc.). In some cases, occasional, repeated participation by the user in naming newly identified zones means that no explicit training on the part of the user is required when the motion detection system 120 is started. Instead, the system can simply improve slowly over time as it is deployed.

[0025] In some implementations, the delineation of new motion zones by the offline adjustment system 110 can be used as input to further train the machine learning system 160 of the motion detection system 120. For example, the machine learning system 160 can use the new motion zones and other existing motion zones to train the motion detection system to distinguish motion occurring in different motion zones (including the new motion zones) and between different motion zones, and to localize the motion to one of the motion zones in space. In some instances, the machine learning system 160 provides a model for mapping instantaneous (or near instantaneous, e.g., over a short time window) observations of channel response data to zone identifiers (including the new zones identified by the offline adjustment system 110), so that the motion detection system will be able to detect user activity in real time. In some implementations, the machine learning model operates as a supervised process in which the model is explicitly provided with zone identifiers that must be matched to the observed channel response data. The zone identifiers can be made available from the offline adjustment system 110, which, together with semi-supervised human intervention, can determine the actual zones and their occupancy intervals in the data set.

[0026] In some instances, additional processing can be performed in the cloud based on the results of the offline adjustment system 110 to make new zones available to the motion detection system 120 and the user device 140. For example, once the machine learning model is trained using a supervised data set, the model output can be provided to a classifier that operates as the forward part of the machine learning model 160. In some instances, the classifier is a process that creates actual zone categories based on observed channel data, and the machine learning model has a feedback loop to train the parameters of the classifier. In some instances, once the classifier is complete, it can be transmitted 150 to the user device 140 as an embedded system service update (e.g., overnight or at other times) so that the motion detection prediction can be immediately available to the user. In some instances, the classifier can continue to run as a cloud service, notifying the user of activity with a slightly higher latency than running on the device.

[0027] In some instances, when the offline adjustment system 110 identifies a new motion zone, the motion detection system 120 is updated so that the motion detection system 120 can identify the motion or presence of objects in the new motion zone. For example, the offline adjustment system 110 can provide distribution parameters or other types of data as follows, which the motion detection system 120 can use to identify motion occurring in the new motion zone based on the wireless signals collected by the motion detection system. As another example, the offline adjustment system 110 can provide motion zone markers or other types of information that allow the motion detection system 120 to provide meaningful output to other systems. For example, the motion zone markers can be used to notify a user, a security system, a power management system, or other entity: where motion is occurring in a home (or other space) being monitored by the motion detection system 120.

[0028] In some implementations, the offline adjustment system 110 or the machine learning system 160 (or both) are operated as a hosted service that supports motion detection systems 120 in many different locations. In other words, the offline adjustment system 110 and the machine learning system 160 are not necessarily dedicated to any individual motion detection system 120. Instead, the offline adjustment system 110 and the machine learning system 160 can receive input data from a plurality of different systems and users and provide output data to a plurality of different systems and users.

[0029] Figure 2 is a diagram showing an example time series of subcarriers. The example shows the type of observation data 130 that may be received from the motion detection system 120 and interpreted by the offline adjustment system 110. For example, the motion detection system 120 may collect channel response data 210 during the motion detection process. In some cases, the channel response data 210 spans several subcarriers (as shown on the x-axis). Here, the channel response data 210 shows a snapshot of the amplitude (vertical axis) of the wireless signal over a frequency range (horizontal axis); the frequency range shown in the figure includes multiple subcarriers, where each subcarrier represents a different frequency component of the wireless signal. In the example shown, the offline adjustment system selects subcarrier 215 from the total channel response for analysis. Figure 2 The example diagram in FIG. 2 shows a time series 220 of selected subcarriers at the observation time provided by the motion detection system 120. In the diagram showing the time series 220, the horizontal axis shows the sampling timeline (sampling is performed at a rate of one sample per 0.1 second); and the vertical axis shows the amplitude of the channel response data for the selected subcarrier at each sampling time point. The channel response values ​​are typically complex values, and the amplitudes of these complex values ​​are shown in the time series 220.

[0030] exist Figure 2In the example shown, the offline adjustment system 110 identifies certain segments 230 in the time series 220 where disturbances in the signal indicate motion activity. Each identified segment 230 of the time series is labeled with a characteristic distribution (e.g., a Gaussian distribution characterized by a mean μ and a standard deviation σ) that can be characterized by statistical parameters corresponding to the characteristic activity zone. In this example, the time series 220 is taken from example channel response data 130 obtained by the motion detection system 120, where the user is walking in different motion zones in the space. As shown, the time series 220 shows that different spatial zones in a home (or other space) have different time signatures and size distributions (e.g., μ1, σ1 230A, μ2, σ2 230B, μ3, σ3 230C), referred to herein as zone distributions, and each of these spatial zones represents a characteristic activity zone. In some instances, by using the zone distribution information, the offline adjustment system 110 can generate a model of zones with characteristic distributions whose parameters can be learned, and then use the model to detect those zones in the time series. Figure 2 In FIG. 2 , the individual zone distributions 230A, 230B, and 230C corresponding to the characteristic activity zones are represented by the parameters (mean μ and standard deviation σ) of a Gaussian distribution. However, other types of distributions and appropriate parameters may be used, for example based on analysis of data or a model, and may be any distribution. Regardless of the distribution applied, Figure 2 It is shown that the activity in various zones is mapped to a characteristic distribution 230, and the offline adjustment system 110 can learn the distribution to know when activity in a certain motion zone is occurring.

[0031] Figure 3 is a diagram illustrating an example adaptive model 300 used by an offline adjustment system to identify new motion regions. The adaptive model 300 may be executed by a computer system that receives data generated by a motion detection system. For example, the adaptive model 300 may be executed by Figure 1 The offline adjustment system 110 shown may be executed by other types of computer systems.

[0032] In some examples, the offline adjustment system will operate on observed channel response data 310 collected from the motion detection system during the course of a day (e.g., as Figure 1As shown). In an implementation, the offline adjustment system includes a random generator that is seeded in some manner to generate a time series that simulates a time series from data collected for a particular subcarrier. In this example, it is initially assumed that there are only two motion zones Z1, Z2. The initial number of motion zones Z generated by the model 300 can be the number of zones set during installation and initialization of the motion detection system in the space. For example, the motion detection system may include two nodes placed in different areas of a home, and each node initially represents a motion zone. In some cases, the motion is confined to the nodes associated with the motion zone. In some instances, the adaptive model 300 allows additional motion zones to be detected without additional equipment.

[0033] The modeled motion zones Z1, Z2 are each associated with hyperparameters h1, h2 representing characteristic moments of the distribution. In implementation, a random generator generates corresponding model distributions D1, D2 (given by corresponding hyperparameters h1, h2 representing characteristic distributions) representing activities in the motion zones for specific motion zones Z1, Z2. In some instances, the model 300 applies another random variable tau representing a time coordinate. In some cases, tau is a time vector comprising multiple time values. In some implementations, tau is used to determine the switching time between motion zones. The switching time is the point in time when the motion is determined to be transitioning between two motion zones. For example, a walking user may leave one motion zone and enter another. The time vector tau indicates the point in time in the entire time series when switching occurs back and forth between motion zones (e.g., from Z2 to Z1).

[0034] In the example model 300, both the modeled distributions D1 and D2 and tau are fed as inputs into the switch 330. The output of the switch 330 is controlled by tau. For example, the switch causes the distribution D1 or D2 to flow through the switch based on the value and identification of tau. In some instances, when one of the distributions D1 and D2 is routed to the output of the switch 330 based on the value of tau, the output distribution is the distribution observed to be active during the time period of the tau, that is, the modeled data matches the distribution D1 or D2 during the time period specified by tau. In some instances, the active distribution at the time indicated by tau indicates the movement associated with the active distribution. In some implementations, the output distribution generated by the switch 330 represents the estimated channel response data fed as input to the Monte Carlo engine 380. As shown, the Monte Carlo engine 380 is also supplied with observed channel response data 310 from the motion detection environment (e.g., home (or other space)) in which the motion detection system 120 is operating throughout the day. In some cases, the Monte Carlo engine 380 processes the observed data and the estimated data, and adapts the parameters of the stochastic model so that the generated stochastic data is consistent with the observed data to an acceptable degree. For example, the data can be adapted to move in the direction of gradient descent to minimize the error. In some cases, if the estimated data can be made consistent with the observed data by adaptation, the offline adjustment system can determine that the parameters of the different distributions (e.g., D1, D2) used to model the data are statistically accurate. In some implementations, the Monte Carlo engine 380 continuously adapts the parameters of each distribution, adjusts the parameters, regenerates the data, measures the similarity with the observed data, and repeats the process when evaluating the observed data.

[0035] When the Monte Carlo engine 380 shows convergence characteristics (such as when the estimated parameters (e.g., h1, h2) remain stable from one iteration to the next, etc.), the offline adjustment system can stop the model and check the distribution of the new motion zone. In some cases, the examination of tau can show which moments are most likely to trigger a change from one distribution to a different distribution (e.g., another motion zone). For example, using tau values ​​and the unique distribution between each tau, the offline adjustment system can determine instances in which a moving object is displaced to a different motion zone. In some cases, using the distribution of each active zone, the offline adjustment system can determine how different two active zones are. For example, if the active zones are sufficiently separated (e.g., their characteristic distributions are sufficiently different from each other), a new motion zone with a specific distribution can be identified. In some instances, the offline adjustment system can prompt the user (e.g., by providing a notification via an interface of the user device) that movement is detected in a unique motion zone, and ask the user if he wants to mark the zone for better position and motion tracking accuracy.

[0036] Figure 4 is shown for update Figure 3 4 is a block diagram of an example process 400 for adapting hyperparameters in the model 300 shown. The process 400 may be performed, for example, by executing Figure 3 The computer system of the adaptive model 300 shown in the example process 400 is executed. In this example process 400, all components of the random model that can be adjusted by the Monte Carlo engine 380 are shown. For example, a branch 420 including N motion zones Z (e.g., Z1, Z2, Z3) is shown. The number of motion zones N is a hyperparameter, which means that the number of motion zones N is not fixed and can be adapted when new motion zones are detected. Therefore, the number of motion zones N is a random variable that can be adjusted by the Monte Carlo engine 380. In some instances, this allows the Monte Carlo engine 380 to flexibly adapt the number of independent motion zones in the model, and also allows the Monte Carlo engine 380 to define updated motion zones when, for example, the user updates the configuration and arrangement of the home (or other space). In some instances, whenever an update occurs in the home, the performance of the instantaneous locator algorithm for detecting motion and the location of the motion may operate with reduced accuracy until the adaptive model 300 is run during an offline adjustment process (e.g., a regular (e.g., daily) update process). At this point, a new motion zone may be identified and the user may be prompted to confirm the motion zone.

[0037] In some examples, one or more random region generators 440 (each representing one or more motion regions) are used to seed with different distributions D, each associated with some hyperparameter h attached thereto. In some implementations, the seeded distribution generates random samples of the output from each generator 440. These random samples begin to generate estimated channel response data. The switch 330 uses tau to determine the output distribution, such as Figure 3 In implementation, the estimated channel response data is ultimately compared with the actual channel response data (eg, observed channel response data 310). Figure 4 As shown, these two sets of data are used as input to train the adaptive model 300 through the Monte Carlo engine 380. The output of the Monte Carlo engine 380 provides the effective number of motor regions N, the characteristic distribution of each motor region (e.g., μ N , σ N) and the switching time tau associated with the time when the transition to the particular motion zone occurs. In some cases, the data output 490 from the Monte Carlo engine 380 is fed back to the random generator, thereby seeding the generation of randomly distributed data. As described above, the adaptive model is repeated until the modeled data converges with the actual observed channel response data 410 after a certain iteration.

[0038] Figure 5 is a diagram illustrating an example process 500 for aggregating link data. In some examples, Figure 3 and 4 The adaptive model 300 can be adapted to the channel response data (eg, Figure 2 The channel response data 210 in each subcarrier (eg, Figure 2 The channel response data 210 is calculated based on the subcarriers 215 in the signal. In some cases, each subcarrier 215 represents how a single frequency in the signal is affected in amplitude and phase by channel changes caused, for example, by human motion or other types of motion in space. In some implementations, the channel response data 210 is calculated over many subcarriers. For example, in the WiFi 802.11a standard, each 20 MHz portion of the spectrum includes 52 effective subcarriers, all of which can be used to calculate the channel response data. These subcarriers can each be input into and modeled by the Monte Carlo engine 380, which then creates different zone distributions D and transition time estimates tau based on each individual subcarrier.

[0039] In some implementations, each transceiver operating in a MIMO system has many different transmission / reception paths, and each transmitter / receiver (Tx / Rx) pair can be referred to as a link in the MIMO system. In a MIMO system, each transmit antenna and each receive antenna in the MIMO system calculates a channel response over the entire subcarrier range for the link. In some instances, since subcarriers from each Tx / Rx pair can be utilized in a Monte Carlo simulation, many Monte Carlo simulations can be started in parallel. For example, Monte Carlo simulations can be run based on each subcarrier within a link of a Tx / Rx pair, and these simulations can each output an estimate of the zone transition time. In some implementations, the aggregation of the outputs of each subcarrier and, in some instances, each link can be used to form a decision on the final transition time between, for example, two different motion zones.

[0040] Figure 5The example process 500 in FIG. 5 shows a clustering mechanism for improving the detection accuracy and reliability of transition times. For example, in some instances, a Monte Carlo simulation will generate a tau grid 510 for each subcarrier of each different link. This is shown in FIG. Figure 5 520 , where the x-axis represents the zone transition estimate tau (e.g., the spike on the time axis in the case of a zone transition is inferred), and each individual square represents a channel estimate on an individual Tx / Rx link of the MIMO system. In the illustration, the 3D structure of the tau grid 510 for each link is folded into a 2D shape 520, where the tau (or time) dimension remains the same, but the data for subcarriers and links has been flattened. In other words, the tau estimates for all subcarriers within a link are collected and aligned with the tau estimates for all subcarriers in another link until all links are exhausted. In some cases, this aggregation process produces data to allow for voting types across subcarriers and across links. For example, referring to Figure 520 , starting from the bottom and upward, it is observed that the first transition estimate 530 is signaled by two subcarriers (all subcarriers from all links). In this first transition, two votes are collected, which in this example are represented by a circle 540 with a larger diameter. Similarly, next transition estimate 532 is signaled by only a single subcarrier, and subsequent transition estimate 534 is signaled by three subcarriers, transition estimate 536 is signaled by one subcarrier, and transition estimate 538 is signaled by four subcarriers, respectively. Each of these transition estimates results in a circle of different weights, e.g., 542, 544, 546, 548, based on the number of subcarriers signaling the transition. In some examples, the transition estimate 536 is signaled by three subcarriers, e.g., 542, 544, 546, 548, based on the number of subcarriers signaling the transition. Figure 5 The offline adjustment system selects K transitions from the data using weights represented by the diameters of circles 540, 542, 544, 546, 548 in FIG. In some implementations, the system selects the most heavily weighted K transition points (e.g., the transition points with the most votes), and these transition points will be used to notify motion zone transitions and new motion zones.

[0041] Figure 8 An example output 800 from an actual experiment conducted using an example implementation of the offline training system described above is shown. Figure 5 As shown in graph 810, the vertical axis corresponds to the estimated time split (e.g., indicated by tau) determined by the Monte Carlo engine (e.g., Monte Carlo engine 380). The horizontal axis represents the individual subcarrier frequencies for which the estimated channel response data is generated. The time series of the individual subcarriers (e.g., Figure 2 The time sequence 220 of subcarriers 215 shown is input to (e.g., at Figure 3The Monte Carlo simulation in the adaptive model 300 shown in FIG. 1 estimates the transition point based on the aggregated change of the subcarriers. Once all subcarriers have been processed, the estimated transition time ( Figure 2 The estimated time splits in the ) are collected together and passed through a 1-dimensional clustering algorithm, whereby (for example, Figure 5 The data for multiple subcarriers and multiple links is flattened as described in . In this example, the clustering algorithm collects all points along the tau axis and finds the largest cluster, and the largest cluster is then considered to be a valid transition point between zones. Figure 820 shows the channel response from one of the subcarriers in the data. Along with the channel response data, the horizontal solid lines in Figure 820 represent zone locations marked by the user, and the vertical dashed lines show zone transition instances discovered by the offline adjustment system. In this example, these results show that the transition points marked by humans (where the horizontal solid lines gradually increase to indicate new zone locations) and the transition points marked by the machine (represented by the vertical dashed lines) meet with high accuracy.

[0042] Figure 6 600 is a diagram showing a flow chart for identifying a new motion region. The method 600 shown in the flow chart can be performed by an offline adjustment system (e.g., Figure 1 In some cases, Figure 6 One or more operations in the operations shown in are implemented as a process including multiple operations, sub-processing or other types of routines. In some cases, the operations may be combined, performed in another order, performed in parallel, iterated or otherwise repeated or performed in other ways.

[0043] The example method 600 can be used as an adaptive model (e.g., Figure 3 to Figure 4 In an implementation, at 610, observed channel response data (e.g., Figure 2 210 described in the above). The observed channel response data is based on the spatial (e.g., Figure 1 Observed channel response data may be associated with a plurality of subcarriers for each wireless link over a period of time. At 620, for each activity zone (e.g., distributions 230A, 230B, 230C) in the observed channel response data, a plurality of subcarriers for each wireless link (e.g., distributions 230A, 230B, 230C) may be generated for each subcarrier in the plurality of subcarriers for each wireless link. Figure 2 The estimated channel response data for the subcarrier 220) described in FIG.

[0044] In some examples, the active region has a specific distribution in the time series of observed channel response data (e.g., Figure 2 In some cases, generating estimated channel response data for each of the plurality of subcarriers for each wireless link for each activity zone in the observed channel response data includes: generating an estimated channel response data for each of the plurality of subcarriers for each wireless link based on a distribution of the activity zone (e.g., a distribution defined by μ1, σ1, μ2, σ2, μ3, σ3, etc.) and a hyperparameter corresponding to a characteristic of the distribution of the activity zone (e.g., Figure 3 The hyper parameters h1, h2 in the example are used to generate, for each activity area, each of the multiple subcarriers for each wireless link (for example, Figure 2 random data (e.g., Figure 3 In some instances, the number of active regions is a hyperparameter (e.g., Figure 4 N regions described in ).

[0045] At 630, transitions are identified between activity regions in estimated channel response data for each of a plurality of subcarriers of the wireless link (eg, as Figure 3 to Figure 4 In some examples, the time vector includes transitions identified between activity regions in the estimated channel response data for each of the plurality of subcarriers for each wireless link and the time of each transition.

[0046] At 640, new motion regions of the motion detection system are detected based on the estimated channel response data for each of the activity regions and the identified transitions. In some cases, detecting new motion regions in the motion detection system includes analyzing random data generated for each of the activity regions (e.g., Figure 3 D1, D2, etc. in ) and the transitions identified between the various activity zones (e.g., Figure 3 to Figure 4 In some examples, (e.g., by Figure 3 The described tau-based switch 330) determines the activity distribution for a given time period. Then, (e.g., by Figure 3 The Monte Carlo engine 350 in FIG. 3 can adjust the hyperparameters of each distribution and use the adjusted hyperparameters to regenerate random data (such as Figure 4 to adapt the model as described in ).

[0047] In some examples of method 600, it is determined whether the random data is correlated with the observed channel response data. In response to determining that the random data is correlated with the observed channel response data, the offline adjustment system can aggregate the identified transitions for each of the plurality of subcarriers for each of the wireless links and determine that at least one of the identified transitions indicates a new motion region (e.g., such as Figure 5 In instances where it is determined that the random data is not correlated with the observed channel response data, it is possible (for example, by Figure 4 The Monte Carlo engine described in ) is used to adapt the hyperparameters of each distribution, and the adjusted hyperparameters can then be used to regenerate random data (e.g., a random generator generates distribution data D for each motion zone Z) until the random data is correlated with the observed channel response data (as in Figure 4 as described in the model in ).

[0048] In some implementations of method 600, a notification is provided to a user of the motion detection system indicating the new motion region. In response to the notification, a confirmation of the new motion region or a rejection of the new motion region is received from the user. In an example, the confirmation of the new motion region includes a mark for the new motion region provided by the user. In some instances, the location of the new motion region is not associated with the location of a particular node of the motion detection system.

[0049] Figure 7 is a block diagram showing an example offline adjustment apparatus 700. The offline adjustment apparatus 700 may be configured as follows Figure 1 The offline adjustment system 110 described in Figure 7 As shown, the example device 700 includes an interface 730, a processor 710, a memory 720, and a power supply unit 740. In some implementations, the interface 730, the processor 710, the memory 720, and the power supply unit 740 of the offline adjustment device are housed together in a common housing or other assembly. In some implementations, one or more components of the offline adjustment device may be separately housed, for example, in separate housings or other assemblies.

[0050] The example interface 730 can communicate (receive, transmit, or both) signals, such as from the motion detection system 120. For example, the interface 730 can be configured to wirelessly receive data, such as a radio frequency (RF) signal formatted according to a wireless communication standard (e.g., Wi-Fi or Bluetooth). In some cases, the interface 730 can be configured to transmit signals, such as to transmit data to a system, server, or other device. In some implementations, the interface 730 can be configured to receive and transmit data via a wired connection (e.g., via Ethernet, etc.). The interface 730 can be implemented as the example interface 730 shown, or can be implemented in other ways (e.g., using other types of components or subsystems). In some implementations, the interface 730 is configured to receive or obtain observation channel response data from the motion detection system. The interface 730 can also be configured to transmit or provide data to other systems or devices, such as the motion detection system 120, the machine learning system 160, or the user device 140, etc.

[0051] The example processor 710 can execute instructions to generate output data, for example, based on data input. The instructions may include a program, code, script, module, or other type of data stored in the memory 720. Additionally or alternatively, the instructions may be encoded as a preprogrammed or reprogrammable logic circuit, logic gate, or other type of hardware or firmware component or module. The processor 710 may be a general-purpose microprocessor or include a general-purpose microprocessor as a special-purpose coprocessor or other type of data processing device. In some cases, the processor 710 performs high-level operations of the device 700. For example, the processor 710 may be configured to execute or interpret software, scripts, modules, programs, functions, executable programs, or other instructions stored in the memory 720. In some instances, the processor 710 may be configured to execute instructions for performing the steps of the method 600, such as, for example, Figure 6 This allows the off-line adjustment system to detect new motion areas of the motion detection system.

[0052] Example memory 720 may include computer-readable storage media, such as volatile memory devices, non-volatile memory devices, or both. Memory 720 may include one or more read-only memory devices, random access memory devices, buffer memory devices, or combinations of these and other types of memory devices. In some instances, one or more components of the memory may be integrated or otherwise associated with other components of the wireless communication device 700. Memory 720 may store instructions executable by processor 710. For example, the instructions may include instructions for obtaining observed channel response data from a motion detection system. The instructions may also include instructions for the following operations: such as by Figures 3 to 5 Medium or Figure 6One or more of the operations described in the illustrated example process 600 generates estimated channel response data for each activity region in the observed channel response data, identifies transitions between activity regions in the estimated channel response data, and detects new motion regions for a motion detection system based on the estimated channel response data for each activity region and the identified transitions. In some examples, the memory 720 may include one or more instruction sets or modules (including the above instructions) to, for example, perform adaptive region modeling 722 and / or detect new motion regions 724.

[0053] The example power supply unit 740 provides power to other components of the offline adjustment device 700. For example, the other components can operate based on power provided by the power supply unit 740 through a voltage bus or other connection. In some implementations, the power supply unit 740 includes a battery or battery system, such as a rechargeable battery. In some implementations, the power supply unit 740 includes an adapter (e.g., an AC adapter) that receives an external power signal (from an external source) and converts the external power signal into an internal power signal regulated for the components of the external sensor device 700. The power supply unit 740 can include other components or operate in another manner.

[0054] Some of the subjects and operations described in this specification (including the structures disclosed in this specification and their structural equivalents) can be implemented in digital electronic circuits, or in computer software, modules, firmware or hardware, or a combination of one or more thereof, for example, as described for the offline adjustment device 700. Some of the subjects described in this specification can be implemented as one or more computer programs (i.e., one or more modules of computer program instructions) encoded on a computer-readable storage medium for execution by a data processing device or for controlling the operation of a data processing device. A computer-readable storage medium can be or be included in a computer-readable storage device, a computer-readable storage substrate, a random or serial access memory array or device, or a combination of one or more thereof. In addition, although a computer-readable storage medium is not a propagation signal, a computer-readable storage medium can be a source or destination of computer program instructions encoded in an artificially generated propagation signal. A computer-readable storage medium can also be one or more separate physical components or media (e.g., multiple CDs, disks or other storage devices), or be included in one or more separate physical components or media. A computer-readable storage medium can include multiple computer-readable storage media. The computer-readable storage devices may be located in the same location (eg, instructions stored in a single storage device) or in different locations (eg, instructions stored in distributed locations).

[0055] A computer program (also known as a program, software, module, software application, script, or code) can be written in any form of programming language including compiled or interpreted languages, declarative or procedural languages, and can be deployed in any form including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program can, but need not, correspond to a file in a file system. A program can be stored in a portion of a file, where the file keeps other programs or data (e.g., one or more scripts stored in a markup language file) in a single file dedicated to the program, or in multiple coordinated files (e.g., a file for storing a portion of one or more modules, subroutines, or code). A computer program can be deployed to execute on one computer, or on multiple computers located at one site or distributed across multiple sites and interconnected by a communication network.

[0056] Some of the processes and logic flows described in this specification can be performed using one or more programmable processors, which execute one or more computer programs to perform actions by operating on input data and generating output. These processes and logic flows can also be performed by special-purpose logic circuits, and the device can also be implemented as special-purpose logic circuits, such as FPGAs (field programmable gate arrays) or ASICs (application-specific integrated circuits).

[0057] For example, processors suitable for executing computer programs include both general-purpose microprocessors and special-purpose microprocessors, and processors in any type of digital computer. Generally, the processor will receive instructions and data from a read-only memory or a random access memory or both. The elements of a computer may include a processor for performing actions according to instructions and one or more memory devices for storing instructions and data. The computer may also include one or more large-capacity storage devices (e.g., non-magnetic drives (e.g., solid-state drives), disks, magneto-optical disks, or optical disks) for storing data, or may be operably coupled to receive or transmit data relative to the one or more large-capacity storage devices, or both. However, the computer need not have such a device. In addition, the computer may be embedded in other devices (e.g., phones, tablet computers, electrical appliances, mobile audio or video players, game consoles, global positioning system (GPS) receivers, Internet of Things (IoT) devices, machine-to-machine (M2M) sensors or actuators, or portable storage devices (e.g., universal serial bus (USB) flash drives)). Devices suitable for storing computer program instructions and data include all forms of non-volatile memory, media and storage devices, including, by way of example, semiconductor memory devices (e.g., EPROM, EEPROM and flash memory devices, etc.), magnetic disks (e.g., internal hard disks and removable disks, etc.), magneto-optical disks, and CD-ROM and DVD-ROM disks. In some cases, the processor and memory may be supplemented by, or incorporated in, special purpose logic circuitry.

[0058] To provide interaction with a user, operations may be implemented on a computer having a display device (e.g., a monitor or other type of display device) for displaying information to the user and a keyboard and pointing device (e.g., a mouse, a trackball, a stylus, a touch-sensitive screen, or other type of pointing device) through which the user can provide input to the computer. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic, voice, or tactile input). In addition, the computer may interact with the user by sending and receiving documents relative to the device used by the user (e.g., by sending web pages to a web browser on the user's client device in response to a request received from the web browser).

[0059] A computer system may include a single computing device, or multiple computers operating in close proximity to or generally remote from each other and typically interacting through a communications network. A communications network may include one or more of a local area network ("LAN") and a wide area network ("WAN"), an Internet (e.g., the Internet), a network including satellite links, and a peer-to-peer network (e.g., an ad hoc peer-to-peer network). The relationship of client and server may arise from computer programs running on the respective computers and having a client-server relationship to each other.

[0060] In a general aspect of some described examples, a new motion region is detected in a motion detection system.

[0061] In a first example, observed channel response data from a motion detection system is obtained. The observed channel response data is based on wireless signals transmitted on multiple wireless links in space, and the observed channel response data is associated with multiple subcarriers for each wireless link over a period of time. Estimated channel response data for each of the multiple subcarriers for each wireless link is generated for each activity zone in the observed channel response data. In some instances, transitions are identified between activity zones in the estimated channel response data for each of the multiple subcarriers for each wireless link. In some cases, new motion zones for the motion detection system are detected based on the estimated channel response data for each activity zone and the identified transitions.

[0062] In some cases, implementations of the first example may include one or more of the following features. The activity zone corresponds to a segment having a specific distribution in a time series of observed channel response data, and generating estimated channel response data for each of a plurality of subcarriers for each wireless link for each activity zone in the observed channel response data includes: generating random data for each of a plurality of subcarriers for each wireless link for each activity zone based on a distribution of the activity zone and a hyperparameter corresponding to a characteristic of the distribution of the activity zone. Detecting new activity zones in a motion detection system based on the estimated channel response data for each activity zone and the identified transitions includes: analyzing the random data generated for each activity zone and the identified transitions between the zones, determining an active distribution, adjusting the hyperparameters of each distribution, and regenerating random data using the adjusted hyperparameters, determining whether the random data is correlated with the observed channel response data, and in response to determining that the random data is correlated with the observed channel response data, aggregating the identified transitions for each of a plurality of subcarriers for each wireless link, and determining that at least one of the identified transitions indicates a new activity zone.

[0063] In some cases, implementations of the first example may also include one or more of the following features. In response to determining that the random data does not correlate with the observed channel response data, hyperparameters of each distribution are adjusted, and random data is regenerated using the adjusted hyperparameters until the random data correlates with the observed channel response data. The time vector includes: transitions identified between activity zones in the estimated channel response data for each of a plurality of subcarriers for each wireless link and the times of each transition. The number of activity zones is a hyperparameter. A notification is provided to a user of the motion detection system indicating a new motion zone, and in response to the notification, a confirmation of the new motion zone or a rejection of the new motion zone is received from the user. The confirmation of the new motion zone includes a mark for the new motion zone provided by the user. The location of the new motion zone is not associated with the location of a particular node of the motion detection system.

[0064] In some implementations, a computer-readable medium stores instructions that, when executed by a data processing device, are operable to perform one or more operations of the first example. In some implementations, a system (e.g., a communication device, a computer system, a combination thereof) includes one or more processors and a memory storing instructions that, when executed by one or more processors, are operable to perform one or more operations of the first example.

[0065] Although this specification contains many details, these details should not be construed as limitations on the scope of what is claimed, but rather as descriptions of features that are unique to a particular example. Certain features described in this specification in the context of separate implementations may also be combined. Conversely, various features described in the context of a single implementation may also be implemented separately in multiple embodiments, or in any suitable sub-combination.

[0066] A number of embodiments have been described. However, it will be appreciated that various modifications may be made. Accordingly, other embodiments are within the scope of the following claims.

Claims

1. A detection method comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each of the wireless links; detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions; as well as The number of motion regions in the adaptive model is adjusted, wherein the number of motion regions is a hyperparameter of the adaptive model.

2. The detection method according to claim 1, wherein: Each activity region corresponds to a segment in the time series of the observed channel response data, and Among them, generating estimated channel response data for each subcarrier among multiple subcarriers for each wireless link includes: generating random data for each subcarrier among multiple subcarriers for each wireless link based on the distribution of the active area and hyperparameters corresponding to the characteristics of each distribution.

3. The detection method according to claim 2, wherein: Detecting new motion regions based on the estimated channel response data and the identified transitions includes: analyzing the random data and the identified transitions between the various activity zones; Adjusting hyperparameters corresponding to the characteristics of each distribution, and regenerating the random data using the adjusted hyperparameters; determining whether the random data is correlated with the observed channel response data; In response to determining that the random data is correlated with the observed channel response data, aggregating the identified transitions for respective ones of a plurality of subcarriers for respective wireless links; and It is determined that at least one of the identified transitions is indicative of a new motion zone.

4. The detection method according to claim 3, comprising: In response to determining that the random data is not correlated with the observed channel response data, hyperparameters corresponding to characteristics of the respective distributions are adjusted, and the random data is regenerated using the adjusted hyperparameters until the random data is correlated with the observed channel response data.

5. The detection method according to any one of claims 1 to 3, comprising: A time vector is generated indicating transitions identified between activity regions in estimated channel response data for each of a plurality of subcarriers for each of the wireless links.

6. The detection method according to any one of claims 1 to 3, comprising: providing a notification indicating the new motion zone to a user device associated with the motion detection system; as well as In response to the notification, a confirmation of the new motion region or a rejection of the new motion region is received from the user device.

7. The detection method according to claim 6, wherein: The confirmation of the new motion zone includes a mark provided by the user for the new motion zone.

8. The detection method according to any one of claims 1 to 3, wherein The location of the new motion area is not associated with any specific node of the motion detection system.

9. A computer system comprising: one or more processors; as well as a memory including instructions that, when executed by the one or more processors, cause the computer system to perform operations including: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; detecting new motion regions of the motion detection system based on the estimated channel response data and the identified transitions; and The number of motion regions in the adaptive model is adjusted, wherein the number of motion regions is a hyperparameter of the adaptive model.

10. The computer system according to claim 9, wherein: Each activity region corresponds to a segment in the time series of the observed channel response data, and Generating estimated channel response data for each of the multiple subcarriers includes generating random data for each of the multiple subcarriers for each wireless link based on the distribution of the active area and hyperparameters corresponding to characteristics of each distribution.

11. The computer system according to claim 10, wherein: Detecting new motion regions based on the estimated channel response data and the identified transitions includes: analyzing the random data and the identified transitions between the various activity zones; Adjusting hyperparameters corresponding to the characteristics of each distribution, and regenerating the random data using the adjusted hyperparameters; determining whether the random data is correlated with the observed channel response data; In response to determining that the random data is correlated with the observed channel response data, aggregating the identified transitions for respective ones of a plurality of subcarriers for respective wireless links; and It is determined that at least one of the identified transitions is indicative of a new motion zone.

12. The computer system of claim 11, wherein the operations include: In response to determining that the random data is not correlated with the observed channel response data, hyperparameters corresponding to characteristics of the respective distributions are adjusted, and the random data is regenerated using the adjusted hyperparameters until the random data is correlated with the observed channel response data.

13. The computer system of any one of claims 9 to 11, the operations comprising generating a time vector indicating transitions identified between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link.

14. The computer system according to any one of claims 9 to 11, wherein the operations include: providing a notification indicating the new motion zone to a user device associated with the motion detection system; as well as In response to the notification, a confirmation of the new motion region or a rejection of the new motion region is received from the user device.

15. The computer system of claim 14, wherein: The confirmation of the new motion zone includes a mark provided by the user for the new motion zone.

16. The computer system according to any one of claims 9 to 11, wherein: The location of the new motion area is not associated with any specific node of the motion detection system.

17. A non-transitory computer-readable medium comprising instructions that, when executed by a data processing device, cause the data processing device to perform operations comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions; as well as The number of motion regions in the adaptive model is adjusted, wherein the number of motion regions is a hyperparameter of the adaptive model.

18. The non-transitory computer readable medium of claim 17, wherein: Each activity region corresponds to a segment in the time series of the observed channel response data, and Among them, generating estimated channel response data for each subcarrier among multiple subcarriers for each wireless link includes: generating random data for each subcarrier among multiple subcarriers for each wireless link based on the distribution of the active area and hyperparameters corresponding to the characteristics of each distribution.

19. The non-transitory computer readable medium of claim 18, wherein: Detecting new motion regions based on the estimated channel response data and the identified transitions includes: analyzing the random data and the identified transitions between the various activity zones; Adjusting hyperparameters corresponding to the characteristics of each distribution, and regenerating the random data using the adjusted hyperparameters; determining whether the random data is correlated with the observed channel response data; In response to determining that the random data is correlated with the observed channel response data, aggregating the identified transitions for respective ones of a plurality of subcarriers for respective wireless links; and It is determined that at least one of the identified transitions is indicative of a new motion zone.

20. The non-transitory computer readable medium of claim 19, the operations comprising: In response to determining that the random data is not correlated with the observed channel response data, hyperparameters corresponding to characteristics of the respective distributions are adjusted, and the random data is regenerated using the adjusted hyperparameters until the random data is correlated with the observed channel response data.

21. The non-transitory computer-readable medium of any one of claims 17 to 19, the operations comprising generating a time vector indicating transitions identified between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link.

22. The non-transitory computer readable medium of any one of claims 17 to 19, the operations further comprising: providing a notification indicating the new motion zone to a user device associated with the motion detection system; as well as In response to the notification, a confirmation of the new motion region or a rejection of the new motion region is received from the user device.

23. The non-transitory computer readable medium of claim 22, wherein: The confirmation of the new motion zone includes a mark provided by the user for the new motion zone.

24. The non-transitory computer readable medium of any one of claims 17 to 19, wherein: The location of the new motion area is not associated with any specific node of the motion detection system.

25. A detection method comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; as well as detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions, wherein each activity region corresponds to a segment in the time series of the observed channel response data, and Among them, generating estimated channel response data for each subcarrier among multiple subcarriers for each wireless link includes: generating random data for each subcarrier among multiple subcarriers for each wireless link based on the distribution of the active area and hyperparameters corresponding to the characteristics of each distribution.

26. A detection method comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions; as well as A time vector is generated indicating transitions identified between activity regions in estimated channel response data for each of a plurality of subcarriers for each of the wireless links.

27. A detection method comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions; providing a notification indicating the new motion zone to a user device associated with the motion detection system; as well as In response to the notification, a confirmation of the new motion region or a rejection of the new motion region is received from the user device.

28. A detection method comprising: Obtaining observed channel response data from a motion detection system, the observed channel response data being based on wireless signals transmitted over a plurality of wireless links in space, the observed channel response data being associated with a plurality of subcarriers for each wireless link over a period of time; generating estimated channel response data for each of a plurality of subcarriers for each wireless link; identifying transitions between activity regions in estimated channel response data for each of a plurality of subcarriers for each wireless link; as well as detecting a new motion region of the motion detection system based on the estimated channel response data and the identified transitions, The position of the new motion area is not associated with any specific node of the motion detection system. 29 . A computer program product comprising a program, which, when executed by a computer, causes the computer to execute the detection method according to any one of claims 1 to 8 and 25 to 28.

Images