WO2022050461A1 - Method and apparatus by which wireless device cooperates with another device to perform wireless sensing on basis of wireless sensing - Google Patents

Abstract

Proposed are a method and an apparatus by which a wireless device cooperates with another device to perform wireless sensing on the basis of wireless sensing in a wireless LAN system. Particularly, a first device performs capabilities negotiation with a second device. The first device receives first determination information from the second device on the basis of the result of the capabilities negotiation. The first device transmits, to the second device, second determination information that is a result of processing the first determination information. The first determination information is prior information required for identification on the basis of wireless sensing. The second determination information is a result of identification on the basis of wireless sensing.

Description

A method and apparatus for a wireless device to perform wireless sensing in cooperation with other devices based on wireless sensing

The present specification relates to a method of identifying a user or a gesture based on wireless sensing, and more particularly, to a method and apparatus in which a wireless device performs wireless sensing in cooperation with another device.

As wireless technologies and sensing methods advance, many studies are using wireless signals (e.g. WiFi) to detect human activity to detect intrusion, recognize daily activities, monitor vital signs associated with finer-grained motion detection, and It has succeeded in realizing various applications such as gesture recognition for user identification.

These applications can support a variety of domains for smart home and office environments, including safety protection, wellness monitoring/management, smart healthcare, and smart appliance interactions.

Human movement affects radio signal propagation (eg reflection, diffraction, and scattering), providing a great opportunity to capture human movement by analyzing the received radio signal. Researchers can extract ready-to-use signal measurements, or employ frequency-modulated signals for frequency shifting. Due to its low cost and non-intrusion detection properties, wireless-based human activity detection has attracted considerable attention and has become a prominent research area in the past decade.

This specification examines the existing wireless sensing system in terms of basic principle, technology and system architecture. Specifically, it describes how wireless signals can be utilized to facilitate a variety of applications including intrusion detection, room occupancy monitoring, daily activity recognition, gesture recognition, vital sign monitoring, user identification and indoor location. Future research directions and limitations using radio signals for human activity detection are also discussed.

The present specification proposes a method and apparatus for a wireless device to perform wireless sensing in cooperation with another device based on wireless sensing.

An example of the present specification proposes a method for a wireless device to perform wireless sensing in cooperation with another device.

This embodiment proposes a method of identifying a user (or gesture) by performing learning and prediction through mutual cooperation when there are a plurality of devices based on wireless sensing. Through this embodiment, it is possible to implement a system that efficiently collects, learns, and predicts signal patterns suitable for the user's home environment even when multiple devices coexist, creating a new paradigm of IoT (Internet of Things) future smart home devices. There is a new effect that can be produced. It is assumed that first and second devices to be described below are devices based on wireless sensing.

A first device (device) performs capabilities negotiation with a second device (Capabilities Negotiation).

The first device receives first decision information from the second device based on the result of the capability negotiation.

The first device transmits second decision information that is a result of processing the first decision information to the second device.

In this case, the first decision information is preliminary information required for identification based on wireless sensing (soft decision), and the second decision information is a result of identification based on the wireless sensing (hard decision).

According to the embodiment proposed in this specification, by performing a learning and prediction method through collaboration between devices based on wireless sensing, a signal pattern suitable for a user's home environment is efficiently collected, learned, There is a new effect of creating a new paradigm of IoT (Internet of Things) future smart home devices as a predictive system can be implemented.

1 shows an example of a transmitting apparatus and/or a receiving apparatus of the present specification.

2 is a conceptual diagram illustrating the structure of a wireless LAN (WLAN).

3 is a view for explaining a general link setup process.

4 shows a flowchart of a WiFi sensing procedure.

5 shows a flow diagram of a general procedure for sensing human activity via a wireless signal.

6 shows a CSI spectrogram according to a human gait.

7 shows a deep learning architecture for user authentication.

8 shows a problem that occurs when a wireless sensing-based device independently performs a procedure for measuring, processing, and predicting a wireless signal.

9 shows a block diagram of a wireless sensing device.

10 is a block diagram of a functional unit of a wireless sensing device.

11 is a block diagram of a wireless sensing device including an interface.

12 is a diagram illustrating a type of cooperative device.

13 shows an example of a procedure in which a wireless sensing device cooperates to perform learning and prediction.

14 shows various examples of an information delivery method.

15 is a diagram illustrating Example 1 in which predictions are made in AI Cloud and the results are shared.

16 is a diagram illustrating Example 2 in which a representative device predicts and shares a result.

17 is a diagram illustrating Example 3 in which predictions are made in each device and the results are shared.

18 is a flowchart illustrating a procedure in which a wireless device performs wireless sensing in cooperation with another device according to the present embodiment.

19 is a flowchart illustrating a procedure in which a wireless device performs wireless sensing in cooperation with another device according to the present embodiment.

20 shows a modified example of a transmitting apparatus and/or a receiving apparatus of the present specification.

In this specification, “A or B (A or B)” may mean “only A”, “only B” or “both A and B”. In other words, in the present specification, “A or B (A or B)” may be interpreted as “A and/or B (A and/or B)”. For example, “A, B or C (A, B or C)” herein means “only A,” “only B,” “only C,” or “any and any combination of A, B and C. combination of A, B and C)”.

A slash (/) or a comma (comma) used herein may mean “and/or”. For example, “A/B” may mean “and/or B”. Accordingly, “A/B” may mean “only A”, “only B”, or “both A and B”. For example, “A, B, C” may mean “A, B, or C”.

As used herein, “at least one of A and B” may mean “only A”, “only B” or “both A and B”. In addition, in this specification, the expression “at least one of A or B” or “at least one of A and/or B” means “at least one It can be interpreted the same as “at least one of A and B”.

Also, in this specification, “at least one of A, B and C” means “only A”, “only B”, “only C” or “of A, B and C”. any combination of A, B and C”. Also, “at least one of A, B or C” or “at least one of A, B and/or C” means may mean “at least one of A, B and C”.

In addition, parentheses used herein may mean “for example”. Specifically, when displayed as “control information (EHT-Signal)”, “EHT-Signal” may be proposed as an example of “control information”. In other words, “control information” of the present specification is not limited to “EHT-Signal”, and “EHT-Signal” may be proposed as an example of “control information”. Also, even when displayed as “control information (ie, EHT-signal)”, “EHT-Signal” may be proposed as an example of “control information”.

In this specification, technical features that are individually described within one drawing may be implemented individually or simultaneously.

The following examples of the present specification may be applied to various wireless communication systems. For example, the following example of the present specification may be applied to a wireless local area network (WLAN) system. For example, the present specification may be applied to the IEEE 802.11a/g/n/ac standard or the IEEE 802.11ax standard. In addition, this specification may be applied to the newly proposed EHT standard or IEEE 802.11be standard. In addition, an example of the present specification may be applied to the EHT standard or a new wireless LAN standard that is an enhancement of IEEE 802.11be. Also, an example of the present specification may be applied to a mobile communication system. For example, it may be applied to a mobile communication system based on Long Term Evolution (LTE) based on the 3rd Generation Partnership Project (3GPP) standard and its evolution. In addition, an example of the present specification may be applied to a communication system of the 5G NR standard based on the 3GPP standard.

Hereinafter, technical features to which the present specification can be applied in order to describe the technical features of the present specification will be described.

1 shows an example of a transmitting apparatus and/or a receiving apparatus of the present specification.

The example of FIG. 1 may perform various technical features described below. 1 relates to at least one STA (station). For example, the STAs 110 and 120 of the present specification are a mobile terminal, a wireless device, a wireless transmit/receive unit (WTRU), a user equipment (UE), It may also be called by various names such as a mobile station (MS), a mobile subscriber unit, or simply a user. The STAs 110 and 120 in the present specification may be referred to by various names such as a network, a base station, a Node-B, an access point (AP), a repeater, a router, and a relay. In the present specification, the STAs 110 and 120 may be referred to by various names such as a receiving device, a transmitting device, a receiving STA, a transmitting STA, a receiving device, and a transmitting device.

For example, the STAs 110 and 120 may perform an access point (AP) role or a non-AP role. That is, the STAs 110 and 120 of the present specification may perform AP and/or non-AP functions. In this specification, the AP may also be indicated as an AP STA.

The STAs 110 and 120 of the present specification may support various communication standards other than the IEEE 802.11 standard. For example, a communication standard (eg, LTE, LTE-A, 5G NR standard) according to the 3GPP standard may be supported. In addition, the STA of the present specification may be implemented in various devices such as a mobile phone, a vehicle, and a personal computer. In addition, the STA of the present specification may support communication for various communication services such as voice call, video call, data communication, and autonomous driving (Self-Driving, Autonomous-Driving).

In this specification, the STAs 110 and 120 may include a medium access control (MAC) conforming to the IEEE 802.11 standard and a physical layer interface for a wireless medium.

The STAs 110 and 120 will be described based on the sub-view (a) of FIG. 1 as follows.

The first STA 110 may include a processor 111 , a memory 112 , and a transceiver 113 . The illustrated processor, memory, and transceiver may each be implemented as separate chips, or at least two or more blocks/functions may be implemented through one chip.

The transceiver 113 of the first STA performs a signal transmission/reception operation. Specifically, IEEE 802.11 packets (eg, IEEE 802.11a/b/g/n/ac/ax/be, etc.) may be transmitted/received.

For example, the first STA 110 may perform an intended operation of the AP. For example, the processor 111 of the AP may receive a signal through the transceiver 113 , process the received signal, generate a transmission signal, and perform control for signal transmission. The memory 112 of the AP may store a signal (ie, a received signal) received through the transceiver 113 , and may store a signal to be transmitted through the transceiver (ie, a transmission signal).

For example, the second STA 120 may perform an intended operation of a non-AP STA. For example, the transceiver 123 of the non-AP performs a signal transmission/reception operation. Specifically, IEEE 802.11 packets (eg, IEEE 802.11a/b/g/n/ac/ax/be, etc.) may be transmitted/received.

For example, the processor 121 of the non-AP STA may receive a signal through the transceiver 123 , process the received signal, generate a transmission signal, and perform control for signal transmission. The memory 122 of the non-AP STA may store a signal (ie, a received signal) received through the transceiver 123 and may store a signal to be transmitted through the transceiver (ie, a transmission signal).

For example, an operation of a device indicated as an AP in the following specification may be performed by the first STA 110 or the second STA 120 . For example, when the first STA 110 is an AP, the operation of the device marked as AP is controlled by the processor 111 of the first STA 110 , and is controlled by the processor 111 of the first STA 110 . Relevant signals may be transmitted or received via the controlled transceiver 113 . In addition, control information related to an operation of the AP or a transmission/reception signal of the AP may be stored in the memory 112 of the first STA 110 . In addition, when the second STA 110 is an AP, the operation of the device indicated by the AP is controlled by the processor 121 of the second STA 120 and controlled by the processor 121 of the second STA 120 . A related signal may be transmitted or received via the transceiver 123 that is used. In addition, control information related to an operation of the AP or a transmission/reception signal of the AP may be stored in the memory 122 of the second STA 110 .

For example, an operation of a device indicated as a non-AP (or User-STA) in the following specification may be performed by the first STA 110 or the second STA 120 . For example, when the second STA 120 is a non-AP, the operation of the device marked as non-AP is controlled by the processor 121 of the second STA 120, and the processor ( A related signal may be transmitted or received via the transceiver 123 controlled by 121 . In addition, control information related to the operation of the non-AP or the AP transmit/receive signal may be stored in the memory 122 of the second STA 120 . For example, when the first STA 110 is a non-AP, the operation of the device marked as non-AP is controlled by the processor 111 of the first STA 110 , and the processor ( Related signals may be transmitted or received via transceiver 113 controlled by 111 . In addition, control information related to the operation of the non-AP or the AP transmission/reception signal may be stored in the memory 112 of the first STA 110 .

In the following specification (transmission / reception) STA, first STA, second STA, STA1, STA2, AP, first AP, second AP, AP1, AP2, (transmission / reception) Terminal, (transmission / reception) device , (transmitting/receiving) apparatus, a device called a network, etc. may refer to the STAs 110 and 120 of FIG. 1 . For example, without specific reference numerals (transmitting/receiving) STA, first STA, second STA, STA1, STA2, AP, first AP, second AP, AP1, AP2, (transmitting/receiving) Terminal, (transmitting) A device indicated by a /receiver) device, a (transmit/receive) apparatus, and a network may also refer to the STAs 110 and 120 of FIG. 1 . For example, in the following example, an operation in which various STAs transmit and receive signals (eg, PPPDUs) may be performed by the transceivers 113 and 123 of FIG. 1 . In addition, in the following example, the operations of the various STAs generating transmission/reception signals or performing data processing or calculation in advance for the transmission/reception signals may be performed by the processors 111 and 121 of FIG. 1 . For example, an example of an operation of generating a transmission/reception signal or performing data processing or operation in advance for a transmission/reception signal is 1) Determining bit information of a subfield (SIG, STF, LTF, Data) field included in a PPDU /Acquisition/configuration/computation/decoding/encoding operation, 2) time resource or frequency resource (eg, subcarrier resource) used for the subfield (SIG, STF, LTF, Data) field included in the PPDU, etc. operation of determining / configuring / obtaining, 3) a specific sequence (eg, pilot sequence, STF / LTF sequence, SIG) used for the subfield (SIG, STF, LTF, Data) field included in the PPDU operation of determining / configuring / acquiring an extra sequence), etc., 4) a power control operation and / or a power saving operation applied to the STA, 5) an operation related to determination / acquisition / configuration / operation / decoding / encoding of an ACK signal may include In addition, in the following example, various information (eg, field/subfield/control field/parameter/power related information) used by various STAs for determination/acquisition/configuration/computation/decoding/encoding of transmit/receive signals is may be stored in the memories 112 and 122 of FIG. 1 .

The device/STA of the sub-view (a) of FIG. 1 described above may be modified as shown in the sub-view (b) of FIG. 1 . Hereinafter, the STAs 110 and 120 of the present specification will be described based on the sub-drawing (b) of FIG. 1 .

For example, the transceivers 113 and 123 illustrated in (b) of FIG. 1 may perform the same function as the transceivers illustrated in (a) of FIG. 1 . For example, the processing chips 114 and 124 illustrated in (b) of FIG. 1 may include processors 111 and 121 and memories 112 and 122 . The processors 111 and 121 and the memories 112 and 122 illustrated in (b) of FIG. 1 are the processors 111 and 121 and the memories 112 and 122 illustrated in (a) of FIG. ) can perform the same function.

As described below, a mobile terminal, a wireless device, a wireless transmit/receive unit (WTRU), a user equipment (UE), a mobile station (MS), a mobile Mobile Subscriber Unit, user, user STA, network, base station, Node-B, access point (AP), repeater, router, relay, receiving device, transmitting device, receiving STA, transmitting STA, Receiving Device, Transmitting Device, Receiving Apparatus, and/or Transmitting Apparatus means the STAs 110 and 120 shown in the sub-drawings (a)/(b) of FIG. ) may mean the processing chips 114 and 124 shown in FIG. That is, the technical features of the present specification may be performed on the STAs 110 and 120 shown in (a)/(b) of FIG. 1, and the processing chip ( 114 and 124). For example, a technical feature in which a transmitting STA transmits a control signal is that the control signals generated by the processors 111 and 121 shown in the sub-drawings (a)/(b) of FIG. 1 are (a) of FIG. ) / (b) can be understood as a technical feature transmitted through the transceivers 113 and 123 shown in (b). Alternatively, the technical feature in which the transmitting STA transmits the control signal is a technical feature in which a control signal to be transmitted to the transceivers 113 and 123 is generated from the processing chips 114 and 124 shown in the sub-view (b) of FIG. can be understood

For example, the technical feature in which the receiving STA receives the control signal may be understood as the technical feature in which the control signal is received by the transceivers 113 and 123 shown in the sub-drawing (a) of FIG. 1 . Alternatively, the technical feature that the receiving STA receives the control signal is that the control signal received by the transceivers 113 and 123 shown in the sub-drawing (a) of FIG. 1 is the processor shown in (a) of FIG. 111, 121) can be understood as a technical feature obtained by. Alternatively, the technical feature for the receiving STA to receive the control signal is that the control signal received by the transceivers 113 and 123 shown in the sub-view (b) of FIG. 1 is the processing chip shown in the sub-view (b) of FIG. It can be understood as a technical feature obtained by (114, 124).

Referring to (b) of FIG. 1 , software codes 115 and 125 may be included in the memories 112 and 122 . The software codes 115 and 125 may include instructions for controlling the operations of the processors 111 and 121 . Software code 115, 125 may be included in a variety of programming languages.

The processors 111 and 121 or the processing chips 114 and 124 shown in FIG. 1 may include an application-specific integrated circuit (ASIC), other chipsets, logic circuits, and/or data processing devices. The processor may be an application processor (AP). For example, the processors 111 and 121 or the processing chips 114 and 124 illustrated in FIG. 1 may include a digital signal processor (DSP), a central processing unit (CPU), a graphics processing unit (GPU), and a modem (Modem). and demodulator). For example, the processors 111 and 121 or the processing chips 114 and 124 shown in FIG. 1 are a SNAPDRAGON™ series processor manufactured by Qualcomm®, an EXYNOSTM series processor manufactured by Samsung®, and a processor manufactured by Apple®. It may be an A series processor, a HELIOTM series processor manufactured by MediaTek®, an ATOMTM series processor manufactured by INTEL®, or a processor enhanced therewith.

In this specification, uplink may mean a link for communication from a non-AP STA to an AP STA, and an uplink PPDU/packet/signal may be transmitted through the uplink. In addition, in this specification, downlink may mean a link for communication from an AP STA to a non-AP STA, and a downlink PPDU/packet/signal may be transmitted through the downlink.

2 is a conceptual diagram illustrating the structure of a wireless LAN (WLAN).

The upper part of FIG. 2 shows the structure of an infrastructure basic service set (BSS) of the Institute of Electrical and Electronic Engineers (IEEE) 802.11.

Referring to the upper part of FIG. 2 , a wireless LAN system may include one or more infrastructure BSSs 200 and 205 (hereinafter, BSSs). The BSSs 200 and 205 are a set of APs and STAs such as an access point (AP) 225 and a station 200-1 (STA1) that can communicate with each other through successful synchronization, and are not a concept indicating a specific area. The BSS 205 may include one or more combinable STAs 205 - 1 and 205 - 2 to one AP 230 .

The BSS may include at least one STA, the APs 225 and 230 providing a distribution service, and a distribution system (DS) 210 connecting a plurality of APs.

The distributed system 210 may implement an extended service set (ESS) 240 that is an extended service set by connecting several BSSs 200 and 205 . The ESS 240 may be used as a term indicating one network in which one or several APs are connected through the distributed system 210 . APs included in one ESS 240 may have the same service set identification (SSID).

The portal 220 may serve as a bridge connecting a wireless LAN network (IEEE 802.11) and another network (eg, 802.X).

In the BSS as shown in the upper part of FIG. 2 , a network between the APs 225 and 230 and a network between the APs 225 and 230 and the STAs 200 - 1 , 205 - 1 and 205 - 2 may be implemented. However, it may be possible to establish a network and perform communication even between STAs without the APs 225 and 230 . A network that establishes a network and performs communication even between STAs without the APs 225 and 230 is defined as an ad-hoc network or an independent basic service set (IBSS).

The lower part of FIG. 2 is a conceptual diagram illustrating the IBSS.

Referring to the lower part of FIG. 2 , the IBSS is a BSS operating in an ad-hoc mode. Since IBSS does not include an AP, there is no centralized management entity that performs a centralized management function. That is, in the IBSS, the STAs 250-1, 250-2, 250-3, 255-4, and 255-5 are managed in a distributed manner. In IBSS, all STAs (250-1, 250-2, 250-3, 255-4, 255-5) can be mobile STAs, and access to a distributed system is not allowed, so a self-contained network network) is formed.

3 is a view for explaining a general link setup process.

In the illustrated step S310, the STA may perform a network discovery operation. The network discovery operation may include a scanning operation of the STA. That is, in order for the STA to access the network, it is necessary to find a network in which it can participate. An STA must identify a compatible network before participating in a wireless network. The process of identifying a network existing in a specific area is called scanning. Scanning methods include active scanning and passive scanning.

3 exemplarily illustrates a network discovery operation including an active scanning process. In active scanning, an STA performing scanning transmits a probe request frame to discover which APs exist nearby while moving channels, and waits for a response. A responder transmits a probe response frame to the STA that has transmitted the probe request frame in response to the probe request frame. Here, the responder may be an STA that last transmitted a beacon frame in the BSS of the channel being scanned. In the BSS, since the AP transmits a beacon frame, the AP becomes the responder. In the IBSS, the STAs in the IBSS rotate and transmit the beacon frame, so the responder is not constant. For example, an STA that transmits a probe request frame on channel 1 and receives a probe response frame on channel 1 stores BSS-related information included in the received probe response frame and channel) to perform scanning (ie, probe request/response transmission/reception on channel 2) in the same way.

Although not shown in the example of FIG. 3 , the scanning operation may be performed in a passive scanning manner. An STA performing scanning based on passive scanning may wait for a beacon frame while moving channels. The beacon frame is one of the management frames in IEEE 802.11, and is periodically transmitted to inform the existence of a wireless network, and to allow a scanning STA to search for a wireless network and participate in the wireless network. In the BSS, the AP plays a role of periodically transmitting a beacon frame, and in the IBSS, the STAs in the IBSS rotate and transmit the beacon frame. When the STA performing scanning receives the beacon frame, it stores information on the BSS included in the beacon frame and records beacon frame information in each channel while moving to another channel. Upon receiving the beacon frame, the STA may store BSS-related information included in the received beacon frame, move to the next channel, and perform scanning on the next channel in the same manner.

The STA discovering the network may perform an authentication process through step SS320. This authentication process may be referred to as a first authentication process in order to clearly distinguish it from the security setup operation of step S340 to be described later. The authentication process of S320 may include a process in which the STA transmits an authentication request frame to the AP, and in response thereto, the AP transmits an authentication response frame to the STA. An authentication frame used for an authentication request/response corresponds to a management frame.

The authentication frame includes an authentication algorithm number, an authentication transaction sequence number, a status code, a challenge text, a Robust Security Network (RSN), and a Finite Cyclic Group), etc. may be included.

The STA may transmit an authentication request frame to the AP. The AP may determine whether to allow authentication for the corresponding STA based on information included in the received authentication request frame. The AP may provide the result of the authentication process to the STA through the authentication response frame.

The successfully authenticated STA may perform a connection process based on step S330. The association process includes a process in which the STA transmits an association request frame to the AP, and in response, the AP transmits an association response frame to the STA. For example, the connection request frame includes information related to various capabilities, a beacon listening interval, a service set identifier (SSID), supported rates, supported channels, RSN, and a mobility domain. , supported operating classes, TIM broadcast request (Traffic Indication Map Broadcast request), may include information on interworking service capability, and the like. For example, the connection response frame includes information related to various capabilities, status codes, Association IDs (AIDs), support rates, Enhanced Distributed Channel Access (EDCA) parameter sets, Received Channel Power Indicator (RCPI), Received Signal to Noise (RSNI). indicator), mobility domain, timeout interval (association comeback time), overlapping BSS scan parameters, TIM broadcast response, QoS map, and the like.

Thereafter, in step S340, the STA may perform a security setup process. The security setup process of step S340 may include, for example, a process of private key setup through 4-way handshaking through an Extensible Authentication Protocol over LAN (EAPOL) frame. .

As the demand for wireless data traffic increases, WiFi networks grow very rapidly as they offer high throughput and are easy to deploy. Recently, CSI (Channel State Information) measured by a WiFi network is widely used for various sensing purposes. In order to better understand the existing WiFi sensing technology and the future WiFi sensing trend, this specification comprehensively reviews the signal processing technology, algorithm, application, and performance results of WiFi sensing using CSI. Different WiFi sensing algorithms and signal processing technologies have their own advantages and limitations and are suitable for different WiFi sensing applications. This specification classifies CSI-based WiFi sensing applications into three categories: sensing, recognition, and estimation according to whether the output is binary/multi-class classification or numeric. With the development and deployment of new WiFi technologies, there will be more WiFi sensing opportunities where objects can move from humans to the environment, animals and objects.

This specification emphasizes the coexistence of three challenges in WiFi sensing: robustness and generalization, privacy and security, and WiFi sensing and networking. In addition, this specification proposes three future WiFi sensing trends: inter-layer network information integration, multi-device cooperation, and fusion of different sensors to enhance the existing WiFi sensing function and enable new WiFi sensing opportunities.

With the growing popularity of wireless devices, WiFi is growing very rapidly. One of the key technologies for WiFi's success is Multiple-Input Multiple-Output (MIMO), which provides high throughput to meet the growing demands of wireless data traffic. MIMO together with OFDM (Orthogonal Frequency-Division Multiplexing) provides channel state information (CSI) for each transmit/receive antenna pair at each carrier frequency. Recently, CSI measurement of WiFi systems is used for various sensing purposes. WiFi sensing reuses the infrastructure used for wireless communication, making deployment easy and low cost. Also, unlike sensor-based and video-based solutions, WiFi sensing does not interfere with lighting conditions.

CSI refers to how a radio path propagates from a transmitter to a receiver at a specific carrier frequency along multiple paths. For WiFi systems with MIMO-OFDM, CSI is a 3D matrix of complex values representing the amplitude attenuation and phase shift of a multipath WiFi channel.

Time series of CSI measurements can be used for other wireless sensing applications by capturing how radio signals travel through surrounding objects and people in time, frequency, and spatial domains. For example, CSI amplitude fluctuations in the time domain can be used for human presence detection, fall detection, motion detection, activity recognition, gesture recognition, and human, activity, gesture, etc. different patterns that can be used for human identification/authentication. has

CSI phase shift in spatial and frequency domains, i.e., transmit/receive antenna and carrier frequencies, is related to signal transmission delay and direction, which can be used for human location and tracking. The CSI phase shift in the time domain can have other dominant frequency components that can be used to estimate the respiration rate. Various WiFi sensing applications have specific requirements for signal processing techniques and classification/estimation algorithms.

This specification proposes signal processing technologies, algorithms, applications, performance results, challenges, and future trends of WiFi sensing through CSI to increase understanding of existing WiFi sensing technologies and gain insight into future WiFi sensing directions.

4 shows a flowchart of a WiFi sensing procedure.

A WiFi signal (eg, CSI measurement value) including a mathematical model, a measurement procedure, an actual WiFi model, a basic processing principle, and an experimental platform is input from the Input stage 410 . Raw CSI measurements are fed to a signal processing module for noise reduction, signal conversion and/or signal extraction as indicated by the Signal Processing stage 420 .

The pre-processed CSI tracking is supplied as a modeling-based, learning-based or hybrid algorithm, such as the Algorithm stage 430, to obtain an output for various WiFi sensing purposes. Depending on the output type, WiFi sensing can be classified into three categories. In the Application stage 440, the detection/recognition application tries to solve the binary/multi-class classification problem, and the estimation application tries to obtain the quantity value of another task.

5 shows a flow diagram of a general procedure for sensing human activity via a wireless signal.

Specifically, the sensing system is a different sensing method (eg, Received Signal Strength Indicator (RSSI), Channel State Information (CSI), Frequency Modulated Carrier Wave (FMCW), and Doppler shift) based on human activity and The related signal change is first extracted. A series of signal preprocessing procedures (eg, filtering, denoising, and correction) are then employed to mitigate the effects of interference, ambient noise, and system offsets. Finally, unique features are extracted and served as machine learning models to perform human activity detection and recognition.

That is, the human activity sensing procedure of FIG. 5 is as follows.

1) Measurements: Measure RSSI, CSI, Doppler shift, etc. as input values

2) Derived Metrics with Human movements: Signal strength variations, Channel condition variations, Frequency shift associated with human body depth, Frequency shift associated with human moving speed

3) Signal Pre-processing: Noise reduction, Signal Time-Frequency Transform, Signal Extraction

4) Feature Extraction: Extracts user ID features using gait cycle, body speed, and human activity

5) Prediction via Machine/Deep learning: Algorithms

6) Application: User identification prediction model Detection, Recognition, Estimation (Intrusion detection, Room occupancy monitoring, Daily activity recognition, Gesture recognition, Vital signs monitoring, User identification, Indoor localization & tracking)

1. Wireless Sensing, Wi-Fi, Machine Learning

<Background of invention>

The IoT future smart home market is changing from device connection-centric to service-centric, and as a result, the need for AI device-based personalization and automation services is increasing. Wireless sensing-based technology, which is one of the element technologies for IoT service of artificial intelligence devices, is being actively developed. Research on user identification by learning the pattern of this signal using

<Background technology and problems>

In order to mount Wireless Sensing-based User Identification technology on commercial products, pre-training and distributing a model for prediction of data collected in Machine Learning (e.g., learning a model to predict dogs and cats in advance) It is difficult to distribute and predict new images that are not used for training.) Wireless Signal is a commercial product because it is impossible to create and pre-distribute a general model as the signal pattern according to the influence of user movement changes even for the same user depending on the environment. For this purpose, it is necessary to create a model through learning suitable for each environment, but pre-learning using supervised used in existing research requires user participation for the collection and labeling of learning data. The practicality of commercialization is low.

Therefore, the present specification proposes a post-learning automation method for wireless sensing-based user identification.

When learning the wireless sensing signal pattern suitable for each environment, the personal electronic device (PED) personal identification information is used to enable post-learning by collecting the correct answer (eg, label) for learning. Learning methods for post-learning can be applied to several methods such as unsupervised, supervised, semi-supervised, and unsupervised/supervised fusion learning.

Through this embodiment, it is possible to implement a system for predicting by learning a signal pattern suitable for the user's home environment, thereby creating a new paradigm of IoT future smart home devices such as artificial intelligence devices that identify people.

<Example of Wi-Fi CSI-based User Identification Study>

An example of a study for learning/prediction using wireless signal refinement, feature extraction, and machine learning using Wi-Fi CSI is as follows.

1) Signal Pre-processing

-> CSI measurement collection - Collect CSI measurement values of 30~52 subcarriers based on 20MHz bandwidth as many as the number of TX/RX antennas.

-> Denoising - Removes noise from signals using algorithms such as PCA (Principal Component Analysis), phase unwrapping, and band-pass Butterworth filter.

-> Transform to Time-Frequency domain - Spectrogram generation using STFT (Shot-Time Fourier Transform) (refer to Fig. 6) -> The denoising waveform is mixed with the reflection form of the human body, and it can be classified by frequency .

6 shows a CSI spectrogram according to a human gait.

Referring to FIG. 6 , torso reflection and leg reflection are illustrated in a CSI spectogram in a time/frequency domain. In this case, the CSI spectogram has a certain cycle time.

2) Feature Extraction

-> The process of extracting features for user identification learning and prediction

-> Use Gait Cycle Time, Movement (or Torso) Speed, Human Activity, etc.

-> Based on the theory that the gait cycle is unique to each person, it is used as a feature of User Identification

-> Example of body velocity estimation method: using the percentile method used in Doppler Radar

-> Example of Human Activity estimation method: Using time domain features (max, min, mean, skewness, kurtiosis, std), which are low level features of CSI, human movements and contours, frequency domain features (spcetrogram energy, percentile frequency) component, spectrogram energy difference) to predict the movement speed of the torso and legs, and express walking or stationary activities using these features.

3) Machine/Deep Learning based training and prediction

-> Learning and prediction through various machine/deep learning-based algorithms

-> Representative Algorithm

i) Supervised Learning: Using machine learning and deep learning learning algorithms such as decision tree-based machine learning classifier, SVM (Support Vector Machine), Softmax classifier, etc.

i)-1 The predictive model is created only by supervised learning, and the unsupervised learning algorithm is used to construct the layers of the supervised learning model (some studies)

-> Learning method

i) Select Training/Evaluation data at a specific ratio by collecting data under specific environmental conditions for each person (eg, Training data : Evaluation data = 8:2) -> Holdout verification

ii) Training data is trained by manually mapping the correct answer (e.g. Label) for each person and using it as an input for the Machine/Deep learning model.

iii) In some studies, auto feature extraction and clustering are performed using unsupervised learning to increase the degree of freedom in the data collection environment, and then user identification is performed using a supervised learning model (eg, Softmax classifier).

Unsupervised learning is a learning method in which only the problem is studied without teaching the answer (label). According to unsupervised learning, a correct answer is found by clustering (a typical example of unsupervised learning) based on the relationship between variables (eg, recommending a YouTuber, classifying animals).

In contrast, supervised learning is a learning method that teaches and studies answers. Supervised learning is divided into regression and classification. Regression is a learning method that predicts outcomes within a range of continuous data (eg, age 0-100). Classification is a learning method that predicts outcomes within a range of discretely separated data (for example, whether a tumor is malignant or benign).

In addition, semi-supervised learning is a method of simultaneously learning data with answers and data without answers, and it is a learning method that studies a lot of data without answers without discarding them.

7 shows a deep learning architecture for user authentication.

The deep learning architecture of FIG. 7 is an example in which auto feature extraction is performed using an autoencoder for each hidden layer, and softmax classification is used for classification.

Referring to FIG. 7 , the supervised learning model constitutes each hidden layer, and the unsupervised learning model is used only for constructing the corresponding layer. Activity Separation, Activity Recognition, and User Authentication of FIG. 7 are all characteristics obtained by auto feature extraction.

1. Wireless Sensing, Wi-Fi, Machine Learning

<Invention Background>

The IoT future smart home market is changing from device connection-centric to service-centric, and as a result, the need for AI device-based personalization and automation services is increasing. Wireless sensing-based technology, which is one of the element technologies for IoT service of artificial intelligence devices, is being actively developed. Research on human recognition and user identification / gesture identification by learning the pattern of this signal using At this time, the wireless sensing-based device independently performs the procedure of measuring, processing, and predicting a wireless signal such as Wi-Fi CSI (channel state information).

<Prior art and problems>

When several wireless sensing-based devices are together, the need for a protocol that can improve performance through collaborative learning and prediction is increasing. In addition, the need for collaboration for cooperative sensing for devices with restrictions on wireless sensing such as resources is increasing.

In addition, it is necessary to define a procedure for mutual sharing and utilization of various information collected by wireless sensing devices. Capabilities supported may differ depending on the device type (for example, a water purifier can collect only wireless signals, and a refrigerator can learn/predict by collecting wireless signals - there are restrictions such as resources) Devices that do not (including legacy devices) need to learn and predict with the help of peripherals that support higher capabilities.

Accordingly, the present specification proposes a wireless sensing-based cooperative architecture protocol and signaling (Cooperation Architecture Protocol & Signaling) method. Briefly describing the cooperative architecture protocol and signaling, first, 1) When there are several Wireless Sensing devices, Capabilities information can be exchanged with each other through mutual negotiation. 2) For exchanging Hard Decision and Soft Decision information, a representative device may be selected according to the negotiation process. 3) When there are multiple Wireless Sensing devices, Hard Decision and Soft Decision information can be exchanged according to the capabilities of the device (or the device status, burden, etc. may be considered). At this time, the hard decision information is the wireless sensing-based identification decision result, and the soft decision information is the prior information required for wireless sensing-based identification (eg, signal raw data, data that has undergone signal preprocessing, learning and input data for learning, etc.) prior data for prediction). 4) An interface between Wireless PHY/MAC and an application can be defined to exchange information between devices for cooperation (by doing so, hard decision and soft decision information can be exchanged without going to the upper layer).

Through the method proposed in this specification, it is possible to implement a system that efficiently collects, learns, and predicts signal patterns suitable for the user's home environment even when multiple devices coexist, thereby creating a new paradigm of IoT future smart home devices. there is.

The existing protocols based on Wireless Sensing and the existing operation methods will be described as follows. 1) The transmitting device transmits a measurable signal such as Wi-Fi CSI (Channel State Information). 2) The receiving device measures the CSI radio signal sent from the transmitting device. 3) The transmitting/receiving device performs wireless signal pre-processing to refine the collected signal. 4) The transmitting/receiving device performs a process of extracting features for learning and prediction (Feature Extraction). 5) The sending/receiving device divides the data set that has undergone Wireless Signal Pre-processing and Feature Extraction into an appropriate ratio (eg 8:2) and uses the large ratio as the data input for learning, and the remaining data is used for evaluation of the learning model. do.

8 shows a problem that occurs when a wireless sensing-based device independently performs a procedure for measuring, processing, and predicting a wireless signal.

In the existing method, a wireless sensing-based device independently performs a procedure of measuring, processing, and predicting a wireless signal such as Wi-Fi CSI (Channel State Information). In this case, supported capabilities may be different depending on the device type. For example, there may be devices that can only collect wireless signals, and when there are devices that can learn/predict by collecting wireless signals, learning/prediction is impossible for devices that can only collect wireless signals (compatibility problem with legacy products). In addition, in the case of a device capable of learning/predicting by collecting wireless signals, there is a problem that the capability of the device capable of only collecting wireless signals is not known, and that the collected wireless signals cannot be received.

8 illustrates a problem with the existing method as a specific embodiment. First, (1) the AP transmits a wireless signal such as Wi-Fi CSI. (2) TV and air conditioner measure this radio signal. (3) TV includes AI (Artificial Intelligence) function and can learn/predict, but air conditioner does not include AI function, so learning/prediction is impossible. (4) At this time, it is assumed that Paul passes between the AP and the TV/air conditioner. (5) TV measures, learns/predicts radio signals such as CSI and recognizes Paul passing in front of the TV, but the air conditioner does not. (6) The result of Paul's learning information is not transmitted from the TV to the air conditioner, and (7) Paul's wireless signal is not transmitted from the air conditioner to the TV. That is, the device can only operate according to its own capabilities, and since the devices do not cooperate with each other, it is impossible to learn and predict with the help of peripheral devices.

9 shows a block diagram of a wireless sensing device.

9 specifically illustrates a functional unit responsible for a wireless sensing-based cooperative architecture protocol and a procedure for sharing information (Cooperation Architecture Protocol & Information).

10 is a block diagram of a functional unit of a wireless sensing device.

10 specifically shows a Function Block for Wireless Sensing Architecture and Negotiation and Signal exchange.

The functional units shown in FIGS. 9 and 10 may be defined as follows.

First, the Wireless PHY/MAC Driver block 10 serves to exchange information with the PHY/MAC layer of the wireless sensing device. The device discovery block 20 serves to discover peripheral devices. The Capabilities Negotiation block 30 serves to negotiate whether the discovered peripheral devices are wireless sensing or not, representative device settings, and decision methods between devices. The Wireless Sensing block 40 serves to transmit and collect wireless signals such as Wi-Fi CSI (Channel State Information). The Signal Pre-Processing block 50 serves to perform CSI Measurement, Phase Offset Calibration, De-Noising, and the like. The Feature Selection & Extraction block 60 serves to select and extract features for learning and prediction. The Machine/Deep Learning block 70 serves to perform Training & Prediction through various Machine/Deep Learning-based algorithms. The information exchange network unit 80 is a wireless network that transmits and receives information of the Device Discovery block 20 , the Capabilities Negotiation block 30 , and the Wireless Sensing block 40 . AI Cloud (90) must include Deep/Machine Learning (70) function, and a Cloud that can perform some or all of Signal Pre-processing (50) and Feature Selection/Extraction (60) functions depending on the connected device configuration is the server The cloud information exchange network unit 100 is a network for exchanging information between the cloud and the wireless sensing device.

11 is a block diagram of a wireless sensing device including an interface.

11 defines an interface for information exchange with the Wireless PHY / MAC Driver. The Soft Decision Interface (110) communicates with the Wireless PHY/MAC Driver (10) PHY/MAC and the Wireless Sensing (40), Signal Pre-Processing (50), Feature Selection/Extraction (60) information in the Wireless Sensing device. It is an interface between applications. The Hard Decision Interface 120 is an interface between the PHY/MAC and the Application for exchanging the Deep/Machine Learning 70 information in the wireless sensing device with the Wireless PHY/MAC Driver 10 .

Decision method is defined below.

Hard Decision information is data that has gone through Wireless Sensing (40) Data collection, Signal Pre-processing (50), Feature Selection/Extraction (60), Machine/Deep Learning (70) (it can be a prediction result determined by AI) can be

Soft Decision information has three types of information, and Soft Decision 1 may be data (which may be raw data) that has been collected through the wireless sensing (40) process. Soft Decision 2 may be data that has gone through the wireless sensing (40) and signal pre-processing (50) processes (it may be in the form of noise removal through signal pre-processing). Soft Decision 3 may be data (which may be a sensing prediction result through AI) that has gone through the wireless sensing (40), signal pre-processing (50), and feature selection/extraction (60) processes.

Decision method may have one or more of the above definitions according to device capabilities. For example, a device that supports AI can share hard decision and soft decision information with other devices, and a device that does not support AI can share only soft decision information with other devices.

Hereinafter, the type of cooperative device is defined.

12 is a diagram illustrating a type of cooperative device.

Level 1 includes functions of Wireless PHY/MAC Driver (10), Device Discovery (20), Capabilities Negotiation (30), and Wireless Sensing (40). Level 1 Device is defined as a device that can collect Wireless Sensing Raw data. Level 1 device can transmit soft decision (sensing raw data) to other devices and receive hard decision (prediction results) from other devices.

Level 2 includes the function of Signal Pre-processing (50) in the function of Level 1 Device. Level 2 Device is defined as a device capable of noise removal and signal refinement through signal pre-processing of the collected sensing data. Level 2 Device can deliver Soft Decision (Sensing Raw Data or Signal Pre-processed Data) to other devices, and can receive Hard Decision (prediction results) from other devices.

Level 3 includes the function of Feature Selection/Extraction (60) in the function of Level 2 Device. Level 3 Device is defined as a device that can select and extract features from refined sensing data to generate input data for machine learning learning/prediction. Level 3 Device can deliver Soft Decision (Sensing Raw Data, Signal Pre-processed Data, or Input Data for learning/prediction) to other devices, and can receive Hard Decision (prediction results) from other devices.

Level 4 includes the function of Deep/Machine Learning (70) in the function of Level 3 Device. Level 4 Device is defined as a device that can perform machine learning learning/prediction by collecting and preprocessing sensing data. Level 4 Device transmits Soft Decision (Sensing Raw Data, Signal Pre-processed Data, or Input Data for learning/prediction) or Hard Decision (prediction result) to other devices, and receives Hard Decision (prediction result) from other devices. can

AI Cloud can play the role of receiving soft decisions from Level 1~4 devices and delivering hard decisions. Signal Pre-Processing (50), Feature Selection / Extraction (60) can be included, and Machine/Deep Learning (70) must be included (to deliver Hard Decision).

That is, both Level 1 to Level 4 devices can receive a hard decision from other devices (AI Cloud or Level 4 Device), and there is a difference in the soft decision information they deliver for each level (in the case of Level 4 devices, hard decision information is also forwardable).

13 shows an example of a procedure in which wireless sensing devices cooperate to perform learning and prediction. 13 shows a protocol and signaling procedure of a cooperative architecture when Device B provides Soft Decision information to Device A. Referring to FIG.

Referring to FIG. 13 , Device A and Device B may search for and detect a device while transmitting and receiving Device Discovery Request/Response through the Device Discovery 20 and the information exchange network unit 80 . For example, when Device A transmits a Device Discovery Request to Device B, Device B may respond with a Device Discovery Response, and Device A may confirm that Device B exists.

After the discovery procedure, Device A and Device B may perform Capabilities Negotiation while transmitting and receiving Capabilities Negotiation Request/Response/Confirm through the Capabilities Negotiation 30 and the information exchange network unit 80 . For example, when Device A transmits a Capabilities Negotiation Request to Device B, Device B may respond with a Capabilities Negotiation Response, and Device A may send a Capabilities Negotiation Confirm to Device B, thereby completing Capabilities Negotiation.

Through Capabilities Negotiation, Device A and Device B can set 1) whether wireless sensing is supported, 2) decision method, and 3) representative device setting. That is, Capabilities information may be exchanged between devices through Capabilities Negotiation, and a representative device may be set.

Specifically, the device that has completed device discovery transmits a Capabilities Negotiation Request to start Capabilities Negotiation between devices. The device that received the Capabilities Negotiation Request transmits its capabilities (Wireless Sensing support, decision method, etc.) to the Capabilities Negotiation Response to the device that sent the Capabilities Negotiation Request. The device receiving the Capabilities Negotiation Response compares its capabilities with its own capabilities, sets the representative device, and transmits its capabilities (Wireless Sensing support, decision method, and representative device setting) on the Capabilities Negotiation Confirm.

In the capability information exchange, whether wireless sensing is supported or not, a device having capabilities of Level 1 or higher delivers information about supporting wireless sensing to the counterpart. Decision method exchange delivers decision information matching one's own capabilities to the other party according to the decision method definition described above.

When setting the representative device, the AI-supported device may be set as the representative device in preference to the non-AI-supported device. In addition, the higher the level, the higher the ranking can be set as a representative device. In the case of the same level, the better the device's performance (device state, burden, etc.), the more it can be set as a representative device. Among devices that do not support AI, if the device is at the same level, the device connected to the AI Cloud 90 may be set as the representative device.

After Capabilities Negotiation, the device may perform an operation for each level according to the capabilities of each device. That is, Device A and Device B perform 1) transmission and reception of Wireless Sensing Data regardless of their capabilities, and perform one or more of 2) Signal Pre-Processing, 3) Feature Selection / Extraction, and 4) Deep/Machine Learning according to their capabilities. can be done

A device may transmit at least one of 5) Soft Decision or 6) Hard Decision information to another device for cooperation. Soft Decision is one of 1) Sensing Data itself collected through Wireless Sensing Data, 2) Signal Data pre-processed through Signal Pre-Processing, and 3) Input Data pre-processed for machine learning learning/prediction through Feature Selection / Extraction. can be delivered in the form of Hard Decision can be the result of predicting a device with AI Capabilities (Level 4 Device or Device connected to AI Cloud) by fusion of the data it collects or the data acquired in the process of cooperation.

The device may transmit 5) Soft Decision or 6) Hard Decision information through the information exchange network unit 80 . The detailed information delivery method is as follows.

A device can deliver soft decision or hard decision information to a device having a higher level function or a device having the same level function according to the capabilities of each device.

In the case of a device including a Level 1 function, Soft Decision 1 may be delivered to a device including Level 1, 2, 3, and 4 functions and to the AI Cloud 90 .

In the case of a device including a Level 2 function, Soft Decisions 1 and 2 may be delivered to a device including Level 2, 3, and 4 functions and to the AI Cloud (90).

In the case of devices including Level 3 functions, Soft Decisions 1, 2, and 3 may be delivered to devices including Level 3 and 4 functions and to the AI Cloud 90 .

In the case of a device including a Level 4 function, Soft Decision 1, 2, 3, and Hard Decision information may be delivered to a device including a Level 4 function and the AI Cloud 90 .

A device or AI Cloud that includes a Level 4 function with Hard Decision can share Hard Decision information with connected devices.

14 shows various examples of an information delivery method.

According to the left view of FIG. 14 , the Level 1 Device may deliver Soft Decision 1 to Level 1, 2, 3, and 4 Devices and the AI Cloud 90 . Level 2 Device may deliver Soft Decision 1 and 2 to Level 2, 3, 4 Device and AI Cloud 90. Level 3 Device may deliver Soft Decision 1, 2, and 3 to Level 3, 4 Device and AI Cloud (90). The Level 4 Device may deliver Soft Decision 1, 2, 3 and Hard Decision information to the Level 4 Device and the AI Cloud 90 . Level 4 Device or AI Cloud can share Hard Decision information with connected devices.

According to the right view of Fig. 14, an example of information delivery Fig. 1 shows an example of sharing a result (or decision) by predicting in AI Cloud. Example of information delivery Figure 2 shows an example of sharing the result (or decision) by predicting from the representative device. Example of information transfer Figure 3 shows an example of sharing the result (or decision) by predicting each device.

15 is a diagram illustrating Example 1 in which predictions are made in AI Cloud and the results are shared.

Referring to FIG. 15 , the Level 2 Device performs peripheral device discovery ( 20 ), and when two Level 1 devices are detected, it performs Capabilities Negotiation ( 30 ) with the detected Device. The devices determine 1) whether wireless sensing is supported, 2) a decision method, and 3) a representative device through the capabilities negotiation (30). In this case, a Level 2 Device with a higher Level ranking is determined as the representative Device.

A Level 2 Device or AP transmits a measurable signal including CSI Information to Level 1 Devices that have performed Capabilities Negotiation (30) (Wireless Sensing (40)). Level 2 Device and Level 1 Devices measure a measurable signal including CSI information (Wireless Sensing (40)).

The received Level 2 Device and Level 1 Devices collect CSI information (40), and then transmit decision information through the information exchange network unit 80 according to the result of negotiation with the representative device (Level 2 Device).

The representative device processes the decision information received through the information exchange network unit 80 as data and delivers it to the AI Cloud 90 through the cloud information exchange network unit 100 . At this time, if the decision information delivered from the representative device is Soft Decision 1, the sensing raw data or CSI Information 40 collected through the wireless sensing data will be delivered to the AI Cloud 90 through the information exchange network unit 80. can

If the decision information delivered from the representative device is Soft Decision 2, the signal data (50) pre-processed through Signal Pre-Processing to the sensing raw data collected through Wireless Sensing Data or CSI Information (40) is transferred to AI Cloud (90) can be passed on to

After processing the decision information of the devices in the AI Cloud 100, the result (Hard Decision) is delivered to the representative device (Level 2 Device) through the cloud information exchange network unit 100, and the representative device (Level 2 Device) is the other device (Level 1 Devices) to share the result through the information exchange network unit 80 .

Earlier, when AI Cloud 100 received Soft Decision 1 from the representative device, AI Cloud 100 performs Signal Pre-processing(50), Feature Extraction(60), Machine/Deep Learning based training and prediction(70) processes. It can process the passed data and deliver it to the representative device.

When AI Cloud(100) receives Soft Decision 2 from the representative device, AI Cloud(100) processes the data that has gone through the Feature Extraction(60), Machine/Deep Learning based training and prediction(70) process and sends the result to the representative device. can pass

16 is a diagram illustrating Example 2 in which a representative device predicts and shares a result.

Referring to FIG. 16 , the Level 4 Device performs peripheral device discovery 20 , and when a Level 2 Device and a Level 1 Device are detected, the Level 4 Device and Capabilities Negotiation 30 are performed. The devices determine 1) whether wireless sensing is supported, 2) a decision method, and 3) a representative device through the capabilities negotiation (30). At this time, a Level 4 Device with a higher Level ranking is determined as the representative Device.

The Level 4 Device or AP transmits a measurable signal including CSI Information to the Level 4 Device, Level 2 Device, and Level 1 Device that have performed Capabilities Negotiation (30) (Wireless Sensing (40)). Level 4 Device, Level 2 Device, and Level 1 Device measure a measurable signal including CSI Information (Wireless Sensing (40)).

The received Level 4 Device, Level 2 Device, and Level 1 Device collect CSI information (40) and then transmit decision information (Soft Decision 1 or 2) to the information exchange network unit according to the result of negotiation with the representative device (Level 4 Device). pass through (80).

The representative device processes the decision information received through the information exchange network unit 80 and shares the result (Hard Decision) through the information exchange network unit 100 .

In the case of Level 2 Device, the information delivery method is as follows. If the result of Negotiation for Level 2 Device is Soft Decision 1, Level 4 Device that received Soft Decision 1 is subjected to Signal Pre-processing(50), Feature Extraction(60), Machine/Deep Learning based training and prediction(70) It can process the data passed through and deliver the result to Level 2 Device. If the result of Negotiation for Level 2 Device is Soft Decision 2, the Level 4 Device that received Soft Decision 2 processes the data that has gone through the Feature Extraction (60), Machine/Deep Learning based training and prediction (70) process and The result can be sent to the device.

In the case of Level 1 Device, the information delivery method is as follows. As the result of Negotiation for Level 1 Device will be Soft Decision 1, Level 4 Device that received Soft Decision 1 takes the course of Signal Pre-processing(50), Feature Extraction(60), Machine/Deep Learning based training and prediction(70) It can process the data that has been passed through and deliver the result to the Level 1 Device.

17 is a diagram illustrating Example 3 in which predictions are made in each device and the results are shared.

Referring to FIG. 17 , the Level 4 Device performs peripheral device discovery ( 20 ), and when two Level 4 devices are detected, it performs Capabilities Negotiation ( 30 ) with the detected devices. The devices determine 1) whether wireless sensing is supported, 2) a decision method, and 3) a representative device through the capabilities negotiation (30). Since the level ranking of each device is the same, a device with good performance, good traffic condition, connected to the AI Cloud 90, or performing Deep/Machine Learning can be a representative device. In this embodiment, it is assumed that the Level 4 Device (1) is determined as the representative device.

Level 4 Device(1) or AP transmits a measurable signal including CSI Information to Level 4 Device(1), Level 4 Device(2), Level 4 Device(3) that performed Capabilities Negotiation(30) (Wireless Sensing (40)). Level 4 Device(1), Level 4 Device(2), and Level 4 Device(3) measure a measurable signal including CSI Information (Wireless Sensing(40)).

Received Level 4 Device(1), Level 4 Device(2), Level 4 Device(3) collect CSI information (40) (Hard Decision) is shared through the information exchange network unit (80). Alternatively, after exchanging and sharing decision information (Soft Decision) according to the status of Level 4 Device(1), Level 4 Device(2), Level 4 Device(3), the Deep/Machine Learning Device predicts the result to inform the decision information (Hard Decision).

Hereinafter, the above-described embodiment will be described with reference to FIGS. 1 to 17 .

18 is a flowchart illustrating a procedure in which a wireless device performs wireless sensing in cooperation with another device according to the present embodiment.

This embodiment proposes a method of identifying a user (or gesture) by performing learning and prediction through mutual cooperation when there are a plurality of devices based on wireless sensing. Through this embodiment, it is possible to implement a system that efficiently collects, learns, and predicts signal patterns suitable for the user's home environment even when multiple devices coexist, creating a new paradigm of IoT (Internet of Things) future smart home devices. There is a new effect that can be produced. It is assumed that first and second devices to be described below are devices based on wireless sensing.

In step S1810, the first device (device) performs capability negotiation with the second device (Capabilities Negotiation).

In step S1820, the first device receives first determination information from the second device based on the result of the capability negotiation.

In step S1830, the first device transmits second decision information that is a result of processing the first decision information to the second device.

In this case, the first decision information is preliminary information required for identification based on wireless sensing (soft decision), and the second decision information is a result of identification based on the wireless sensing (hard decision).

Before performing the capability negotiation, the first device may perform device discovery to find the second device. When the first device transmits a device discovery request to the second device, the second device may transmit a device discovery response to the first device, through which the first device confirms the existence of the second device can

The first device may exchange capability information with the second device and determine a representative device based on the capability negotiation. The capability information may include whether the wireless sensing is supported and the first determination information.

The first decision information may be determined as one of first to third soft decisions based on the levels of the first and second devices. The first soft decision may be raw data of the radio signal. The second soft determination may be data obtained by pre-processing the radio signal. The third soft determination may be input data extracted from data pre-processed on the radio signal.

The second decision information may be a result learned and predicted based on machine learning or deep learning for the first, second, or third soft decision.

The level of the first and second devices may be determined as one of the first to fourth levels based on the capability information.

When the level of the second device is the first level, the first determination information may be the first soft determination. When the level of the second device is the second level, the first determination information may be determined as the first or second software. When the level of the second device is the third level, the first determination information may be determined as the first or second or third software. When the level of the second device is the fourth level, the first determination information may be determined as the first, second, or third software, or may be the second determination information. That is, the second device sends the first decision information (Soft Decision) or the second decision information (Hard Decision) to a device (the first device) having a function of a higher level or the same level based on the capability information. can pass

The representative device may be determined based on a device level, device performance, or whether it supports or is connected to an Artificial Intelligence Cloud (AI Cloud).

When the first device is determined as the representative device, an information transfer process is as follows. If the first decision information is the first soft decision, the second decision information is data that has undergone a learning and prediction process based on signal preprocessing, feature extraction, and machine learning or deep learning in the first decision information. can When the first decision information is the second soft decision, the second decision information may be data that has undergone a learning and prediction process based on the feature extraction and the machine learning or the deep learning in the first decision information. . When the first decision information is the third soft decision, the second decision information may be data that has undergone a learning and prediction process based on the deep learning in the first decision information. That is, according to the received soft decision information, the first device may transmit result information (second decision information) to the second device through each data processing.

When the first device is connected to the AI cloud, the information delivery process is as follows. The first device may transmit the first determination information to the AI cloud. The first device may receive the second determination information from the AI cloud. The second determination information may be obtained by the AI cloud performing a learning and prediction process based on the machine learning or the deep learning on the first determination information. That is, in the above-described embodiment, after the AI cloud processes the decision information transmitted by the devices, result information (second decision information) may be delivered to the first device. The first device may share result information received from the AI cloud with the second device.

As another example, when both the first and second devices have the same level as the fourth level, each of the first and second devices acquires result information (second decision information) based on the result of the capability negotiation You can share it across devices.

After completing the capability negotiation, the first device may transmit a radio signal including channel state information (CSI) information to the second device. The second device may collect and measure the wireless signal.

The first and second devices may include a wireless PHY and MAC driver, a soft decision interface, and a hard decision interface.

The first decision information may be transmitted to the wireless PHY and MAC driver through the soft decision interface. The second decision information may be transmitted to the wireless PHY and MAC driver through the hard decision interface. Accordingly, the first and second devices may transmit the first and second determination information to the PHY/MAC without going through an upper layer.

When the first device shares the second decision information with the second device, the device can know both the learning through cooperation and the predicted result. The first and second devices may identify a user or a gesture based on the wireless sensing result.

19 is a flowchart illustrating a procedure in which a wireless device performs wireless sensing in cooperation with another device according to the present embodiment.

This embodiment proposes a method of identifying a user (or gesture) by performing learning and prediction through mutual cooperation when there are a plurality of devices based on wireless sensing. Through this embodiment, it is possible to implement a system that efficiently collects, learns, and predicts signal patterns suitable for the user's home environment even when multiple devices coexist, creating a new paradigm of IoT (Internet of Things) future smart home devices. There is a new effect that can be produced. It is assumed that first and second devices to be described below are devices based on wireless sensing.

In step S1910, the second device (device) performs capability negotiation with the first device (Capabilities Negotiation).

In step S1920, first determination information is transmitted to the first device based on the result of the capability negotiation.

In step S1930, the second device receives second decision information, which is a result of processing the first decision information, from the first device.

In this case, the first decision information is preliminary information required for identification based on wireless sensing (soft decision), and the second decision information is a result of identification based on the wireless sensing (hard decision).

Before performing the capability negotiation, the first device may perform device discovery to find the second device. When the first device transmits a device discovery request to the second device, the second device may transmit a device discovery response to the first device, through which the first device confirms the existence of the second device can

The first device may exchange capability information with the second device and determine a representative device based on the capability negotiation. The capability information may include whether the wireless sensing is supported and the first determination information.

The first decision information may be determined as one of first to third soft decisions based on the levels of the first and second devices. The first soft decision may be raw data of the radio signal. The second soft determination may be data obtained by pre-processing the radio signal. The third soft determination may be input data extracted from data pre-processed on the radio signal.

The second decision information may be a result learned and predicted based on machine learning or deep learning for the first, second, or third soft decision.

The level of the first and second devices may be determined as one of the first to fourth levels based on the capability information.

When the level of the second device is the first level, the first determination information may be the first soft determination. When the level of the second device is the second level, the first determination information may be determined as the first or second software. When the level of the second device is the third level, the first determination information may be determined as the first or second or third software. When the level of the second device is the fourth level, the first determination information may be determined as the first, second, or third software, or may be the second determination information. That is, the second device sends the first decision information (Soft Decision) or the second decision information (Hard Decision) to a device (the first device) having a function of a higher level or the same level based on the capability information. can pass

The representative device may be determined based on a device level, device performance, or whether it supports or is connected to an Artificial Intelligence Cloud (AI Cloud).

When the first device is determined as the representative device, an information transfer process is as follows. If the first decision information is the first soft decision, the second decision information is data that has undergone a learning and prediction process based on signal preprocessing, feature extraction, and machine learning or deep learning in the first decision information. can When the first decision information is the second soft decision, the second decision information may be data that has undergone a learning and prediction process based on the feature extraction and the machine learning or the deep learning in the first decision information. . When the first decision information is the third soft decision, the second decision information may be data that has undergone a learning and prediction process based on the deep learning in the first decision information. That is, according to the received soft decision information, the first device may transmit result information (second decision information) to the second device through respective data processing.

When the first device is connected to the AI cloud, the information delivery process is as follows. The first device may transmit the first determination information to the AI cloud. The first device may receive the second determination information from the AI cloud. The second determination information may be obtained by the AI cloud performing a learning and prediction process based on the machine learning or the deep learning on the first determination information. That is, in the above-described embodiment, after the AI cloud processes the decision information transmitted by the devices, result information (second decision information) may be transmitted to the first device. The first device may share result information received from the AI cloud with the second device.

As another example, when both the first and second devices have the same level as the fourth level, each of the first and second devices acquires result information (second decision information) based on the result of the capability negotiation You can share it across devices.

After completing the capability negotiation, the first device may transmit a radio signal including channel state information (CSI) information to the second device. The second device may collect and measure the wireless signal.

The first and second devices may include a wireless PHY and MAC driver, a soft decision interface, and a hard decision interface.

The first decision information may be transmitted to the wireless PHY and MAC driver through the soft decision interface. The second decision information may be transmitted to the wireless PHY and MAC driver through the hard decision interface. Accordingly, the first and second devices may transmit the first and second determination information to the PHY/MAC without going through an upper layer.

When the first device shares the second decision information with the second device, the device can know both the learning through cooperation and the predicted result. The first and second devices may identify a user or a gesture based on the wireless sensing result.

2. Device configuration

20 shows a modified example of a transmitting apparatus and/or a receiving apparatus of the present specification.

Each device/STA of the sub-drawings (a)/(b) of FIG. 1 may be modified as shown in FIG. 20 . The transceiver 630 of FIG. 20 may be the same as the transceivers 113 and 123 of FIG. 1 . The transceiver 630 of FIG. 20 may include a receiver and a transmitter.

The processor 610 of FIG. 20 may be the same as the processors 111 and 121 of FIG. 1 . Alternatively, the processor 610 of FIG. 20 may be the same as the processing chips 114 and 124 of FIG. 1 .

The memory 150 of FIG. 20 may be the same as the memories 112 and 122 of FIG. 1 . Alternatively, the memory 150 of FIG. 20 may be a separate external memory different from the memories 112 and 122 of FIG. 1 .

Referring to FIG. 20 , the power management module 611 manages power for the processor 610 and/or the transceiver 630 . The battery 612 supplies power to the power management module 611 . The display 613 outputs the result processed by the processor 610 . Keypad 614 receives input to be used by processor 610 . A keypad 614 may be displayed on the display 613 . SIM card 615 may be an integrated circuit used to securely store an international mobile subscriber identity (IMSI) used to identify and authenticate subscribers in mobile phone devices, such as mobile phones and computers, and keys associated therewith. .

Referring to FIG. 20 , the speaker 640 may output a sound related result processed by the processor 610 . The microphone 641 may receive a sound related input to be used by the processor 610 .

The technical features of the present specification described above may be applied to various devices and methods. For example, the above-described technical features of the present specification may be performed/supported through the apparatus of FIGS. 1 and/or 20 . For example, the above-described technical features of the present specification may be applied only to a part of FIGS. 1 and/or 20 . For example, the technical features of the present specification described above are implemented based on the processing chips 114 and 124 of FIG. 1 , or implemented based on the processors 111 and 121 and the memories 112 and 122 of FIG. 1 , or , may be implemented based on the processor 610 and the memory 620 of FIG. 20 . For example, the device of the present specification is a device for identifying a user or a gesture based on wireless sensing, wherein the device includes a memory and a processor operatively coupled to the memory, wherein the processor includes a second device Performs capability negotiation with the second device, receives first decision information from the second device based on the result of the capability negotiation, and sends second decision information that is a result of processing the first decision information to the second device 2 to the device. The first determination information is prior information required for identification based on wireless sensing, and the second determination information is a result of identification based on the wireless sensing.

The technical features of the present specification may be implemented based on a CRM (computer readable medium). For example, CRM proposed by the present specification is at least one computer readable medium including at least one computer readable medium including instructions based on being executed by at least one processor.

The CRM, performing capability negotiation with the second device (Capabilities Negotiation); receiving first decision information from the second device based on a result of the capability negotiation; and transmitting second decision information, which is a result of processing the first decision information, to the second device. The instructions stored in the CRM of the present specification may be executed by at least one processor. At least one processor related to CRM in the present specification may be the processors 111 and 121 or the processing chips 114 and 124 of FIG. 1 , or the processor 610 of FIG. 20 . Meanwhile, the CRM of the present specification may be the memories 112 and 122 of FIG. 1 , the memory 620 of FIG. 20 , or a separate external memory/storage medium/disk.

The technical features of the present specification described above are applicable to various applications or business models. For example, the above-described technical features may be applied for wireless communication in a device supporting artificial intelligence (AI).

Artificial intelligence refers to a field that studies artificial intelligence or methodologies that can create it, and machine learning refers to a field that defines various problems dealt with in the field of artificial intelligence and studies methodologies to solve them. do. Machine learning is also defined as an algorithm that improves the performance of a certain task through constant experience.

An artificial neural network (ANN) is a model used in machine learning, and may refer to an overall model having problem-solving ability, which is composed of artificial neurons (nodes) that form a network by combining synapses. An artificial neural network may be defined by a connection pattern between neurons of different layers, a learning process that updates model parameters, and an activation function that generates an output value.

The artificial neural network may include an input layer, an output layer, and optionally one or more hidden layers. Each layer includes one or more neurons, and the artificial neural network may include neurons and synapses connecting neurons. In the artificial neural network, each neuron may output a function value of an activation function for input signals, weights, and biases input through synapses.

Model parameters refer to parameters determined through learning, and include the weight of synaptic connections and the bias of neurons. In addition, the hyperparameter refers to a parameter that must be set before learning in a machine learning algorithm, and includes a learning rate, the number of iterations, a mini-batch size, an initialization function, and the like.

The purpose of learning the artificial neural network can be seen as determining the model parameters that minimize the loss function. The loss function may be used as an index for determining optimal model parameters in the learning process of the artificial neural network.

Machine learning can be classified into supervised learning, unsupervised learning, reinforcement learning, and semi-supervised learning according to a learning method.

Supervised learning refers to a method of training an artificial neural network in a state in which a label for the training data is given, and the label is the correct answer (or result value) that the artificial neural network should infer when the training data is input to the artificial neural network. can mean Unsupervised learning may refer to a method of training an artificial neural network in a state where no labels are given for training data. Reinforcement learning can refer to a learning method in which an agent defined in an environment learns to select an action or sequence of actions that maximizes the cumulative reward in each state.

Among artificial neural networks, machine learning implemented as a deep neural network (DNN) including a plurality of hidden layers is also called deep learning, and deep learning is a part of machine learning. Hereinafter, machine learning is used in a sense including deep learning.

In addition, the above-described technical features can be applied to the wireless communication of the robot.

A robot can mean a machine that automatically handles or operates a task given by its own capabilities. In particular, a robot having a function of recognizing an environment and performing an operation by self-judgment may be referred to as an intelligent robot.

Robots can be classified into industrial, medical, home, military, etc. depending on the purpose or field of use. The robot may be provided with a driving unit including an actuator or a motor to perform various physical operations such as moving the robot joints. In addition, the movable robot includes a wheel, a brake, a propeller, and the like in the driving unit, and may travel on the ground or fly in the air through the driving unit.

In addition, the above-described technical features may be applied to a device supporting extended reality.

The extended reality is a generic term for virtual reality (VR), augmented reality (AR), and mixed reality (MR). VR technology provides only CG images of objects or backgrounds in the real world, AR technology provides virtual CG images on top of images of real objects, and MR technology is a computer that mixes and combines virtual objects in the real world. graphic technology.

MR technology is similar to AR technology in that it shows both real and virtual objects. However, there is a difference in that in AR technology, a virtual object is used in a form that complements a real object, whereas in MR technology, a virtual object and a real object are used with equal characteristics.

XR technology can be applied to HMD (Head-Mount Display), HUD (Head-Up Display), mobile phone, tablet PC, laptop, desktop, TV, digital signage, etc. can be called

The claims described herein may be combined in various ways. For example, the technical features of the method claims of the present specification may be combined and implemented as an apparatus, and the technical features of the apparatus claims of the present specification may be combined and implemented as a method. In addition, the technical features of the method claim of the present specification and the technical features of the apparatus claim may be combined to be implemented as an apparatus, and the technical features of the method claim of the present specification and the technical features of the apparatus claim may be combined and implemented as a method.

Claims (20)

1. in a wireless LAN system

   performing, by a first device, capability negotiation with a second device;

   receiving, by the first device, first decision information from the second device based on a result of the capability negotiation; and

   and transmitting, by the first device, second decision information that is a result of processing the first decision information to the second device,

   The first determination information is prior information required for identification based on wireless sensing, and

   The second determination information is a result of identification based on the wireless sensing

   method.
2. According to claim 1,

   finding, by the first device, the second device by performing device discovery; and

   Further comprising the step of transmitting, by the first device, a radio signal including CSI (Channel State Information) information to the second device

   method.
3. 3. The method of claim 2,

   The method further comprising, by the first device, exchanging capability information with the second device based on the capability negotiation and determining a representative device,

   The capability information includes whether to support the wireless sensing and the first determination information,

   The first decision information is determined as one of first to third soft decisions based on the levels of the first and second devices,

   The first soft decision is raw data of the radio signal,

   The second soft decision is data pre-processed on the radio signal,

   The third soft decision is input data extracted from data pre-processed into the radio signal,

   The second decision information is a result of learning and predicted based on machine learning or deep learning in the first, second, or third soft decision.

   method.
4. 4. The method of claim 3,

   The level of the first and second devices is determined as one of the first to fourth levels based on the capability information,

   When the level of the second device is the first level, the first determination information is determined as the first software,

   When the level of the second device is the second level, the first determination information is determined as the first or second software;

   When the level of the second device is the third level, the first determination information is determined as the first or second or third software;

   When the level of the second device is the fourth level, the first determination information is determined as the first, second, or third software or becomes the second determination information.

   method.
5. 5. The method of claim 4,

   The representative device is determined based on device level, device performance, or whether it supports or is connected to AI Cloud (Artificial Intelligence Cloud).

   method.
6. 6. The method of claim 5,

   When the first device is determined as the representative device,

   If the first decision information is the first soft decision, the second decision information is data that has undergone a signal preprocessing, feature extraction, and learning and prediction process based on the first decision information and the machine learning or deep learning, and ,

   If the first decision information is the second soft decision, the second decision information is data that has undergone a learning and prediction process based on the feature extraction and the machine learning or the deep learning in the first decision information,

   If the first decision information is the third soft decision, the second decision information is data that has undergone a learning and prediction process based on the deep learning in the first decision information.

   method.
7. 6. The method of claim 5,

   When the first device is connected to the AI cloud,

   transmitting, by the first device, the first determination information to the AI cloud; and

   The method further comprising the step of receiving, by the first device, the second determination information from the AI cloud,

   The second decision information is obtained by performing a learning and prediction process based on the machine learning or the deep learning on the first decision information by the AI cloud.

   method.
8. According to claim 1,

   The first and second devices include a wireless PHY and MAC driver, a soft decision interface and a hard decision interface,

   The first decision information is transmitted to the wireless PHY and MAC driver through the soft decision interface,

   The second decision information is transmitted to the wireless PHY and MAC driver through the hard decision interface.

   method.
9. In the first device in the wireless LAN system,

   Memory;

   transceiver; and

   a processor operatively coupled with the memory and the transceiver, the processor comprising:

   perform Capabilities Negotiation with the second device;

   receive first decision information from the second device based on a result of the capability negotiation; and

   Transmitting second decision information, which is a result of processing the first decision information, to the second device,

   The first determination information is prior information required for identification based on wireless sensing, and

   The second determination information is a result of identification based on the wireless sensing

   first device.
10. 10. The method of claim 9,

    the processor searches for the second device by performing device discovery; and

    The processor transmits a radio signal including CSI (Channel State Information) information to the second device

    first device.
11. 11. The method of claim 10,

    The processor exchanges capability information with the second device based on the capability negotiation and determines a representative device,

    The capability information includes whether to support the wireless sensing and the first determination information,

    The first decision information is determined as one of first to third soft decisions based on the levels of the first and second devices,

    The first soft decision is raw data of the radio signal,

    The second soft decision is data pre-processed on the radio signal,

    The third soft decision is input data extracted from data pre-processed into the radio signal,

    The second decision information is a result of learning and predicted based on machine learning or deep learning in the first, second, or third soft decision.

    first device.
12. 12. The method of claim 11,

    The level of the first and second devices is determined as one of the first to fourth levels based on the capability information,

    When the level of the second device is the first level, the first determination information is determined as the first software,

    When the level of the second device is the second level, the first determination information is determined as the first or second software;

    When the level of the second device is the third level, the first determination information is determined as the first or second or third software;

    When the level of the second device is the fourth level, the first determination information is determined as the first, second, or third software or becomes the second determination information.

    first device.
13. 13. The method of claim 12,

    The representative device is determined based on device level, device performance, or whether it supports or is connected to AI Cloud (Artificial Intelligence Cloud).

    first device.
14. 14. The method of claim 13,

    When the first device is determined as the representative device,

    If the first decision information is the first soft decision, the second decision information is data that has undergone a signal preprocessing, feature extraction, and learning and prediction process based on the first decision information and the machine learning or deep learning, and ,

    If the first decision information is the second soft decision, the second decision information is data that has undergone a learning and prediction process based on the feature extraction and the machine learning or the deep learning in the first decision information,

    If the first decision information is the third soft decision, the second decision information is data that has undergone a learning and prediction process based on the deep learning in the first decision information.

    first device.
15. 14. The method of claim 13,

    When the first device is connected to the AI cloud,

    the processor transmits the first determination information to the AI cloud; and

    The processor receives the second determination information from the AI cloud,

    The second decision information is obtained by performing a learning and prediction process based on the machine learning or the deep learning on the first decision information by the AI cloud.

    first device.
16. 10. The method of claim 9,

    The first and second devices include a wireless PHY and MAC driver, a soft decision interface and a hard decision interface,

    The first decision information is transmitted to the wireless PHY and MAC driver through the soft decision interface,

    The second decision information is transmitted to the wireless PHY and MAC driver through the hard decision interface.

    first device.
17. in a wireless LAN system

    performing, by a second device, capability negotiation with the first device;

    transmitting, by the second device, first decision information to the first device based on the result of the capability negotiation; and

    Receiving, by the second device, second decision information that is a result of processing the first decision information from the first device,

    The first determination information is prior information required for identification based on wireless sensing, and

    The second determination information is a result of identification based on the wireless sensing

    method.
18. In the second device in the wireless LAN system,

    Memory;

    transceiver; and

    a processor operatively coupled with the memory and the transceiver, the processor comprising:

    perform capability negotiation with the first device;

    send first decision information to the first device based on the result of the capability negotiation; and

    Receive second decision information, which is a result of processing the first decision information, from the first device,

    The first determination information is prior information required for identification based on wireless sensing, and

    The second determination information is a result of identification based on the wireless sensing

    second device.
19. In at least one computer-readable recording medium comprising an instruction based on being executed by at least one processor,

    performing capability negotiation with a second device;

    receiving first decision information from the second device based on a result of the capability negotiation; and

    Transmitting second decision information that is a result of processing the first decision information to the second device,

    The first determination information is prior information required for identification based on wireless sensing, and

    The second determination information is a result of identification based on the wireless sensing

    recording medium.
20. A device in a wireless LAN system, comprising:

    Memory; and

    a processor operatively coupled with the memory, the processor comprising:

    perform capabilities negotiation with the second device;

    receive first decision information from the second device based on a result of the capability negotiation; and

    Transmitting second decision information, which is a result of processing the first decision information, to the second device,

    The first determination information is prior information required for identification based on wireless sensing, and

    The second determination information is a result of identification based on the wireless sensing

    Device.

Images