CN113225144B - Wireless sensing method based on channel state information decomposition - Google Patents

Wireless sensing method based on channel state information decomposition Download PDF

Info

Publication number
CN113225144B
CN113225144B CN202110493738.6A CN202110493738A CN113225144B CN 113225144 B CN113225144 B CN 113225144B CN 202110493738 A CN202110493738 A CN 202110493738A CN 113225144 B CN113225144 B CN 113225144B
Authority
CN
China
Prior art keywords
csi amplitude
csi
amplitude
image
information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110493738.6A
Other languages
Chinese (zh)
Other versions
CN113225144A (en
Inventor
颜俊
刘广磊
曹艳华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing University of Posts and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Posts and Telecommunications filed Critical Nanjing University of Posts and Telecommunications
Priority to CN202110493738.6A priority Critical patent/CN113225144B/en
Publication of CN113225144A publication Critical patent/CN113225144A/en
Application granted granted Critical
Publication of CN113225144B publication Critical patent/CN113225144B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/024—Guidance services
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management

Abstract

The invention provides a wireless sensing method based on channel state information decomposition.A static information of a CSI amplitude of a target at each reference position is measured in an off-line stage, and the CSI amplitudes of the target at different positions and different actions are measured simultaneously; constructing a CSI image by using the CSI amplitude, performing classification learning by using CNN (CNN) to obtain a position classification model, simultaneously subtracting the CSI amplitude static information on the position point of the CSI amplitude from the CSI amplitude under each action to obtain CSI amplitude dynamic information, and establishing the CSI amplitude dynamic information image; and performing classification learning by using the CNN to obtain an action classification model. In an online stage, a CSI amplitude image is constructed by the received CSI amplitude, and position estimation is carried out by using a position classification model; and then subtracting the CSI amplitude static information at the estimation position from the received CSI amplitude to construct a CSI amplitude dynamic information image, and then estimating the action by using an action classification model.

Description

Wireless sensing method based on channel state information decomposition
Technical Field
The invention relates to a wireless sensing method based on channel state information decomposition, in particular to a method for realizing position estimation and action identification of a target under the condition of no equipment by mainly utilizing Channel State Information (CSI) of a WiFi signal through a deep learning algorithm, and belongs to the technical field of positioning and navigation.
Background
It is understood that research on the estimation of the position and state of a person has been identified as an important way to perceive the surroundings and to explore the cognitive abilities of humans, which has found widespread use in everyday life, such as environmental monitoring, person tracking, intelligent healthcare, etc., where a more typical application is a fall detection system designed for the elderly living alone. Traditional personnel's position, position estimation mainly rely on wearing relevant check out test set, or rely on the camera control, but these equipment are not the cost is expensive, portable, be intelligent, need personnel to detect scheduling problem inadequately.
With the continuous development of wireless communication, corresponding wireless communication technologies are developed. Research shows that the wireless signal not only can transmit data, but also can wirelessly sense the surrounding environment through signal transformation, such as indoor positioning, personnel identification and the like. Therefore, Channel State Information (CSI) using WiFi signals has received a lot of attention from many researchers under the advantages of low cost and high stability. In a complex environment, signals cannot propagate along straight lines, attenuated signals formed by path loss, scattering, multipath fading and obstacle occlusion are generated, and the attenuated signals can be used as quantitative characteristics of channel frequency response to reflect characteristic information of the environment. Compared with the traditional signal strength receiving mode, the orthogonal frequency division multiplexing multi-channel subcarrier is utilized to ensure that the received signal stability is stronger, and the contained amplitude frequency and phase frequency information is richer.
Through retrieval, Chinese patent with publication number CN112153736A discloses a method for identifying human actions and estimating positions based on channel state information, which comprises the steps of firstly utilizing a convolutional neural network CNN to perform action-based classification learning, extracting depth characteristic information on two axis positions of a CSI image through the CNN, and utilizing a support vector machine SVM to perform regression learning, wherein the action identification precision of the method cannot achieve an ideal effect.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a wireless sensing method based on channel state information decomposition, which overcomes the disadvantages of the prior art, and has the advantages of high positioning accuracy, further improved action recognition rate, and lower implementation cost.
The invention provides a wireless sensing method based on channel state information decomposition, which consists of an off-line stage and an on-line stage. Wherein the off-line phase comprises the steps of:
step 101, when a target is statically positioned at each reference point, acquiring static information of Channel State Information (CSI) amplitude in a WiFi signal; meanwhile, on different reference position points, when the target makes different actions, the wireless network card is used for receiving CSI amplitude information;
102, constructing a CSI amplitude image by utilizing a time domain, a space domain and a frequency domain of CSI amplitude information, and constructing a CSI amplitude dynamic information image under each action by utilizing static and dynamic additivity of the CSI amplitude;
103, using the CSI amplitude image as a fingerprint of a reference position, and using a Convolutional Neural Network (CNN) to perform classification learning to obtain a classification learning model based on the position;
step 104, using the CSI amplitude dynamic information image as a fingerprint of the action, and using a convolutional neural network to perform classification learning to obtain a classification learning model based on the action;
the online phase comprises the following steps:
step 201, constructing a new CSI amplitude image according to the method of the step 102 in the off-line stage from the received CSI amplitude measurement value;
step 202, taking the CSI amplitude image constructed in the step 201 as an input, and bringing the CSI amplitude image into a classification learning model based on the position obtained in the step 103 in an off-line stage to obtain a target position estimation value;
and step 203, subtracting the CSI amplitude static information at the estimation position from the received CSI amplitude measurement value, constructing a CSI amplitude dynamic information image, and estimating the target action by using the action-based classification learning model.
The method of the invention is divided into an off-line stage and an on-line stage. In an off-line stage, measuring static information of CSI amplitude of a target at each reference position, and simultaneously measuring CSI amplitude of the target at different positions and different actions; then, constructing a CSI image by using the CSI amplitude, performing classification learning by using a Convolutional Neural Network (CNN) to obtain a position classification model, simultaneously subtracting the CSI amplitude static information on the position point of the CSI amplitude from the CSI amplitude under each action to obtain CSI amplitude dynamic information, and establishing a CSI amplitude dynamic information image; and performing classification learning by using the CNN to obtain an action classification model. In an online stage, constructing a CSI amplitude image by using the received CSI amplitude, and estimating the position by using a position classification model; and then subtracting the CSI amplitude static information at the estimation position from the received CSI amplitude to construct a CSI amplitude dynamic information image, and then performing action estimation by using an action classification model. According to the invention, the static information of each action is removed by locking the position state, so that the effect of extracting the dynamic CSI information is achieved, the precision of action identification can be displayed and improved, and the method has the advantages of simple structure and low cost.
The further technical scheme of the invention is as follows:
in the offline-stage step 102, constructing the CSI magnitude image specifically includes the following steps:
step 1021, when the target is at different positions and acts, using the first antennas of the receiver and the transmitter as the receiving end and the transmitting end, extracting the amplitude information of the 1 data streams generated, and then extracting N K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 XN K The vector of (a); finally, the obtained vectors are spliced based on lines through the data packet to form N P ×N K Of CSI amplitude matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets;
step 1022, preprocessing the CSI amplitude matrix, namely removing abnormal values in the CSI amplitude matrix through hampel filtering, filtering out impulse noise through median filtering, and retaining matrix edge information;
and 1023, rendering each element in the CSI amplitude matrix into different colors by using a linear mapping method to obtain a CSI amplitude image.
In the offline-stage step 102, constructing a CSI amplitude dynamic information image specifically includes the following steps:
step 1024, subtracting the CSI amplitude static information of the target at the reference position from the CSI amplitude obtained when the target is at a certain reference position and at a certain action, so as to obtain CSI amplitude dynamic information corresponding to the action;
step 1025, when the target is in different positions and actions, the first antennas of the receiver and the transmitter are used as the receiving end and the transmitting end, the amplitude information of the 1 generated data streams is extracted, and then N data streams are processed K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 xN K The vector of (a);and then carrying out line-based splicing operation on the obtained vectors through the data packet to form N P ×N K Of the CSI amplitude dynamic information matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets; preprocessing the CSI amplitude dynamic information matrix, namely removing abnormal values in the CSI amplitude dynamic information matrix through hampel filtering, filtering pulse noise through median filtering, and reserving matrix edge information; and finally, rendering each element in the CSI amplitude dynamic information matrix into different colors by using a linear mapping method to obtain a CSI amplitude dynamic information image.
In the online stage step 203, a specific method for estimating the target action is as follows:
2031, according to the estimated value of the target position obtained in the online stage 202, subtracting the CSI amplitude static information corresponding to the estimated position from the obtained CSI amplitude measured value to obtain CSI amplitude dynamic information, and constructing a new CSI amplitude dynamic information image according to the method of the offline stage 102;
step 2032, the CSI amplitude dynamic information image constructed in step 2031 is taken as input and brought into the classification learning model based on the motion obtained in the offline stage step 104 to obtain the target motion.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
(1) the invention utilizes the channel state information of the WiFi signal to carry out the joint judgment of the position and the action of the personnel, thereby improving the practicability and the convenience of the related technology;
(2) the image preprocessing is carried out through the hampel filtering and the median filtering, and the effect of removing the abnormal value of the image is obvious;
(3) according to the invention, based on the fingerprint database of the standing actions of the personnel, the action information of the personnel is better extracted by subtracting the amplitude difference matrix of the standing actions, background useless information of CSI is removed, the characteristic identification is more favorably carried out, and the action identification rate is obviously improved.
In a word, the method forms the CSI amplitude image and the CSI amplitude dynamic information image by collecting the CSI amplitude information, estimates the target action based on the specific framework of the position and action classification learning of the CNN network, and has the advantages of simple realization and high estimation performance.
Drawings
The invention is further described below with reference to the accompanying drawings.
Fig. 1 is a schematic block diagram of the present invention.
Fig. 2 shows CSI images at different positions in the same operation according to the present invention.
Fig. 3 shows CSI images at the same location and in different operations according to the present invention.
FIG. 4 is a performance diagram of the location classification of the present invention.
Detailed Description
The technical scheme of the invention is further explained in detail by combining the drawings as follows: the present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the protection authority of the present invention is not limited to the following embodiments.
As shown in fig. 1, a wireless sensing method based on channel state information decomposition includes the following specific steps:
step 1, off-line stage
Step 101, perception data acquisition
And when the target is statically positioned at each reference point, measuring static information of the amplitude of Channel State Information (CSI) in the received WiFi signal by using the wireless network card. Meanwhile, when the target makes different actions at different reference position points, the wireless network card is used for receiving CSI amplitude information.
Step 102, constructing a CSI amplitude image and a CSI amplitude dynamic information image
And constructing a CSI amplitude image by utilizing the time domain, the space domain and the frequency domain of the CSI amplitude information. And constructing a CSI amplitude dynamic information image under each action by using the static and dynamic additivity of the CSI amplitude.
The method for constructing the CSI amplitude image specifically comprises the following steps:
step 1021, when the target is at different positions and actions, the first antenna of the receiver and the first antenna of the transmitter are setAs receiving end and transmitting end, 1 data stream generated extracts its amplitude information, and then N K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 XN K The vector of (a); finally, the obtained vectors are spliced on the basis of rows through the data packet to form N P ×N K Of CSI amplitude matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets;
step 1022, preprocessing the CSI amplitude matrix, namely removing abnormal values in the CSI amplitude matrix through hampel filtering, filtering out impulse noise through median filtering, and retaining matrix edge information;
and 1023, rendering each element in the CSI amplitude matrix into different colors by using a linear mapping method to obtain a CSI amplitude image.
The method for constructing the CSI amplitude dynamic information image specifically comprises the following steps:
step 1024, subtracting the CSI amplitude static information of the target at the reference position from the CSI amplitude obtained when the target is at a certain reference position and at a certain action, to obtain CSI amplitude dynamic information corresponding to the action;
step 1025, when the target is in different positions and actions, the first antennas of the receiver and the transmitter are used as the receiving end and the transmitting end, the amplitude information of the 1 generated data streams is extracted, and then N data streams are processed K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 XN K The vector of (a); then, the obtained vectors are spliced on the basis of lines through the data packet to form N P ×N K Of the CSI amplitude dynamic information matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets; preprocessing the CSI amplitude dynamic information matrix, namely removing abnormal values in the CSI amplitude dynamic information matrix through hampel filtering, filtering pulse noise through median filtering, and keeping matrix edge information; and finally, rendering each element in the CSI amplitude dynamic information matrix into different colors by using a linear mapping method to obtain a CSI amplitude dynamic information image.
Step 103, classification learning based on position
And (3) performing classification learning by using the CSI amplitude image constructed in the step (102) as a fingerprint of a reference position and using a Convolutional Neural Network (CNN) to obtain a classification learning model based on the position.
Step 104, Classification learning based on actions
And (4) taking the dynamic image of the CSI amplitude image constructed in the step (102) as a fingerprint of the action, and performing classification learning by using a convolutional neural network to obtain a classification learning model based on the action.
Step 2, on-line stage
Step 201, CSI image construction
And (3) constructing a new CSI amplitude image according to the received CSI amplitude measurement value and the method of the off-line stage step 102, namely constructing the CSI amplitude image by using the time domain, the space domain and the frequency domain of the CSI amplitude measurement value.
Step 202, target position estimation
And taking the CSI amplitude image constructed in the step 201 as an input, and bringing the CSI amplitude image into the classification learning model based on the position obtained in the step 103 in the off-line stage to obtain a target position estimation value.
And step 203, subtracting the CSI amplitude static information at the estimation position from the received CSI amplitude measurement value, constructing a CSI amplitude dynamic information image, and estimating the target action by using the action-based classification learning model.
The specific method for target motion estimation is as follows:
step 2031, according to the target position estimation value obtained in step 202, subtracting the CSI amplitude static information corresponding to the estimated position from the obtained CSI amplitude measurement value to obtain CSI amplitude dynamic information. Meanwhile, a new CSI amplitude dynamic information image is constructed according to the method in step 102, that is, a CSI amplitude dynamic information image under each action is constructed by using static and dynamic additivity of CSI amplitude.
Step 2032, taking the CSI amplitude dynamic information image constructed in step 2031 as input, and substituting into the classification learning model based on motion obtained in step 104 to obtain the target motion.
In fig. 1, each position point performs 5 different actions, receives the CSI values through a computer as a receiver, and then constructs a CSI image through the amplitude-related information of the CSI values. Fig. 2 illustrates two different images of the same motion (fig. 2 a is an image of position 1 of the same motion, and fig. 2 b is an image of position 2 of the same motion). Fig. 3 illustrates two different motion amplitude difference images at the same position (fig. 3 a is the amplitude difference image of motion 1 at the same position, fig. 3 b is the amplitude difference image of motion 2 at the same position). Fig. 4 illustrates a performance analysis of location classification based on a Convolutional Neural Network (CNN) in the present invention, and it can be found from the figure that, as the number of samples increases, the location estimation effect is the best, and when the number of samples is 30000, the location estimation accuracy reaches the highest, which is 98.75%, and thus, the effect of directly performing location estimation first is significant. Comparing the effect of using the amplitude difference to process with the effect of not using the amplitude difference, and finding that the amplitude difference and the effect of the motion process are the same when the number of samples is 7500, the accuracy of motion recognition obtained without using the amplitude difference is 92.95%, and after the amplitude difference is processed, the accuracy is obviously improved to 96.92%, and is improved by 4%, so that the effect of extracting motion characteristics by using the amplitude difference is obvious, and the algorithm performance is very excellent.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can understand that the modifications or substitutions should be included in the scope of the present invention, and therefore, the scope of the present invention should be subject to the protection scope of the claims.

Claims (2)

1. A wireless sensing method based on channel state information decomposition is characterized by comprising an off-line stage and an on-line stage, wherein the off-line stage comprises the following steps:
step 101, when a target is statically positioned at each reference point, acquiring static information of CSI amplitude in a WiFi signal; meanwhile, when the target makes different actions on different reference position points, the CSI amplitude information is received by using the wireless network card;
102, constructing a CSI amplitude image by utilizing a time domain, a space domain and a frequency domain of CSI amplitude information, and constructing a CSI amplitude dynamic information image under each action by utilizing static and dynamic additivity of the CSI amplitude; the method for constructing the CSI amplitude image specifically comprises the following steps:
step 1021, when the target is at different positions and acts, taking the first antennas of the receiver and the transmitter as the receiving end and the transmitting end, extracting the amplitude information of the 1 generated data streams, and then extracting N K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 XN K The vector of (a); finally, the obtained vectors are spliced based on lines through the data packet to form N P ×N K Of CSI amplitude matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets;
step 1022, preprocessing the CSI amplitude matrix, namely removing abnormal values in the CSI amplitude matrix through hampel filtering, filtering out impulse noise through median filtering, and retaining matrix edge information;
1023, rendering each element in the CSI amplitude matrix into different colors by using a linear mapping method to obtain a CSI amplitude image;
the method for constructing the CSI amplitude dynamic information image specifically comprises the following steps:
step 1024, subtracting the CSI amplitude static information of the target at the reference position from the CSI amplitude obtained when the target is at a certain reference position and at a certain action, to obtain CSI amplitude dynamic information corresponding to the action;
step 1025, when the target is in different positions and actions, the first antennas of the receiver and the transmitter are used as the receiving end and the transmitting end, the amplitude information of the 1 generated data streams is extracted, and then N data streams are processed K The data stream amplitude values of the sub-carriers are arranged in a row to form 1 XN K The vector of (a); then, the obtained vectors are spliced on the basis of lines through the data packet to form N P ×N K Of the CSI amplitude dynamic information matrix, where N K 、N P Respectively representing the number of subcarriers and the number of data packets; preprocessing the CSI amplitude dynamic information matrix, namely removing C through hampel filteringAbnormal values in the SI amplitude dynamic information matrix are filtered out of pulse noise through median filtering, and matrix edge information is reserved; finally, rendering each element in the CSI amplitude dynamic information matrix into different colors by using a linear mapping method to obtain a CSI amplitude dynamic information image;
103, using the CSI amplitude image as a fingerprint of a reference position, and using a convolutional neural network to perform classification learning to obtain a classification learning model based on the position;
step 104, using the CSI amplitude dynamic information image as a fingerprint of the action, and using a convolutional neural network to perform classification learning to obtain a classification learning model based on the action;
the online phase comprises the steps of:
step 201, constructing a new CSI amplitude image from the received CSI amplitude measurement value;
step 202, taking the CSI amplitude image constructed in the step 201 as input, and bringing the input into a classification learning model based on positions to obtain a target position estimation value;
and step 203, subtracting the CSI amplitude static information at the estimated position from the received CSI amplitude measured value, constructing a CSI amplitude dynamic information image, and estimating the target action by using the action-based classification learning model.
2. The method for wireless sensing based on channel state information decomposition according to claim 1, wherein in the online stage step 203, a specific method for target action estimation is as follows:
step 2031, according to the target position estimation value obtained in the online stage step 202, subtracting the CSI amplitude static information corresponding to the estimated position from the obtained CSI amplitude measurement value to obtain CSI amplitude dynamic information, and constructing a new CSI amplitude dynamic information image according to the method of the offline stage step 102;
step 2032, the CSI amplitude dynamic information image constructed in step 2031 is taken as input and brought into the classification learning model based on the motion obtained in the offline stage step 104 to obtain the target motion.
CN202110493738.6A 2021-05-07 2021-05-07 Wireless sensing method based on channel state information decomposition Active CN113225144B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110493738.6A CN113225144B (en) 2021-05-07 2021-05-07 Wireless sensing method based on channel state information decomposition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110493738.6A CN113225144B (en) 2021-05-07 2021-05-07 Wireless sensing method based on channel state information decomposition

Publications (2)

Publication Number Publication Date
CN113225144A CN113225144A (en) 2021-08-06
CN113225144B true CN113225144B (en) 2022-07-26

Family

ID=77091214

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110493738.6A Active CN113225144B (en) 2021-05-07 2021-05-07 Wireless sensing method based on channel state information decomposition

Country Status (1)

Country Link
CN (1) CN113225144B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110475221A (en) * 2019-07-15 2019-11-19 南京邮电大学 A kind of personnel's action recognition and location estimation method based on channel state information
CN112135344A (en) * 2020-09-11 2020-12-25 南京邮电大学 CSI (channel State information) and DCNN (distributed channel neural network) -based equipment-free target positioning method
CN112153736A (en) * 2020-09-14 2020-12-29 南京邮电大学 Personnel action identification and position estimation method based on channel state information

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11675066B2 (en) * 2019-08-08 2023-06-13 Syracuse University Motion detection and classification using ambient wireless signals

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110475221A (en) * 2019-07-15 2019-11-19 南京邮电大学 A kind of personnel's action recognition and location estimation method based on channel state information
CN112135344A (en) * 2020-09-11 2020-12-25 南京邮电大学 CSI (channel State information) and DCNN (distributed channel neural network) -based equipment-free target positioning method
CN112153736A (en) * 2020-09-14 2020-12-29 南京邮电大学 Personnel action identification and position estimation method based on channel state information

Also Published As

Publication number Publication date
CN113225144A (en) 2021-08-06

Similar Documents

Publication Publication Date Title
CN110475221B (en) Personnel action identification and position estimation method based on channel state information
CN112153736B (en) Personnel action identification and position estimation method based on channel state information
CN106846729B (en) Tumble detection method and system based on convolutional neural network
CN110502105B (en) Gesture recognition system and method based on CSI phase difference
CN107749143A (en) A kind of indoor occupant fall detection system and method through walls based on WiFi signal
CN108038419B (en) Wi-Fi-based indoor personnel passive detection method
CN103606248A (en) Automatic detection method and system for human body falling-over
Jannat et al. Efficient Wi-Fi-based human activity recognition using adaptive antenna elimination
CN111597991A (en) Rehabilitation detection method based on channel state information and BilSTM-Attention
CN104757968A (en) Statistical evaluation method for intermediate data of paroxysmal conditions of children's absence epilepsy
Cao et al. Wi-Wri: Fine-grained writing recognition using Wi-Fi signals
CN114222202B (en) Environment self-adaptive activity detection method and system based on WiFi CSI
Yang et al. Unsupervised detection of apnea using commodity RFID tags with a recurrent variational autoencoder
Samadh et al. Indoor localization based on channel state information
CN109802973A (en) Method and apparatus for detection flows
Wang et al. RFID & vision based indoor positioning and identification system
CN108268894A (en) A kind of human body based on network Visual Graph is towards detection method
CN112949487A (en) WiFi-based ship cab personnel number detection method and system
CN114048773A (en) Behavior identification method and system based on transfer learning and WiFi
Bahadori et al. ReWiS: Reliable Wi-Fi sensing through few-shot multi-antenna multi-receiver CSI learning
CN114707563A (en) Human body action recognition method based on Wi-Fi channel state information
CN113743374B (en) Personnel identity recognition method based on channel state information respiratory perception
Hu et al. ResFi: WiFi-enabled device-free respiration detection based on deep learning
CN113225144B (en) Wireless sensing method based on channel state information decomposition
CN109316192B (en) Sitting posture detection method and device based on mobile network channel state information

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant