CN108489509A - A kind of non-contact single step-recording method and system based on commercial Wi-Fi - Google Patents
A kind of non-contact single step-recording method and system based on commercial Wi-Fi Download PDFInfo
- Publication number
- CN108489509A CN108489509A CN201810573883.3A CN201810573883A CN108489509A CN 108489509 A CN108489509 A CN 108489509A CN 201810573883 A CN201810573883 A CN 201810573883A CN 108489509 A CN108489509 A CN 108489509A
- Authority
- CN
- China
- Prior art keywords
- csi
- wave
- crest
- commercial
- noise
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C22/00—Measuring distance traversed on the ground by vehicles, persons, animals or other moving solid bodies, e.g. using odometers, using pedometers
- G01C22/006—Pedometers
Abstract
The invention discloses a kind of non-contact single step-recording methods and system, method based on commercial Wi Fi to include：Participant carries out repeating motion according to preset action request, and the CSI signals of participant are acquired by commercial Wi Fi；The sum of the mean absolute deviation that d (t) is all amplitude datas in a time slide window is defined, the noise level in the t periods is sought according to d (t), noise signal is screened out from CSI signals；The static component that each subcarrier in CSI signals is rejected by your filter of Han Consulting, other noises are removed using Sa Weiciji Gray's filters；Single signal is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis；Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：It is detected according to four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss.System includes：The receiving terminal of microprocessor, the transmitting terminal of WiFi and WiFi.
Description
Technical field
The present invention relates to computer networks, are related specifically to feature extraction, and Data Mining more particularly to one kind are based on
The non-contact single step-recording method and system of commercial Wi-Fi.
Background technology
With the development of wireless technology, wireless signal can be not only used for transmission data, can also be used to perceive environment,
Direct projection, reflection and scattering occurs on the wireless signal of WiFi equipment generation object different around under indoor environment, finally arrives
Up to receiving device, therefore wireless signal carries the information of ambient enviroment.
By establishing the relationship of the wave character of channel state information and goal task in receiving device physical layer, Ke Yijin
Row indoor positioning, Activity recognition, security monitoring and medical monitoring etc..Previous Activity recognition system, they are generally using photograph
Camera, wearable sensor and software radio carry out pursuit movement information, although these equipment captures action is accurate
Degree is high, but expensive, and universality is low.
Other systems then utilize the method for machine learning to CSI (the Channel State in commercial WiFi equipment
Information radio channel status informations) signal is trained, finally corelation behaviour is identified using model, but model
Not only time consumption for training, and it is larger to the dependence of feature, so the use of trained method being difficult to develop lightweight, and be good for
Strong user interface.
The present invention will use conventional commercial Wi-Fi equipment to carry out step number estimation, it is cheap, and universality is high；In skill
Unsupervised method will be used to estimate the step number of single movement in art, it efficiently, and is not necessarily to training.
Invention content
The present invention provides a kind of non-contact single step-recording method and system based on commercial Wi-Fi, by the present invention in that
It is handled with a series of signal and handles Wi-Fi signal with data mining technology, realizing can be right using conventional Wi-Fi equipment
Single people carries out meter step, described below：
A kind of non-contact single step-recording method based on commercial Wi-Fi, the method includes：
Participant carries out repeating motion according to preset action request, and the CSI that participant is acquired by commercial Wi-Fi believes
Number；
The sum of the mean absolute deviation that d (t) is all amplitude datas in a time slide window is defined, is asked according to d (t)
The noise level N (t) in the t periods is taken, noise signal is screened out from CSI signals；
The static component that each subcarrier in CSI signals is rejected by your filter of Han Consulting, is filtered using Sa Weiciji-Gray
Wave device further removes other noises；
Single signal is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis；
Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：According to
Four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss are detected.
Wherein,
Wherein, ap(n) indicate that the data packet index corresponding to subcarrier p is the amplitude at n, P is sub-carrier indices
Maximum value (being herein 90), N are the set of all data packet indexes in sliding window, and L is the length of sliding window, and E is sliding
The mean value of all data packet amplitudes in window.
Further, four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss are respectively：
(1) corrugation pitch：
Interval time between step when based on normal person's running in place, the minimum 0.32s of corrugation pitch is set, maximum spacing is
0.7s；
(2) crest height：
The maximum speed of leg when based on normal person's running in place is arranged the minimum 0.6m/s of crest height, is up to 1.3m/
s；
(3) wave crest protrusion：
The difference of the maximum speed and minimum speed of leg and a large amount of experimental result, are arranged wave when based on normal person's jog in place
Overshooting goes out minimum 0.11m/s, is up to 0.92m/s；
(4) wave peak width：
List leg, which is raised to, when based on normal person's jog in place falls the undergone time, herein the reference standard of wave peak width
For the half of crest height, in conjunction with abundant experimental results, it is minimum 0.1s that the value, which is arranged, is up to 0.5s.
Another embodiment, a kind of non-contact single step counting system based on commercial Wi-Fi, the system comprises：Microprocessor
The receiving terminal of device, the transmitting terminal of WiFi and WiFi,
The transmitting terminal and receiving terminal of the WiFi is put on the ground, and transmitting terminal and receiving terminal are point-blank；
It collects on the receive side after completing CSI data, CSI data are sent to microprocessor by receiving terminal by ICP/IP protocol
On device, microprocessor handles CSI data by MATLAB；
That is, rejecting the static component of each subcarrier in CSI signals by your filter of Han Consulting, Sa Weiciji-lattice are used
Thunder filter further removes other noises；One is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis
Signal；Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：According to wave crest
Four spacing, crest height, wave crest protrusion, wave peak width constraintss are detected.
The advantageous effect of technical solution provided by the invention is：
1, the present invention can be realized indoor meter using conventional Wi-Fi equipment and walk, and cheap and universality is high.
2, data processing method proposed by the present invention, can be used for other field, have good versatility；The present invention
New approaches, new concept are provided for the exploitation of immersion game station.
3, on the one hand the meaning of meter step is the action of specification running in place, on the other hand can increase running in place itself
Interest, body-building is become into a kind of amusement game, for example, can in advance to use the system that the present invention designs carry out the time and
Step number is set, and then completes scheduled step number at the appointed time, white-collar job person can be played in office game in 3 minutes with
It releives pressure, kinsfolk can usually carry out the match of jog in place step number to promote emotion using the scrappy time.
4, it is contemplated that the present invention may be use with medical rehabilitation, such as：Patient does quantitative movement daily, to reach the mesh of rehabilitation
's.
Description of the drawings
Fig. 1 is a kind of flow chart of the non-contact single step-recording method based on commercial Wi-Fi；
Fig. 2 is contrast schematic diagram before and after CSI amplitude noise reductions；
Wherein, (a) is original CSI signals；(b) it is the CSI signals after noise reduction.
Linear dependence schematic diagrames of the Fig. 3 between CSI subcarriers；
Fig. 4 is the relation schematic diagram because of subnumber and factor accumulation contribution rate；
Wherein, (a) is the contribution rate of accumulative total of common factor number；(b) it is the influence of common factor coefficient sub-carrier.Scheme (a)
In 1-4 be respectively common factor number.
Fig. 5 is Maximum-likelihood estimation and results of factor analysis schematic diagram；
Wherein, (a) is common factor number when being 2, maximum likelihood estimation of the common factor for each subcarrier；(b) be because
The amplitude of CSI signals after son analysis.
Fig. 6 is step number estimated result schematic diagram；
Fig. 7 is Current speed frequency distribution histogram.
Fig. 8 is the structural schematic diagram of immersion games system.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, embodiment of the present invention is made below further
It is described in detail on ground.
Embodiment 1
A kind of non-contact single step-recording method based on commercial Wi-Fi, this method include：
1) participant carries out repeating motion according to preset action request, and the CSI that participant is acquired by commercial Wi-Fi believes
Number；
2) the sum of the mean absolute deviation that d (t) is all amplitude datas in a time slide window is defined, according to d (t)
The noise level N (t) in the t periods is sought, noise signal is screened out from CSI signals；
3) static component that each subcarrier in CSI signals is rejected by your filter of Han Consulting, uses Sa Weiciji-Gray
Filter further removes other noises；
4) dimension-reduction algorithm of factorial analysis is used to extract single signal from the CSI signals after noise reduction；
5) it carries out time frequency analysis, feature extraction and wave crest to single signal to detect, medium wave peak detection is specially：Root
It is detected according to four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss.
In conclusion the embodiment of the present invention is handled by using a series of signal handles Wi-Fi with data mining technology
Signal, meter step can be carried out using conventional Wi-Fi equipment to single people by realizing.
Embodiment 2
The embodiment of the present invention proposes a kind of non-contact single step-recording method based on commercial Wi-Fi, referring to Fig. 1, the party
Method includes the following steps：
One, CSI signals are acquired by commercial Wi-Fi
1) using two commercial Wi-Fi equipments as transmitting terminal and receiving terminal；
Such as：Router persistently emits wireless signal as transmitting terminal, another equipment is for example：Laptop, as
Receiving terminal, it is continuous to receive CSI signals.
When specific implementation, the embodiment of the present invention is not limited the model of above-mentioned device, as long as above-mentioned function can be realized
Device.
2) when game starts, participant firstly the need of body erect, open with shoulder with wide by both feet, both arms body both sides from
It is so sagging；
3) and then one leg is bent, and foot is raised up to knee height, homonymy arm swing backward, offside arm is put forward
It is dynamic；Another one leg is slight curving, and foot stays on the ground, and weight concentrates on toe；
4) leg lifted is put back into initial position, then lifts another one leg, repeat the above action, both sides are alternately repeated action
To step number is recommended, (when specific implementation, the embodiment of the present invention is not limited step number, according in practical application set
It is fixed).
During participant plays, WiFi signal is reflected by participant generates unique, fluctuating change
CSI signals, in receiving terminal since WiFi signal is influenced by multipath effect, so games system can utilize signal processing skill
Art obtains the characteristic information of participant's jog in place, such as；The movement velocity of leg, the variation characteristic etc. of leg signal, finally
Games system estimates the step number of single participant's jog in place using the characteristic information of these jog in place.
Two, motion detection
Amplitude information is extracted from the WiFi signal collected first, then needs monitoring to jog and whether starts.In reality
It is observed that the fluctuation of the time series of CSI signals caused by people's movement is bigger than background noise, in order to measure during testing
Change this fluctuation, defines the sum of the mean absolute deviation that d (t) is all amplitude datas in a time slide window：
Wherein, ap(n) indicate that the data packet index corresponding to subcarrier p is the amplitude at n, P is sub-carrier indices
Maximum value (being herein 90), N are the set of all data packet indexes in sliding window, and L is the length of sliding window, and E is sliding
The mean value of all data packet amplitudes in window.
Due to when indoor nobody moves, the fluctuation of CSI signals is mainly caused by noise, and noise level with
Time is slowly varying, so tracking the variation of noise level using the algorithm of dynamic threshold.
First, the mean absolute deviation d (t) in a time slide window, the embodiment of the present invention are calculated according to formula (1)
The time slide window used is 200 sampled points, and about 0.2s uses index sliding average in this time slide window
To update the noise level N (t) in the t periods：
N (t)=(1- αn)N(t-1)+αn×d(t) (2)
Wherein, αnFor constant coefficient.
Herein, factor alpha is arranged in the embodiment of the present inventionnIt is 0.15.Once or terminate movement when, the time of CSI signals
Violent fluctuation can occur at this moment for sequence, and in addition combined with experimental data, it is noise that monitor value, which is arranged, in the embodiment of the present invention
At 3 times of horizontal updated value N ' (t), it is believed that jogging starts, and and then the embodiment of the present invention can carry out at noise reduction data
Reason.
Three, Noise reducing of data
Include from the CSI data obtained in first part：Static component, low-frequency disturbance and pulse noise, on the one hand
It is asynchronous by clock, caused by the interference of radio wave and the variation of transmitting terminal transmission power；On the other hand, in actual environment
In, people, due to the shake of parts of body, can also make to be mixed into different degrees of low frequency noise in CSI data during jogging
And high frequency noise, the extraction for existing for motion feature of these noises increase difficulty, this requirement, which will be not only effectively removed, makes an uproar
Sound, and to maintain waveform not to be changed as far as possible after denoising, traditional IIR (infinite impulse response) low-pass filtering can not
It maintains waveform constant well after the filtering, herein and is not suitable for, since pulse noise band is roomy in CSI data, energy
Height, so bandpass filtering can not be effectively removed noise.
In order to solve these problems, the embodiment of the present invention is picked using Hampel Filter (Han Consulting's that filter) first
Except the static component of each subcarrier in CSI data specifically can be divided into following two step：
1) big window is set, and the Hampel Filter of small threshold value obtain the static component of each subcarrier；
Wherein, big window is set, and the concrete operation step of the Hampel Filter of small threshold value is those skilled in the art institute
Known, the embodiment of the present invention does not repeat this.
2) above-mentioned static component is subtracted in original CSI data；Then, using Savizky-Golay filter (Sas
Wei Ciji-Gray's filter) further remove other noises；
Wherein, Savizky-Golay filter be it is a kind of in time domain based on Local Polynomial least square fitting
The step of filtering method can keep CSI amplitude signal waveforms not to be changed while removing noise, specific filtering is this
Well known to field technology personnel, the embodiment of the present invention does not repeat this.
When specific implementation, it can also realize that above-mentioned step, the present invention are real using other filters or filtering method
It is without limitation to apply example.
Four, single signal extraction
It is handled by data above, the embodiment of the present invention has obtained 60 row comprising two pairs of antennas after noise reduction process
CSI subcarrier datas necessarily contain the information of many overlappings in these data, in addition if divided each subcarrier
Analysis, not only process is cumbersome, but also each subcarrier can not reflect complete running information, in order to remove redundancy and
It polymerize all subcarrier informations in certain several dimension, with comprehensive reflection running process, the embodiment of the present invention uses Factor minute
(process of specific dimensionality reduction is known to those skilled in the art, the embodiment of the present invention to solve the problems, such as this for the dimension-reduction algorithm of analysis
This is not repeated).
Traditional PCA (principal component analysis) although and factorial analysis is all dimension-reduction algorithm, specifically, the two will solve
Certainly the problem of, theoretical foundation, method of estimation, the specific various aspects such as calculate have an essential distinction, it is popular for PCA be analysis dimension
Spend attribute main component expression, and factorial analysis be analyze attribute in common portion expression, factorial analysis more suitable for
The concealed structure of Mining Multidimensional subcarrier behind, in addition, classical nonlinear reductive dimension algorithm such as ISOMAP (Isometric Maps algorithm),
LLE (Local Liner Prediction), MVU (maximum difference deployment algorithm) etc., it is unfavorable on the one hand since time complexity is higher
In timely responding to for system, on the other hand, under conditions of ignoring noise, amplitude of the CSI different sub-carriers under the same time
Meet linear relationship, so nonlinear dimension-reduction algorithm is not appropriate for using herein.
Five, step number is estimated
A, time frequency analysis
Indoors in environment, wireless device (i.e. transmitting terminal) emits radiofrequency signal, and radiofrequency signal is by multipath, different
It is reflected on object, finally reaches receiving terminal, therefore these radiofrequency signals carry the information in relation to external environment, by dividing
The radiofrequency signal received is analysed, action, breathing etc. can be monitored, specifically, when people jogs in situ, different body parts
Movement velocity is different, and each section body part can be by signal reflex, and the signal of these reflections mixes to form CSI waveforms,
Waveform carries all information of jogging, and effective running feature how is extracted from CSI waveforms, is that one faced chooses
War, it is known that when people jogs in situ, different movements of parts of the body speed are different, reflected through remarkable different body parts
Wireless signal there is different frequencies, formally, the relationship of speed of CSI frequencies and people's movement meets following relational expression：
F=2v/ λ (3)
Wherein, v indicates that movement velocity, f indicate that CSI frequencies, λ indicate the wavelength of wireless signal.Therefore in order to detach original place
The different frequency CSI signals of different body part reflections, the embodiment of the present invention will be extracted by factorial analysis during jogging
Single running signal carry out time frequency analysis.
The embodiment of the present invention carries out time frequency analysis, the window size of FFT using Short Time Fourier Transform technology to CSI waveforms
Determine the temporal resolution and frequency resolution of Short Time Fourier Transform, big window correspond to high frequency resolution and it is small when
Between resolution ratio, so selection one suitable window be even more important.
Since people is when jogging, the corresponding CSI frequency ranges of speed that leg moves up and down are in 30-40Hz, the variation of speed
In 10ms, therefore the length (FFT size) that the embodiment of the present invention selects to do Fast Fourier Transform (FFT) is 512 sampled points, sliding
The step-length of window is 16 sampled points, and temporal resolution can reach 10.6ms, and frequency resolution can reach 2.92Hz, in this way
Time frequency resolution for tracking running signal variation it is most suitable.
Particularly, the embodiment of the present invention is analyzed using energy spectral density (Power Spectral Density) under frequency domain
The amplitude information of CSI, PSD convert the time series of each subcarriers of CSI to its Energy distribution in frequency domain, it can know
There is very strong signal energy not under which frequency, formally, length of time series is under i-th of common factor of N
PSD can be calculated as follows：
Herein, ciIndicate the CSI amplitude sequence vectors under i-th of common factor, in order to build more life-like spectrogram,
The embodiment of the present invention selects preceding 50 energy spectral densities point to represent the energy of 0-146Hz.
B, feature extraction
After Short Time Fourier Transform, the embodiment of the present invention obtains the data of 3 dimensions：Time, frequency and
PSD.It is known that when people jogs in situ, the speed of leg is all faster than the movement velocity at other positions of body, then how
It obtains under jog in place situation, the maximum value of leg movement velocity
In order to solve this problem, the embodiment of the present invention carries out the extraction of leg frequency using method of percentiles, for given
Frequency f, percentile is defined as follows：
Herein, P (f, t) indicates that PSDs of the FFT at time t, frequency f, Per (f, t) indicate FFT at time t, frequency
Energy accumulation percentage less than f, fmaxIndicate that the maximum value of frequency, the two meet f≤fmax。
Estimation for leg speed, experimental setup percentile meet Per (f, t) >=90%.By method of percentiles, divide
It separates out the corresponding CSI frequency estimations of leg speed and then curve is smoothed using Robust Loess, due to
The relationship of CSI frequencies and the speed of people's movement meets formula f=2v/ λ, so just having been obtained by equivalence transformation as shown in Figure 6
Leg velocity estimation curve, what the maximum value of leg speed indicated is single leg from being raised to peak and fall back to surface process again
The maximum value of middle leg speed.
C, wave crest detects
Identification of the algorithm of Previous work medium wave peak monitoring for wave crest, is based primarily upon at 2 points, and crest value first is more than it
The sampled value of two consecutive points.Secondly the distance between setting adjacent peaks have to be larger than a certain threshold value, meet two above item
The point of part is wave crest, although this method can remove some spurious peaks such as the 9th wave crest point, such method is still
Spurious peaks, the 24th wave crest point as shown in FIG. 6 are will produce, although the point meets the two above conditions, its wave crest
The protrusion (Peak Prominence) of peak height (Peak Height) and wave crest is simultaneously unsatisfactory for real standard.
In order to remove spurious peaks, general scene and abundant experimental results of the embodiment of the present invention based on jog in place, again
Following constraints is increased for wave crest：
(1) corrugation pitch (Peak Distance)：
Interval time between step when based on normal person's running in place, it is minimum that corrugation pitch is arranged in the embodiment of the present invention
0.32s, maximum spacing are 0.7s.
(2) crest height (Peak Height)：
The minimum 0.6m/ of crest height is arranged in the maximum speed of leg when based on normal person's running in place, the embodiment of the present invention
S is up to 1.3m/s.
(3) wave crest protrusion (Peak Prominence)：
The difference of the maximum speed and minimum speed of leg and a large amount of experimental result, the present invention when based on normal person's jog in place
Embodiment is arranged wave crest and protrudes minimum 0.11m/s, is up to 0.92m/s.
(4) wave peak width (Peak Width)：
List leg, which is raised to, when based on normal person's jog in place falls the undergone time, herein the reference standard of wave peak width
For the half of crest height, in conjunction with abundant experimental results, it is minimum 0.1s that the value, which is arranged, in the embodiment of the present invention, is up to
0.5s。
The embodiment of the present invention removes spurious peaks using above four wave crest features, then counts in leg velocity estimation curve
Legal wave crest estimated value of the number as common factor servant's jog in place step number, the embodiment of the present invention is by asking
The average value for having wave crest number under common factor, the estimated value of the step number as people this jog in place.
In conclusion the embodiment of the present invention is handled by using a series of signal handles Wi-Fi with data mining technology
Signal, meter step can be carried out using conventional Wi-Fi equipment to single people by realizing, and meet a variety of need in practical application
It wants.
Embodiment 3
Below in conjunction with the accompanying drawings in the embodiment of the present invention 1 and 2 effect and effect be shown.
This example provides the embodiment of invention for based on CSI data processings, is as follows：
Using a laptop as WiFi access points, i.e. transmitting terminal, another laptop as receiving terminal,
Two notebooks have been respectively mounted Intel 5300NIC and Ubuntu 14.04LTS desktop edition systems, and receiving terminal has 3 antennas, hair
Penetrating end has 3 antennas, and it is a wavelength (5.2cm) often to hold the distance between 3 antennas, and on straight line, WiFi
Transmitting terminal and receiving terminal put on the ground, for the two at a distance of 3m, the transmission rate of data packet is 1024HZ, transmitting terminal and receiving terminal
Point-blank, the link of transmission is operated on 165 channels that frequency range is 5.825GHz, and the embodiment of the present invention selects 5GHz's
Frequency range without select 2.4GHz frequency range the reason of be 5GHz frequency ranges wavelength it is shorter, short wavelength has higher point to movement velocity
Resolution.After using Linux CSI tool to collect and complete CSI data on the receive side, receiving terminal is by ICP/IP protocol by CSI
Data, which are sent to, to be configured on Intel i7-5600U 2.6GHz computers, and CSI data are handled finally by MATLAB.
When the present invention proceeds by monitoring to activity first, according to formula (2), once monitor that activity starts, it will be right
Data carry out noise reduction process.
Fig. 2 presents an example of Noise reducing of data, before and after 30 sub-carrier amplitude Noise reducing of data of CSI
Waveform can be seen that original CSI amplitude datas high frequency noise and static component containing there are many, be carried using the embodiment of the present invention
After the noise-reduction method gone out, static component is removed efficiently with most of noise, in addition from the whole shape of signal, width
In terms of degree and smoothness, illustrates the noise-reduction method of the embodiment of the present invention, not only effectively remove in CSI amplitude datas
Noise, while can also preferably maintain the waveform of original CSI amplitude datas.In next processing, which will use
In the extraction of single signal.
After the completion of noise reduction, the embodiment of the present invention carries out dimensionality reduction to the data usage factor analysis after noise reduction, and Fig. 3 shows 60
Linear relationship between row subcarrier, it can be seen that have extremely strong linearly related pass in same root antenna between subcarrier
It is that correlativity minimum is also greater than 0.3 between different antennae, so this method is located using linear dimension-reduction algorithm factorial analysis
60 row CSI subcarrier datas of reason, the number c for initializing common factor first is the number of antenna pair, is herein 2 (c≤60).So
Afterwards using 60 row subcarrier datas as input, and by Factor load-matrix estimation calculate common factor accumulation contribution rate come
Determine the number of common factor, the offline accumulation contribution rate that the present invention uses is 85%, if the contribution rate of accumulative total because of subnumber is more than
This is offline, then sets common factor number as current value, otherwise increase common factor number, until offline more than the contribution rate of setting.
Fig. 4, Fig. 5 illustrate the processing procedure of usage factor analysis, and Fig. 4 (a) is presented because subnumber and factor accumulation are contributed
The relationship of rate can see, and with the increase of factor number, the accumulation contribution rate of the factor is rising rapidly, when being 2 because of subnumber,
The accumulation contribution rate of the factor 1 and the factor 2 is to 91.77%, and later with the increase because of subnumber, factor contribution rate of accumulative total slowly increases
It grows and tends towards stability, so Systematic selection is 2 because of subnumber.
Fig. 4 (b) present because subnumber be 2 when, the relationship of CSI subcarriers and common factor, for preceding 30 subcarriers, the
The load of two common factors is larger, and illustrate the reflection of second common factor is 30 row subcarrier informations of first antenna,
It can be construed to first antenna factor, for rear 30 subcarriers, the load of first common factor is larger, illustrates first
What common factor reacted is 30 row subcarrier informations of second antenna, can be construed to second antenna factor, two factors
Contribution rate to initial data population variance is respectively 47.3538% and 44.4187%, accumulation contribution rate to 91.7725%；
Fig. 5 (a) is from the point of view of the Maximum-likelihood estimation of special variance matrix, and the special variance of each subcarrier is smaller, only No.1
The Maximum-likelihood estimation of special variance corresponding with two sub-carriers has been more than 0.1, so Heywood phenomenons are not occurred,
The models fitting effect of this two common factor of explanation is very good.As Fig. 5 (b) shows first that usage factor is analyzed
The dimensionality reduction of antenna factor is results, it can be seen that 30 after the waveform and noise reduction obtained by first antenna factor after dimensionality reduction arrange CSI
The Overall waveform of subcarrier is consistent.The follow-up embodiment of the present invention will be to input the step for estimating single jog in place with dimensionality reduction result
Number.
Then, the embodiment of the present invention uses Short Time Fourier Transform, carries out feature extraction.1-47 indicates leg speed in Fig. 6
Maximum of points, as seen from the figure：Tester's leg left first leaves ground, starts jog in place, and right rear left-leg speed is accelerated,
Left leg speed reaches maximum value when 0.74s, and then left leg slows, and completes the first step until left leg falls back to ground, later
Right leg lifts (left and right leg rotation time interval is extremely short), and then right leg speed is accelerated, and when 1.15s, right leg speed reached maximum value,
Then speed slows down, and falls back to ground and completes second step, and left leg lifts therewith ..., so repeats, and running action continue for big
About 20s, two step interval times were 0.41s, if we take in sometime window the average value being spaced between (such as 2s) step, as
Tester interval time between the step of running in place within the time period, interval time is shorter between step, and movement velocity is faster.From
From the point of view of the distribution of leg speed maximum value, as shown in fig. 7, the maximum value of leg speed is mainly distributed in 0.7m/s between 1m/s,
Corresponding frequency (is converted) in 28Hz between 40Hz by 2v/ λ, and estimated value is very close to truth (30Hz-40Hz), reflection
This method can effectively estimate the maximum value of leg speed in jog in place scene.Cause leg speed maximum value distributional difference
When reason is mainly that people runs in situ, the maximum height that each leg lifts is different, in general, when both legs rotation is very fast, leg
The height lifted can be relatively low, and the maximum value of leg speed can also reduce.
Finally, the embodiment of the present invention to leg velocity estimation curve by carrying out wave crest monitoring to obtain the step of jog in place
Number.
Embodiment 4
A kind of non-contact single step counting system based on commercial Wi-Fi, referring to Fig. 8, which includes：Microprocessor, hair
Penetrate end and receiving terminal.
In embodiments of the present invention, a kind of similar cool run has been designed and Implemented using not modified business WiFi equipment
Immersion games system, it can carry out meter step, and games system includes：Two WiFi equipments, equipment (such as：Router,
As transmitting terminal) persistently emit wireless signal, another equipment (such as：Laptop, as receiving terminal) constantly receive nothing
Line signal.
Microprocessor is to the CSI signals by commercial Wi-Fi acquisition participants；It is a time slide window to define d (t)
The sum of the mean absolute deviation of interior all amplitude datas seeks the noise level N (t) in the t periods according to d (t), believes from CSI
Noise signal is screened out in number；The static component that each subcarrier in CSI signals is rejected by your filter of Han Consulting, uses Sa Weici
Base-Gray's filter further removes other noises；It is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis
Single signal；Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：According to
Four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss are detected.
To the model of each device in addition to doing specified otherwise, the model of other devices is not limited the embodiment of the present invention,
As long as the device of above-mentioned function can be completed.
It will be appreciated by those skilled in the art that attached drawing is the schematic diagram of a preferred embodiment, the embodiments of the present invention
Serial number is for illustration only, can not represent the quality of embodiment.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (4)
1. a kind of non-contact single step-recording method based on commercial Wi-Fi, which is characterized in that the method includes：
Participant carries out repeating motion according to preset action request, and the CSI signals of participant are acquired by commercial Wi-Fi；
The sum of the mean absolute deviation that d (t) is all amplitude datas in a time slide window is defined, t is sought according to d (t)
Noise level N (t) in period screens out noise signal from CSI signals；
The static component that each subcarrier in CSI signals is rejected by your filter of Han Consulting, uses Sa Weiciji-Gray's filter
Further remove other noises；
Single signal is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis；
Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：According to wave crest
Four spacing, crest height, wave crest protrusion, wave peak width constraintss are detected.
2. a kind of non-contact single step-recording method based on commercial Wi-Fi according to claim 1, which is characterized in that
Wherein, ap(n) indicate that the data packet index corresponding to subcarrier p is the amplitude at n, P is the maximum value of sub-carrier indices
(being herein 90), N are the set of all data packet indexes in sliding window, and L is the length of sliding window, and E is in sliding window
All data packet amplitudes mean value.
3. a kind of non-contact single step-recording method based on commercial Wi-Fi according to claim 1, which is characterized in that institute
Stating four corrugation pitch, crest height, wave crest protrusion, wave peak width constraintss is respectively：
(1) corrugation pitch：
Interval time between step when based on normal person's running in place, the minimum 0.32s of corrugation pitch is set, maximum spacing is
0.7s；
(2) crest height：
The maximum speed of leg when based on normal person's running in place is arranged the minimum 0.6m/s of crest height, is up to 1.3m/s；
(3) wave crest protrusion：
The difference of the maximum speed and minimum speed of leg and a large amount of experimental result when based on normal person's jog in place, setting wave crest are prominent
Go out minimum 0.11m/s, is up to 0.92m/s；
(4) wave peak width：
List leg, which is raised to, when based on normal person's jog in place falls the undergone time, and the reference standard of wave peak width is wave herein
The half of peak heights, in conjunction with abundant experimental results, it is minimum 0.1s that the value, which is arranged, is up to 0.5s.
4. a kind of non-contact single step counting system based on commercial Wi-Fi, which is characterized in that the system comprises：Microprocessor,
The transmitting terminal of WiFi and the receiving terminal of WiFi,
The transmitting terminal and receiving terminal of the WiFi is put on the ground, and transmitting terminal and receiving terminal are point-blank；
It collects on the receive side after completing CSI data, CSI data are sent to microprocessor by receiving terminal by ICP/IP protocol
On, microprocessor handles CSI data by MATLAB；
That is, rejecting the static component of each subcarrier in CSI signals by your filter of Han Consulting, filtered using Sa Weiciji-Gray
Wave device further removes other noises；Single signal is extracted from the CSI signals after noise reduction using the dimension-reduction algorithm of factorial analysis；
Time frequency analysis, feature extraction and wave crest detection are carried out to single signal, medium wave peak detection is specially：According to corrugation pitch,
Four crest height, wave crest protrusion, wave peak width constraintss are detected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810573883.3A CN108489509A (en) | 2018-06-06 | 2018-06-06 | A kind of non-contact single step-recording method and system based on commercial Wi-Fi |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810573883.3A CN108489509A (en) | 2018-06-06 | 2018-06-06 | A kind of non-contact single step-recording method and system based on commercial Wi-Fi |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108489509A true CN108489509A (en) | 2018-09-04 |
Family
ID=63342264
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810573883.3A Pending CN108489509A (en) | 2018-06-06 | 2018-06-06 | A kind of non-contact single step-recording method and system based on commercial Wi-Fi |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108489509A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657572A (en) * | 2018-12-04 | 2019-04-19 | 重庆邮电大学 | Goal behavior recognition methods after a kind of wall based on Wi-Fi |
| CN109818695A (en) * | 2019-03-08 | 2019-05-28 | 重庆邮电大学 | Short wave communication half-blindness frequency-selective system and frequency-selecting method based on CW signal |
| CN110013252A (en) * | 2019-04-18 | 2019-07-16 | 北京邮电大学 | Obtain method, apparatus, electronic equipment and the readable storage medium storing program for executing of respiratory state |
| CN110300399A (en) * | 2019-06-24 | 2019-10-01 | 西北大学 | A kind of short distance multi-user concealed communication method and system based on Wi-Fi network interface card |
| CN113203754A (en) * | 2021-05-13 | 2021-08-03 | 合肥工业大学 | Material inspection method based on wireless commercial equipment |
| CN113419212A (en) * | 2021-06-16 | 2021-09-21 | 南京邮电大学 | Equipment-free positioning method based on gray level co-occurrence matrix and factor analysis feature extraction |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105844216A (en) * | 2016-03-11 | 2016-08-10 | 南京航空航天大学 | Detection and matching mechanism for recognition of handwritten letters using WiFi signals |
| CN106323330A (en) * | 2016-08-15 | 2017-01-11 | 中国科学技术大学苏州研究院 | Non-contact-type step count method based on WiFi motion recognition system |
| CN106658590A (en) * | 2016-12-28 | 2017-05-10 | 南京航空航天大学 | Design and implementation of multi-person indoor environment state monitoring system based on WiFi channel state information |
| CN106772219A (en) * | 2017-03-08 | 2017-05-31 | 南京大学 | Indoor orientation method based on CSI signals |
-
2018
- 2018-06-06 CN CN201810573883.3A patent/CN108489509A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105844216A (en) * | 2016-03-11 | 2016-08-10 | 南京航空航天大学 | Detection and matching mechanism for recognition of handwritten letters using WiFi signals |
| CN106323330A (en) * | 2016-08-15 | 2017-01-11 | 中国科学技术大学苏州研究院 | Non-contact-type step count method based on WiFi motion recognition system |
| CN106658590A (en) * | 2016-12-28 | 2017-05-10 | 南京航空航天大学 | Design and implementation of multi-person indoor environment state monitoring system based on WiFi channel state information |
| CN106772219A (en) * | 2017-03-08 | 2017-05-31 | 南京大学 | Indoor orientation method based on CSI signals |
Non-Patent Citations (3)
| Title |
|---|
| 吴革: ""浅谈主成份分析与因子分析方法的区别"", 《山西统计》 * |
| 张煜东: ""降维技术与方法综述"", 《四川兵工学报》 * |
| 石洪波: ""因子分析降维对分类性能的影响研究"", 《中北大学学报(自然科学版)》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657572A (en) * | 2018-12-04 | 2019-04-19 | 重庆邮电大学 | Goal behavior recognition methods after a kind of wall based on Wi-Fi |
| CN109818695A (en) * | 2019-03-08 | 2019-05-28 | 重庆邮电大学 | Short wave communication half-blindness frequency-selective system and frequency-selecting method based on CW signal |
| CN110013252A (en) * | 2019-04-18 | 2019-07-16 | 北京邮电大学 | Obtain method, apparatus, electronic equipment and the readable storage medium storing program for executing of respiratory state |
| CN110300399A (en) * | 2019-06-24 | 2019-10-01 | 西北大学 | A kind of short distance multi-user concealed communication method and system based on Wi-Fi network interface card |
| CN110300399B (en) * | 2019-06-24 | 2020-07-21 | 西北大学 | Close-range multi-user covert communication method and system based on Wi-Fi network card |
| CN113203754A (en) * | 2021-05-13 | 2021-08-03 | 合肥工业大学 | Material inspection method based on wireless commercial equipment |
| CN113419212A (en) * | 2021-06-16 | 2021-09-21 | 南京邮电大学 | Equipment-free positioning method based on gray level co-occurrence matrix and factor analysis feature extraction |
| CN113419212B (en) * | 2021-06-16 | 2022-08-23 | 南京邮电大学 | Equipment-free positioning method based on gray level co-occurrence matrix and factor analysis feature extraction |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108489509A (en) | A kind of non-contact single step-recording method and system based on commercial Wi-Fi | |
| US11766182B2 (en) | Systems and methods for real-time signal processing and fitting | |
| JP6272405B2 (en) | A method for measuring athletic activity using sensors | |
| CN107506684A (en) | Gait recognition method and device | |
| CN107273827B (en) | Electrocardiosignal R wave detection method and device | |
| CN110168316B (en) | Foot step counting method, device and equipment based on step frequency | |
| CN109189207A (en) | A kind of detection method and device of wearable device wearing state | |
| CN109124620A (en) | A kind of atrial fibrillation detection method, device and equipment | |
| Das et al. | Strength training: A fitness application for indoor based exercise recognition and comfort analysis | |
| CN109620260A (en) | Psychological condition recognition methods, equipment and storage medium | |
| KR20210001575A (en) | Real-time cardiac rate detection apparatus in noisy environment and method thereof | |
| CN113743374B (en) | Personnel identity recognition method based on channel state information respiratory perception | |
| CN114707563A (en) | Human body action recognition method based on Wi-Fi channel state information | |
| WO2020228320A1 (en) | Blood pressure estimation method and device | |
| CN112364770B (en) | Commercial Wi-Fi-based human activity recognition and action quality evaluation method | |
| CN107526064A (en) | Adaptive LFM modulated parameter estimating methods based on two dimensional character | |
| CN114145735A (en) | Motion planning scheme generation method, system and equipment based on BCG (Block-coded BCG) signals | |
| Chen et al. | WiTT: Modeling and the evaluation of table tennis actions based on WIFI signals | |
| CN106323330A (en) | Non-contact-type step count method based on WiFi motion recognition system | |
| CN110414479B (en) | Drinking behavior sensing method and continuous and discontinuous action segmentation and identification method | |
| CN110123328B (en) | Breathing frequency detection method based on wireless identification | |
| CN109922726A (en) | State obtains computer, state adquisitiones and information processing system | |
| CN114947754A (en) | Method and apparatus for determining sleep data | |
| CN108652604A (en) | It is a kind of based on electrocardiosignal without air bag blood pressure detecting method and system | |
| Zhou et al. | A Robust Respiration Detection System via Similarity-Based Selection Mechanism Using WiFi |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180904 |