CN106604394A - CSI-based indoor human body motion speed judgment model - Google Patents

Abstract

The invention provides a CSI-based indoor human body motion speed judgment model. The model comprises velocity magnitude and velocity direction, and is a solution to indoor positioning and tracking. The method mainly utilizes the characteristic of wireless signal multipath propagation and is characterized by obtaining a CSI ( channel state information) data, wherein the CSI data carries information reflecting external environmental characteristics, comprising actions produced when a person walks; according to the obtained CSI data, through related algorithms, removing noise and extracting walking information; obtaining velocity magnitude information through short-time Fourier transform, and obtaining velocity direction through a music algorithm; and determining walking direction information of the person through multiple-AP joint deployment, and thus complete speed information in the walking process of the person can be obtained. The speed model can realize high-precision indoor positioning and tracking.

Description

A kind of judgement indoor human body movement velocity model based on CSI

Technical field

The invention discloses a kind of utilization WiFi signal obtains the people that angle and cycle time information when people walks are realized Speed of moving body model, is mainly used in solving indoor human body positioning and tracks, and positions especially for the human body under WiFi environment And tracking problem.The invention belongs to wireless aware technical field.

Background technology

Indoors under environment, the radio wave that signal transmitter is produced is passed via mulitpaths such as direct projection, reflection, scatterings Broadcast, multipath superposed signal is formed at signal receiver.Multipath superposed signal is propagated physical space by it to be affected, and carries reflection The information of environmental characteristic.

Traditional identifying system is mainly divided into three kinds：The identifying system of view-based access control model technology, sensor-based identification System, the identifying system based on special hardware equipment.The identifying system of view-based access control model technology is mainly had no light and illumination The restriction of intensity；Sensor-based identifying system needs deployment or carry sensors, and cost is big, and inconvenient；It is based on The identifying system of special hardware equipment needs to dispose special hardware, hardly results in popularization.Based on the identification system that WiFi equipment is unrelated System overcomes the restriction of traditional system, it is only necessary to dispose an existing business WiFi equipment as AP, and one or more WiFi receiving devices.Present WiFi equipment is found everywhere, including mobile phone, intelligent television, router etc. housed device, all may be used Using the part as system, the popularization for system provides possibility.

By way of D.Halperin et al. is changing firmware, based on CSI (Channel State Information, channel condition information) environment perception technology developed rapidly.The Massachusetts Institute of Technology, Washington are big , Stanford University, Duke University, Hong Kong University of Science and Thchnology, Xi'an Communications University, Tsing-Hua University etc. are in ACM SIGCOMM, ACM MobiCom, IEEE INFOCOM, IEEE Trans.On Mobile Computing, IEEE Trans.On Parallel Deliver many on the famous academic conference of the computer networks such as and Distributed Systems and mobile computing field and periodical Related paper, has done many good tries, including wireless location, action recognition, gesture identification etc..With RSSI (Received Signal Strength Indicator, received signal strength indicator device) is compared, because CSI data are certain Multipath transmisstion is featured in degree, CSI can be regarded as the upgrade version of RSSI.CSI includes many Jie as physical layer information The sightless channel information of matter MAC layer.On the one hand, CSI can obtain multiple subcarriers simultaneously from a packet Frequency response, so as to more subtly portray frequency-selective channel；On the other hand, CSI can both measure each subcarrier Amplitude, can also measure the phase information of each subcarrier.CSI causes the common Wifi equipment to a certain extent can be from time domain On roughly distinguish propagation path, provide accurate resolution ratio for identifying system from multiple angles.

In addition, with RSSI (Received Signal Strength Indicator, received signal strength indicator device) phase Than because CSI (Channel State Information, channel condition information) data feature to a certain extent multipath Propagate, CSI can be regarded as the upgrade version of RSSI., used as physical layer information, including many media access control layers can not for CSI The channel information seen.On the one hand, CSI can simultaneously obtain the frequency response of multiple subcarriers from a packet, so as to more Plus subtly portray frequency-selective channel；On the other hand, CSI can both measure the amplitude of each subcarrier, can also measure The phase information of each subcarrier.CSI causes common WiFi equipment roughly to distinguish biography from time domain to a certain extent Path is broadcast, accurate resolution ratio is provided for identifying system from multiple angles.

The content of the invention

[goal of the invention]：It is an object of the invention to provide a kind of utilization Wi-Fi signal obtains angle when people walks and week The rate pattern that phase temporal information is realized, for solving indoor human body positioning and tracking, especially for the people under WiFi environment Body is positioned and tracking problem.Indoor human body positioning under WiFi environment and tracking system can be realized by model proposed by the present invention System, improves efficiency and precision.

[technical scheme]：The present invention causes what wireless signal multipath changed using the position of multiple stage AP according to human motion Feature, extraction rate information.The present invention gathers first a series of seasonal effect in time series CSI data and carries out denoising, Ran Houcong CSI extracting data features, detect the beginning of people's walking, and obtain velocity magnitude and the velocity attitude that people walks, its final mesh Mark is to realize a rate pattern realized using WiFi signal by the method, and solves to be directed to WiFi environment using the system Under be indoor human body positioning and tracking problem.The present invention program mainly includes herein below：

1) human motion velocity magnitude is judged

Judge that human motion speed is related to the noise processed of data, waveform cutting, Short Time Fourier Transform process, determination The velocity magnitude of human motion.

Noise processed：Effectively remove noise for system design with realize it is critical that.The CSI of system acquisition Data not only include significantly noisy noise, but also comprising trickle noise.LPF algorithm is traditional main Signal antinoise method, including Butterworth filter, Chebyshev filter etc..The present invention is filtered using Butterworth Device.Because Butterworth filter will not be very serious destruction CSI data flows in regard to velocity information.But Butterworth is filtered Ripple device can not well remove trickle noise.For this problem, the present invention is related using the data of CSI different channels Property, denoising is carried out using principal component analysis technology, in addition, the introducing of principal component analysis technology, does not only reach well Denoising effect, and dimension-reduction treatment has been carried out to CSI data, amount of calculation is reduced, improve running efficiency of system.

Waveform cutting：Waveform cutting needs the stage for being syncopated as people's walking, so as to extract the letter for needing in each time threshold Breath.The technology used in the present invention route is different according to the action sequence of people's walking, the produced action of walking caused by institute Sequence waveform is different from other action variances, the different actions according to corresponding to variance judges each time threshold, so as to accurate Carry out waveform cutting.

Determine human motion velocity magnitude：It is analyzed by the CSI data to collecting, it has been found that CSI energy bags Include two parts：Static CSI energy values, dynamic CSI energy variations, wherein indoor static object produce stable CSI energy values, Dynamic CSI energy variations are because human motion is produced.Its physical relationship is as follows：

By above-mentioned analysis, the CSI energy values that can be obtained human motion speed and collect are present for relation.Use Short Time Fourier Transform, by the time-domain information of CSI frequency domain information is changed into, and according to the relation of frequency and wavelength people is can be obtained by The size of speed of moving body.

2) human motion velocity attitude is judged

The determination of human motion velocity attitude need use multiple stage AP special arrangement, experimental situation figure as shown in Figure 1, Using music algorithms, every receiver direction of arrival degree is determined, the speed side of human motion is determined by multiple stage receiver To.But the use of the premise of music algorithms is multi-antenna array, family expenses WiFi equipment antenna amount is unsatisfactory for the bar of multiple antennas Part, so needing the CSI data to obtaining to do the process of correlation.

[beneficial effect]：The invention provides a kind of utilization WiFi signal obtains angle when people walks and cycle time letter The rate pattern that breath is realized, for solving the problems, such as indoor location and tracking.Other existing commercial WiFi equipment and at present WiFi universal situation, is that the popularization of the present invention creates condition.Then the present invention determines as the indoor human body under WiFi environment Position and the application of tracking, have expanded the application of wireless aware.Finally, design of the invention ensure that the true of whole mechanism Property.

The rate pattern realized using Wi-Fi signal is had the advantage that：

(1) efficiency high.

(2) high precision.

(3) versatility, autgmentability are strong.

[description of the drawings]

Fig. 1 is experimental situation figure proposed by the present invention；

Fig. 2 system flow charts；

Fig. 3 music algorithm schematic diagrames

Fig. 4 direction of travel determines schematic diagram

[specific embodiment]

Specific introduction is done to the present invention below in conjunction with accompanying drawing and instantiation.

The present invention is mainly made up of a wireless signal transmitter and two wireless signal receivers.Wireless signal transmitter It is existing business WiFi equipment, wireless signal receiver is equipped with the equipment (notebook computer) of Intel5300.Wireless signal Receiver constantly receives the wireless signal of transmitter transmission and extracts the CSI data of reflection surrounding environment feature, and sample rate is 2500/per second.Two wireless signal receivers are used to determine the velocity attitude of human motion, while reducing measurement human body speed The error of size, experimental situation figure as shown in Figure 1, has gathered data and CSI data has been processed, overall flow such as accompanying drawing 2 Shown, specific implementation process is as follows：

Step 1：Denoising is carried out to CSI data

Step 1.1：Using Butterworth filter denoising.The characteristics of Butterworth filter is that the frequency in passband is rung Answer curve flat to greatest extent, without fluctuating, and it is zero to be then gradually reduced in suppressed frequency band.Using this feature, Butterworth filtering Device can enter denoising to the CSI data for gathering, and remove most noise.

Step 1.2：Using PCA (principal component analysis) denoising.According to noise data different channels be it is incoherent and by The characteristics of different channels are correlations, PCA can further eliminate trickle noise to the data of the mobile generation of people, and drop The low data dimension of CSI, improves the recognition efficiency of system.Its detailed process is as follows：

1) the CSI data of different channels are deducted into its mean value, forms standard CSI matrix.

2) Eigen Covariance matrix is sought.

3) characteristic value and characteristic vector of covariance are asked.

4) characteristic value is sorted according to order from big to small, selects maximum of which k, then by its corresponding k

Characteristic vector is respectively as Column vector groups into eigenvectors matrix.

5) sample point is projected in the characteristic vector of selection.

Step 2：Waveform is divided：

Step 2.1：Sliding window W is set, absolute mean deviation of each group of channel data in j-th window is calculated：

Step 2.2：Calculate all channels absolute mean deviation and：

Step 2.3：The beginning that be regarded as another action of the given threshold beyond certain threshold value

Step 3：Determine velocity magnitude：

The information in the CSI time domains for obtaining is transformed on frequency domain using Short Time Fourier Transform, in order to accurate Judge the velocity magnitude of human motion, further reduce error.We can use formula using the percentage algorithm in Radar Technology It is expressed as follows：

Wherein P (f, t) represents that cumulative percentage F (f, t) represents frequency size, and it is minimum more than 95% that we take P (f, t) Frequency, according to frequency and the relation of human motion speed, obtains the velocity magnitude of human motion, can be expressed as follows with formula：

Wherein λ represents wavelength size, and f represents sub-carrier frequencies

Step 4：Determine velocity attitude：

Step 4.1：We determine direction of arrival degree using CSI phase informations, during gathered data, multiple Receiving terminal may produce sampling frequency deviation, so as to produce extra error.So we first have to remove sampling time deviation Error.We remove sampling time deviation using the method for linear fit, can be expressed as follows with formula：

ψ¹The π f of (m, n)=ψ (m, n)+2_σ(n-1)τ_{S, i}

Step 4.2：Multi-path influence is eliminated, WiFi signal is propagated indoors, by factors such as wall, furniture, ground, people Affect, multipath effect can be produced, cause the signal for receiving to there are other unrelated interruptions, it would be desirable to from the signal for receiving Extract because signal changes (directapath signal) caused by people's walking process, remove irrelevant signal.Directapath signal is deposited In a feature：Signal is directly received through human body reflection by signal receiver, there is no again the situation of secondary reflection, so signal Arrival time is most short.We are judged according to time of arrival (toa), most short time of arrival (toa) are chosen, so as to remove multipath Impact, obtain need directapath signal.

Step 4.3：Using music algorithms as shown in Figure 3, need to build multi-antenna array, the CSI data that we obtain It is the time series data of 36 subcarriers, needs to do CSI data corresponding conversion, can proper use of music algorithms The angle of signal arrival is obtained, building process is as follows：

Step 4.4：Using the music algorithms angle for obtaining signal arrival as follows：

Step 4.5：Determine the side that people walks using a WiFi signal transmitter and two WiFi signal receiver joints To as shown in Figure 4, dotted line represents WiFi signal, and the commissarial direction of travel of red arrow solid line, blue extended line represents letter Number reflection path extended line, when people in the process of walking, WiFi transmission signals are received respectively through human body reflection by WiFi signal Device R1, R2 are received, and the reception signal extended line of R1 and R2 intersects at a point, so that it is determined that the direction of current state servant walking. As people, in the process of walking position constantly changes, and WiFi signal transmitter persistently sends WiFi signal, through human body reflection Persistently captured by two WiFi signal receivers, so that it is determined that the direction of the whole walking process of people changes.

Claims (10)

1. the judgement indoor human body movement velocity model of CSI is based on, it is characterised in that：

1) this model extracts velocity information when people walks using WiFi data for the first time；

2) effective waveform segmentation algorithm is proposed, the beginning and end of walking is accurately identified

3) this model proposes the algorithm of effective extraction rate size

4) this model proposes the algorithm in effective extraction rate direction；

2. the judgement indoor human body movement velocity model of CSI is based on as claimed in claim 1, it is characterised in that the system first Secondary utilization WiFi signal extracts people's walking information indoors, and using CSI data the side of traditional view-based access control model is effectively avoided Method, it is to avoid the danger of privacy leakage, for solving indoor human body positioning and tracking, especially for the human body under WiFi environment Positioning and tracking problem, the invention belongs to wireless aware technical field.

3. waveform cutting method as claimed in claim 1, needs to be syncopated as the time interval that indoor walking motion occurs, and finds out The time threshold that each action occurs, it is characterised in that the otherness that the system is produced using human body difference action to waveform, according to Variance is effectively cut.

Step 1：Sliding window W is set, absolute mean deviation of each group of channel data in j-th window is calculated：

Step 2：Calculate all channels absolute mean deviation and：

${\mathrm{\ΔD}}_j={\mathrm{\Σ}}_2^p{\mathrm{\Δd}}_j\[k\]$

Step 3：The beginning that be regarded as another action of the given threshold beyond certain threshold value

4. the as claimed in claim 1 algorithm for determining velocity magnitude, its feature will obtained using Short Time Fourier Transform Information in CSI time domains is transformed on frequency domain, in order to accurately judge the velocity magnitude of human motion, is further reduced and is missed Difference.We can be expressed as follows using the percentage algorithm in Radar Technology with formula：

$P(f,t)=\frac{{\mathrm{\Σ}}_0^{f}F(f,t)}{{\mathrm{\Σ}}_0^{f_{\max }}F(f,t)}$

Wherein P (f, t) represents that cumulative percentage F (f, t) represents frequency size, and we take lowest frequencies of the P (f, t) more than 95% Rate, according to frequency and the relation of human motion speed, obtains the velocity magnitude of human motion, can be expressed as follows with formula：

$\mathrm{\υ}=\frac{1}{2}\mathrm{\λ}f$

Wherein λ represents wavelength size, and f represents sub-carrier frequencies

5. the algorithm of velocity attitude is determined as claimed in claim 1, it is characterised in that with reference to music algorithms, accurately judge letter Number angle of arrival：

Step 1：Remove sampling frequency deviation

Step 2：Eliminate multi-path influence

Step 3：Reconfigure CSI data matrixes

Step 4：Direction of arrival degree is obtained using music algorithms

Step 5：Multiple devices joint determines velocity attitude

6. it is as claimed in claim 5 to remove sampling frequency deviation, it is characterised in that using the method for linear fit, formula can be used It is expressed as follows：

${\mathrm{\τ}}_{s,i}={\mathrm{argmin\Σ}}_{m,n=1}^{M,N}{\left(\mathrm{\ψ}\right(m,n)+2{\mathrm{\πf}}_{\mathrm{\δ}}(n-1\left)\mathrm{\ρ}\right)}^2$

ψ¹The π f of (m, n)=ψ (m, n)+2_σ(n-1)τ_{S, i}

7. it is as claimed in claim 5 to eliminate multi-path influence, it is characterised in that the characteristics of being propagated using WiFi indoor signals, find The method of most short directapath, determines directapath signal, eliminates multi-path influence.WiFi signal is propagated indoors, by wall, The impact of the factors such as furniture, ground, people, can produce multipath effect, cause the signal for receiving to there are other unrelated interruptions, we Needs are extracted from the signal for receiving because signal changes (directapath signal) caused by people's walking process, are removed unrelated Signal.There is a feature in directapath signal：Signal is directly received through human body reflection by signal receiver, is not existed again The situation of reflection, so time of arrival (toa) is most short.We are judged according to time of arrival (toa), choose most short signal and arrive Up to the time, so as to remove the impact of multipath, directapath signal is obtained.

8. CSI data matrixes are rebuild as claimed in claim 5, it is characterised in that carried containing 36 sons using CSI packets The information of ripple, each subcarrier may be considered an antenna, rebuild CSI data matrixes, as shown in the figure：

$\left[\begin{array}{c}{\mathrm{csi}}_{1,1,}\\ .\\ .\\ .\\ {\mathrm{csi}}_{1,30}\\ {\mathrm{csi}}_{2,1}\\ .\\ .\\ .\\ {\mathrm{csi}}_{2,30}\\ {\mathrm{csi}}_{3,1}\\ .\\ .\\ .\\ {\mathrm{csi}}_{3,30}\end{array}\right]=\left[\begin{array}{ccccccc}{\mathrm{csi}}_{1,1}& {\mathrm{csi}}_{1,2}& \mathrm{...}& {\mathrm{csi}}_{1,16}& {\mathrm{csi}}_{2,1}& \mathrm{...}& {\mathrm{csi}}_{2,16}\\ {\mathrm{csi}}_{1,2}& {\mathrm{csi}}_{1,3}& \mathrm{...}& {\mathrm{csi}}_{1,17}& {\mathrm{csi}}_{2,2}& \mathrm{...}& {\mathrm{csi}}_{2,17}\\ .& .& .& .& .& .& .\\ .& .& .& .& .& .& .\\ .& .& .& .& .& .& .\\ {\mathrm{csi}}_{1,16}& {\mathrm{csi}}_{1,16}& \mathrm{...}& {\mathrm{csi}}_{1,30}& {\mathrm{csi}}_{2,16}& \mathrm{...}& {\mathrm{csi}}_{2,30}\\ {\mathrm{csi}}_{2,1}& {\mathrm{csi}}_{2,2}& \mathrm{...}& {\mathrm{csi}}_{2,16}& {\mathrm{csi}}_{3.1}& \mathrm{...}& {\mathrm{csi}}_{3,16}\\ {\mathrm{csi}}_{2,2}& {\mathrm{csi}}_{2,3}& \mathrm{...}& {\mathrm{csi}}_{2,17}& {\mathrm{csi}}_{3,2}& \mathrm{...}& {\mathrm{csi}}_{3,17}\\ .& .& .& .& .& .& .\\ .& .& .& .& .& .& .\\ .& .& .& .& .& .& .\\ {\mathrm{csi}}_{2,16}& {\mathrm{csi}}_{2,16}& \mathrm{...}& {\mathrm{csi}}_{2,30}& {\mathrm{csi}}_{3,16}& \mathrm{...}& {\mathrm{csi}}_{3,30}\end{array}\right]$

9. as claimed in claim 5 direction of arrival degree is obtained using music algorithms, it is characterised in that using multiple receivers Direction of arrival is respectively obtained, so as to judge the direction that people moves.

$P_{mu}(\mathrm{\θ},\mathrm{\τ})=\frac{1}{a^H(\mathrm{\θ},\mathrm{\τ}){\mathrm{EE}}^{H}a(\mathrm{\θ},\mathrm{\τ})}$

10. as claimed in claim 5 using multiple devices joint determination velocity attitude, it is characterised in that using multiple devices Joint deployment, is intersected at a point, so that it is determined that velocity attitude using the extended line of two reflected signals.When people in the process of walking When position constantly changes, WiFi signal transmitter persistently sends WiFi signal, reflects by two WiFi signals through human body Receiver is persistently captured, so that it is determined that in the whole walking process of people direction change.