CN109547127A - A kind of angle detecting method of bowing based on CSI signal strength in wifi - Google Patents
A kind of angle detecting method of bowing based on CSI signal strength in wifi Download PDFInfo
- Publication number
- CN109547127A CN109547127A CN201811630390.5A CN201811630390A CN109547127A CN 109547127 A CN109547127 A CN 109547127A CN 201811630390 A CN201811630390 A CN 201811630390A CN 109547127 A CN109547127 A CN 109547127A
- Authority
- CN
- China
- Prior art keywords
- bowing
- csi
- angle
- data
- wifi
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L1/00—Arrangements for detecting or preventing errors in the information received
- H04L1/02—Arrangements for detecting or preventing errors in the information received by diversity reception
- H04L1/06—Arrangements for detecting or preventing errors in the information received by diversity reception using space diversity
- H04L1/0618—Space-time coding
- H04L1/0675—Space-time coding characterised by the signaling
- H04L1/0693—Partial feedback, e.g. partial channel state information [CSI]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
A kind of angle detecting method of bowing based on CSI signal strength in wifi, comprising the following steps: step 1, configure AP and MP；Construct experimental situation；The wifi router for preparing commercial common wireless network card driving, supporting 802.11n or the above agreement；Configure CSI environment；Step 2, data are acquired；Step 3, data processing；By analyzing and handling channel state information CSI, the characteristic value for embodying movement of bowing is extracted, harmonic-model is established；Angle motion of bowing finally is sorted out using classifier, identifies simultaneously correcting sitting postures；Realize the synchronous adjustment of user's waist and cervical vertebra；By angle reflection waist bending degree of bowing；The method of the present invention is simple, and without carrying any sensor device, real-time is higher.
Description
Technical field
The invention belongs to computer vision fields, and in particular to a kind of angle of bowing based on CSI signal strength in wifi
Detection method.
Background technique
Computer becomes essential study and media of communication in daily life and work, improper using posture, can draw
Play the health problems such as cervical vertebra, lumbar vertebrae.Posture problem is used for computer, existing detection method is mainly based upon computer view
Feel, sensor etc., these methods need extras to support more, and system cost is higher；And method based on computer vision is wanted
Partes corporis humani position information in image is extracted, is unfavorable for protecting the individual privacy of subjects.
Summary of the invention
To overcome above-mentioned the deficiencies in the prior art, the CSI signal strength based on wifi that the object of the present invention is to provide a kind of
Angle detecting method of bowing, can timely remind the angle problem of bowing of user oneself, help to be corrected in time, can
It solves the problems, such as to bow for a long time and bring.
To achieve the above object, the technical solution adopted by the present invention is that: a kind of CSI signal strength based on wifi is bowed
Angle detecting method, comprising the following steps:
Step 1, AP and MP is configured, Ubuntu system, the driving of installing wireless network card, configuration are compiled in experimental calculation machine
CSI environment, experimental calculation machine install Ubuntu system as systemic vectors and signal receiver MP, experimental calculation machine；By wifi
Router is as signal emitting-source AP；
Step 2, original CSI is acquired, angle measurement experiment of bowing is carried out between configured good AP and MP, is first acquired just
Normal angular range data of bowing, then acquire improper angle-data of bowing；
Step 3, abnormal data is handled, period size, amplitude and the frequency in each subcarrier can be extracted in CSI
Rate, due to the influence of ambient enviroment, can generate some abnormal datas in CSI collection process, for accurately to collected number
According to being analyzed, it is necessary to handle these existing abnormal datas, specific treatment process are as follows:
1) data prediction and denoising, the original CSI got are the three-dimensional matrices of X × Y × 30, to the three-dimensional square
Battle array carries out dimension-reduction treatment and extracts the data for representing angle of bowing, and since CSI is easily affected by environment, uses wavelet filter
Method carries out denoising to it；
2) anomaly data detection, since bow angle measurement selection is angular range, too high or too low angle of bowing
It can not describe the problem, these angles are considered as abnormal data, and accurately to analyze these data, based on threshold test, these are different
Regular data；
3) characteristic is extracted, and period size, frequency and the amplitude of each subcarrier are extracted in CSI, passes through analysis
Data obtain period size, frequency and amplitude as characteristic, resettle harmonic-model, obtain most preferably estimating for phase and amplitude
Evaluation；
4) classifier is classified, to realize that the identification and classification of angular range of bowing will be handled using algorithm of support vector machine
Good CSI is input in preparatory trained classifier, angle of bowing is identified, so that corresponding sitting posture is exported, if being currently at
Incorrect sitting posture, can instant alerts user adjustment.
Wireless network card described in step 1 drives model Intel 5300.
Wifi router described in step 1 supports 802.11n or the above agreement.
Representative described in step 3 bow angle data using the denoising of wavelet filter method.
Classifier described in step 4 is divided into three classes: angle classifier of normally bowing, the excessively high angle classifier and too low of bowing
Angle of bowing classifier, the corresponding sitting posture of the classifier of these angles of bowing are respectively normal sitting position, sit down weak and limp, seat of lying prone.
Period size is extracted according to the following steps in each subcarrier:
1) Wavelet Denoising Method processing is carried out to collected CSI；
2) frequency, the amplitude in each subcarrier are extracted；
3) harmonic-model y is established to each CSI sub-carrier signal(t)=Ai sin(wit+τi), wherein y(t)For subcarrier
Signal, t are time variable；Using Least Square Method, amplitude A is obtainediWith phase τiBest estimate.By amplitude AiWith
Phase τiBest estimate substitute into harmonic-model obtained in estimation signal and collected sub-carrier signal error be smaller
, remember that this error is Ki, then each period size be
Compared with prior art, the beneficial effects of the present invention are:
The present invention does not need to carry any sensor device, facilitates tester to test, while will not relate to tester
Privacy concern, system is simple, be easy deployment, have certain practicability.Using commercial common wireless network card and without route
By device, mobile device etc. is can be used to replace in the wireless router in experiment, because the wifi that router provides does not need to test
Card, as long as being connectable to wifi, so that it may realize this method.
System is simple, does not need to carry additional special hardware device, is easy deployment, has certain practicability.
Detailed description of the invention
Fig. 1 is flow chart of the invention.
Specific embodiment
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments.
Proposed by the present invention a kind of based on CSI sitting posture detecting method in wifi, basic principle is: different human limbs is living
Kinetic energy enough causes different CSI signal change patterns, it will be observed that signal change pattern to be mapped as different human limb living
It is dynamic, it will be able to be identified in real time.
To use wifi signal real time automatic detection sitting posture method in the actual environment, need to solve the problems, such as following four:
1) CSI in wifi signal stream is how to embody sitting posture movement, how to determine head angle or waist angle in sitting posture；2) sitting posture
Does related physical feeling angle use its exact value or range? 3) for complicated actual environment, how in wifi wireless signal
Movement of bowing automatically and precisely is identified in stream? 4) relationship of bow angle and sitting posture is determined.
In response to the above problems, solution of the present invention is as follows: if 1) there is no any movement of bowing between AP and MP, acquisition
To CSI waveform diagram after processing there is no apparent amplitude to change, if bowing movement therebetween, CSI is after processing
The peak value of waveform diagram may get higher or be lower, but the height of peak change is different, and difference angle of bowing corresponds to peak value
The height of variation is different.2) according to actual environment and experimental result various aspects the considerations of, angle of bowing in this experiment takes one
Range, and this angular range should be maintained within 20 degree, angular range is excessive or too small all result can be made not accurate enough；
3) further, when doing this experiment, it should guarantee that data fluctuations are only generated by movement of bowing as far as possible.Respectively under the same environment
Collected data are handled, and the difference of observation comparison waveform diagram, existing difference is exactly caused by bowing and acting.4) not
The corresponding waist bending degree of same angle of bowing is different, extracts due to waist and the common caused CSI of headwork variation
Signal intensity characteristic value.And the angular range that provides normally to bow corresponds to a kind of correct sitting posture, and improper angle of bowing
The sitting posture of the corresponding mistake of range.Improper angular range of bowing includes being greater than normally to bow angular range and to be less than normal
It bows angular range, so that corresponding sitting posture is exactly seat of sitting down weak and limp and lie prone, both sitting postures can all lead to flank pain.
Referring to Fig. 1, a kind of angle detecting method of bowing based on CSI signal strength in wifi, comprising the following steps:
Step 1, AP and MP is configured, Ubuntu system is compiled on the experimental calculation machine prepared, common commercialization is installed
Wireless network card drives (Intel 5300), configures CSI environment, prepares the wifi router for supporting 802.11n or the above agreement；
Step 2, initial data is acquired, angle measurement experiment of bowing is carried out between configured good AP and MP, is first acquired
Normal angular range data of bowing, then improper angle-data of bowing is acquired, because will form a chain between each pair of AP and MP
Collected CSI is saved in experimental calculation machine by road using channel state information tool CSI Tool；Wifi router is used for
Wifi signal is issued, experimental calculation machine, that is, receiving end is for receiving CSI in wifi and CSI being sent to setting for signal processing unit
Standby (experimental calculation machine)；AP is wifi router, for generating wifi signal；MP is the experimental calculation machine configured with CSI environment,
For receiving wifi signal, the channel state information CSI in wifi signal is acquired；
Step 3, abnormal data is handled, period size, amplitude and the frequency in each subcarrier can be extracted in CSI
Rate is easy to be affected by the ambient, generates some abnormal datas when acquiring CSI, in order to accurately analyze acquisition
Information, it is necessary to handle these abnormal datas；Specific treatment process are as follows:
1) data prediction and denoising, the original CSI of acquisition are the three-dimensional matrices of X × Y × 30, wherein X expression connects
By antenna number, Y indicates transmission antenna number, and 30 indicate 30 subcarriers in an ofdm system；Firstly, being dropped to three-dimensional matrice
The data that can represent angle measurement of bowing are extracted after dimension processing, it is contemplated that existing high-frequency noise need to use low-pass filtering
Device is to its denoising；High-frequency noise included in CSI is got rid of using wavelet filter method；
2) anomaly data detection, the present invention construct three groups of test sets, and angular range of normally bowing crosses low angle and bows range
With excessively high angular range of bowing, the anomaly data detection of this step refer to bow it is too high or too low low present in angular range
Brilliance degree, too high or too low angle of bowing belong to abnormal data in daily angular range of bowing, this step is different to these
Regular data is detected, and according to the requirement of experimental situation and experimental result, is detected using local outlier factor detection algorithm (LOF)
Outlier factor present in CSI；When testing, if normally bowing in angular range defined, CSI is that smoothly, do not have
Too big fluctuating；But if angle of bowing is too low or excessively high, and channel state information CSI just will appear obvious variation；At this
In the process, angle of bowing is too low or excessively high occurred data variation is exactly abnormal data, needs to carry out these abnormal datas
Detection is compared and discusses by analyzing these abnormal datas, then with the data for angle appearance of normally bowing；
(since bow angle measurement selection is angular range, too high or too low angle of bowing can not be described the problem,
These existing angles are considered as abnormal data.For accurate these data of analysis, these abnormal numbers based on threshold test
According to)
3) characteristic is extracted, and after completing data exception detection, needs to extract the characteristic value in abnormal data, CSI
In include multiple subcarriers frequency, amplitude and period size, while extracting the frequency for the multiple subcarriers for including in CSI,
Amplitude and period size obtain period size, frequency and amplitude by analyzing data, characteristic are regarded as, in conjunction with frequency
Collected CSI establishes harmonic-model y to each CSI sub-carrier signal(t)=Ai sin(wit+τi), wherein y(t)Indicate each
CSI sub-carrier signal, t are time variable, obtain amplitude A using Least Square Method according to collected every CSIiWith
Phase τiBest estimate；By amplitude AiWith phase τiBest estimate substitute into harmonic-model obtained in estimation signal and
Collected sub-carrier signal error be it is lesser, remember this error be Ki, then the period size of each subcarrier be
4) classifier is classified, and the present invention classifies to the characteristic extracted using algorithm of support vector machine, makes
With algorithm of support vector machine, i.e., will treated CSI inputs to preparatory trained classifier, classifier bows output angle
Range, i.e., corresponding output sitting posture；Preparatory trained classifier is divided into three classes: normally bow angle classifier, excessively high angle of bowing
It spends classifier and too low angle classifier of bowing, the corresponding sitting posture of the classifier of these angles of bowing is respectively normal sitting position, paralysis
It sits, seat of lying prone；If this classifier is too high or too low angle classifier of bowing, corresponding sitting posture is seat of sitting down weak and limp or lie prone, and system is just
It can be automatically reminded to the sitting posture that the user needs to adjust oneself in time.
Wireless network card described in step 1 drives model Intel 5300.
Wifi router described in step 1 supports 802.11n or the above agreement.
Representative described in step 3 bow angle data using the denoising of wavelet filter method.
Classifier described in step 4 is divided into three classes: angle classifier of normally bowing, the excessively high angle classifier and too low of bowing
Angle of bowing classifier, the corresponding sitting posture of the classifier of these angles of bowing are respectively normal sitting position, sit down weak and limp, seat of lying prone.
Period size is extracted according to the following steps in each subcarrier:
1) Wavelet Denoising Method processing is carried out to collected CSI；
2) frequency, the amplitude in each subcarrier are extracted；
3) harmonic-model y is established to each CSI sub-carrier signal(t)=Ai sin(wit+τi), wherein y(t)For subcarrier
Signal, t are time variable；Using Least Square Method, amplitude A is obtainediWith phase τiBest estimate.By amplitude AiWith
Phase τiBest estimate substitute into harmonic-model obtained in estimation signal and collected sub-carrier signal error be smaller
, remember that this error is Ki, then each period size be
The CSI treatment process includes extracting characteristic period size and test and bowing angle and to identify sitting posture, is extracted
The characteristic period size extracted, is inputed to classifier trained in advance by the characteristic for reflecting movement of bowing out, point
Class device judges to bow angular range and identifies sitting posture.
Step 2.1: acquisition CSI data do different angle motions of bowing, by collected CSI number between AP and MP
According to being saved in experimental calculation machine.Wherein, AP is wifi router, and for generating wifi signal, MP is configured with CSI environment
Experimental calculation machine acquires the channel state information CSI data in wifi signal for receiving wifi signal；
Step 2.2: processing CSI data, Data Dimensionality Reduction pretreatment, Wavelet Denoising Method extract the period for representing movement of bowing
Size establishes harmonic-model, and period size is substituted into trained classifier in advance and is classified, identifies different angles of bowing
Degree movement, adjusts sitting posture.
During this method is realized, hardware that the angle detecting method of bowing based on CSI signal strength in wifi uses
Respectively there are 2 classes with software:
Hardware
1) equipped with Intel 5300 wireless network card driving experimental calculation machine, this computer can be used as systemic vectors and
Signal receiver MP, experimental calculation machine need to be equipped with Ubuntu system；
2) router is as signal emitting-source AP.
Software
1) using Linux 802.11n CSI Tool kit from the channel status information that physical layer acquires need using
MATLAB carries out signal processing；
2) experimental calculation machine work system is Ubuntu11.04, is guaranteed the repair free of charge using CSI information collection tool under linux system
Change computer radio trawl performance.
Claims (5)
1. a kind of angle detecting method of bowing of the CSI signal strength based on wifi, which comprises the following steps:
Step 1, AP and MP is configured, Ubuntu system is compiled in experimental calculation machine, the driving of installing wireless network card configures CSI ring
Border, experimental calculation machine install Ubuntu system as systemic vectors and signal receiver MP, experimental calculation machine；By wifi router
As signal emitting-source AP；
Step 2, original CSI is acquired, angle measurement experiment of bowing is carried out between configured good AP and MP, first acquisition is normal
It bows angular range data, then acquires improper angle-data of bowing；
Step 3, abnormal data is handled, period size, amplitude and the frequency in each subcarrier can be extracted in CSI,
In CSI collection process, due to the influence of ambient enviroment, some abnormal datas can be generated, for accurately to collected data into
Row analysis, it is necessary to these existing abnormal datas be handled, specific treatment process are as follows:
1) data prediction and denoising, the original CSI got are the three-dimensional matrices of X × Y × 30, to the three-dimensional matrice into
Row dimension-reduction treatment extracts the data for representing angle of bowing, and since CSI is easily affected by environment, uses wavelet filter method
Denoising is carried out to it；
2) anomaly data detection, since bow angle measurement selection is angular range, too high or too low angle of bowing cannot
Enough describing the problem, these angles are considered as abnormal data, accurately to analyze these data, these abnormal numbers based on threshold test
According to；
3) characteristic is extracted, and period size, frequency and the amplitude of each subcarrier are extracted in CSI, by analyzing data
Period size, frequency and amplitude are obtained as characteristic, resettles harmonic-model, obtains the best estimate of phase and amplitude
Value；
4) classifier is classified, to realize the bow identification and classification of angular range, using algorithm of support vector machine, by what is handled well
CSI is input in preparatory trained classifier, angle of bowing is identified, so that corresponding sitting posture is exported, if being currently at not just
True sitting posture, can instant alerts user adjustment.
2. a kind of angle detecting method of bowing of CSI signal strength based on wifi according to claim 1, feature exist
In wireless network card described in step 1 drives model Intel5300.
3. a kind of angle detecting method of bowing of CSI signal strength based on wifi according to claim 1, feature exist
In wifi router described in step 1 supports 802.11n or the above agreement.
4. a kind of angle detecting method of bowing of CSI signal strength based on wifi according to claim 1, feature exist
In, representative described in step 3 bow angle data using the denoising of wavelet filter method.
5. a kind of angle detecting method of bowing of CSI signal strength based on wifi according to claim 1, feature exist
In classifier described in step 4 is divided into three classes: angle classifier of normally bowing, excessively high bow and too low are bowed at angle classifier
Angle classifier, the corresponding sitting posture of the classifier of these angles of bowing are respectively normal sitting position, sit down weak and limp, seat of lying prone.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811630390.5A CN109547127A (en) | 2018-12-29 | 2018-12-29 | A kind of angle detecting method of bowing based on CSI signal strength in wifi |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811630390.5A CN109547127A (en) | 2018-12-29 | 2018-12-29 | A kind of angle detecting method of bowing based on CSI signal strength in wifi |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109547127A true CN109547127A (en) | 2019-03-29 |
Family
ID=65830934
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811630390.5A Pending CN109547127A (en) | 2018-12-29 | 2018-12-29 | A kind of angle detecting method of bowing based on CSI signal strength in wifi |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109547127A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111027416A (en) * | 2019-11-21 | 2020-04-17 | 电子科技大学 | Position self-adaptive action recognition method based on commercial WiFi |
| WO2020228324A1 (en) * | 2019-05-14 | 2020-11-19 | 广东康云科技有限公司 | Wifi-based object detection method and system, and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105962946A (en) * | 2016-06-23 | 2016-09-28 | 华中科技大学 | Non-contact human body sleeping posture safety detection method and system |
| CN106108904A (en) * | 2016-06-23 | 2016-11-16 | 华中科技大学 | A kind of contactless human body respiration parameter method for real-time measurement and system |
-
2018
- 2018-12-29 CN CN201811630390.5A patent/CN109547127A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105962946A (en) * | 2016-06-23 | 2016-09-28 | 华中科技大学 | Non-contact human body sleeping posture safety detection method and system |
| CN106108904A (en) * | 2016-06-23 | 2016-11-16 | 华中科技大学 | A kind of contactless human body respiration parameter method for real-time measurement and system |
Non-Patent Citations (2)
| Title |
|---|
| WENFENG HE等: "WiFi-based Gesture Recognition System", 《 2015 24TH INTERNATIONAL CONFERENCE ON COMPUTER COMMUNICATION AND NETWORKS (ICCCN)》 * |
| 何文锋: "基于WIFI的手势识别研究", 《CNKI全国优秀硕士论文全文库 信息科技》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020228324A1 (en) * | 2019-05-14 | 2020-11-19 | 广东康云科技有限公司 | Wifi-based object detection method and system, and storage medium |
| CN111027416A (en) * | 2019-11-21 | 2020-04-17 | 电子科技大学 | Position self-adaptive action recognition method based on commercial WiFi |
| CN111027416B (en) * | 2019-11-21 | 2022-03-15 | 电子科技大学 | Position self-adaptive action recognition method based on commercial WiFi |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106022378B (en) | Sitting posture judgment method and based on camera and pressure sensor cervical spondylosis identifying system | |
| JP5848551B2 (en) | Learning device, learning device control method, detection device, detection device control method, and program | |
| CN105678222B (en) | A kind of mobile device-based Human bodys' response method | |
| CN104143079B (en) | The method and system of face character identification | |
| CN105022981B (en) | A kind of method, device and mobile terminal detecting human eye health state | |
| CN106108904B (en) | A kind of contactless human body respiration parameter method for real-time measurement and system | |
| CN105310696B (en) | A kind of fall detection model building method and corresponding fall detection method and device | |
| CN109766796A (en) | A kind of depth pedestrian detection method towards dense population | |
| CN113435283B (en) | Ultra-wideband radar identity recognition method based on breath sample space | |
| CN103718175A (en) | Apparatus, method, and medium detecting object pose | |
| CN106491129B (en) | A kind of Human bodys' response system and method based on EOG | |
| CN109547127A (en) | A kind of angle detecting method of bowing based on CSI signal strength in wifi | |
| CN106137130A (en) | A kind of sleep period recognition methods based on Audio Processing | |
| CN107480716A (en) | Method and system for identifying saccade signal by combining EOG and video | |
| CN103902978A (en) | Face detection and identification method | |
| Chen et al. | WiFace: Facial expression recognition using Wi-Fi signals | |
| CN109740522A (en) | A kind of personnel's detection method, device, equipment and medium | |
| Benocci et al. | A context-aware smart seat | |
| CN108958489A (en) | A kind of interesting image regions Rapid Detection method based on brain electricity and eye tracker | |
| Kwon et al. | Hands-free human activity recognition using millimeter-wave sensors | |
| CN108371806B (en) | Gymnastics training device and training method thereof | |
| CN110309767A (en) | In vivo detection equipment, recognition methods, device and storage medium | |
| CN105962946A (en) | Non-contact human body sleeping posture safety detection method and system | |
| CN107480635A (en) | Glance signal identification method and system based on bimodal classification model fusion | |
| CN106725466A (en) | A kind of bow-backed alarm set and based reminding method based on surface myoelectric technology |
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 |
Application publication date: 20190329 |
|
| RJ01 | Rejection of invention patent application after publication |