CN109409216B - Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection - Google Patents
Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection Download PDFInfo
- Publication number
- CN109409216B CN109409216B CN201811086897.9A CN201811086897A CN109409216B CN 109409216 B CN109409216 B CN 109409216B CN 201811086897 A CN201811086897 A CN 201811086897A CN 109409216 B CN109409216 B CN 109409216B
- Authority
- CN
- China
- Prior art keywords
- state information
- channel state
- window
- human body
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Geophysics And Detection Of Objects (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a speed self-adaptive indoor human body detection method based on subcarrier dynamic selection.A training stage comprises the acquisition and pretreatment of channel state information; denoising the channel state information; extracting a channel state information characteristic value; the channel state information is classified and trained, and the detection stage comprises channel state information preprocessing; extracting characteristic values in the window: extracting characteristic values according to window segmentation, wherein the characteristic values are used as verification data of human body detection; and (5) detecting a human body. The method can prevent the condition of inaccurate detection result caused by the fact that the slow movement has no obvious influence on the channel state information, and effectively solves the problem of high detection report missing rate when the slow moving human body exists in the detection environment.
Description
Technical Field
The invention relates to a self-adaptive indoor human body detection method, in particular to a speed self-adaptive indoor human body detection method based on subcarrier dynamic selection, and belongs to the field of human body detection based on channel state information.
Background
With the rapid development of the internet of things in recent years, intelligent equipment and systems are developed and applied in a large quantity, and safety and convenience are brought to life of people. Among them, the human body detection system has a wide application prospect in homes and businesses. The traditional human body detection technology needs large-scale physical equipment deployment or requires a user to wear a special detection instrument, which causes the limitation of the traditional human body detection. With the popularization of WiFi, a human body detection method based on wireless devices is developed. Wherein the channel state information becomes the object of the researchers. When the human body detection is studied by using the channel state information, a certain subcarrier or an average value of all subcarriers is generally adopted as detection data. If a slowly moving human body exists in the detection environment, the interference to the channel state information is small due to the small action amplitude, and the situation of report missing is easily caused. Aiming at the problems, a speed self-adaptive human body detection method based on channel state information is provided by analyzing the sensitivity difference of subcarriers.
Disclosure of Invention
Aiming at the prior art, the technical problem to be solved by the invention is to provide a speed self-adaptive human body detection method based on channel state information by analyzing the sensitivity difference of subcarriers.
In order to solve the technical problem, the invention discloses a speed self-adaptive indoor human body detection method based on subcarrier dynamic selection, which comprises a training stage and a detection stage, wherein the training stage comprises the following steps:
step (1): acquiring and preprocessing channel state information: collecting channel state information captured by a wireless network card, preprocessing the channel state information, and decomposing the channel state information in a complex form into two sets of amplitude and phase;
step (2): denoising channel state information: carrying out noise removal processing on the channel state information preprocessed in the step (1);
and (3): channel state information characteristic value extraction: extracting a characteristic value of channel state information in a time window, and taking the characteristic value as a judgment standard of human body detection;
and (4): and (3) channel state information classification training: performing early-stage classification training by taking the extracted characteristic values as classification standards to generate a classifier, and preparing for a detection stage;
wherein the detection stage comprises the following steps:
step (1): preprocessing channel state information: the method comprises decomposition processing, denoising processing and window segmentation;
step (2): extracting characteristic values in the window: extracting characteristic values according to window segmentation, wherein the characteristic values are used as verification data of human body detection;
and (3): human body detection: and inputting the characteristic values into a classifier generated in a training stage for inspection.
The invention has the beneficial effects that:
the existing human body detection method based on the channel state information has accurate detection results when a normally moving human body exists in the environment, and cannot accurately detect when a slowly moving human body exists in the environment. Aiming at the problems, the invention mainly utilizes the difference of subcarrier sensitivity and a dynamic subcarrier selection strategy to ensure that the subcarrier selection strategy meets the condition of speed self-adaption and improves the accuracy rate of detection. The method mainly aims at the condition that the influence of a slow moving human body in the detection environment on the channel state information is not obvious so as to influence the detection result. For the human body which normally moves in the environment, a conventional detection method is adopted during detection. And for the human body which does not obviously move in the environment, a dynamic subcarrier selection strategy is adopted during detection. Therefore, the situation that the detection result is inaccurate due to the fact that the slow movement has no obvious influence on the channel state information can be prevented. The problem that the detection report missing rate is high when a slow-moving human body exists in the detection environment is effectively solved.
Drawings
FIG. 1 is an overall flow chart of the present invention.
Fig. 2 is a flow chart of the detection process.
FIG. 3 is a diagram of a random forest classification process.
Fig. 4 is a graph of the conventional moving human body detection method and the detection rate result of the present invention at different moving speeds.
FIG. 5 is a graph of the detection rate results of the present invention for different time window sizes.
Fig. 6 is a diagram showing the result of the detection rate of the present invention under the condition that different numbers of subcarriers are selected during the selection of the dynamic subcarriers.
Detailed Description
The following describes embodiments of the present invention with reference to the drawings.
As shown in FIG. 1, the whole human body detection method is divided into two parts, namely an early training stage and a later detection stage. The early training phase includes:
1) acquiring and preprocessing channel state information: and collecting the channel state information captured by the wireless network card, preprocessing the channel state information, and decomposing the channel state information in a complex form into two sets of amplitude and phase.
2) Denoising channel state information: and performing denoising processing on the collected channel state information.
3) Channel state information characteristic value extraction: and extracting characteristic values of the channel state information in the time window, wherein the characteristic values are used as judgment standards for human body detection.
4) And (3) channel state information classification training: and performing early-stage classification training by taking the extracted characteristic values as classification standards to prepare for later-stage detection.
The later detection stage comprises:
1) preprocessing channel state information: the method comprises decomposition processing, denoising processing and window segmentation.
2) Extracting characteristic values in the window: and extracting characteristic values according to window segmentation, wherein the characteristic values are used as verification data of human body detection.
3) Human body detection: and inputting the extracted characteristic values into a classifier trained before for checking.
FIG. 2 is a flow chart of the detection section of the present invention:
1) the channel state information is collected by the MIMO technology, and 30 subcarriers contained in each channel state information are processed into a set separated by amplitude and phase. The denoised channel state is not enough to automatically judge whether a human body exists or not, so that the denoised channel state information needs to be subjected to characteristic value extraction, and then the characteristic value is used for judging whether the human body exists or not in a self-adaptive manner. And (3) carrying out window segmentation according to the initially set window size, and calculating expectation, standard deviation, average energy and extreme value difference of window occurrence of amplitude information in the window according to a random single subcarrier by using statistical characteristics in time domain analysis.
In the process of extracting the features, it is assumed that the length of each window after the signal state information is subjected to window segmentation is n, and the amplitude of the ith channel state information is xiThe extraction mode of each feature is as follows:
a. it is desired that: for indicating the data set trend of the CSI. The expression is shown below, wherein xiIs the amplitude of the ith point of the channel state information, and n is the number of points in the time window.
b. Standard deviation: the degree of data dispersion used to represent the CSI. The expression is shown below, wherein xiIs the amplitude of the ith point of the channel state information, n is the number of points in the time window, and μ is the average value of all the channel state information amplitudes in the window.
c. Average energy: the expression is shown below, wherein
Represents the square of the amplitude of the ith point of the channel state information, and n is the total number of points in the time window.
d. Extreme difference: representing the difference between the maximum and minimum values of the amplitude within the window. The expression is shown below, wherein xmaxAmplitude of the maximum point within the window, xminIs the amplitude of the minimum point within the window.
D＝xmax-xmin
After the sampled channel state information is divided into the sizes of the appointed time windows, the extraction analysis is carried out according to the characteristic extraction mode, the characteristic value in each time window is calculated, and the standards of the static environment and the dynamic environment are calculated according to statistics.
2) The invention introduces a machine learning classification method to solve the limitation that the threshold value obtained by the conventional method is obtained by a large amount of experiments and artificial calculation comparison. Therefore, the human body detection method has self-adaptability and operability.
Firstly, a training set is divided into a static environment and a dynamic environment, the characteristic values in the training set are extracted by the method, and the obtained characteristic values are put into a classifier to obtain a classification result. In the invention, a random forest classifier is adopted to classify the characteristic values of the collected channel state information. The random forest classification algorithm is the expansion of a decision tree, and different data sets are formed by combining with a Bagging strategy. The random forest uses decision tree as a base learning device, when the indoor human body detection based on the channel state information is trained, various detection characteristics described above are formed into a plurality of decision trees, the decision trees are mutually independent, and the decision trees formed by the characteristics are formed into the random forest. FIG. 3 is a diagram of a random forest classification process.
In the process of random forest classification, the selection of the samples is that a certain amount of data is extracted from the sample set to serve as a root node sample of each decision tree. Then randomly extracting a certain number of attributes to form nodes of the decision tree, comparing the attributes through an algorithm, and taking the proper attributes as split nodes.
During testing, the characteristics of the test sample are input into the classifier, each base decision tree in the classifier provides a judgment result of the classifier, and the final output result is obtained by voting by combining the judgment results provided by each decision tree through a random forest so as to determine the final category.
3) The sample data is classified by a random forest classifier, and thus after a feature value set of the test sample is obtained, the sample data is input into the classifier. If the final category is someone, an alarm is raised.
4) And if the final classification of the classifier is unmanned, judging that the environment is unmanned, and continuously detecting the next window.
FIG. 4 is a diagram of a conventional mobile human detection method and detection rate of the present invention at different mobile speeds; FIG. 5 is a graph of the detection rate of the present invention at different time window sizes; fig. 6 shows the detection rate of the present invention when different numbers of subcarriers are selected during dynamic subcarrier selection. By combining the results of the three figures, it can be seen that the mobile human body detection of the invention achieves a better detection effect, especially when a slow moving human body exists in the environment.
Claims (1)
1. A speed self-adaptive indoor human body detection method based on subcarrier dynamic selection is characterized in that: the method comprises a training stage and a detection stage, wherein the training stage comprises the following steps:
step (1): acquiring and preprocessing channel state information: collecting channel state information captured by a wireless network card, preprocessing the channel state information, and decomposing the channel state information in a complex form into two sets of amplitude and phase;
step (2): denoising channel state information: carrying out noise removal processing on the channel state information preprocessed in the step (1);
and (3): channel state information characteristic value extraction: extracting a characteristic value of channel state information in a time window, and taking the characteristic value as a judgment standard of human body detection;
the characteristic values comprise expectation, standard deviation, average energy of amplitude information in the window and extreme value difference of window occurrence;
data set trends are expected to be used to represent CSI, specifically:
wherein x isiThe amplitude of the ith point of the channel state information is obtained, and n is the number of points in a time window;
the standard deviation is used for representing the data dispersion degree of the CSI, and specifically includes:
the average energy is specifically:
extreme difference: for representing the difference between the maximum and minimum values of the amplitude within the window, in particular:
D＝xmax-xmin
wherein xmaxAmplitude of the maximum point within the window, xminThe amplitude of the minimum point in the window; and (4): and (3) channel state information classification training: performing early-stage classification training by taking the extracted characteristic values as classification standards to generate a classifier, and preparing for a detection stage; the classifier is a random forest classifier, the random forest is a decision tree-based learner, when the indoor human body detection based on the channel state information is trained, the characteristic values form decision trees, the decision trees are mutually independent, and then the decision trees form the random forest;
wherein the detection stage comprises the following steps:
step (1): preprocessing channel state information: the method comprises decomposition processing, denoising processing and window segmentation;
step (2): extracting characteristic values in the window: extracting characteristic values according to window segmentation to form a detection characteristic set, wherein the characteristic values comprise expectation, standard deviation, average energy of amplitude information in a window and extreme value difference of the window; the characteristic value is used as verification data of human body detection;
and (3): human body detection: inputting the characteristic values into a random forest classifier generated in a training stage for inspection, wherein each decision tree in the classifier provides a judgment result of the decision tree, the final output result is obtained by voting by combining the judgment results provided by each decision tree through a random forest so as to determine a final class, and if the final class is someone, an alarm is given; and if the final classification of the classifier is unmanned, judging that the environment is unmanned, and continuously detecting the next window.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811086897.9A CN109409216B (en) | 2018-09-18 | 2018-09-18 | Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811086897.9A CN109409216B (en) | 2018-09-18 | 2018-09-18 | Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109409216A CN109409216A (en) | 2019-03-01 |
| CN109409216B true CN109409216B (en) | 2022-04-05 |
Family
ID=65464971
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811086897.9A Active CN109409216B (en) | 2018-09-18 | 2018-09-18 | Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409216B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110011741A (en) * | 2019-03-29 | 2019-07-12 | 河北工程大学 | Personal identification method and device based on wireless signal |
| CN111481203B (en) * | 2020-05-22 | 2023-05-05 | 哈尔滨工程大学 | Indoor static passive human body detection method based on channel state information |
| CN116304915B (en) * | 2023-05-16 | 2023-08-29 | 山东科技大学 | WiFi-based contactless action recognition method, system and laboratory device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012061325A1 (en) * | 2010-11-01 | 2012-05-10 | Rearden, Llc | Systems and methods to coordinate transmissions in distributed wireless systems via user clustering |
| CN105158727A (en) * | 2015-06-18 | 2015-12-16 | 哈尔滨工程大学 | Enhanced indoor passive human body positioning method |
| CN106411433A (en) * | 2016-09-08 | 2017-02-15 | 哈尔滨工程大学 | WLAN-based fine-grained indoor passive intrusion detection method |
| CN107968689A (en) * | 2017-12-06 | 2018-04-27 | 北京邮电大学 | Perception recognition methods and device based on wireless communication signals |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107347210B (en) * | 2017-07-04 | 2019-10-29 | 江苏先安科技有限公司 | A kind of precision target localization method based on channel state information |
| CN107994960B (en) * | 2017-11-06 | 2020-11-27 | 北京大学（天津滨海）新一代信息技术研究院 | Indoor activity detection method and system |
-
2018
- 2018-09-18 CN CN201811086897.9A patent/CN109409216B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012061325A1 (en) * | 2010-11-01 | 2012-05-10 | Rearden, Llc | Systems and methods to coordinate transmissions in distributed wireless systems via user clustering |
| CN105158727A (en) * | 2015-06-18 | 2015-12-16 | 哈尔滨工程大学 | Enhanced indoor passive human body positioning method |
| CN106411433A (en) * | 2016-09-08 | 2017-02-15 | 哈尔滨工程大学 | WLAN-based fine-grained indoor passive intrusion detection method |
| CN107968689A (en) * | 2017-12-06 | 2018-04-27 | 北京邮电大学 | Perception recognition methods and device based on wireless communication signals |
Non-Patent Citations (3)
| Title |
|---|
| Robust WLAN-based Indoor Fine-grained Intrusion Detection;Jiguang Lv et al.;《2016 IEEE》;20161231;第1-6页 * |
| R-TTWD: Robust Device-Free Through-The-Wall Detection of Moving Human With WiFi;Hai Zhu et al;《IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS》;20170531;第35卷(第5期);第1092、1094-1095页 * |
| 基于信道状态信息的无源室内定位;吴哲夫等;《哈尔滨工程大学学报》;20170831;第38卷(第8期);第1329-1333页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109409216A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109409216B (en) | Speed self-adaptive indoor human body detection method based on subcarrier dynamic selection | |
| CN110287552B (en) | Motor bearing fault diagnosis method and system based on improved random forest algorithm | |
| CN107657088B (en) | Rolling bearing fault diagnosis method based on MCKD algorithm and support vector machine | |
| CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
| CN106845556A (en) | A kind of fabric defect detection method based on convolutional neural networks | |
| CN111238843B (en) | Fan health evaluation method based on rapid spectrum kurtosis analysis | |
| CN107305774A (en) | Speech detection method and device | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| CN109946080B (en) | Mechanical equipment health state identification method based on embedded circulation network | |
| CN110929842B (en) | Accurate intelligent detection method for non-cooperative radio signal burst time region | |
| CN112528774B (en) | Intelligent unknown radar signal sorting system and method in complex electromagnetic environment | |
| CN108334902A (en) | A kind of track train equipment room smog fireproof monitoring method based on deep learning | |
| Pourhomayoun et al. | Bioacoustic signal classification based on continuous region processing, grid masking and artificial neural network | |
| CN107563403B (en) | Working condition identification method for high-speed train operation | |
| CN111289251A (en) | Rolling bearing fine-grained fault identification method | |
| CN111600878A (en) | Low-rate denial of service attack detection method based on MAF-ADM | |
| CN106708009A (en) | Ship dynamic positioning measurement system multiple-fault diagnosis method based on support vector machine clustering | |
| CN106951924B (en) | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm | |
| CN107341519B (en) | Support vector machine identification optimization method based on multi-resolution analysis | |
| CN116881712A (en) | Electromagnetic pulse signal identification method for movable cracks of concrete dam | |
| CN111209891A (en) | Deep neural network-based bearing working condition detection method and system | |
| CN116466408A (en) | Artificial neural network superbedrock identification method based on aeromagnetic data | |
| CN110071884A (en) | A kind of Modulation Recognition of Communication Signal method based on improvement entropy cloud feature | |
| CN112134634B (en) | Random forest algorithm-based spectrum sensing method, system and medium | |
| Kats et al. | Features extraction from non-destructive testing data in cyber-physical monitoring system of construction facilities |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |