CN110176968B - Jump phenomenon correction method for WiFi human behavior recognition - Google Patents

Jump phenomenon correction method for WiFi human behavior recognition Download PDF

Info

Publication number
CN110176968B
CN110176968B CN201910420441.XA CN201910420441A CN110176968B CN 110176968 B CN110176968 B CN 110176968B CN 201910420441 A CN201910420441 A CN 201910420441A CN 110176968 B CN110176968 B CN 110176968B
Authority
CN
China
Prior art keywords
csi
data
mode
segments
segmentation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910420441.XA
Other languages
Chinese (zh)
Other versions
CN110176968A (en
Inventor
邓昀
张庆俊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guilin University of Technology
Original Assignee
Guilin University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guilin University of Technology filed Critical Guilin University of Technology
Priority to CN201910420441.XA priority Critical patent/CN110176968B/en
Publication of CN110176968A publication Critical patent/CN110176968A/en
Application granted granted Critical
Publication of CN110176968B publication Critical patent/CN110176968B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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

Abstract

The invention discloses a jump phenomenon correction method for WiFi human behavior recognition, which is characterized by comprising the following steps of: 1) identifying a trip point; 2) obtaining the optimal initial replacement position s (s is more than or equal to 1 and less than or equal to 2m-N-1), the number N of segmentation segments and the specific segmentation point { x1，x2，...，xn‑1}; 3) obtaining the optimal segmentation result { x1，x2，...，xn‑1}; 4) any two CSI data segments to be input
Figure DDA0002065837750000011
And
Figure DDA0002065837750000012
constructing a corresponding time series; 5) acquiring multi-scale shape information corresponding to the time sequence; 6) computing any two CSI data segments
Figure DDA0002065837750000013
And

Description

Jump phenomenon correction method for WiFi human behavior recognition
Technical Field
The invention relates to a jump phenomenon correction method in human behavior recognition, in particular to a jump phenomenon correction method used in WiFi human behavior recognition.
Background
With the rapid development of wireless technology and artificial intelligence, human behavior recognition based on WiFi signals is receiving more and more extensive attention from academia. At present, the WiFi human behavior recognition technology is successfully applied to the fields of daily behavior perception, gesture recognition, heartbeat and respiration detection, falling detection and the like. The action of the human body can affect the WiFi signal in transmission, so that the signal is subjected to phenomena of reflection, refraction, diffraction, scattering and the like. By studying the changes of the received CSI data, the current actions performed by the human body can be identified. According to different modes, WiFi human body behavior recognition based on models and modes can be divided.
In 2009, by simulating the communication behavior of fireflies, Yang proposes a new meta-heuristic method, called standard fireflies algorithm, which can find global and local optima of the objective function at the same time, however, the standard fireflies algorithm uses fixed random parameters in the optimization process, and emphasizes exploration rather than development, so that the convergence rate is slow, and the algorithm is easy to fall into local optima. Intuitively, in the early stages of the standard firefly algorithm, exploration should play a more important role to identify promising search regions, however, as the evolution process continues, development should be emphasized in order to fine-tune the solution. In view of these features, Jianan Huang et al proposed a Switch-Mode firefly algorithm that focused on exploration and then turned to development. Fixed random parameters are adopted in the exploration mode, and gradually reduced random parameters are adopted in the development mode, so that the conversion condition from exploration to development is automatically determined.
Through observing the spectrogram of a received signal through a large number of experiments, a ubiquitous jump phenomenon is found, namely, some jump points exist in the spectrogram, the conversion of troughs and peaks occurs on the left side and the right side of the jump points, and through further analyzing and researching the characteristics of received data, the jump phenomenon is found to influence the value of a characteristic value provided in the human behavior recognition based on a mode, so that inaccurate human behavior classification recognition is caused.
The definition of similarity in time series was proposed by Agrawal et al in 1993 and is described as follows: suppose two time series S are given1And S2A similarity metric function Dist (S)1,S2) If the sequence S1And S2Satisfies Dist (S)1,S2) If epsilon is less than or equal to epsilon, then the time sequence S is called1And S2Are similar, where ε is the time series similarity threshold, Dist (S)1,S2) Denotes S1And S2The similarity of the two time series is measured by the distance, and whether the two time series are similar or not depends on whether the variation trends are consistent or not. Due to timeThe inter-sequence data has the characteristics of high dimensionality, multiple data types and the like, certain differences may exist between any two sequences, and the factors influencing the differences are as follows: noise, amplitude translation and scaling, linear drift, discontinuities, time axis scaling, etc. Therefore, it is important how to find a suitable similarity measure for different time series data.
Disclosure of Invention
The invention aims to provide a jump phenomenon correction method for WiFi human behavior recognition, aiming at the defects of the prior art. The method can correct the influence of the jump phenomenon on the characteristic value provided by the human behavior recognition based on the mode, thereby improving the accuracy of the human behavior recognition.
The technical scheme for realizing the purpose of the invention is as follows:
different from the prior art, the jump phenomenon correction method for WiFi human behavior recognition comprises the following steps:
1) identifying a trip point: for a received set of CSI data { CSI of length L1,CSI2,...,CSINL is more than or equal to 1 and less than or equal to N, and calculating the Pearson correlation coefficient between all adjacent CSI data frames, namely the No. i CSI data frame CSIiAnd the No. i-1 CSI data frame CSIi-1The Pearson correlation coefficient of (A) is calculated as formula (1):
Figure GDA0002938741310000021
wherein the content of the first and second substances,
Figure GDA0002938741310000022
for the No. i CSI data frame CSIiIs determined by the average value of (a) of (b),
Figure GDA0002938741310000023
for the No. i CSI data frame CSIiStandard deviation of (D), if
Figure GDA0002938741310000024
Is less than 0, a trip point occurs at CSI data frame number m,
Figure GDA0002938741310000025
the length range of the normal data segment is more than or equal to L1M-1 or less, and the length range of the abnormal data segment is m-L or less2N or less, and normal data segment is { CSI1,CSI2,...,CSIm-1The abnormal data segment is { CSI }m,CSIm+1,...,CSIN}，
Figure GDA0002938741310000026
Represents a ceiling operation;
2) obtaining the optimal initial replacement position s, s is more than or equal to 1 and less than or equal to 2m-N-1, the number of segmentation segments N and the specific segmentation point { x1,x2,...,xn-1}: the independent variables of the outer layer Switch-Mode firefly algorithm are the initial position s of the replacement data segment, the number n of the segmentation segments and specific segmentation points { x1,x2,...,xn-1And obtaining values of each group of independent variables by iterative traversal, namely, in each iteration, firstly obtaining a replacement data segment { CSI (channel state information) based on the current initial replacement position ss,CSIs+1,...,CSIs+N-mIs used to replace the anomalous data segment CSIm,CSIm+1,...,CSINAnd then, according to the segmentation segment number n, an optimal specific segmentation point { x ] is obtained by using an inner layer Switch-Mode firefly algorithm1,x2,...,xn-1Dividing the corrected group of CSI data into segments, and averaging the similarities among the data segments
Figure GDA0002938741310000031
At the maximum, simultaneously, the optimal objective function value of the inner-layer Switch-Mode firefly algorithm is used as the objective function value of the current iteration of the outer-layer Switch-Mode firefly algorithm, finally, after the rest operation of the current iteration of the outer-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the outer-layer Switch-Mode firefly algorithm is completed, the initial position s, the segmentation segment number n and the specific segmentation point { x ] of the optimal replacement data segment are used as the basis1,x2,...,xn-1Get a corrected group of CSI data { CSI1,CSI2,...,CSIN}；
3) Obtaining the optimal segmentation result { x1,x2,...,xn-1}: the independent variable of the inner layer Switch-Mode firefly algorithm is a specific segmentation point { x1,x2,...,xn-1And obtaining by iterating and traversing the values of each group of independent variables, namely in each iteration, correcting a group of CSI data { CSI data according to a specific segmentation point1,CSI2,...,CSINDivide into n CSI data sections
Figure GDA0002938741310000032
Then, the similarity between any two CSI data segments is calculated according to the SimSiData model
Figure GDA0002938741310000033
Finally, the average value of all similarity results is obtained
Figure GDA0002938741310000034
Meanwhile, the average result is used as an objective function value of the current iteration of the inner-layer Switch-Mode firefly algorithm, finally, after the residual operation of the current iteration of the inner-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the inner-layer Switch-Mode firefly algorithm is completed, the optimal segmentation point { x } is obtained based on the obtained optimal segmentation point1,x2,...,xn-1Dividing the corrected group of CSI data, then calculating the similarity between any two CSI data segments according to an SimSiData model, finally obtaining the maximum average value of the similarity results between all the data segments after the corrected group of CSI data is divided in segments,
step 2) and step 3) are specifically, namely, the outer layer Switch-Mode firefly algorithm determines the initial replacement position s and the segmentation segment number n, and the inner layer Switch-Mode firefly algorithm determines the specific segmentation result x under the condition of the given segmentation segment number n1,x2,...,xn-1；
4) Any two CSI data segments to be input
Figure GDA0002938741310000041
And
Figure GDA0002938741310000042
configured as a corresponding time series: let the 56 sub-carriers of the tth data frame have values of
Figure GDA0002938741310000043
The ith CSI data segment
Figure GDA0002938741310000044
There are a total of M data frames in the frame, will
Figure GDA0002938741310000045
The CSI data in the CSI data segment are spliced end to end according to the serial numbers of the subcarriers to form a time sequence, and the time sequence corresponding to the CSI data segment is
Figure GDA0002938741310000046
5) Acquiring multi-scale shape information corresponding to the time sequence: time series using multi-scale discrete wavelets
Figure GDA0002938741310000047
Decomposing to obtain approximate coefficient sequence (CA)k-1,...,CA2,CA1Extracting key points of each approximate coefficient, and generating a key point sequence (MM)k-1,...,MM2,MM1Calculating the relative change of adjacent points in the key point sequence to perform symbolization to obtain a symbolized expression sequence { S }k-1,...,S2,S1Coding the symbolized representation sequence to obtain multi-scale shape information { W }k-1,...,W2,W1}；
6) Computing any two CSI data segments
Figure GDA0002938741310000048
And
Figure GDA0002938741310000049
similarity results of (c): definition of
Figure GDA00029387413100000410
For the ith CSI data segment
Figure GDA00029387413100000411
Corresponding multi-scale shape information, and calculating the similarity of any two CSI data segments is shown in formula (2):
Figure GDA00029387413100000412
wherein the content of the first and second substances,
Figure GDA00029387413100000413
and
Figure GDA00029387413100000414
for any two of the CSI data segments,
Figure GDA00029387413100000415
and
Figure GDA00029387413100000416
is a time sequence XiAnd Xi+1Xcom (j) is used to record whether two elements in the j scale are equal, if so, xcom (j) is 1, otherwise, xcom (j) is 0, weight (j) is the weight of the j scale, and weight (j) is 5/6k-j+1。
In the technical scheme, the similarity between any two CSI data segments is calculated by adopting an SimCiData model, and a nested Switch-Mode firefly algorithm is utilized to calculate the similarity from a normal data segment { CSI1,CSI2,...,CSIm-1Find the optimal replacement data segment in { CSI }s,CSIs+1,...,CSIs+N-mThe corrected group of CSI data is used for replacing the abnormal data section, and the whole body has continuity and is based on correctionThe characteristic value obtained by the CSI data tends to be stable, and the influence of the jump phenomenon on the characteristic value provided in the mode-based human behavior recognition is corrected.
The technical scheme is suitable for WiFi human behavior recognition based on the mode.
The method can correct the influence of the jump phenomenon on the characteristic value provided by the human behavior recognition based on the mode, thereby improving the accuracy of the human behavior recognition.
Detailed Description
The present invention will be further illustrated with reference to the following examples, but is not limited thereto.
Example (b):
a jump phenomenon correction method for WiFi human behavior recognition comprises the following steps:
1) identifying a trip point: for a received set of CSI data { CSI of length L1,CSI2,...,CSINL is more than or equal to 1 and less than or equal to N, and calculating the Pearson correlation coefficient between all adjacent CSI data frames, namely the No. i CSI data frame CSIiAnd the No. i-1 CSI data frame CSIi-1The Pearson correlation coefficient of (A) is calculated as formula (1):
Figure GDA0002938741310000051
wherein the content of the first and second substances,
Figure GDA0002938741310000052
for the No. i CSI data frame CSIiIs determined by the average value of (a) of (b),
Figure GDA0002938741310000053
for the No. i CSI data frame CSIiStandard deviation of (D), if
Figure GDA0002938741310000054
Is less than 0, a trip point occurs at the mth CSI data frame, m being
Figure GDA0002938741310000055
The length range of the normal data segment is more than or equal to L1M-1 or less, and the length range of the abnormal data segment is m-L or less2N or less, and normal data segment is { CSI1,CSI2,...,CSIm-1The abnormal data segment is { CSI }m,CSIm+1,...,CSIN}，
Figure GDA0002938741310000056
Represents a ceiling operation;
2) obtaining the optimal initial replacement position s, s is more than or equal to 1 and less than or equal to 2m-N-1, the number of segmentation segments N and the specific segmentation point { x1,x2,...,xn-1}: the independent variables of the outer layer Switch-Mode firefly algorithm are the initial position s of the replacement data segment, the number n of the segmentation segments and specific segmentation points { x1,x2,...,xn-1And obtaining values of each group of independent variables by iteration traversal, namely, in each iteration, firstly obtaining a replacement data section based on the current initial replacement position s
Figure GDA0002938741310000057
I for replacing anomalous data segments { CSIm,CSIm+1,...,CSINAnd then, according to the segmentation segment number n, an optimal specific segmentation point { x ] is obtained by using an inner layer Switch-Mode firefly algorithm1,x2,...,xn-1Dividing the corrected group of CSI data into segments, and averaging the similarities among the data segments
Figure GDA0002938741310000061
At the maximum, simultaneously, the optimal objective function value of the inner-layer Switch-Mode firefly algorithm is used as the objective function value of the current iteration of the outer-layer Switch-Mode firefly algorithm, finally, after the rest operation of the current iteration of the outer-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the outer-layer Switch-Mode firefly algorithm is completed, the initial position s, the segmentation segment number n and the specific segmentation point { x ] of the optimal replacement data segment are used as the basis1,x2,...,xn-1Get correctedSet of CSI data { CSI1,CSI2,...,CSIN}；
3) Obtaining the optimal segmentation result { x1,x2,...,xn-1}: the independent variable of the inner layer Switch-Mode firefly algorithm is a specific segmentation point { x1,x2,...,xn-1And obtaining by iterating and traversing the values of each group of independent variables, namely in each iteration, correcting a group of CSI data { CSI data according to a specific segmentation point1,CSI2,...,CSINDivide into n CSI data sections
Figure GDA0002938741310000062
Then, the similarity between any two CSI data segments is calculated according to the SimSiData model
Figure GDA0002938741310000063
Finally, the average value of all similarity results is obtained
Figure GDA0002938741310000064
Meanwhile, the average result is used as an objective function value of the current iteration of the inner-layer Switch-Mode firefly algorithm, finally, after the residual operation of the current iteration of the inner-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the inner-layer Switch-Mode firefly algorithm is completed, the optimal segmentation point { x } is obtained based on the obtained optimal segmentation point1,x2,...,xn-1Dividing the corrected group of CSI data, then calculating the similarity between any two CSI data segments according to an SimSiData model, finally obtaining the maximum average value of the similarity results between all the data segments after the corrected group of CSI data is divided in segments,
in this example, the outer layer Switch-Mode firefly algorithm determines the starting replacement position s and the number of segmentation segments n, and the inner layer Switch-Mode firefly algorithm determines the specific segmentation result x1, x for a given number of segmentation segments n2,...,xn-1；
4) Any two CSI data segments to be input
Figure GDA0002938741310000065
And
Figure GDA0002938741310000066
configured as a corresponding time series: let the 56 sub-carriers of the tth data frame have values of
Figure GDA0002938741310000067
The ith CSI data segment
Figure GDA0002938741310000068
There are a total of M data frames in the frame, will
Figure GDA0002938741310000071
The CSI data in the CSI data segment are spliced end to end according to the serial numbers of the subcarriers to form a time sequence, and the time sequence corresponding to the CSI data segment is
Figure GDA0002938741310000072
5) Acquiring multi-scale shape information corresponding to the time sequence: time series using multi-scale discrete wavelets
Figure GDA0002938741310000073
Decomposing to obtain approximate coefficient sequence (CA)k-1,...,CA2,CA1Extracting key points of each approximate coefficient, and generating a key point sequence (MM)k-1,...,MM2,MM1Calculating the relative change of adjacent points in the key point sequence to perform symbolization to obtain a symbolized expression sequence { S }k-1,...,S2,S1Coding the symbolized representation sequence to obtain multi-scale shape information { W }k-1,...,W2,W1}；
6) Computing any two CSI data segments
Figure GDA0002938741310000074
And
Figure GDA0002938741310000075
similarity results of (c): definition of
Figure GDA0002938741310000076
For the ith CSI data segment
Figure GDA0002938741310000077
Corresponding multi-scale shape information, and calculating the similarity of any two CSI data segments is shown in formula (2):
Figure GDA0002938741310000078
wherein the content of the first and second substances,
Figure GDA0002938741310000079
and
Figure GDA00029387413100000710
for any two of the CSI data segments,
Figure GDA00029387413100000711
and
Figure GDA00029387413100000712
is a time sequence XiAnd Xi+1Xcom (j) is used to record whether two elements in the j scale are equal, if so, xcom (j) is 1, otherwise, xcom (j) is 0, weight (j) is the weight of the j scale, and weight (j) is 5/6k-j+1。
In this example, it is assumed that the trip point occurs at
Figure GDA00029387413100000713
At each CSI data frame, there are: for a received set of CSI data { CSI) of length L (1 ≦ L ≦ N)1,CSI2,...,CSINCalculating the Pearson correlation coefficient between all adjacent CSI data frames
Figure GDA00029387413100000714
Identifying that a jumping point occurs at the mth CSI data frame, and the jumping point divides the CSI data into normal data segments { CSI data1,CSI2,...,CSIm-1And abnormal data segment CSIm,CSIm+1,...,CSINFinding out an optimal replacement data segment from the normal data segment for replacing the abnormal data segment, and correcting the influence of the jump phenomenon on the characteristic value provided in the human behavior recognition based on the mode;
obtaining the optimal replacement data segment { CSI (channel state information) by adopting a nested Switch-Mode firefly algorithms,CSIs+1,...,CSIs+N-mIs used to replace the anomalous data segment CSIm,CSIm+1,...,CSINGet a corrected group of CSI data { CSI1,CSI2,...,CSINAnd the inner layer Switch-Mode firefly algorithm is used for solving the optimal segmentation result { x ] under the condition of a given segmentation segment number n1,x2,...,xn-1}；
Calculating any two CSI data segments by adopting SimSiData model
Figure GDA0002938741310000081
And
Figure GDA0002938741310000082
similarity between: firstly, the data in the CSI data segment are spliced end to end according to the serial numbers of the subcarriers to construct a corresponding time sequence
Figure GDA0002938741310000083
Then, the corresponding approximate coefficient sequence { CA ] is obtained in turnk-1,...,CA2,CA1}, Key Point sequence { MMk-1,...,MM2,MM1H and a symbolized representation sequence Sk-1,...,S2,S1Finally, obtaining multi-scale shape information { W }k-1,...,W2,W1Finally, obtaining a common part xcom (j) (j is more than or equal to 1 and less than or equal to k-1) of the multi-scale shape information corresponding to the two CSI data sections, and then obtaining similarity by combining with a proportional weight factor weight (j)And (4) performing a sexual result.
The example is applicable to pattern-based WiFi human behavior recognition.

Claims (1)

1. A jump phenomenon correction method for WiFi human behavior recognition is characterized by comprising the following steps:
1) identifying a trip point: for a received set of CSI data { CSI of length L1,CSI2,...,CSINL is more than or equal to 1 and less than or equal to N, and calculating the Pearson correlation coefficient between all adjacent CSI data frames, namely the No. i CSI data frame CSIiAnd the No. i-1 CSI data frame CSIi-1The Pearson correlation coefficient of (A) is calculated as formula (1):
Figure FDA0002938741300000011
wherein the content of the first and second substances,
Figure FDA0002938741300000012
for the No. i CSI data frame CSIiIs determined by the average value of (a) of (b),
Figure FDA0002938741300000013
for the No. i CSI data frame CSIiStandard deviation of (D), if
Figure FDA0002938741300000014
Is less than 0, a trip point occurs at the mth CSI data frame, m being
Figure FDA0002938741300000015
The length range of the normal data segment is more than or equal to L1M-1 or less, and the length range of the abnormal data segment is m-L or less2N or less, and normal data segment is { CSI1,CSI2,...,CSIm-1The abnormal data segment is { CSI }m,CSIm+1,...,CSIN}，
Figure FDA0002938741300000016
Represents a ceiling operation;
2) obtaining the optimal initial replacement position s, s is more than or equal to 1 and less than or equal to 2m-N-1, the number of segmentation segments N and the specific segmentation point { x1,x2,...,xn-1}: the independent variables of the outer layer Switch-Mode firefly algorithm are the initial position s of the replacement data segment, the number n of the segmentation segments and specific segmentation points { x1,x2,...,xn-1And obtaining values of each group of independent variables by iterative traversal, namely, in each iteration, firstly obtaining a replacement data segment { CSI (channel state information) based on the current initial replacement position ss,CSIs+1,...,CSIs+N-mIs used to replace the anomalous data segment CSIm,CSIm+1,...,CSINAnd then, according to the segmentation segment number n, an optimal specific segmentation point { x ] is obtained by using an inner layer Switch-Mode firefly algorithm1,x2,...,xn-1Dividing the corrected group of CSI data into segments, and averaging the similarities among the data segments
Figure FDA0002938741300000017
At the maximum, simultaneously, the optimal objective function value of the inner-layer Switch-Mode firefly algorithm is used as the objective function value of the current iteration of the outer-layer Switch-Mode firefly algorithm, finally, after the rest operation of the current iteration of the outer-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the outer-layer Switch-Mode firefly algorithm is completed, the initial position s, the segmentation segment number n and the specific segmentation point { x ] of the optimal replacement data segment are used as the basis1,x2,...,xn-1Get a corrected group of CSI data { CSI1,CSI2,...,CSIN}；
3) Obtaining the optimal segmentation result { x1,x2,...,xn-1}: the independent variable of the inner layer Switch-Mode firefly algorithm is a specific segmentation point { x1,x2,...,xn-1And obtaining by iterating and traversing the values of each group of independent variables, namely in each iteration, correcting a group of CSI data { CSI data according to a specific segmentation point1,CSI2,...,CSINDivide into n CSI data sections
Figure FDA0002938741300000021
Then, the similarity between any two CSI data segments is calculated according to the SimSiData model
Figure FDA0002938741300000022
Finally, the average value of all similarity results is obtained
Figure FDA0002938741300000023
Meanwhile, the average result is used as an objective function value of the current iteration of the inner-layer Switch-Mode firefly algorithm, finally, after the residual operation of the current iteration of the inner-layer Switch-Mode firefly algorithm is completed, the next iteration is continued, and if the iteration of the inner-layer Switch-Mode firefly algorithm is completed, the optimal segmentation point { x } is obtained based on the obtained optimal segmentation point1,x2,...,xn-1Dividing the corrected group of CSI data, then calculating the similarity between any two CSI data segments according to an SimSiData model, and finally obtaining the maximum average value of the similarity results between all the data segments after the corrected group of CSI data is divided in segments;
4) any two CSI data segments to be input
Figure FDA0002938741300000024
And
Figure FDA0002938741300000025
configured as a corresponding time series: let the 56 sub-carriers of the tth data frame have values of
Figure FDA0002938741300000026
The ith CSI data segment
Figure FDA0002938741300000027
There are a total of M data frames in the frame, will
Figure FDA0002938741300000028
The CSI data in the CSI data segment are spliced end to end according to the serial numbers of the subcarriers to form a time sequence, and the time sequence corresponding to the CSI data segment is
Figure FDA0002938741300000029
5) Acquiring multi-scale shape information corresponding to the time sequence: time series using multi-scale discrete wavelets
Figure FDA00029387413000000210
Decomposing to obtain approximate coefficient sequence (CA)k-1,...,CA2,CA1Extracting key points of each approximate coefficient, and generating a key point sequence (MM)k-1,...,MM2,MM1Calculating the relative change of adjacent points in the key point sequence to perform symbolization to obtain a symbolized expression sequence { S }k-1,...,S2,S1Coding the symbolized representation sequence to obtain multi-scale shape information { W }k-1,...,W2,W1}；
6) Computing any two CSI data segments
Figure FDA00029387413000000211
And
Figure FDA00029387413000000212
similarity results of (c): definition of
Figure FDA00029387413000000213
For the ith CSI data segment
Figure FDA00029387413000000214
Corresponding multi-scale shape information, and calculating the similarity of any two CSI data segments is shown in formula (2):
Figure FDA00029387413000000215
wherein the content of the first and second substances,
Figure FDA0002938741300000031
and
Figure FDA0002938741300000032
for any two of the CSI data segments,
Figure FDA0002938741300000033
and
Figure FDA0002938741300000034
is a time sequence XiAnd Xi+1Xcom (j) is used to record whether two elements in the j scale are equal, if so, xcom (j) is 1, otherwise, xcom (j) is 0, weight (j) is the weight of the j scale, and weight (j) is 5/6k-j+1。
CN201910420441.XA 2019-05-20 2019-05-20 Jump phenomenon correction method for WiFi human behavior recognition Active CN110176968B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910420441.XA CN110176968B (en) 2019-05-20 2019-05-20 Jump phenomenon correction method for WiFi human behavior recognition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910420441.XA CN110176968B (en) 2019-05-20 2019-05-20 Jump phenomenon correction method for WiFi human behavior recognition

Publications (2)

Publication Number Publication Date
CN110176968A CN110176968A (en) 2019-08-27
CN110176968B true CN110176968B (en) 2021-04-06

Family

ID=67691640

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910420441.XA Active CN110176968B (en) 2019-05-20 2019-05-20 Jump phenomenon correction method for WiFi human behavior recognition

Country Status (1)

Country Link
CN (1) CN110176968B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112668439A (en) * 2020-12-23 2021-04-16 中国科学技术大学 Human body posture estimation method and device based on wireless signals

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0955746A (en) * 1995-08-11 1997-02-25 Fujitsu Ltd Phase jump prevention system for cbr signal
CN104951757A (en) * 2015-06-10 2015-09-30 南京大学 Action detecting and identifying method based on radio signals
CN106066995A (en) * 2016-05-25 2016-11-02 西安交通大学 A kind of wireless unbundling human body behavioral value algorithm
CN106228121A (en) * 2016-07-15 2016-12-14 华中科技大学 Gesture feature recognition methods and device
CN108553108A (en) * 2018-03-05 2018-09-21 叶伟 It is a kind of based on the human action of CSI signals in Wi-Fi and the detection method of breathing and system

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0955746A (en) * 1995-08-11 1997-02-25 Fujitsu Ltd Phase jump prevention system for cbr signal
CN104951757A (en) * 2015-06-10 2015-09-30 南京大学 Action detecting and identifying method based on radio signals
CN106066995A (en) * 2016-05-25 2016-11-02 西安交通大学 A kind of wireless unbundling human body behavioral value algorithm
CN106228121A (en) * 2016-07-15 2016-12-14 华中科技大学 Gesture feature recognition methods and device
CN108553108A (en) * 2018-03-05 2018-09-21 叶伟 It is a kind of based on the human action of CSI signals in Wi-Fi and the detection method of breathing and system

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
A Switch-Mode Firefly Algorithm for Global Optimization;Huang Jian 等;《IEEE Access》;20180917;第54177-54184页 *
Wi-Fi CSI-Based Behavior Recognition: From Signals and Actions to Activities;Zhu Wang 等;《 IEEE Communications Magazine》;20180517;第109-115页 *
基于WLAN信道状态信息的人体行为识别算法研究;王向勇;《中国优秀硕士学位论文全文数据库 信息科技辑》;20180415;I138-3041 *

Also Published As

Publication number Publication date
CN110176968A (en) 2019-08-27

Similar Documents

Publication Publication Date Title
CN115659160B (en) Data quality measurement method for digital twin model optimization
US8427346B2 (en) Adaptive compression
JP6622329B2 (en) Target value estimation system, target value estimation method, and target value estimation program
CN109255395B (en) Service life prediction method of ball screw pair
CN109376892B (en) Equipment state prediction method based on life cycle stage of equipment
CN110705722A (en) Diagnostic model for industrial equipment fault diagnosis and construction method and application thereof
CN110176968B (en) Jump phenomenon correction method for WiFi human behavior recognition
Omitaomu et al. Online support vector regression with varying parameters for time-dependent data
JP4265296B2 (en) Change point detection apparatus, change point detection method, and change point detection program
CN117454096B (en) Motor production quality detection method and system
CN113569758B (en) Time sequence action positioning method, system, equipment and medium based on action triplet guidance
KR102100386B1 (en) Method for Kalman filtering using measurement noise recommendation, and recording medium thereof
CN108415880B (en) Line loss characteristic analysis method based on sample entropy and wavelet transformation
CN110032585B (en) Time sequence double-layer symbolization method and device
Rivero et al. Using genetic algorithms and k-nearest neighbour for automatic frequency band selection for signal classification
WO2014141682A1 (en) Sensor terminal and signal acquisition method
CN116595465A (en) High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement
US20220207382A1 (en) Apparatus and method for refining data and improving performance of behavior recognition model by reflecting time-series characteristics of behavior
He et al. A two-stage deep graph clustering method for identifying the evolutionary patterns of the time series of animation view counts
JP2018156619A (en) Global search device and program for consecutive optimization problem
CN113792879A (en) Case reasoning attribute weight adjusting method based on introspection learning
US20220300676A1 (en) Information processing apparatus and information processing method
JP7081708B1 (en) Neural network weight reduction device, neural network weight reduction method and program
CN117194892A (en) Method and device for determining peak value and valley value of signal and electronic equipment
US20230053174A1 (en) Data amount sufficiency determination device, data amount sufficiency determination method, learning model generation system, trained model generation method, and medium

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
EE01 Entry into force of recordation of patent licensing contract
EE01 Entry into force of recordation of patent licensing contract

Application publication date: 20190827

Assignee: Guilin zhouyou Technology Co.,Ltd.

Assignor: GUILIN University OF TECHNOLOGY

Contract record no.: X2022450000016

Denomination of invention: A Jump Correction Method for WiFi Human Behavior Recognition

Granted publication date: 20210406

License type: Common License

Record date: 20221116