WO2018006798A1

WO2018006798A1 - Wireless sensing-based temperature detection method and system

Info

Publication number: WO2018006798A1
Application number: PCT/CN2017/091670
Authority: WO
Inventors: 伍楷舜; 王璐; 张翔
Original assignee: 深圳大学
Priority date: 2016-07-05
Filing date: 2017-07-04
Publication date: 2018-01-11
Also published as: CN107294621B; CN106198868A; CN107294621A

Abstract

A wireless sensing-based temperature detection method and system. The method is utilized to acquire, by a wireless signal acquisition device, a wireless signal transmitted from a transmitter, and evaluate channel state information to achieve the goal of detecting ambient humidity. The method comprises acquiring, by a wireless signal acquisition device arranged in a specific layout, a wireless signal transmitted from a transmitter. The wireless sensing-based temperature detection method has advantages of using existing wireless networks and equipment without adding another specific detection apparatus, providing high coverage.

Description

Method and system for detecting humidity based on wireless sensing

Technical field

The present invention relates to the field of wireless detection, and in particular, to a method and system for detecting humidity based on wireless sensing.

Background technique

In recent years, wireless network technology has developed rapidly, and people have tried to apply it to all aspects of life. At the same time, with the development of the times and the progress of life, people are paying more and more attention to the relationship between themselves and nature. As an environmental parameter, atmospheric humidity strongly influences the natural economy and plays an important role in various environmental processes. For example, the growth of various crops is inseparable from the proper humidity. In combination with current hotspots, wireless sensing technology is used to conduct research on atmospheric humidity monitoring, so that technology can better serve life.

Existing humidity measurement methods can be mainly divided into two types. First, the look-up method, the common hygrometer is the method used. The general principle is as follows: Since the temperature sensitive bubble of the wet bubble thermometer is wrapped with cotton yarn, the lower end of the cotton yarn is immersed in water, and the evaporation of water makes the temperature indication of the wet bubble thermometer always lower than the temperature indication of the dry bubble thermometer, and this The temperature difference is related to the speed at which the water evaporates (ie the relative humidity at that time). The relative humidity of the air can be found from the table or curve based on the readings of the two thermometers. Second, the attenuation method is based on the principle that the electromagnetic wave is attenuated by the influence of water vapor when it is transmitted in the air. By measuring the RSSI (Received Signal Strength Indicator) of the transmitting end and the receiving end, the humidity is estimated according to the degree of attenuation. .

Summarize the above two methods. Method one, the physical phenomenon is used to measure the humidity, and the human error is large. And when we need to know the humidity values at the same time, it is necessary to place multiple hygrometers, so the cost is too high. In the second method, the effect of water vapor on signal transmission is small. Therefore, in actual operation, it is necessary to measure the signal strength after transmission over a long distance (up to several kilometers). Over long distances, other weather factors, such as rain, snow, and fog, increase the likelihood of signal effects, resulting in limited measurements. Moreover, the long distance itself increases the inconvenience of measurement.

Summary of the invention

In order to overcome the deficiencies in the existing measurement methods referred to above, the present invention provides a method and system for humidity detection based on wireless sensing. Using a widely deployed wireless signal transmitting device, such as a commercial router, in a specific layout, through a collected CSI (Channel State Information) data, a series of algorithms are used to analyze the data to detect the surrounding environment. Humidity information. This method has a good realistic foundation: wireless networks have been widely deployed. Compared with the previous methods, the measurement cost is reduced and the applicability is improved.

The invention is achieved by the following technical solutions:

A wireless sensing based humidity detection method includes the following steps:

S1: The wireless receiving end receives the wireless signal from the wireless transmitting end, and evaluates channel state information; wherein each wireless channel includes multiple subcarriers, and the channel state information of the wireless channel includes CIS values of multiple subcarriers in the channel. average value;

S2, CSI data feature extraction, output target mode, that is, performing data processing on the obtained state information of multiple wireless channels, and obtaining a feature parameter set characterizing an environment space in which the plurality of wireless channels are located;

S3. Using the SVM to classify the target mode outputted in step S2 to detect the humidity information of the surrounding environment, including: inputting the obtained feature parameter set into a classification model corresponding to the characteristic parameter of the ambient humidity and the wireless channel set in advance. , classify or categorize it to obtain environmental humidity;

S4: feedback the response information of the detection result, adjust the parameters of the classification algorithm, and further improve the accuracy, including adjusting the parameters of the classification model according to the parameter values and the detection results of the feature parameter set obtained by the current detection.

As a further improvement of the present invention, the step S1 further includes:

S11. Simultaneously measure and obtain CSI values of consecutive T subcarriers among N subcarriers in the radio channel in a wireless channel.

S12. The average value of the obtained T CSI values is used as channel state information of the wireless channel.

S13. Perform the foregoing steps simultaneously or sequentially in the M wireless channels to obtain M channel state information.

Where M, N and T are both natural numbers greater than 1, and N is greater than T.

As a further improvement of the present invention, the step S2 further includes:

S21. Performing, for each of the obtained M channel state information, a mean value, a normalized standard deviation, an average absolute deviation, a quartile range, and a signal entropy value, respectively, and obtaining an average value of the M channel state information, Normalized standard deviation, mean absolute deviation, quartile range, and signal entropy;

S22. The mean value, the normalized standard deviation, the average absolute deviation, the quartile range, and the signal entropy value obtained in the above are placed into a vector set to obtain a feature parameter set of the current detection.

As a further improvement of the present invention, the predetermined classification model includes a mapping set of the feature parameter set, the mapping set includes a plurality of feature parameter sets obtained by setting the ambient humidity and the setting The set of elements of the ambient humidity; the step S3 further comprising:

S31. Compare the currently obtained feature parameter set with the feature parameter set in the classification model one by one, and find a set element that is closest to all the feature parameter values in the currently obtained feature parameter set;

S32. Select an ambient humidity in the set element as the current ambient humidity.

As a further improvement of the present invention, the set environmental humidity values in the map set are linearly distributed within the measurement range.

As a further improvement of the present invention, adjusting the classification model includes forming the currently acquired feature parameter set and the currently selected environmental humidity into the set element and adding to the mapping set.

The invention also provides a wireless sensing based humidity detection system, comprising:

a CSI obtaining module, configured to obtain channel state information of each of the plurality of wireless channels; wherein each wireless channel includes multiple subcarriers, and channel state information of the wireless channel includes CIS values of multiple subcarriers in the channel average value;

a feature parameter set obtaining module, configured to perform data processing on the obtained status information of the plurality of wireless channels, to obtain a feature parameter set that characterizes an environment space where the plurality of wireless channels are located;

a detection module module, configured to input the obtained feature parameter set into a classification model corresponding to the preset characteristic parameter of the ambient humidity and the wireless channel, and classify or categorize the same, thereby obtaining an environmental humidity;

The feedback module is configured to adjust parameters of the classification model according to parameter values and detection results in the feature parameter set obtained by the current detection.

As a further improvement of the present invention, the CSI acquisition module includes:

a sensing unit, configured to collect initial channel state data, where the initial channel state data includes CSI values of N subcarriers in M spatial streams, and M and N are natural numbers greater than 1;

The data processing unit is configured to obtain an average value of CSI values of T consecutive subcarriers at the same time point for each spatial stream, and use the average value as channel state information, and T is a natural number greater than 1 and less than N.

As a further improvement of the present invention, the feature parameter obtaining module further includes:

The calculating unit is configured to perform feature extraction on the CSI data, and the extracted features include: mean, normalized standard deviation, average absolute deviation, quartile range, and signal entropy value;

Establishing a model unit for pre-establishing a high-dimensional feature model for changing channel state information caused by various humidity in the set space as a training sample based on statistical learning theory;

The detecting unit is configured to map the output target mode to the high-dimensional feature model of the support vector machine, and separate the target humidity class.

As a further improvement of the present invention, the feedback module includes: the feedback module further comprising: a feedback for the current environment humidity detection response information, and adjusting the high-dimensional feature model of the support vector machine.

The invention has the beneficial effects that CSI is a better evaluation of the wireless propagation channel with respect to RSSI; the present invention utilizes the advantages of CSI to design a wireless sensing based humidity detection system and enables the correction module to be more Accurately detect the humidity information of the surrounding environment; the detection method is based on the existing wireless network and equipment, and performs humidity detection work. It is not necessary to install other specific detection equipment in the detected environment, and has high popularity.

DRAWINGS

1 is a schematic diagram of a system configuration of wireless sensing based humidity detection according to an embodiment of the present invention;

2 is a schematic flow chart showing the implementation of the humidity detecting method of the present invention;

3 is a schematic flow chart showing an implementation of a humidity detecting method according to an embodiment of the present invention;

4 is a skeleton diagram of a humidity detecting system according to an embodiment of the present invention.

detailed description

To facilitate the understanding of those skilled in the art, the present invention will be further described in conjunction with the drawings and examples.

Support Vector Machine is the SVM

S1. The wireless receiving end receives the wireless signal from the wireless transmitting end, and evaluates channel state information.

S2, CSI data feature extraction, output target mode;

S3. The target mode output in step S2 is classified by using SVM to detect humidity information of the surrounding environment;

S4: feedback the response information of the detection result, adjust the parameters of the classification algorithm, and further improve the accuracy.

The step S1 evaluating channel state information includes:

S11. Acquire initial channel state data, based on a multiple input multiple output technology, where the initial channel state data includes CSI values of N subcarriers in M spatial streams, and both M and N are natural numbers greater than 1.

S12. For each spatial stream, obtain an average value of CSI values of T consecutive subcarriers at the same time point, and use the average value as channel state information, where T is a natural number greater than 1 and less than N;

S13. Filtering channel state information by using a data filtering technology to reduce interference of channel state information caused by object movement in a surrounding environment.

The step S2 includes extracting five features of the mean value Mean, the normalized standard deviation The Normalized Standard Deviation, the mean absolute deviation Mean Absolute Deviation, the quartile range Interquartile Range, and the signal entropy value Signal Entropy.

The classification algorithm in which the step S3 classifies according to the characteristics of the CSI data is a support vector machine SVM method.

The step S3 includes:

S31. Based on the statistical learning theory, pre-establishing a high-dimensional feature model that changes channel state information caused by various humidity in the set space as a training sample;

S32. The target mode output in step S2 is classified by using SVM to detect the humidity information of the surrounding environment.

As a further improvement of the present invention, in the step S4, the response information for the humidity detection result is fed back, and the high-dimensional feature model of the support vector machine is adjusted.

In summary, in the embodiment, in the above humidity detecting method, a connection of a wireless channel is established between a wireless receiving end and a wireless transmitting end, and channel state information is evaluated; wherein each wireless channel includes multiple subcarriers. The channel state information of the wireless channel includes an average value of CIS values of the plurality of subcarriers in the channel. After the connection is established, performing data processing on the obtained state information of the multiple wireless channels to obtain the plurality of wireless channels. The characteristic parameter set of the environment space; the obtained feature parameter set is input into a classification model corresponding to the characteristic parameter of the environment humidity and the wireless channel, and is classified or attributed to obtain the environmental humidity; The parameter values and detection results in the feature parameter set obtained in this test are adjusted for the parameters of the classification model.

The specific process includes: simultaneously measuring and obtaining CSI values of consecutive T subcarriers among N subcarriers in the radio channel in a wireless channel; and obtaining an average value of the obtained T CSI values as channel state information of the radio channel; The above steps are performed simultaneously or sequentially in the M wireless channels to obtain M channel state information; wherein, M, N, and T are both natural numbers greater than 1, and N is greater than T.

Then, the obtained M channel state information are respectively calculated for the mean value, the normalized standard deviation, the average absolute deviation, the quartile range, and the signal entropy value, and the mean values of the M channel state information are respectively obtained. Normalized standard deviation, mean absolute deviation, quartile range, and signal entropy values; the mean, normalized standard deviation, mean absolute deviation, quartile range, and signal entropy values obtained above are placed into a vector set , get the feature parameter set of this test.

In this embodiment, since the preset classification model includes a mapping set of the feature parameter set, the mapping set includes a plurality of feature parameter sets obtained by setting the ambient humidity and the setting. The set of elements of the ambient humidity; therefore, in the subsequent steps, the parameters of the currently obtained feature parameter set and the feature parameter set in the classification model may be compared one by one to find all the feature parameters of the currently obtained feature parameter set. The closest collection element; then select the ambient humidity in the collection element to be the current ambient humidity.

In this embodiment, the set environmental humidity values in the mapping set are linearly distributed within the measurement range. Adjusting the classification model includes forming the currently acquired feature parameter set and the currently selected environmental humidity into the set element and adding to the mapping set.

The CSI obtaining module is configured to receive the wireless signal from the wireless transmitting end and evaluate the channel state information, and the detecting module is configured to extract the feature according to the CSI data collected by the CSI acquiring module, perform classification matching, and detect the current environment. Humidity information;

a feedback module, configured to compare the detected result of the detecting module with a known category, and if the deviation occurs, perform correction, thereby making the classification algorithm more accurate;

A display module that displays the detected results.

The CSI obtaining module includes:

a sensing unit, configured to collect initial channel state data, where the initial channel state data includes CSI values of N subcarriers in M spatial streams, and M and N are natural numbers greater than 1; data processing a unit, configured to obtain an average value of CSI values of T consecutive subcarriers at the same time point for each spatial stream, and use the average value as channel state information, and T is a natural number greater than 1 and less than N;

A filtering unit is configured to filter channel state information by using a data filtering technology.

The detecting module includes:

The feedback module includes: a feedback for the current environment humidity detection response information, and a high-dimensional feature model of the support vector machine; the display module includes but is not limited to a mobile terminal screen, a palmtop computer, or a liquid crystal display.

In an embodiment, a method for wireless sensing based humidity detection, the steps comprising:

S2, CSI data feature extraction, the collected CSI data for Mean, The Normalized Standard Deviation, Mean Absolute Deviation, Interquartile Range and Signal Entropy processing, output target mode;

Specifically, in step S1, evaluating channel state information includes:

When the system of the present invention starts to work, the wireless transmitting end propagates the wireless network signal, and the wireless receiving end (such as a computer equipped with a network card) in a specific area collects CSI as initial channel state data, and then performs data processing.

Please refer to FIG. 1, which is an experimental layout diagram of the entire experimental scene.

The step S2 includes performing Mean, the Normalized Standard Deviation, the Mean Absolute Deviation, the Interquartile Range, and the signal entropy value (Signal) of the CSI data. Entropy) A total of five aspects of feature extraction.

The step S3 includes:

The step S4 includes: feeding back response information for the humidity detection result, and adjusting a high-dimensional feature model of the support vector machine.

As shown in the flow chart of FIG. 2, four important steps of the detection method of the present invention are disclosed, including: channel state evaluation, CSI data feature extraction, humidity classification, and feedback detection results.

In another embodiment, as shown in FIG. 3, the present invention further provides an implementation flow of a humidity detecting method according to an embodiment, and the steps thereof include:

S301. The wireless receiving end receives the wireless signal from the wireless transmitting end, and simultaneously collects initial channel state data.

S302. Take a CSI average value of the combined subcarriers as channel state information.

S303. Filter channel state information.

S304. Perform feature extraction on channel state information.

S305. Output target mode;

S306. Mapping the target mode to a high-dimensional feature model of the support vector machine;

S307. Perform classification by using a support vector machine.

S308, whether the humidity information of the current environment is detected, if yes, step S309 is performed, otherwise returns to step S301;

S309: Display detection results in the display module, adjust parameters, and optimize detection and classification algorithms.

The present invention also provides a humidity detecting system, as shown in FIG. 4, comprising:

The CSI obtaining module 41 is configured to receive, by the wireless receiving end, a wireless signal from the wireless transmitting end, and evaluate channel state information;

The detecting module 42 is configured to extract a feature according to the CSI data collected by the CSI acquiring module, output a target mode, and map the output target mode to a high-dimensional feature model of the support vector machine to separate the target humidity class;

The feedback module 43 is configured to compare the detected result of the detecting module with a known category, and if a deviation occurs, perform correction, thereby making the classification algorithm more accurate;

The display module 44 can use the display screen of the mobile phone or other display module to display the detected result.

The CSI obtaining module includes:

The sensing unit 411 is configured to collect initial channel state data, where the initial channel state data includes CSI values of N subcarriers in M spatial streams, and M and N are natural numbers greater than 1;

The data processing unit 412 is configured to obtain an average value of CSI values of T consecutive subcarriers at the same time point for each spatial stream, and use the average value as channel state information, and T is a natural number greater than 1 and less than N. ;

The filtering unit 413 is configured to filter channel state information by using a data filtering technology.

The detecting module includes:

The calculating unit 421 is configured to perform feature extraction on the CSI data, and the extracted features include: Mean, Normalized Standard Deviation, Mean Absolute Deviation, Interquartile Range) and signal entropy (Signal Entropy), output target mode;

Establishing a model unit 422, configured to pre-establish a high-dimensional feature model that changes channel state information caused by various humidity in the set space as a training sample based on statistical learning theory;

The humidity identification unit 423 is configured to map the output target mode to the high-dimensional feature model of the support vector machine, and separate the target humidity class.

The humidity detection system of the present invention further includes a feedback module 43 for feeding back response information for the current environmental humidity detection and adjusting the high dimensional feature model of the support vector machine.

The humidity detecting system of the present invention further includes a display module 44 for displaying the final detection result, and the display module is a mobile terminal screen, a palmtop computer, a liquid crystal display, and other display devices (such as projectors, etc.) for displaying content.

The above is a further detailed description of the present invention in combination with the specific preferred embodiments, and the specific embodiments of the present invention are not limited to the above description. It will be apparent to those skilled in the art that <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt;

Claims

A wireless sensing-based humidity detecting method, comprising the steps of:

S1: The wireless receiving end receives the wireless signal from the wireless transmitting end, and evaluates channel state information; wherein each wireless channel includes multiple subcarriers, and the channel state information of the wireless channel includes CIS values of multiple subcarriers in the channel. average value;

S2, CSI data feature extraction, output target mode, that is, performing data processing on the obtained state information of multiple wireless channels, and obtaining a feature parameter set characterizing an environment space in which the plurality of wireless channels are located;

S3. Using the SVM to classify the target mode outputted in step S2 to detect the humidity information of the surrounding environment, including: inputting the obtained feature parameter set into a classification model corresponding to the characteristic parameter of the ambient humidity and the wireless channel set in advance. , classify or categorize it to obtain environmental humidity;

S4: feedback the response information of the detection result, adjust the parameters of the classification algorithm, and further improve the accuracy, including adjusting the parameters of the classification model according to the parameter values and the detection results of the feature parameter set obtained by the current detection.
The wireless sensing-based humidity detecting method according to claim 1, wherein the step S1 further comprises:

S11. Simultaneously measure and obtain CSI values of consecutive T subcarriers among N subcarriers in the radio channel in a wireless channel.

S12. The average value of the obtained T CSI values is used as channel state information of the wireless channel.

S13. Perform the foregoing steps simultaneously or sequentially in the M wireless channels to obtain M channel state information.

Where M, N and T are both natural numbers greater than 1, and N is greater than T.
The wireless sensing-based humidity detecting method according to claim 2, wherein the step S2 further comprises:

S21. Performing, for each of the obtained M channel state information, a mean value, a normalized standard deviation, an average absolute deviation, a quartile range, and a signal entropy value, respectively, and obtaining an average value of the M channel state information, Normalized Standard deviation, mean absolute deviation, quartile range, and signal entropy;

S22. The mean value, the normalized standard deviation, the average absolute deviation, the quartile range, and the signal entropy value obtained in the above are placed into a vector set to obtain a feature parameter set of the current detection.
The wireless sensing-based humidity detecting method according to claim 3, wherein the predetermined classification model includes a mapping set of the feature parameter set, and the mapping set includes a plurality of settings a set of characteristic parameters obtained under ambient humidity and a set element of the set ambient humidity; the step S3 further comprising:

S31. Compare the currently obtained feature parameter set with the feature parameter set in the classification model one by one, and find a set element that is closest to all the feature parameter values in the currently obtained feature parameter set;

S32. Select an ambient humidity in the set element as the current ambient humidity.
The wireless sensing-based humidity detecting method according to claim 4, wherein the set environmental humidity value in the mapping set is linearly distributed within the measurement range.
The wireless sensing-based humidity detecting method according to claim 5, wherein the adjusting the classification model comprises forming the currently obtained feature parameter set and the currently selected environmental humidity into the set element, and adding the set element In the mapping set.
A wireless sensing based humidity detection system, comprising:

a CSI obtaining module, configured to obtain channel state information of each of the plurality of wireless channels; wherein each wireless channel includes multiple subcarriers, and channel state information of the wireless channel includes CIS values of multiple subcarriers in the channel average value;

a feature parameter set obtaining module, configured to perform data processing on the obtained status information of the plurality of wireless channels, to obtain a feature parameter set that characterizes an environment space where the plurality of wireless channels are located;

a detection module module, configured to input the obtained feature parameter set into a classification model corresponding to the preset characteristic parameter of the ambient humidity and the wireless channel, and classify or categorize the same, thereby obtaining an environmental humidity;

The feedback module is configured to adjust parameters of the classification model according to parameter values and detection results in the feature parameter set obtained by the current detection.
The wireless sensing-based humidity detection system according to claim 7, wherein the CSI acquisition module comprises:

a sensing unit, configured to collect initial channel state data, where the initial channel state data includes CSI values of N subcarriers in M spatial streams, and M and N are natural numbers greater than 1;

The data processing unit is configured to obtain an average value of CSI values of T consecutive subcarriers at the same time point for each spatial stream, and use the average value as channel state information, and T is a natural number greater than 1 and less than N.
The wireless sensor-based humidity detecting system according to claim 8, wherein the feature parameter obtaining module further comprises:

The calculating unit is configured to perform feature extraction on the CSI data, and the extracted features include: mean, normalized standard deviation, average absolute deviation, quartile range, and signal entropy value;

Establishing a model unit for pre-establishing a high-dimensional feature model for changing channel state information caused by various humidity in the set space as a training sample based on statistical learning theory;

The detecting unit is configured to map the output target mode to the high-dimensional feature model of the support vector machine, and separate the target humidity class.
The wireless sensing-based humidity detecting system according to claim 9, wherein the feedback module further comprises: a feedback high-dimensional feature model of the support vector machine for feeding back response information for the current environmental humidity detection.