[go: up one dir, main page]

WO2018133264A1 - 一种人体室内定位自动检测方法及系统 - Google Patents

一种人体室内定位自动检测方法及系统 Download PDF

Info

Publication number
WO2018133264A1
WO2018133264A1 PCT/CN2017/084427 CN2017084427W WO2018133264A1 WO 2018133264 A1 WO2018133264 A1 WO 2018133264A1 CN 2017084427 W CN2017084427 W CN 2017084427W WO 2018133264 A1 WO2018133264 A1 WO 2018133264A1
Authority
WO
WIPO (PCT)
Prior art keywords
human body
positioning
wireless signal
channel state
information
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.)
Ceased
Application number
PCT/CN2017/084427
Other languages
English (en)
French (fr)
Inventor
伍楷舜
黄梓琪
王璐
明仲
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.)
Shenzhen University
Original Assignee
Shenzhen University
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 Shenzhen University filed Critical Shenzhen University
Publication of WO2018133264A1 publication Critical patent/WO2018133264A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
    • G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
    • G01S5/0257—Hybrid positioning

Definitions

  • the present invention relates to positioning technology, and in particular to a method for automatically detecting indoor positioning of a human body, and to a system for implementing the method.
  • the detection system built by these methods has its own shortcomings.
  • the visual positioning detection system it is necessary to mount a high-resolution camera in a detection environment in which the subject is located to take a large number of images, and then to track the positioning by the captured images.
  • installing the camera will in some way infringe on the personal privacy of the subject.
  • the visual positioning detection system cannot work effectively in dark conditions.
  • the wavelength should not be too short, otherwise it will be greatly affected by scattering, reflection, and multipath, which makes the positioning accuracy low.
  • other sounds or other environmental factors around the subject can easily interfere with the operation of the environmental device, resulting in lower accuracy.
  • UWB Ultra Wideband, carrierless communication technology
  • UWB pulse radio technology for indoor positioning, although the average transmission power is very low, can be "quiet coexistence" with other wireless communication systems, low energy consumption, low cost, good confidentiality, anti- Multipath interference and other advantages, but at the same time, the pulse UWB system has low spectrum utilization and is not suitable for high data rate transmission. Another important reason is UWB positioning. It is necessary to be additionally tagged by people and objects (relative to the natural label of the terminal such as a mobile phone), and these methods will cause a certain degree of inconvenience to the life of the subject.
  • the present invention provides a method for automatically detecting indoor positioning of a human body, and a system for realizing the automatic detection method for indoor positioning of the human body.
  • the method for automatically detecting indoor positioning of a human body comprises the following steps:
  • S1 the wireless receiving end receives the wireless signal transmitted by the wireless signal transmitter, and uploads the wireless signal information to the server;
  • the present invention is further improved, and further includes a parameter correction step A: correcting the parameters used in the algorithm detected in step S3 by using a sensor attached to the human body.
  • the present invention is further improved.
  • the sensor includes a sensor attached to the mobile phone, and the sensor included in the mobile phone includes a gravity sensor, an acceleration sensor, and a gyroscope.
  • the invention is further improved, and the parameter correction method comprises:
  • A1 setting the start and end of the human body motion, recording the motion characteristics with the sensor, and recording the time period;
  • A2 The sensor calculates the number of steps of human motion during this time period
  • A3 The speed of the human body is calculated by the time period and the number of steps, and the parameters of the detection algorithm are corrected.
  • step S2 evaluating channel state information includes the following steps:
  • S21 Collect channel state data, where the channel state data includes CSI values of M subcarriers in N spatial streams, where N and M are both natural numbers greater than one;
  • S22 For each spatial stream, obtain an average value of CSI values of P consecutive subcarriers at the same time point, and use the average value as channel state information, and P is a natural number greater than 1 and less than M;
  • step S3 includes the following steps:
  • S32 receiving wireless signal transmitter coordinates in the network layer and channel state information CSI from the physical layer;
  • step S3 further includes step S34: adaptively modifying parameters in the positioning algorithm in step S33 by using a fingerprint card algorithm.
  • step S3 further includes step S35: establishing a database, and positioning the bit Set the sample to map to the radio map.
  • step S35 a statistical machine learning is used to create a determination region for the channel response mode, and the received wireless signals are cross-correlated for location identification.
  • the method for the statistical machine to learn to identify the location includes the following steps:
  • B2 statistics the cross-correlation of the received wireless signals of the overall statistical information of the channel responses at each grid point;
  • B4 The Mahalanobis distance is calculated according to the multidimensional Gaussian distribution and the maximum likelihood estimation method, and the human body position is determined by the best match with the radio signal in the radio map in the indoor propagation model.
  • the present invention also provides a system for implementing the method, comprising: a wireless signal transmitter for transmitting a wireless signal; and a wireless receiving end for receiving a wireless signal transmitted by the wireless signal transmitter and uploading the wireless signal information to the server, Obtaining the positioning result returned by the server; the server is configured to receive the wireless signal information of the wireless receiving end, and evaluate the channel state information, detect the positioning information of the human body according to the change of the channel state information, and then return the calculated positioning result to the wireless receiving end.
  • the invention has the beneficial effects that the detection accuracy of the detected action is 84% to 94% in an indoor environment with less decoration, and in an indoor environment with more decorations, the detection is performed.
  • Accuracy rate can reach 78%; can accurately locate the human body indoors, and use the system's self-learning function to deal with false positives, further reduce the false positive rate;
  • based on the existing wireless network and equipment, carry out indoor Inspection work no need to install other specific detection equipment in the tested environment, can be used in any environment spread over the wireless network, has a very high popularity, and the testee does not need to carry any additional sensing equipment, avoiding being detected
  • the inconvenience caused by carrying the testing equipment provides convenience for their lives.
  • FIG. 1 is a schematic structural diagram of a system according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of a server implementation process of the present invention
  • FIG. 3 is a flow chart of a method according to an embodiment of the present invention.
  • the automatic indoor positioning automatic detection system of the present invention comprises a wireless signal transmitter for transmitting a wireless signal, and a wireless receiving end for receiving a wireless signal transmitted by the wireless signal transmitter and uploading the wireless signal information to the server.
  • the wireless receiving end is a wireless network card
  • the wireless signal transmitter is a wireless router
  • the server is a mobile phone
  • the method is based on a radio propagation mechanism in an indoor environment, and establishes a wireless signal and a human body movement.
  • the relationship between the distances only needs to use the existing wireless network equipment of the home, that is, the change of the wireless signal caused by the distance of the detected person can be analyzed, thereby obtaining a precise indoor positioning position.
  • the rich channel state information of the wireless network can be collected through the wireless network card.
  • the system has a plurality of antennas for transmitting and receiving wireless signals, respectively; the wireless network card used by the system can receive channel state information.
  • the number of the wireless signal transmitters is one or two, and the number of the wireless receiving ends is one or more. Preferably, the number of wireless signal transmitters and wireless receiving ends is 2 or 3 respectively.
  • the first wireless receiving end receives the CSI from the first wireless signal transmitter (the abbreviation of Channel State Information, that is, channel state information, in the field of wireless communication, CSI is the channel attribute of the communication link, and describes the signal in each transmission path.
  • the weakening factor the second wireless receiving end receives the CSI from the second wireless signal transmitter. In the environment being tested, the subject does not need to carry other additional equipment.
  • the system will use the CSI received by the two wireless receiving ends to detect the motion of the detected object, and thereby determine the motion or position of the detected object, such as whether it falls or the like.
  • the present invention uses the channel state information CSI of the wireless network as an indicator.
  • CSI can describe the propagation path of a signal under the combined influence of time delay, amplitude attenuation, and phase shift.
  • the present invention establishes a link between CSI and human body travel distance.
  • a particular indoor environment such as a room
  • the present invention uses Orthogonal Frequency Division Multiplex (OFDM) to obtain CSI in the form of a subcarrier.
  • OFDM Orthogonal Frequency Division Multiplex
  • the method for automatically detecting the indoor positioning of the human body based on the above system includes the following steps:
  • the wireless receiving end receives the wireless signal transmitted by the wireless signal transmitter, and uploads the wireless signal information to the server.
  • the wireless signal transmitter will propagate the wireless network signal
  • the wireless receiving end (such as the mobile terminal) in a specific area collects the CSI as the initial channel state data and uploads it to the server. After the data processing through the server.
  • the server receives the wireless signal information of the wireless receiving end, and evaluates the channel state information, where the evaluating the channel state information includes the following steps:
  • S21 Collect channel state data, where the channel state data includes CSI values of M subcarriers in N spatial streams, where N and M are both natural numbers greater than one;
  • S22 For each spatial stream, obtain an average value of CSI values of P consecutive subcarriers at the same time point, and use the average value as channel state information, and P is a natural number greater than 1 and less than M;
  • the present invention uses a 3 ⁇ 3 Multiple-Input Multiple-Out-put (MIMO) as an example.
  • MIMO Multiple-Input Multiple-Out-put
  • the initial channel state data obtained during the sensing phase is divided into 9 spatial streams, and each There will be 30 subcarriers in a stream. Changes in the distance traveled by the human body will affect the data contained in different spatial streams, and will have a similar effect on all subcarriers in each spatial stream.
  • environmental factors such as temperature, room settings
  • the CSI values of 30 subcarriers in each independent spatial stream are combined and combined into a single channel state information.
  • the present invention utilizes a data filtering technique and a moving average method, specifically, using a weighted moving average to smooth the channel state information processed by the above to reduce noise in the data. .
  • the server detects the positioning information of the human body according to the change of the channel state information.
  • the processing method for detecting human body positioning information includes the following steps:
  • S32 receiving wireless signal transmitter coordinates in the network layer and channel state information CSI from the physical layer;
  • step S3 further includes step S34: adaptively modifying parameters in the positioning algorithm in step S33 by using a fingerprint card algorithm. In order to achieve precise positioning with a small amount of training samples.
  • the present invention develops an indoor propagation model based on CSI information and distance.
  • the first step is to collect multiple grouped CSI data at two points to train the environmental factor and the path loss fading index of the indoor propagation model; the second step: test the efficiency of the parameter estimation using the CSI collected at the third point.
  • an indoor propagation model based on CSI information and distance is established, and the distance between the unknown point and the known point can be calculated by the CSI information.
  • the unknown point must be located on a sphere with a known radius as the center of the sphere.
  • the unknown point is at the intersection of the three sphere circumferences.
  • a point on the back side of the reception can be discarded, so that the position of the unknown point can be accurately measured.
  • the present invention uses a fingerprint method to correct the positioning algorithm.
  • Beacon information from a wireless signal transmitter is received at each sample point location, the beacon information including channel response information for a plurality of subcarriers.
  • the mobile receiver receives 30 sets of CSI information at the same time.
  • step S3 of this example further includes step S35: establishing a database to map the located location samples to the radio map.
  • step S35 this example uses statistical machine learning to create a decision region for the channel response mode and cross-correlate the received wireless signals for the mode of position recognition.
  • the algorithm is located using a maximum likelihood estimate with some probability distribution assumptions.
  • the eigenvector follows a multidimensional Gaussian distribution characterized by two statistical parameters (ie, mean vector and covariance matrix).
  • the decision region can be defined by the Mahalanobis distance.
  • the machine learning phase is equivalent to calculating the average vector and covariance matrix of the sample data during statistical machine learning.
  • the method for implementing the statistical machine learning to identify the location is as follows:
  • the cross-correlation of the channel response information of the received wireless signal at the sampling point is calculated.
  • the relevant statistical properties can be described by the statistical properties of the autocorrelation function of the transmitted signal and the cross-correlation function of the channel response.
  • the statistical properties of the autocorrelation function of the transmitted signal can be considered to be constant as it is determined by the modulation and transmission filters of the signal.
  • the cross-correlation function of the channel response depends on the changing environment and the location of the transmitter. Therefore, the cross-correlation fluctuations are mainly affected by the cross-correlation of the channel response.
  • This example uses statistical machine learning, where fluctuations are used as statistical properties of the channel response. This means that the proposed method needs to statistically implement cross-correlation of received signals with overall statistical information about the channel response at each grid point, as follows:
  • the eigenvector is actually assumed by the central limit theorem
  • a multidimensional Gaussian distribution characterized by two statistical parameters (ie, mean vector and covariance matrix) is followed.
  • the decision region can be defined by the Mahalanobis distance.
  • the machine learning phase is equivalent to calculating the average vector and covariance matrix of the sample data during statistical machine learning.
  • This example assumes a eigenvector
  • the multidimensional Gaussian distribution is followed, and the Mahalanobis distance can be calculated by using the maximum likelihood estimation method.
  • the learning data ie the mean vector and the covariance matrix, are defined as follows:
  • H is the conjugate transpose of the matrix (also known as Hermitian transposition), for Average vector, To of The covariance matrix defines the probability function forming decision area as:
  • p(xxx) is the intrinsic format of a probability function.
  • p refers to the probability of the probability function of the decision region.
  • ⁇ ) means the probability of seeking under the condition of ⁇ .
  • the average vector And covariance matrix Both are functions of the antennas k and l and the position information ⁇ .
  • an evaluation function corresponding to the decision area can be defined.
  • the most suspicious location can be estimated as the location with the highest likelihood, ie, the best match between the radio signals in the radio map in the previous training model to infer its location.
  • the present invention further includes a parameter correction step A: correcting the parameters used in the algorithm detected in step S3 by using a sensor attached to the human body.
  • the sensor of this example is preferably a sensor that is provided by the mobile phone, such as a gravity sensor, an acceleration sensor, a gyroscope, and the like that are provided by the mobile phone.
  • the parameter correction method includes:
  • A1 setting the start and end of the human body motion, recording the motion characteristics with the sensor, and recording the time period;
  • A2 The sensor calculates the number of steps of human motion during this time period
  • A3 The speed of the human body is calculated by the time period and the number of steps, and the parameters of the detection algorithm are corrected.
  • the shaking mobile phone action is used as a flag for setting the start and end of the movement, and the time period is recorded at the start and end time points, and the distance moved by the time period is calculated by using the mobile phone's own sensor.
  • the distance and angle information collected are used to calculate the position after the movement, and the positioning algorithm is corrected to make the positioning more accurate.
  • the invention further comprises a step S4: the server returns the calculated positioning result to the wireless receiving end. After the server calculates the positioning result, the positioning result is returned to the wireless receiving end, such as a wireless network card, and other required terminals can obtain positioning information from the wireless network card.
  • the wireless receiving end such as a wireless network card, and other required terminals can obtain positioning information from the wireless network card.
  • the invention is based on the radio propagation mechanism in the indoor environment, establishes the channel state information CSI and the human body moving distance, and judges the distance of the human body movement through the change of the CSI, thereby effectively calculating the position of the human body indoor, which has the following beneficial effects: In indoor environments with less artifacts (such as laboratories), the detection accuracy of detected actions is 84% to 94%, while in indoor environments with more decorations (such as dormitory), the detection accuracy can reach 78%. . It can accurately locate the human body indoors, and use the self-learning function of the system to deal with false alarms and further reduce the false alarm rate.
  • the present invention performs indoor testing on the basis of existing wireless networks and devices.
  • testee does not need to carry any additional sensing equipment, and the testee is not required to carry the detection equipment. The inconvenience caused by it has facilitated its life.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

一种人体室内定位自动检测方法及系统，包括如下步骤：S1：无线接收端接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器；S2：服务器接收到无线接收端的无线信号信息，并评估信道状态信息；S3：服务器根据信道状态信息的变化检测人体的定位信息；S4：服务器将计算的定位结果返回无线接收端。还包括参数校正步骤A：利用人体自带的传感器对步骤S3中检测的算法中用到的参数进行校正。本方法及系统能够在室内对人体进行精准的定位，并利用系统的自学习功能处理误报情况，进一步降低误报率；可以在遍布无线网络的任何环境中使用，同时被检测者不需要携带任何额外传感设备，非常便利。

Description

一种人体室内定位自动检测方法及系统 技术领域
本发明涉及定位技术，尤其涉及一种人体室内定位自动检测方法，还涉及一种实现所述方法的系统。
背景技术
近几年来，室内位置信息在人们的日常生活中扮演着越来越重要的作用，定位服务市场发展迅速，定位服务需求量迅速增长。精确的定位对于公共安全、商业应用以及军事应用都具有非常重要的意义。然而室内环境非常复杂,信号传播会受到墙壁、隔板、天花板等障碍物的阻挡,引起信号发生反射、折射、衍射现象,发射信号经过多条路径、以不同的时间到达接收端,出现多径传播现象和非视距效应,使得室内定位极具挑战性。因此，我们急切需要找到一种能够方便并有效的人体定位的方法。为了位置的准确检测，人们提出了利用视觉、可见光或者声音来进行检测的方法。然而采用这些方法搭建的检测系统本身都存在着种种不足。在利用视觉的定位检测系统，需要在被检测者所处的检测环境中安装有高分辨率的摄像头来拍摄大量影像，进而通过所拍摄的影像来追踪定位。然而，安装摄像头在某种程度下会侵犯到被检测者的个人隐私。与此同时，由于光线因素的影响，利用视觉的定位检测系统不能在黑暗条件下有效的工作。这些系统都存在着受环境因素影响大的缺点，并且这些系统所存在的局限性都为准确并方便的进行对人体定位的检测造成了障碍。而利用可见光进行定位的系统中，特定的设备仪器需要事先被安装在被检测者所处的环境中。传送光线不能够被阻挡，否则基本不具备通信穿透能力。而波长也不能太短，否则会导致受散射、反射、多径的影响很大，从而使定位精确度很低。而利用声音进行定位的系统中，被检测者周围的其他声音或其他环境因素很容易干扰环境设备的工作，从而造成较低的准确度。这些系统都存在着受环境因素影响大的缺点，并且这些系统所存在的局限性都为准确人体室内定位检测造成了障碍。
随着无线通信技术的发展，越来越多的无线设备被人们应用到生活当中。因此，利用无线通信技术来定位人体位置被认作是一种有效可行的方法。现有的无线通信技术提出了多种人体室内定位的方法。比如通过蓝牙技术实现室内定位，该方法具较高准确率，但使用该方法，被检测者需要携带额外标签，仍会造成不便。而利用UWB(Ultra Wideband，无载波通信技术)脉冲无线电技术实现室内定位，虽然平均发射功率很低，可以与其他无线通信系统“安静的共存”，有低能耗、低成本、保密性好、抗多径干扰等优点，但同时，脉冲UWB系统频谱利用率较低，不适合高数据率传输。还有一个重要的原因就是UWB定位 需要被定位人和物额外佩戴标签(相对于手机等终端是天然的标签而言)，这些方法都会给被检测者的生活造成一定程度的不便。
发明内容
为解决现有技术中的问题，本发明提供一种人体室内定位自动检测方法，还提供一种实现所述人体室内定位自动检测方法的系统。
本发明人体室内定位自动检测方法包括如下步骤：
S1：无线接收端接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器；S2：服务器接收到无线接收端的无线信号信息，并评估信道状态信息；S3：服务器根据信道状态信息的变化检测人体的定位信息；S4：服务器将计算的定位结果返回无线接收端。
本发明作进一步改进，还包括参数校正步骤A：利用人体自带的传感器对步骤S3中检测的算法中用到的参数进行校正。
本发明作进一步改进，所述传感器包括手机自带的传感器，手机自带的传感器包括重力传感器、加速度传感器、陀螺仪。
本发明作进一步改进，所述参数校正方法包括：
A1：设置人体运动的开始与结束，利用传感器记录运动特征，并记录此时间段；
A2：传感器计算此时间段内人体运动的步数；
A3：通过时间段和步数计算人体运动速度，并对检测算法的参数进行校正。
本发明作进一步改进，在步骤S2中，评估信道状态信息包括如下步骤：
S21：采集信道状态数据，所述信道状态数据包括N个空间流中的M个子载波的CSI值，其中，N和M均为大于1的自然数；
S22：对每一空间流，求取在同一时间点上的P个连续子载波的CSI值的平均值，将此平均值作为信道状态信息，P为大于1小于M的自然数；
S23：对信道状态信息进行平滑处理。
本发明作进一步改进，步骤S3的处理方法包括如下步骤：
S31：基于统计学习理论，预先建立以设定空间内由信道状态信息CSI与距离作为训练样本的室内传播模型；
S32：接收网络层中的无线信号发射器坐标和来自物理层的信道状态信息CSI；
S33：应用三边定位法和室内传播模型完成人体的定位。
本发明作进一步改进，步骤S3的处理方法还包括步骤S34：通过指纹卡算法自适应修改步骤S33中定位算法中的参数。
本发明作进一步改进，步骤S3的处理方法还包括步骤S35：建立数据库，将定位的位 置样本映射到无线电地图。
本发明作进一步改进，步骤S35中，采用统计机器学习创建用于信道响应模式的判定区域，并将接收的无线信号互相关用于位置识别，所述统计机器学习识别位置的方法包括如下步骤：
B1：计算采样点的接收无线信号的信道响应信息的互相关；
B2：统计每个网格点处的信道响应的整体统计信息的接收无线信号的互相关；
B3：计算天线之间的互相关系数；
B4：根据多维高斯分布和最大似然估计法计算马氏距离，并与室内传播模型中无线电地图中的无线电信号之间的最佳匹配来确定人体位置。
本发明还提供一种实现所述方法的系统，包括无线信号发射器：用于发射无线信号；无线接收端：用于接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器，获取服务器返回的定位结果；服务器：用于接收无线接收端的无线信号信息，并评估信道状态信息，根据信道状态信息的变化检测人体的定位信息，然后把将计算的定位结果返回无线接收端。
与现有技术相比，本发明的有益效果是：在装饰物较少的室内环境中，被检测动作的检测准确率为84％～94％，而在装饰物较多的室内环境中，检测准确率能够达到78％；能够在室内对人体进行精准的定位，并利用系统的自学习功能处理误报情况，进一步降低误报率；在现有的无线网络及设备的基础上，进行室内的检测工作，被检测环境中无需安装其他特定的检测设备，可以在遍布无线网络的任何环境中使用，具有极高的普及性，同时被检测者不需要携带任何额外传感设备，避免了被检测者携带检测设备所造成的不便，为其生活提供了便利。
附图说明
图1为本发明一种实施例系统结构示意图；
图2为本发明服务器实现流程简图；
图3为本发明一实施例方法流程图。
具体实施方式
下面结合附图和实施例对本发明做进一步详细说明。
如图1所示，本发明人体室内定位自动检测系统包括无线信号发射器：用于发射无线信号；无线接收端：用于接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器，获取服务器返回的定位结果；服务器：用于接收无线接收端的无线信号信息，并评估信道状态信息，根据信道状态信息的变化检测人体的定位信息，然后把将计算的定位 结果返回无线接收端。
在实际应用中，优选地，所述无线接收端为无线网卡，所述无线信号发射器为无线路由器，所述服务器为手机，该方法基于室内环境下的无线电传播机制，建立无线信号和人体移动距离的关系，只需要使用家庭现有的无线网络设备，即能够通过对被检测者距离而造成的无线信号的改变进行分析，从而得到精准的室内定位位置。在特定的室内环境中，可通过无线网卡收集无线网络的丰富的信道状态信息。在本发明中，系统分别有多根天线来发送和接收无线信号；系统所使用的无线网卡可以接收信道状态信息。所述无线信号发射器的数目为一个或两个以上，所述无线接收端的数目为一个以上。优选地，无线信号发射器和无线接收端的数目分别为2个或者3个。如图1所示，被检测环境中存在两个无线信号发射器(第一无线信号发射器和第二无线信号发射器)和两个无线接收端(第一无线接收端和第二无线接收端)。其中第一无线接收端接收来自第一无线信号发射器的CSI(Channel State Information的缩写，即信道状态信息，在无线通信领域，CSI就是通信链路的信道属性，描述了信号在每条传输路径上的衰弱因子)，第二无线接收端接收来自第二无线信号发射器的CSI。在被检测环境中，被检测者无需携带其他额外设备。系统将利用两个无线接收端所接收的CSI来对被检测者的动作进行检测，并从而判断被检测者的动作或者位置，比如是否摔倒等。
为了建立无线信号和人体移动距离的联系，本发明采用无线网络的信道状态信息CSI作为指示物。CSI能够描述出在时间延迟、振幅衰减和相位转移的共同影响之下，一个信号的传播途径。基于室内环境下的无线电传播模型，本发明建立了CSI与人体移动距离之间的联系。在一个特定的室内环境中，如一个房间，存在一条主要传播路径和多个因为周围环境，比如天花板、地板和墙等的影响而产生的反射路径。当被检测者处于该房间内，他的身体会产生多条散射路径；当被检测者在该环境中保持静止状态时，处于该环境内的接收端会接收稳定的传播能量；而当被检测者移动到另一位置时，由人体影响而产生散射路径的散射点会迅速的改变位置，而这种突变会导致接收端所收到的能量发生变化。而通过这种变化，本发明将会判断出人体移动距离的变化。本发明利用正交频分载波复用(Orthogonal Frequency Division Multiplex,OFDM)得到以子载波(subcarrier)形式存在的CSI。而使用这种方法得到的CSI与人体动作之间建立联系会提高对动作判断的准确率。
如图2和图3所示，本发明基于上述系统的人体室内定位自动检测方法包括如下步骤：
S1：无线接收端接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器。当本发明的系统开始工作时，无线信号发射器会传播无线网络信号，同时处于特定区域内的无线接收端(如手机端)会收集CSI作为初始信道状态数据，并上传至服务器，然 后通过服务器进行数据处理。
S2：服务器接收到无线接收端的无线信号信息，并评估信道状态信息，其中，所述评估信道状态信息包括如下步骤：
S21：采集信道状态数据，所述信道状态数据包括N个空间流中的M个子载波的CSI值，其中，N和M均为大于1的自然数；
S22：对每一空间流，求取在同一时间点上的P个连续子载波的CSI值的平均值，将此平均值作为信道状态信息，P为大于1小于M的自然数；
S23：对信道状态信息进行平滑处理。
具体地，本发明使用3×3的多入多出技术(Multiple-Input Multiple-Out-put，MIMO)为例，在感应阶段得到的初始信道状态数据会被分成9个空间流，而在每一个流中会有30个子载波。人体移动距离的变化会对不同的空间流所包含的数据产生影响，而对每一个空间流中的所有子载波则会产生相似的影响。同时，实验表明，环境因素(如温度，房间的设置)也会造成收集的CSI有所起伏。因此，在本发明中，每一个独立的空间流中的30个子载波的CSI值被合并汇总成一个单独的信道状态信息。优选地，对每一个空间流，先求得连续5个子载波的CSI平均值，而且对9个空间流分别取同一时间点的CSI平均值作为信道状态信息。为了减小环境因素的干扰，本发明利用数据过滤技术和滑动平均方法，具体地，使用加权移动平均法(weighted moving average)对通过上述处理的信道状态信息进行平滑处理，以减少数据中的噪声。
S3：服务器根据信道状态信息的变化检测人体的定位信息。所述检测人体定位信息的处理方法包括如下步骤：
S31：基于统计学习理论，预先建立以设定空间内由信道状态信息CSI与距离作为训练样本的室内传播模型；
S32：接收网络层中的无线信号发射器坐标和来自物理层的信道状态信息CSI；
S33：应用三边定位法和室内传播模型完成人体的定位。
所述步骤S3的处理方法还包括步骤S34：通过指纹卡算法自适应修改步骤S33中定位算法中的参数。以达到利用少量的训练样本完成精准定位的效果。
根据已有的无线电传播模型，本发明开发了一个基于CSI信息与距离的室内传播模型。通过一种简单的基于监督学习的快速训练方法，并通过三边定位方法来进行定位。第一步：在两个点处收集多个分组的CSI数据以训练环境因子和室内传播模型的路径损耗衰落指数；第二步：使用在第三点处收集的CSI来测试参数估计的效率。通过以上办法，建立一个基于CSI信息与距离的室内传播模型，通过CSI信息可以算出未知点距离和已知点的距离。未知点必然位于以已知点为球心的，距离为半径的球上。只要测出未知点和三个已知点的 距离，则未知点在三个球圆周的相交处，当相交处为两个点时，因有接收方向，故有一个处于接收背面的点可以舍去，从而准确的测出未知点的位置。
为了提高准确度和减少训练样本，本发明采用指纹法来纠正定位算法。在每个采样点位置接收来自无线信号发射器的信标信息，该信标信息包含了多个子载波的信道响应信息。手机接收端会同时收集30组CSI信息。
此外，本例的步骤S3的处理方法还包括步骤S35：建立数据库，将定位的位置样本映射到无线电地图。
步骤S35中，本例采用统计机器学习来创建用于信道响应模式的判定区域，并将接收的无线信号互相关用于位置识别的模式。
如图4所示，利用带有一些概率分布假设的最大似然估计来定位算法。通过中心极限定理，实际上假定特征向量遵循由两个统计参数(即平均值向量和协方差矩阵)为表征的多维高斯分布。利用多维高斯分布及其协方差矩阵，决定区域可以由马氏距离来定义。机器学习阶段等效于在统计机器学习过程中计算样本数据的平均向量和协方差矩阵。
具体地，所述统计机器学习识别位置的方法实现过程如下：
首先，计算采样点的接收无线信号的信道响应信息的互相关。相关的统计特性可以通过发射信号的自相关函数和信道响应的互相关函数的统计特性来描述。
Figure PCTCN2017084427-appb-000001
其中，
Figure PCTCN2017084427-appb-000002
表示第i根天线在位置ψ的接收信号，
Figure PCTCN2017084427-appb-000003
和
Figure PCTCN2017084427-appb-000004
分别表示信道响应和发射信号，
Figure PCTCN2017084427-appb-000005
表示为零均值高斯噪声，1≤m≤M,1≤τ≤m,
Figure PCTCN2017084427-appb-000006
为动作起始和结束之间的时间段。
发射信号的自相关函数的统计特性可以被认为是恒定的，因为它由信号的调制和发射滤波器确定。另一方面，信道响应的互相关函数取决于变化的环境以及发射机的位置。因此，互相关的波动主要受到信道响应的互相关性的影响。本例采用统计机器学习，其中波动被用作信道响应的统计特性。这意味着所提出的方法需要是统计地实现具有关于每个网格点处的信道响应的整体统计信息的接收信号的互相关，统计公式如下：
Figure PCTCN2017084427-appb-000007
本例采集2N+1个互相关的离散的样本，即-N≤n≤N。因此，
Figure PCTCN2017084427-appb-000008
具有2N+1个维度，其中N由最大时延扩展确定。由于手机端自带两根距离很相近的天线，所以k和l可看成一个相同的传感器。所以本例互相关系数的学习数据库可以定义为：
Figure PCTCN2017084427-appb-000009
利用带有一些概率分布假设的最大似然估计法来改善这个算法。通过中心极限定理，实际上假定特征向量
Figure PCTCN2017084427-appb-000010
遵循由两个统计参数(即平均值向量和协方差矩阵)为表征的多维高斯分布。利用多维高斯分布及其协方差矩阵，决定区域可以由马氏距离来定义。机器学习阶段等效于在统计机器学习过程中计算样本数据的平均向量和协方差矩阵。本例假设特征向量
Figure PCTCN2017084427-appb-000011
遵循多维高斯分布，以及可以通过采用最大似然估计法来计算马氏距离。然后，如下定义学习数据，即平均值向量和协方差矩阵：
Figure PCTCN2017084427-appb-000012
Figure PCTCN2017084427-appb-000013
其中H表示为矩阵的共轭转置(又称Hermitian转置)，
Figure PCTCN2017084427-appb-000014
为
Figure PCTCN2017084427-appb-000015
的平均向量，
Figure PCTCN2017084427-appb-000016
为以
Figure PCTCN2017084427-appb-000017
的
Figure PCTCN2017084427-appb-000018
协方差矩阵，将概率函数形成决策区域定义为：
Figure PCTCN2017084427-appb-000019
p(xxx)为一个概率函数的固有格式，其实p指的是决策区域概率函数的概率是多少，p(x|ψ)表达的意思为在ψ的条件下，所求的概率是多少。
在该方法中，平均向量
Figure PCTCN2017084427-appb-000020
和协方差矩阵
Figure PCTCN2017084427-appb-000021
都是天线k与l和位置信息ψ的函数。当无线信号发射器的位置ψ已知，即可以定义对应决策区域的评估函数。在预测位置的时候，最可疑位置可以被估计为具有最高似然性的位置，即与先前训练模型中无线电地图中的无线电信号之间的最佳匹配来推断其位置。
其中，本发明还包括参数校正步骤A：利用人体自带的传感器对步骤S3中检测的算法中用到的参数进行校正。
本例的传感器优选手机自带的传感器，比如手机自带的传重力传感器、加速度传感器、陀螺仪等等。
具体地，所述参数校正方法包括：
A1：设置人体运动的开始与结束，利用传感器记录运动特征，并记录此时间段；
A2：传感器计算此时间段内人体运动的步数；
A3：通过时间段和步数计算人体运动速度，并对检测算法的参数进行校正。
以手机传感器为例，以摇晃手机动作作为设置移动开始与结束的标志，并在开始与结束时间点记录此时间段，并利用手机自带传感器计算此时间段移动的距离。通过收集回来的距离和角度信息来计算移动后的位置，对定位算法进行校正，从而使定位更加准确。
本发明还包括步骤S4：服务器将计算的定位结果返回无线接收端。服务器计算好定位结果后，将定位结果返回无线接收端，比如无线网卡，其他需要的终端可以从无线网卡中获取定位信息。
本发明基于室内环境下的无线电传播机制，将信道状态信息CSI与人体移动距离建立联系，通过CSI的变化判断人体移动的距离，从而有效的计算人体室内的位置，其具有以下有益效果：在装饰物较少的室内环境(如实验室)中，被检测动作的检测准确率为84％～94％，而在装饰物较多的室内环境(如宿舍)中，检测准确率也能够达到78％。能够在室内对人体进行精准的定位，并利用系统的自学习功能处理误报情况，进一步降低误报率；本发明是在现有的无线网络及设备的基础上，进行室内的检测工作，被检测环境中无需安装其他特定的检测设备，可以在遍布无线网络的任何环境中使用，具有极高的普及性，同时被检测者不需要携带任何额外传感设备，避免了被检测者携带检测设备所造成的不便，为其生活提供了便利。
以上所述之具体实施方式为本发明的较佳实施方式，并非以此限定本发明的具体实施范围，本发明的范围包括并不限于本具体实施方式，凡依照本发明所作的等效变化均在本发明的保护范围内。

Claims (10)

  1. 一种人体室内定位自动检测方法，其特征在于包括如下步骤：
    S1：无线接收端接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器；
    S2：服务器接收到无线接收端的无线信号信息，并评估信道状态信息；
    S3：服务器根据信道状态信息的变化检测人体的定位信息；
    S4：服务器将计算的定位结果返回无线接收端。
  2. 根据权利要求1所述的人体室内定位自动检测方法，其特征在于：在步骤S3执行后，计算出定位信息后才执行，其包括参数校正步骤A：利用人体自带的传感器对步骤S3中检测的算法中用到的参数进行校正。
  3. 根据权利要求2所述的人体室内定位自动检测方法，其特征在于：所述传感器包括手机自带的传感器，手机自带的传感器包括重力传感器、加速度传感器、陀螺仪。
  4. 根据权利要求3所述的人体室内定位自动检测方法，其特征在于：所述参数校正方法包括：
    A1：设置人体运动的开始与结束，利用传感器记录运动特征，并记录此时间段；
    A2：传感器计算此时间段内人体运动的步数；
    A3：通过时间段和步数计算人体运动速度，并对检测算法的参数进行校正。
  5. 根据权利要求1-4任一项所述的人体室内定位自动检测方法，其特征在于：在步骤S2中，评估信道状态信息包括如下步骤：
    S21：采集信道状态数据，所述信道状态数据包括N个空间流中的M个子载波的CSI值，其中，N和M均为大于1的自然数；
    S22：对每一空间流，求取在同一时间点上的P个连续子载波的CSI值的平均值，将此平均值作为信道状态信息，P为大于1小于M的自然数；
    S23：对信道状态信息进行平滑处理。
  6. 根据权利要求5所述的人体室内定位自动检测方法，其特征在于：步骤S3的处理方法包括如下步骤：
    S31：基于统计学习理论，预先建立以设定空间内由信道状态信息CSI与距离作为训练样本的室内传播模型；
    S32：接收来自发射端物理层的信道状态信息CSI；
    S33：应用三边定位法和室内传播模型完成人体的定位。
  7. 根据权利要求6所述的人体室内定位自动检测方法，其特征在于：步骤S3的处理方法还包括步骤S34：通过指纹卡算法自适应修改步骤S33中定位算法中的参数。
  8. 根据权利要求7所述的人体室内定位自动检测方法，其特征在于：步骤S3的处理方法还包括步骤S35：建立数据库，将定位的位置样本映射到无线电地图。
  9. 根据权利要求8所述的人体室内定位自动检测方法，其特征在于：步骤S35中，采用统计机器学习创建用于信道响应模式的判定区域，并将接收的无线信号互相关用于位置识别，所述统计机器学习识别无线电地图中的位置的方法包括如下步骤：
    B1：计算采样点的接收无线信号的信道响应信息的互相关；
    B2：统计每个网格点处的信道响应的整体统计信息的接收无线信号的互相关；
    B3：计算天线之间的互相关系数；
    B4：根据多维高斯分布和最大似然估计法计算马氏距离，并与室内传播模型中无线电地图中的无线电信号之间的最佳匹配来确定人体位置。
  10. 一种实现权利要求1-9任一项所述人体室内定位自动检测方法的系统，其特征在于：包括无线信号发射器：用于发射无线信号；
    无线接收端：用于接收无线信号发射器发射的无线信号，并将无线信号信息上传至服务器，获取服务器返回的定位结果；
    服务器：用于接收无线接收端的无线信号信息，并评估信道状态信息，根据信道状态信息的变化检测人体的定位信息，然后把将计算的定位结果返回无线接收端。
PCT/CN2017/084427 2017-01-18 2017-05-16 一种人体室内定位自动检测方法及系统 Ceased WO2018133264A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710035562.3A CN106802404B (zh) 2017-01-18 2017-01-18 一种人体室内定位自动检测方法及系统
CN201710035562.3 2017-01-18

Publications (1)

Publication Number Publication Date
WO2018133264A1 true WO2018133264A1 (zh) 2018-07-26

Family

ID=58984472

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/084427 Ceased WO2018133264A1 (zh) 2017-01-18 2017-05-16 一种人体室内定位自动检测方法及系统

Country Status (2)

Country Link
CN (1) CN106802404B (zh)
WO (1) WO2018133264A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110082716A (zh) * 2019-04-29 2019-08-02 徐州医科大学 一种医院复杂环境室内定位系统及定位方法
CN114285500A (zh) * 2021-12-14 2022-04-05 电子科技大学 一种uwb室内定位信道质量评估方法

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109495193A (zh) * 2017-09-12 2019-03-19 成都华为技术有限公司 信号传输方法、相关装置及系统
CN107909771B (zh) * 2017-10-08 2020-04-10 南京邮电大学 一种基于无线传感网络的人员跌倒报警系统及其实现方法
CN107816985B (zh) * 2017-10-31 2021-03-05 南京阿凡达机器人科技有限公司 人体检测装置及方法
CN108534779B (zh) * 2018-03-09 2020-09-08 华中科技大学 一种基于轨迹纠正和指纹改进的室内定位地图构建方法
CN110609254B (zh) * 2018-06-15 2022-09-27 富士通株式会社 基于无线信号的动作检测方法、检测装置和电子设备
WO2020145949A1 (en) 2019-01-08 2020-07-16 Google Llc Motion detection using wireless local area networks
JP7183829B2 (ja) * 2019-01-31 2022-12-06 株式会社Soken 車両用位置推定システム
CN110502105B (zh) * 2019-07-08 2020-12-11 南京航空航天大学 一种基于csi相位差的手势识别系统及识别方法
DE102019220630A1 (de) 2019-12-30 2021-07-01 Airbus Defence and Space GmbH System und Verfahren zur Objektlokalisierung in einer Innenumgebung
CN111398893B (zh) * 2020-05-14 2021-11-23 南京工程学院 一种基于无线定位的网格人体模型测量装置及方法
TWI741821B (zh) * 2020-10-07 2021-10-01 廣達電腦股份有限公司 具有重力感測器輔助定位的電子裝置
CN113422660B (zh) * 2021-05-14 2022-07-19 山东科技大学 基于无线信号的步数检测方法
CN113692022B (zh) * 2021-07-09 2024-06-11 深圳市赛云数据有限公司 一种基于空信道预判利用的运营商通讯系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103606248A (zh) * 2013-09-30 2014-02-26 广州市香港科大霍英东研究院 一种人体摔倒自动检测方法及系统
CN104333903A (zh) * 2014-09-17 2015-02-04 北京邮电大学 基于rssi和惯性测量的室内多目标的定位系统和方法
CN104812061A (zh) * 2015-03-24 2015-07-29 成都希盟泰克科技发展有限公司 一种基于mimo-ofdm信道状态信息的室内测距及定位方法
WO2015119635A1 (en) * 2014-02-10 2015-08-13 Hewlett Packard Development Company, L.P. Distance estimation
CN105979485A (zh) * 2016-05-11 2016-09-28 南京邮电大学 基于信道状态信息的室内环境下人员检测方法
CN106332277A (zh) * 2016-09-05 2017-01-11 中南大学 一种基于信道状态信息分布的室内定位方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103596266B (zh) * 2013-11-26 2017-06-27 无锡市中安捷联科技有限公司 一种人体检测和定位的方法、装置及系统
CN104267439A (zh) * 2014-08-20 2015-01-07 哈尔滨工程大学 一种无监督人体检测与定位的方法
CN106154222B (zh) * 2016-06-20 2018-06-12 北京大学 一种利用无线电射频信号检测人的行走方向的方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103606248A (zh) * 2013-09-30 2014-02-26 广州市香港科大霍英东研究院 一种人体摔倒自动检测方法及系统
WO2015119635A1 (en) * 2014-02-10 2015-08-13 Hewlett Packard Development Company, L.P. Distance estimation
CN104333903A (zh) * 2014-09-17 2015-02-04 北京邮电大学 基于rssi和惯性测量的室内多目标的定位系统和方法
CN104812061A (zh) * 2015-03-24 2015-07-29 成都希盟泰克科技发展有限公司 一种基于mimo-ofdm信道状态信息的室内测距及定位方法
CN105979485A (zh) * 2016-05-11 2016-09-28 南京邮电大学 基于信道状态信息的室内环境下人员检测方法
CN106332277A (zh) * 2016-09-05 2017-01-11 中南大学 一种基于信道状态信息分布的室内定位方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110082716A (zh) * 2019-04-29 2019-08-02 徐州医科大学 一种医院复杂环境室内定位系统及定位方法
CN114285500A (zh) * 2021-12-14 2022-04-05 电子科技大学 一种uwb室内定位信道质量评估方法
CN114285500B (zh) * 2021-12-14 2023-03-24 电子科技大学 一种uwb室内定位信道质量评估方法

Also Published As

Publication number Publication date
CN106802404B (zh) 2018-10-16
CN106802404A (zh) 2017-06-06

Similar Documents

Publication Publication Date Title
WO2018133264A1 (zh) 一种人体室内定位自动检测方法及系统
CN103596266B (zh) 一种人体检测和定位的方法、装置及系统
CN104502982B (zh) 一种细粒度免校验的室内被动人体检测方法
Choi et al. Deep learning based NLOS identification with commodity WLAN devices
CN108387940B (zh) 一种基于Wi-Fi无线感知的室内移动物体检测方法
CN110475221B (zh) 一种基于信道状态信息的人员动作识别和位置估计方法
CN103606248B (zh) 一种人体摔倒自动检测方法及系统
CN104883732B (zh) 一种增强型室内无源被动人体定位方法
CN107968689A (zh) 基于无线通信信号的感知识别方法及装置
WO2016197648A1 (zh) 一种基于无线信号的动作检测和识别的方法
CN108038419B (zh) 基于Wi-Fi的室内人员被动检测方法
CN112333653A (zh) 基于WiFi信道状态信息的身份智能识别方法、系统
CN104267439A (zh) 一种无监督人体检测与定位的方法
Liu et al. A research on CSI-based human motion detection in complex scenarios
Shi et al. Human activity recognition using deep learning networks with enhanced channel state information
CN106411433A (zh) 一种基于wlan的细粒度室内被动入侵检测方法
Xun et al. Depthwise separable convolution based passive indoor localization using CSI fingerprint
CN114222202A (zh) 一种基于WiFi CSI的环境自适应活动检测方法及系统
CN111770527A (zh) 基于二维特征的可视与非可视信道识别方法及装置
CN108122310A (zh) 一种基于WiFi信道状态信息和动态时间规整的人流量统计方法
Tang et al. Human behavior recognition based on WiFi channel state information
Choi et al. Simultaneous crowd estimation in counting and localization using WiFi CSI
Man et al. PWiG: A phase-based wireless gesture recognition system
CN105785454A (zh) 基于信道频域响应的室内运动检测方法
CN113783639B (zh) 无接触感知边界确定方法、系统、计算机设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17892094

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205N DATED 30/08/2019)

122 Ep: pct application non-entry in european phase

Ref document number: 17892094

Country of ref document: EP

Kind code of ref document: A1