WO2018122817A1

WO2018122817A1 - Method for detecting human traffic in public place by using wi-fi probe

Info

Publication number: WO2018122817A1
Application number: PCT/IB2017/058547
Authority: WO
Inventors: 杜豫川; 岳劲松; 暨育雄
Original assignee: 同济大学; 杜豫川; 许军
Priority date: 2016-12-30
Filing date: 2017-12-30
Publication date: 2018-07-05
Also published as: GB201909414D0; CN109716358A; CN109479206B; WO2018122815A1; CN109644360A; GB201905910D0; GB2569752B; CN109644320B; GB201711411D0; CN109644360B; CN109716358B; WO2018122816A1; CN109479206A; CN109644320A; GB2569752A; WO2018122588A1; GB201909415D0

Abstract

A method for detecting a human traffic in a public place by using Wi-Fi probe. By comparing the detection area of the Wi-Fi probe with the area of each functional region in the public place, different layout solutions of the probe are provided; MAC layer information of a mobile device is collected, and track information of pedestrians in the public place is obtained by making comparisons on the detection results from multiple aspects.

Description

Method for detecting human flow in public places using WI-FI probe

Technical field

The invention belongs to the technical field of WI-FI data acquisition and pedestrian flow detection, and particularly relates to a method for detecting human flow in a public place by using a WI-FI probe. By comparing the detection area of the WI-FI probe with the area of each functional area in the public place, different probe layout schemes are given, the MAC layer information of the mobile device is collected, and the detection results are analyzed in many aspects to obtain pedestrians. Trajectory information in public places. Background technique

Large-scale shopping malls, transportation hubs, tourist resorts and other places often experience large passenger flow. Especially during the peak passenger flow, a large number of pedestrians flooding into the main roads will cause certain safety hazards and affect the operational efficiency in these places. Therefore, real-time detection of pedestrian traffic on the road is of great significance. According to the test results, the actual flow of people can be estimated scientifically, and reliable traffic flow data can be provided for the corresponding security personnel, so that reasonable measures can be taken to ensure shopping malls, transportation hubs and tourist vacations. Normal operation in the district. At present, the detection methods for human traffic are becoming more and more diverse. According to the types of detection technologies, they can be roughly classified into the following categories:

(1) Manual investigation method: Manual investigation is the most traditional method for counting passenger flow. The method is simple and can be superimposed on the standard of manual judgment. However, due to its high requirements for investigators, large counting errors, and low data quality, the data collection work after the investigation is heavy, the data system is not systematic and cannot provide real-time data, and currently it cannot meet the growth of traffic demand. Intensive places are difficult and inefficient in real time.

(2) Gate-type passenger flow counting: The gate is a channel blocking device (channel management equipment), which is used to manage the flow of people and regulate pedestrian access. It is mainly used in subway gate system and toll gate gate system. Its most basic and core function is to realize the entrance channel of only one person at a time, which can be used for various charging and access control occasions. This method is low in cost and accurate in quantity. However, when the service crowd has a large amount of baggage parcels, the method is less efficient, hinders the evacuation of pedestrians in an emergency, and is not conducive to the inconvenience of movement. People's travel. And the method of detecting human flow data is only a certain section, and it is necessary to arrange a plurality of sections to grasp the distribution of people flow, and the floor space is large. (3) Pedal-type passenger flow counting: The pressure plate passenger flow statistic meter is installed on the ground of the inspection area, and the trigger pressure sensor information is automatically recorded when the pedestrian passes. This type of instrument can be roughly divided into two categories, one is based on the "human treading data model mode" for counting and direction judgment, and the other is based on "passenger pedal profile". The method reduces the influence on the passenger flow operation and is simple to install, but the detection accuracy is low, and the components of the pressing system are easily damaged, and the maintainability is poor.

(4) Infrared passenger flow counting: Infrared passenger flow counting can be divided into passive infrared passenger flow counting and active infrared passenger flow counting. Passive infrared passenger flow counting uses a pyroelectric infrared probe that can avoid interference from other objects and can only detect signals from the human body. When someone passes, the infrared sensor can detect a certain change caused by the infrared spectrum of the human body, trigger a pulse signal, and then judge the number of people according to the number of pulse signals. The active infrared type transmits a custom wavelength infrared ray to cover a certain area through the transmitting head, and the number of passengers is recognized by the light reflected by the passenger detected by the sensor. The active infrared passenger flow counter overcomes the shortcomings of the passive infrared passenger flow count affected by the environment and light. However, since it uses the simple judgment of the number of pulses to determine the number of people, the statistical accuracy is low, and for many people at the same time. The situation passed is even more difficult to measure. Moreover, the direction of the passenger flow cannot be discriminated by only the infrared method, and the cost of the detection device is high, and it is not suitable for a wide range of use.

(5) Video passenger flow count: Video passenger flow count can be divided into single-eye video passenger flow count and binocular video passenger flow count. Video passenger flow technology captures video images by installing cameras in critical channels, and captures passenger flow counts using image processing counts such as image segmentation, artificial neural networks, and stereo image analysis. However, the method started late and the technology is not yet mature. Moreover, the implementation cost and maintenance cost are high, and it is difficult to solve the problem of individual flow segmentation when the flow of people is dense, so the accuracy is low.

(6) WI-FI probe passenger flow detection: WI-FI probe passenger flow detection is achieved by deploying the WI-FI network in the detection area to obtain the MAC address of the mobile device that enables WI-FI function, thereby realizing passenger flow counting. The WI-FI-based passenger flow statistical method is simple in operation, reasonable in equipment cost, small in impact by non-line-of-sight factors, high in flexibility, and capable of acquiring a large amount of statistical data at the same time, and has great advantages in the flow statistics under dense passenger flow. In-depth analysis of the data content acquired by the probe can obtain characteristic data such as the flow stop time and streamline flow direction. And this detection method supports the cloud platform in subsequent operations, and the data application can be extended to the marketing layer. It is currently widely used in large commercial areas, tourist attractions, playgrounds and other places.

WI-FI probe technology: The WI-FI probe can detect the MAC address of the mobile device with WI-FI enabled. The principle includes: WI-FI is based on IEEE802. l la/b/g/n protocol, In the standard agreement, There are two working modes of wireless access point and client. The protocol also specifies various wireless data frame types such as Beacon, Ack, Data and Probe. When the client connects to the wireless access point, the data frame and the response frame are interacted, and the wireless access point periodically sends the Beacon. When the client is not connected to the wireless access point, the client also continuously sends the Probe frame. Probing to a nearby wireless access point. The WI-FI probe is a wireless access point that captures information about nearby clients based on various wireless data frames. It can intercept the MAC layer information of mobile clients opened by WI-FI within a certain range, including MAC address and signal. Receive intensity values, timestamps, etc.

However, manual survey methods, gate counting methods, pedal counting methods, infrared counting methods, and video counting methods all have disadvantages such as requiring more labor, expensive equipment, and a large floor space. Currently, the more popular WI is based on WI. -FI's passenger flow detection method has the following problems:

(1) The mobile device's unique MAC address is detected by the probe on the premise that the mobile device's WI-FI needs to be open. In reality, the proportion of mobile devices that open WI-FI in the crowd is low and unknown. Therefore, under normal circumstances, the difference between the traffic volume detected by WI-FI and the actual passenger traffic is large, and the effect is not satisfactory from the detection amount.

(2) The multi-path phenomenon and reflection phenomenon of the wireless detection signal emitted by the mobile device during the capture by the probe, and the multipath phenomenon and reflection phenomenon of the wireless signal will attenuate the signal strength, resulting in the detection of the probe. Received signal strength values (RSSI) have varying degrees of attenuation and are not even detectable in severe cases. Therefore, the detection rate of WI-FI is also low.

(3) Due to the basic characteristics of low detection rate, it is impossible to directly use the test results to calculate the passenger flow. Therefore, it is necessary to establish a suitable prediction model between the detected amount and the actual amount, so as to improve the accuracy of predicting the actual value from the detected value, and also to meet the high accuracy of the predictive model under the continuous fluctuation of pedestrian flow.

(4) Since most of the inspection equipment is installed at the entrance and exit of the public area, it is generally only possible to obtain the inbound and outbound data of the passenger flow, and it is impossible to restore the flow situation between the functional areas of the entire public place, that is, the trajectory information.

At present, the research on WI-FI passenger flow statistics is limited, mainly focusing on the accurate research on the received signal strength value (RSSI) to explore the accurate positioning of indoor pedestrians, as well as the passenger flow under the existing indoor WI-FI system. Description of characteristic parameters such as density and passenger flow trajectory. How to effectively improve the detection rate of WI-FI passenger flow statistics, how to deploy WI-FI probes to achieve better detection results and how There is still a lack of research on the trajectory information of pedestrians through data analysis. Explanation of terms

In order to make the description of the present invention more accurate and clear, various terms appearing in the present invention are explained as follows:

WI-FI probe: A wireless access point that captures information about nearby mobile devices based on various wireless data frames. It can intercept MACXMedia Access Control layer information of mobile clients opened by WI-FI within a certain range, including MAC address, signal reception strength value, time stamp, etc.

Detection area: The effective detection area of the WI-FI probe is generally a spherical area with a probe centered on the radius of 50-100 meters.

Functional area: The area of a public area that provides different functions in a public place;

Detection period: The length of the unit detection time used to detect pedestrians on the road using the WI-FI probe. Pedestrian mobile devices: WI-FI-enabled electronic devices that are carried by pedestrians, such as smart phones, hand-held computers, IPADs, etc.;

MAC address: The Media Access Control address, which translates to media access control, is the physical address of each mobile device, the hardware address, and the location used to define the network device. Expressed as a unique series of 12 characters consisting of numbers and letters;

MAC address raw data: All MAC address data strips detected by the WI-FI probe;

Invalid MAC address data: In the original MAC address data, it is not a MAC address data strip within the scope of the road to be studied;

Valid MAC address data: In the original MAC address data, the MAC address data strip in the range of the road to be studied;

Detecting human traffic data: same as valid MAC address data;

Estimating the human flow data: According to the function model between the detected human flow data and the actual human flow data, there is actual human flow data estimated by detecting the human flow data;

Duration of stay: The length of time that a MAC address data stays in a functional area;

Trajectory reconstruction: Analyze the detection result of a mobile device's MAC address and reconstruct its trajectory between functional areas in public places. Summary of the invention

It is an object of the present invention to provide a method of detecting human flow in a public place using a WI-FI probe. The specific detection method is to compare the detection area of the WI-FI probe with the area of each functional area in the public place, and give different probe layout schemes, complete the collection of the MAC layer information of the mobile device, and perform various aspects on the detection result. Analysis, get the trajectory information of pedestrians in public places. When using the WI-FI probe to detect human flow, the present invention mainly solves the following three problems:

(1) Since the area of the functional area in the public place is large or small, determining the number of WI-FI probes required is the first problem to be solved according to the area of the functional area. When the number of probes is large, the wireless signal emitted by the mobile device has multipath phenomenon and reflection phenomenon during the propagation process, which may cause the signal received strength (RSSI) captured by the probe to be attenuated to different degrees, or even Detected. Therefore, the present invention provides a layout scheme of the multi-probe in the functional area on the basis of exploring the influence of the spatial layout of the plurality of probes on the detection result of the wireless signal, thereby greatly reducing the propagation process of the wireless signal. The influence of multipath phenomenon and reflection phenomenon on the detection results.

(2) The effective detection range of the WI-FI probe is a spherical area with a certain length as a radius centered on the device.

Therefore, when the detection area is larger than the area of the functional area, the mobile device outside the function area to be detected is also detected to cause the invalid data to exist in the detection result. Therefore, the present invention needs to set scientific data screening standards to eliminate these invalid data, thereby ensuring the reliability of the test results.

(3) When the pedestrian flow changes, the degree of multipath and reflection of the wireless signal is different, resulting in a significant change in the detection rate under the given data screening criteria as the flow rate changes. The present invention provides a calculation model suitable for predicting the actual amount of the detected amount in the case of a constantly changing human flow rate, thereby improving the prediction accuracy.

(4) When analyzing a specific MAC address, analyze the trajectory information between the functional areas in the public place, including a reasonable analysis method when there are two functional areas detecting a certain MAC address at the same time. The probability that the MAC address actually passes through each functional area is given, so that the respective probability sizes are given when the trajectory nodes cannot be visually determined. In order to solve the above problems, the technical solutions adopted by the present invention include: (1) When using multiple WI-FI probes to detect pedestrian flow, according to the geometric characteristics of the functional area, in order to ensure more detection area as much as possible, and to consider the cost of routing the probe, give a better multi-probe Deployment plan.

(2) When collecting the original test data, the union of the test results of each probe should be taken to count the number of mobile device MAC addresses detected during a certain period of time.

(3) In order to effectively eliminate invalid interference data, it is necessary to design a pre-experiment to determine the criteria for data screening. The pre-experiment is carried out on the pedestrian road to be tested, ensuring that the arrangement form of the plurality of probes is the same as that when detecting the flow rate of the person. In the effective detection range of the probe, a plurality of smart devices with known MAC addresses are used, and are randomly displaced for a period of time. After that, the received signal strength value of the MAC address data detected by the probe is statistically analyzed to determine the minimum value of the received signal strength within the required detection range, which is used as a data screening standard to exclude pedestrians in areas outside the required detection range. Mobile device MAC address data.

(4) Due to the constant changes in pedestrian traffic, the detection rate also changes. The invention directly discusses the relationship between the pedestrian detection value and the actual value when determining the flow prediction model. Firstly, the probe is used to detect the human flow rate, and the actual human flow rate is manually counted, and the design experiment and data processing are performed. A variety of functions are established to determine the actual human flow rate and the detected human flow rate, and the function value is used to estimate the actual value based on the detected value, thereby improving the detection accuracy.

(5) When reconstructing the trajectory, some MAC address data is detected in two adjacent functional areas at a certain moment. In this case, according to the existing probe detection data, the MAC address is given. This moment belongs to the probability of each functional area, which facilitates subsequent analysis. When exploring the layout of the WI-FI probe in the functional area of the public place, since the probe itself has a detection area of radius r, and the functional area is generally considered to be a rectangle having a length of & and a width b, the probe is There is a relationship between the detection area and the area of the functional area, which directly affects the number of probes required and the layout form. Therefore, according to the relationship between the probe detection radius r and the length of the functional region, three layout schemes of the probe are given, as shown in Fig. 1:

1) When a certain length of the functional area is smaller than the probe detection radius, ie a<r or b<r, a WI-FI probe is placed at the center of the functional area;

2) When one side of the functional area is much longer than the probe detection radius, ie a»r or b»r a set of WI-FI probes on the longer side of the energy zone;

3) When both sides of the functional area are much larger than the probe detection radius, ie a»r and b»r, a set of WI-FI probes are arranged along the diagonal of the functional area; Less than the probe detection radius, that is, when the first probe layout scheme is adopted, since the probe detection area has a portion beyond the functional area, theoretically, MAC address data outside the functional area is detected, and these data are not subject to Within the functional area of the study, it becomes invalid interference data. In determining the criteria for rejecting invalid interference data, the present invention provides a data screening method based on received signal strength values: providing a pre-experiment to explore received signal strength values (RSSI) and mobile devices to a given detection site Corresponding relationship between the distances between the probes, so as to determine the minimum value of the corresponding signal receiving intensity according to the spatial extent of the area to be detected in the actual test place, as a data screening line, filtering out the area to be detected from the original data Interfere with data. In the analysis of the test results, it is necessary to count the average stay duration of all valid MAC address data in each functional area. If a valid MAC address only appears once, there is no duration and no count is counted. If a valid MAC address occurs multiple times, the duration of the stay is the time difference between the last detected time and the first detected time. In determining the functional relationship between the actual human flow and the detected human flow, the present invention adopts one of the following three functional models between detecting the human flow data and the actual human flow data:

1) Average detection rate model: The ratio of the detected person flow in each detection period to the corresponding actual person flow rate is taken as the detection rate, and the average detection rate weighted by the detection rate of each detection period is obtained, which is used to describe the detected flow rate and The relationship between actual human traffic;

2) Segmentation detection rate model: The detected person flow data in each detection period is used as an indicator, and the detected person flow data is divided into a plurality of intervals, and the detection rate in each interval is obtained, thereby establishing the detection person in each interval. The relationship between traffic and detection rate;

3) Cubic spline interpolation model: Using cubic spline interpolation function to fit the tester in each detection period Wherein, when the average detection rate model is used, and the average detection rate is obtained by weighting the detection rate of each detection period, the specific weighting method is: when the actual human flow rate of the detection period 1 is, the detection rate is ^; the actual period of the detection period 2 The flow rate of the person is, the detection rate is ^; ......; the actual person flow of the detection period n

V ₁ W ₁ +V ₂ W ₂ + --+V _n W ₇

Is I;, the detection rate is ^, then the weighted average detection rate is W:

v ₁ +v ₂ +-+v _n wherein, when the cubic spline interpolation model is adopted, the cubic spline interpolation function S (x) given by the present invention has a natural boundary condition of 0, that is,

S"(x ₀ ) = 0

S" (xJ = 0

When the WI-FI probe is used to detect the flow of people in a public place, the detection period used needs to be determined according to the pedestrian characteristics in the functional area to be detected, and may be taken as 10 min, 30 min or lh as the unit time of data collection and statistics. length. When reconstructing a pedestrian trajectory, if a plurality of functional areas can detect a certain MAC address data, it is only necessary to determine the trajectory according to the chronological order in which the MAC address is detected; if a certain MAC address data is at a certain time At the same time, when it is detected by the WI-FI probes in two adjacent functional areas, the MAC address is actually calculated according to the estimated traffic data and the average stay duration in the two functional areas. The probability of the region, and gives the probability of two pedestrian trajectories. Wherein, the estimated person flow data is calculated by the function data of the detected person flow data; the average stay duration refers to the average value of the stay duration of the plurality of valid MAC address data in a certain detection period, for only one detection record The valid MAC address data is not considered when calculating the average stay duration. Based on the estimated human traffic data and the average stay duration, when a certain MAC address data appears in the detection results of the two functional areas at the same time, the specific method for calculating the probability that the MAC address data belongs to the two functional areas is -

1) Mark the two functional areas as A and B respectively; 2) Calculate the estimated person flow data of A and B in the detection period in which the MAC address data appears, respectively, and record the average stay duration of the valid MAC address in the detection period in which the MAC address data appears in A and B respectively. , respectively recorded as 7^ and Γβ _;

4b3) Calculate the probability that the MAC address data actually belongs to A is /^ = ., , , which belongs to B

QA A+QB B

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a schematic diagram of a probe layout scheme in a functional area. Based on the relationship between the probe detection radius and the length of the functional area, three probe placement schemes are given.

Figure 2 is a schematic diagram of data screening pre-experiment. Pre-experimental probe placement scheme that eliminates invalid data.

Figure 3 is a schematic diagram showing the results of data screening experiments. An analysis method for detection data in a data screening standard based on received signal strength values.

Figure 4 is a flow chart of processing MAC address raw data.

Figure 5 is a schematic diagram of the calculation of the trajectory probability distribution when a certain MAC address is detected simultaneously in the adjacent functional area during trajectory reconstruction. detailed description

In the invention, when exploring the layout form of the WI-FI probe in the functional area of the public place, since the probe itself has a detection area with a radius r, the functional area can generally be regarded as a rectangle having a length a and a width b, so The detection area of the probe has a size relationship with the area of the functional area, which directly affects the number of probes required and the layout form. Therefore, according to the relationship between the probe detection radius r and the length of the functional region, three layout schemes of the probe are given, as shown in Fig. 1:

In the first scheme, when a certain length of the functional area is smaller than the detection radius of the probe, that is, a<r or b<r, a WI-FI probe is disposed at the center of the functional area;

In the second scheme, when one side of the functional area is longer than the probe detection radius, that is, a»r or b»r, a set of WI-FI probes are arranged along the longer side of the functional area; In the third scheme, the two side lengths of the functional area are much larger than the probe detection radius, that is, a»r and b»r, then a set of WI-FI probes are arranged along the diagonal of the functional area; The third method uses the method of arranging the probe along the diagonal of the rectangle because when the flow rate of the person is large, the multipath and the emission phenomenon of the wireless signal emitted by the mobile device during the propagation process are obvious, resulting in serious attenuation of the signal strength. By arranging the probes diagonally, it is ensured that probes are arranged in the middle and both sides of the functional area, which can effectively disperse the signal receiving points and more comprehensively receive data from the inside and the outside of the functional area, that is, to some extent. The signal attenuation caused by multipath and reflection phenomenon is reduced; on the other hand, the probe is arranged along the diagonal line to ensure that the probe has a certain distance in the direction of the two sides of the functional area, which can effectively increase the probe. The overall effective detection area of the needle increases the detection time and effectively reduces the probability that the mobile device will not signal when the pedestrian passes through the detection area, that is, the detection rate is increased. The pre-experiment of the present invention determines the data screening standard based on the received signal strength value. The specific content of the pre-experiment is as follows: The layout of the three probes is as shown in FIG. 2, with the probe as the center and the functional region as the short side length. In the area where one-half of the radius is used, multiple mobile devices that turn on the WI-FI function are used to simulate the movement of the pedestrian. After a period of detection, the detection results of each probe are counted. The present invention analyzes the received signal strength value data obtained in the preliminary experiment as shown in Fig. 3, indicating that the received signal strength value obeys a normal distribution, and the present invention takes a 90% confidence interval to determine the final data screening line. In the MAC address raw data, the received signal strength value is smaller than the MAC address data of the screening line, which is considered to be invalid MAC address data. As shown in FIG. 4, the present invention provides a process of processing MAC address raw data of a functional area, thereby obtaining valid MAC address data, and calculating an average stay duration of the functional area by counting the staying duration of the valid MAC address data. Wherein, when a valid MAC address data has multiple detection records in the probe detection result of a certain functional area, the duration of the stay is the time difference between the last detection time and the initial detection time; and the effective MAC address data in the function area The average length of stay is the arithmetic mean of all dwell durations. The invention provides a method for establishing the relationship between the actual value of the human flow and the detected value by using the average detection rate model, that is, the specific weighting method for obtaining the average detection rate by weighting the detection rate of each detection period is: when the actual detection period 1 is The flow rate of the person is, the detection rate is ^; the actual person flow rate of the detection period 2 is, the detection rate is ^; ...; the actual person flow rate of the detection period n is I;, the detection rate is ^, then the weighted average detection rate is W= ¹ 2 ₊ ...+ .

The present invention establishes a cubic spline interpolation function to fit the relationship between the actual value of the pedestrian flow and the detected value. The specific method is as follows:

In the experiment, the n data will be obtained, and the mobile devices detected in each group of data are respectively counted.

The number of MAC addresses is recorded as χ. · ···χ _η , corresponding to the interval [χ. , χ^] on each node, and manually count the actual person traffic corresponding to each node as y. , ...; that is, the correspondence between each node is determined to be fOJ =; ^. Then, the cubic spline interpolation function S(x) can be constructed as follows.

Remember hj = xj― x _l , S"(x _; ) = Mj, then

( ^Χ ) = 3⁄4 ^Μ 7-ι + 3⁄47^ ^W ; + W + ^c 2 (2)

μ _] Μ _] _ ₁ + 2Mj + YjM _j+1 = dj, j = 1,2..., n-1 (5) where in equation (5):

_7/ = - (6) ⁿ _j+ h _j+1

μ!. = 1— γ,- =—^― (7)

h _j+ h _j+1 ^a J - h _j+ h _j+1 { h _j+1 hj ) - ^b / ;-i'7' 7+iJ W Combines the natural boundary condition S"(x.) = M. = 0 And S"(xJ = Μ„ = 0, (5) can be written in matrix form:

According to the equations α) - (9), the cubic spline interpolation function can be calculated as follows:

i S (x), X G [ Q. I]

S2 ( )' ^Χ G [Χι, ^Χ 2]

The invention uses the correction parameter α to correct the established relationship between the detected human flow rate and the actual human flow function, and the modified parameter e is obtained by repeatedly experimenting with a certain mobile device over a certain WI-FI detection area. The WI- of the mobile device in the experiment. The FI function is turned on, and the MAC address of the mobile device, the number of repeated passes, and the like are recorded, and the experimental results are processed as follows:

If the mobile device passes N times in the detection area of the probe, and the MAC address of the mobile device exists n times in the probe detection result, that is, n times is detected, it is considered that the WI-FI probe pair opens the WI- The detection rate of the FI-enabled mobile device is α, and c = then the final cubic spline interpolation function model is needed.

Before S(x), the correction parameter α is added, that is, the modified cubic spline interpolation function is S(x)' = S(x)/a.

The invention uses the modified parameter β to correct the established relationship between the detected human flow rate and the actual human flow function, and the modified parameter is obtained through a questionnaire survey of the pedestrian on the road to be tested. The main content of the questionnaire is to investigate the pedestrians on the road to be tested. The number of mobile devices, the specific correction method is:

If the result of the questionnaire shows that the proportion of pedestrians carrying two mobile devices is a, then the correction parameter β = 1 + a, then the correction parameter β is required to be multiplied before the final cubic spline interpolation function model S(x). The latter cubic spline interpolation function is S(x)' = S(x) ■ β. As shown in FIG. 5, when reconstructing a pedestrian trajectory, when a certain MAC address data appears in the detection result of two functional areas at the same time, a specific method for calculating the probability that the MAC address data belongs to two functional areas is :

1) Mark the two functional areas as A and B respectively; 2) Calculate the estimated person flow data of A and B within the detection period in which the MAC address data appears, respectively, and record and calculate the average stay duration of the valid MAC address of A and B in the detection period in which the MAC address data appears. , respectively, as 7^ and Γ _β;

3) Calculate the probability that the MAC address data actually belongs to A is /^ = ., , , the probability of belonging to B

QA A+QB B

Claims

Claim

1. A method of detecting human flow in a public place using a WI-FI probe, comprising the following steps:

1) Data collection: In the functional areas of the public place, the WI-FI probe is used to obtain the MAC address raw data of the pedestrian mobile device in each detection period of the detection area; and the actual human flow data is manually collected; The area is a rectangular area with a length a and a width b. The specific layout of the WI-FI probe is based on the relationship between the probe detection radius r and the length of the functional area. The following three probe layout schemes are adopted. As a proposed probe layout:

La) When a certain length of the functional area is smaller than the probe detection radius, ie a<r or b<r, a WI-FI probe is placed at the center of the functional area;

Lb) When one side of the functional area is much longer than the probe detection radius, ie a» _r or b»r, a set of WI-FI probes are placed along the longer side of the functional area;

Lc) when both sides of the functional area are much larger than the probe detection radius, ie a» _r and b»r, a set of WI-FI probes are arranged along the diagonal of the functional area;

2) Data screening: screening the raw data of the MAC address based on the received signal strength value, eliminating invalid MAC address data, obtaining valid MAC address data of the pedestrian mobile device, detecting the traffic data, and recording each detected traffic The length of time the data stays in the functional area;

3) data processing: establishing a function model between the detected person flow data and the actual person flow data for the pedestrian mobile device effective MAC address data;

4) Trajectory reconstruction: For pedestrians passing through different functional areas in public places, reconstruct their trajectories according to the results of probe detection, and reconstruct the pedestrian trajectory by one of the following two methods:

4a) When a certain functional area can detect a certain MAC address data, it is only necessary to determine the trajectory according to the chronological order in which the MAC address is detected;

4b) When a certain MAC address data is detected by the WI-FI probe in two adjacent functional areas at a certain time, the MAC is calculated based on the detected traffic data and the average stay duration in the two functional areas. The probability that the address actually belongs to each functional area at this moment, giving the probability of two pedestrian trajectories.

2. The method for detecting human traffic in a public place using the WI-FI probe according to claim 1, wherein when the proposed probe deployment scheme adopts la), step 2) is based on the received signal strength value in the data screening. The screening method is: by designing a pre-experiment, finding a minimum value of the received signal strength value corresponding to the pedestrian mobile device effective MAC address data, as a standard of data filtering, the received signal strength value in the MAC address original data is less than The standard MAC address data is culled.

3. The method according to claim 1, wherein the function model described in step 3) is one of the following three: 3a) Average detection rate model: The ratio of the detected human flow rate to the corresponding actual human flow rate in each detection period is used as the detection rate, and the average detection rate W weighted by the detection rate of each detection period is obtained, which is used to describe the detection flow rate. Relationship with actual person flow;

3b) Segmentation detection rate model: The detected person flow data in each detection period is used as an indicator, and the detected person flow data is divided into a plurality of intervals, and the detection rate in each interval is obtained, thereby establishing a detection person in each interval. The relationship between traffic and detection rate;

3c) Cubic spline interpolation model: The cubic spline interpolation function is used to fit the relationship between the detected human flow and the actual human flow in each detection period, and the natural boundary condition has a value of 0.

4. The method for detecting a person's traffic in a public place using the WI-FI probe according to claim 1, wherein: the detecting period is determined according to a characteristic of a human flow in an actual functional area, and the detecting period is determined. 10min> 30min or lh.

5. The method for detecting human traffic in a public place using a WI-FI probe according to claim 2, wherein: the pre-experiment of the data filtering based on the received signal strength value is as follows: In the region, using la) for the proposed probe deployment scheme, a mobile device using multiple known MAC addresses is active in a region centered on the probe and radiused by half of the short side of the functional region, and the statistical probe is Test results.

6. The method for detecting human traffic in a public place using the WI-FI probe according to claim 1, wherein: the duration of staying in step 2) means that each valid MAC address data is within a certain functional area. The time difference between the first test and the last test.

7. The method for detecting human traffic in a public place using the WI-FI probe according to claim 1, wherein: when 4b) is used to reconstruct a pedestrian trajectory, said average stay duration refers to a certain detection. The average value of the stay duration of multiple valid MAC address data during the period; for valid MAC address data with only one detection record, the calculation of the average stay duration is not considered.

8. The method for detecting a person's traffic in a public place using the WI-FI probe according to claim 3, wherein: when the average detection rate model is used, the average detection rate is obtained by weighting the detection rate of each detection period. The specific weighting method is: when the actual human flow rate of the detection period 1 is the detection rate W _{1 ;} the actual human flow rate of the detection period 2 is V ₂ , the detection rate is W ₂ ;...1; the actual human flow rate of the detection period n is V _n , the detection rate is W _n , then the averaged detection rate after weighting

9. The method for detecting human traffic in a public place using a WI-FI probe according to any one of claims 1 to 8, characterized in that: when 4b) is used to reconstruct a pedestrian trajectory, when a certain MAC address data appears at the same time In the detection results of the two functional areas, the specific method for calculating the probability that the MAC address data belongs to two functional areas is: 4bl) mark the two functional areas as A and B respectively;

4b2) Calculate the estimated person flow data of A and B within the detection period in which the MAC address data appears, respectively, as 3⁄4 sum; calculate the average of the effective MAC addresses of A and B in the detection period in which the MAC address data appears. long residence time, and are referred to as 7 ^ _Γ β;

4b3) Calculate the probability that the MAC address data actually belongs to A is /^ = _n ^ ^A l _n ^A _τ , which belongs to the general structure of _B = ^QbTb

QATA+QBTB