KR102145717B1 - Indoor positioning method and system using machine learning - Google Patents

Abstract

The present invention discloses an indoor positioning method and system using machine learning. The indoor positioning method according to the present invention includes the steps of measuring a received signal strength (RSSI) of a plurality of positioning signal transmitter signals for each of a plurality of areas defined indoors; A data preprocessing step of converting a value excluding zero (0) among the RSSI measurement data of the entire area into an absolute value complement; Learning by executing a machine learning algorithm using preprocessed RSSI data as training data; And inputting RSSI information of a plurality of positioning signal transmitters measured by the positioning target terminal into an input node of a machine learning algorithm to determine the location of the positioning target terminal.
According to the present invention, the primary data preprocessing for converting the RSSI measurement data measured in each area of the positioning target area to the complement of the absolute value and the secondary data for converting the RSSI value marked with zero (0) into a predetermined positive value After performing the pre-processing, by using it as machine learning learning data, the effect of greatly improving the accuracy of indoor positioning using machine learning can be obtained.

Description

Indoor positioning method and system using machine learning

The present invention relates to an indoor positioning method using machine learning, and more particularly, to an indoor positioning method and system in which positioning accuracy is greatly improved by performing appropriate preprocessing on data provided for machine learning.

In recent years, mobile terminals such as smartphones and tablets, or vehicle terminals, are equipped with a GPS (Global Positioning System) module so that the location of the terminal can be checked, and this allows users to receive various location-based services linked to their location. have.

However, since the GPS module receives a signal from a GPS satellite and calculates the location of the terminal, it cannot be used for positioning indoors where satellite signals cannot be received, and therefore, positioning must be performed indoors by a method other than GPS.

Until now, various indoor positioning methods are known, among which the Wi-Fi positioning method using a wireless signal from a Wi-Fi AP (Access Point) installed indoors is the most widely known.

Wi-Fi positioning methods are largely divided into cell ID methods, triangulation methods, and fingerprinting methods.

The Cell ID method is a method of determining the location of the terminal with the highest signal among the received signals as the location of the terminal. Although it is the simplest method, it has a disadvantage of very low accuracy because a positioning error of about 50m occurs.

The triangulation method is a method of receiving three or more AP signals distributed indoors, converting the received signal strength indication (RSSI) of each AP signal into a distance, and then calculating the location of the terminal. However, indoors, a positioning error of about 20m occurs due to reflection, refraction, and absorption of radio signals, so there is still a problem of low accuracy.

In the fingerprinting method, a number of cells are set in an indoor space, and the received signal strength (RSSI) of each AP signal directly measured in each area is converted into a DB to establish a kind of map, and This is a method of estimating the reference point of the area showing the most similar signal pattern as the location of the terminal by comparing the RSSI value of each received AP signal with the DB.

This fingerprinting method is known to have much higher indoor positioning accuracy than the cell ID method or triangulation method, but it is still necessary to further improve the positioning accuracy because an error of more than 5m occurs based on 80% accuracy.

However, even if the indoor Wi-Fi AP signal is measured at the same location, the RSSI fluctuation is large due to various factors, and as the number of APs increases, it becomes difficult to accurately determine the similarity of the signal pattern. There is a limit to raising the ideal.

Korean Patent Publication No. 10-2015-0089633 (published on Aug. 5, 2015)

The present invention was devised from this background, and an object of the present invention is to provide a method that can further improve positioning accuracy in an indoor positioning method based on fingerprinting.

In order to achieve this object, an aspect of the present invention includes measuring the received signal strength (RSSI) of a plurality of positioning signal transmitter signals for each of a plurality of areas defined indoors; A data preprocessing step of converting a value excluding zero (0) among the RSSI measurement data of the entire area into an absolute value complement; Learning by executing a machine learning algorithm using preprocessed RSSI data as training data; Provides an indoor positioning method comprising the step of determining a location of a positioning target terminal by inputting RSSI information of a plurality of positioning signal transmitters measured by a positioning target terminal into an input node of a machine learning algorithm.

In the indoor positioning method according to an aspect of the present invention, the data preprocessing step converts the unconverted value (0) to a positive number smaller than the minimum value among the converted values after converting the RSSI measurement data to the complement of the absolute value. May include a process.

In addition, in the indoor positioning method according to an aspect of the present invention, after the data preprocessing step, data augmentation may be performed on the preprocessed RSSI data to provide sufficient learning data.

Another aspect of the present invention is a positioning system that determines the location of a positioning target terminal by using the received signal strength (RSSI) of a radio signal transmitted from a plurality of positioning signal transmitters installed indoors. A first data preprocessing unit for converting a value excluding zero (0) among the RSSI data of the plurality of positioning signal transmitters measured every time into an absolute value complement; A machine learning algorithm that performs machine learning by using preprocessed RSSI data as training data; Provides an indoor positioning system including a positioning unit that determines the location of the positioning target terminal by inputting RSSI information of a plurality of positioning signal transmitters measured by a positioning target terminal into a machine learning algorithm.

In an indoor positioning system according to another aspect of the present invention, a second data preprocessing unit converts a value (0) that is not converted to an absolute value complement into a positive number smaller than a minimum value among values converted to the absolute value complement. It may include a data preprocessor.

In addition, the indoor positioning system according to another aspect of the present invention may include a data expansion unit that performs data augmentation on pre-processed RSSI data in order to provide sufficient learning data.

In addition, in the indoor positioning system according to another aspect of the present invention, the positioning signal transmitter may be a Wi-Fi AP (Access Point).

According to the present invention, the primary data preprocessing for converting the RSSI measurement data measured in each area of the positioning target area to the complement of the absolute value and the secondary data for converting the RSSI value marked with zero (0) into a predetermined positive value After performing the pre-processing, by using it as machine learning learning data, the effect of greatly improving the accuracy of indoor positioning using machine learning can be obtained.

1 is a schematic configuration diagram of an indoor positioning system according to an embodiment of the present invention
2 is a diagram illustrating a state in which an indoor positioning target area is divided into a plurality of areas
3 is a block diagram of a machine learning unit
4 is a diagram illustrating a configuration of an artificial neural network of a deep learning algorithm
5 is a diagram illustrating a preprocessing process of learning data
6 is a flow chart showing an indoor positioning method according to an embodiment of the present invention

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings.

For reference, in the present specification, when a certain part includes or includes a certain component, it means that other components may be further included or provided, rather than excluding other components unless specifically stated to the contrary. In addition, when one element connects, combines, or communicates with another element, it is not only directly connected, coupled, or communicated with another element, but also indirectly connected with another element in the middle. This includes cases of combining, or communicating. In addition, when one component is directly connected or directly coupled to another component, it means that no other component is intervened. In addition, since the drawings attached to the present specification are merely illustrative for easy understanding of the subject matter of the invention, it should be noted in advance that the scope of the present invention is not limited thereto.

An indoor positioning system according to an embodiment of the present invention includes a plurality of positioning signal transmitters AP1, AP2, AP3, AP4, and positioning signal transmitters AP1, AP2, as shown in the schematic configuration diagram of FIG. 1. A positioning target terminal 100 capable of receiving a wireless signal transmitted from AP3 and AP4, and a positioning device 300 communicating with the positioning target terminal 100 through the communication network 200 may be included.

The positioning signal transmitters AP1, AP2, AP3, and AP4 are devices that periodically transmit a positioning wireless signal, and may be a Wi-Fi AP installed for a wireless Internet service. However, the positioning signal transmitters (AP1, AP2, AP3, AP4) are not necessarily limited to Wi-Fi APs, and APs and repeaters that support other types of wireless communication can also be used as positioning signal transmitters.

When a Wi-Fi AP is used as a positioning signal transmitter (AP1, AP2, AP3, AP4), a beacon signal periodically transmitted from the Wi-Fi AP may be used as a positioning wireless signal.

The positioning target terminal 100 is an electronic device carried by a person or mounted on a mobile means, a wireless signal transmission/reception module, a memory for storing programs and protocols for data transmission/reception and processing, and for calculation and control by executing various programs. A processor or the like may be provided.

For example, the positioning target terminal 100 may be an electronic device such as a smartphone, a tablet PC, a handheld PC, a PDA, a PMP, a notebook computer, or the like. However, since the positioning target terminal 100 is not necessarily limited to a small portable terminal, a digital TV, laptop computer, desktop computer, CCTV, office equipment, home appliances, industrial equipment, etc. that are fixedly installed in the positioning target area are also the positioning target terminal 100. Can be included in

The positioning target terminal 100 receives the positioning radio signal transmitted from a plurality of positioning signal transmitters (AP1, AP2, AP3, AP4) installed around or inside the positioning target area, and receives signal strength (RSSI) for each signal transmitter. ) The information is transmitted to the positioning device 300.

The communication method between the positioning target terminal 100 and the positioning device 300 is not limited. For example, when the positioning signal transmitters AP1, AP2, AP3, and AP4 are Wi-Fi APs, they may communicate with the positioning device 300 through this. It goes without saying that it is also possible to communicate with the positioning device 300 through another communication device in place of the positioning signal transmitters (AP1, AP2, AP3, AP4).

The communication network 200 relays communication between the positioning target terminal 100 and the positioning device 300, and may be configured as a wired network, a wireless network, or a suitable combination thereof, which is known or developed later.

The positioning device 300 acquires RSSI information of the positioning signal transmitters (AP1, AP2, AP3, AP4) measured by the positioning target terminal 100, locates the location of the corresponding terminal, and determines the positioning result of the positioning target terminal 100. ) Or to a preset location-based service providing server.

Since the indoor positioning method according to an embodiment of the present invention is based on a fingerprinting method, the positioning target area is divided into multiple areas (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ) as illustrated in FIG. 2. It is divided and the positioning device 300 is the received signal strength (RSSI) of the positioning signal transmitters (AP1, AP2, AP3, AP4) measured for each area (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ). ) The location of the positioning target terminal 100 is determined using data.

The positioning device 300 according to an embodiment of the present invention may include a machine learning unit 310, a positioning unit 330, a storage unit 350, and a communication unit 370.

The machine learning unit 310 performs machine learning for location positioning using a machine learning algorithm and learning data, and in an embodiment of the present invention, as shown in the block diagram of FIG. 3, the first data preprocessing A unit 311, a second data preprocessor 312, a data expansion unit 313, and a machine learning algorithm 314 may be included.

The machine learning algorithm 314 performs machine learning by using the learning data, thereby increasing the accuracy of the output value.

The kind of machine learning algorithm 314 is not particularly limited. As an example, machine learning for indoor positioning may be performed using algorithms such as Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Random Forest. As another example, a machine for indoor positioning using deep learning algorithms based on artificial neural networks (ANNs) such as DNN (Deep Neural Network), CNN (Convolutional Neural Network), and RNN (Recurrent Neural Network). You can also run.

The artificial neural network (ANN) of the deep learning algorithm has one or more hidden layers between the input layer and the output layer as illustrated in FIG. 4, and learning through the deep learning algorithm is a node of an arbitrary first layer. This is a process of optimizing the weights ([W]1, ..., [W]n) assigned between the (neuron) and the node (neuron) of the second layer.

The deep learning algorithm applies the gradient reduction method, etc., based on the error between the output value of the output node and the target value at each learning cycle or at a set cycle, and each weight ([W]1, ... ,[W ]n), the accuracy of the output value can be continuously improved.

In an embodiment of the present invention, the received signal strength (RSSI) of each signal transmitter (AP1, AP2, AP3, AP4, ..., APn) received by the positioning target terminal 100 at an arbitrary position at each input node of the input layer. ) Is input, and the output node of the output layer may output the location (C _xy ) of the positioning target terminal 100 corresponding to the input RSSI information.

In order to increase the accuracy of indoor positioning using machine learning, it is necessary to determine an optimal weight applied to the machine learning algorithm 314 through learning, and for this purpose, appropriate and sufficient learning data must be provided.

In order to provide appropriate learning data, in the embodiment of the present invention, each signal transmitter (AP1, AP2, AP3, AP4, ... ,) measured for each zone (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ) APn)'s RSSI measurement data is not used as it is, but is used as learning data for machine learning after a predetermined preprocessing is performed on the measurement data.

First, the first data preprocessing unit 311 of the machine learning unit 310 performs primary data preprocessing of converting the RSSI measurement data into the complement (even number) of the absolute value, as illustrated in FIG. 5.

In general, the received signal strength (RSSI) of a wireless signal is expressed as a negative number such as -30dBm, -50dBm, etc., and the smaller the absolute value, the greater the signal strength. However, according to the experiment, it was found that the positioning accuracy was improved by more than 10% when the data converted to the absolute value's complement (even number) was used as the learning data, rather than using the measured data as it is.

Complement (even number) refers to a number in which the sum of each other is in a certain relationship, and in the embodiment of the present invention, numbers in a relationship with a sum of 100 are assumed to be in a complementary relationship, but the sum of two numbers in a conservative relationship is 100 Instead, it can be replaced with another value.

In an embodiment of the present invention, for example, if the measured data is -30dBm, the absolute value is 30, and thus the first data preprocessor 311 converts it to the complement 70 of the absolute value. If the measured data is -70dBm, the absolute value is 70, so convert it to the complement of the absolute value of 30.

However, when performing the primary data preprocessing, it should be noted that if the measured data is 0, the primary data preprocessing should not be performed.

If the measured data is displayed as 0, the AP signal is not received or the AP signal whose RSSI is less than the threshold is ignored for the processor efficiency of the terminal. If measured data 0 is converted to an absolute value of 100, the weakest signal This is because is converted into the strongest signal.

Fig. 5(b) shows the first preprocessed data according to this criterion.

Next, the second data preprocessing unit 312 of the machine learning unit 310 converts the unconverted value (0) from the first preprocessed data into an arbitrary positive number.

The RSSI information of each signal transmitter (AP1, AP2, AP3, AP4, ..., APn) measured in each area (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ) of the positioning target area is machine learning. It is used as the learning data of the algorithm. If the RSSI value is 0, it is recognized that the signal of a specific AP is not weak in the relevant area, but that there is no specific AP at all and learning proceeds.

However, if learning proceeds by recognizing that there is no specific AP in a specific area, overfitting, which is an excessive conditional limitation, may result in lowering of positioning accuracy.

In an embodiment of the present invention, in order to prevent such an overfitting problem, a value (0) that is not converted in the first preprocessing process is converted into a predetermined positive number, thereby preventing a decrease in positioning accuracy.

In the embodiment of the present invention, all 0s are converted to 10 as shown in Fig. 5(c), but the present invention is not limited thereto. However, since the RSSI value of zero (0) means that the received signal is very weak, the converted number is preferably the lowest among all RSSI values.

5(b) shows that the lowest value other than 0 is 30 (C23/AP4, C24/AP4) among all the first preprocessed RSSI data, so the RSSI value zero (0) is naturally converted to a value sufficiently smaller than 30. It should be, and in the embodiment of the present invention, it was converted to 10 for this reason. It goes without saying that it is not limited thereto and may be converted to a positive number other than 10.

Meanwhile, for machine learning, a sufficient amount of training data must be provided, and for this, at least thousands of sets of data illustrated in FIG. 5 must be prepared.

However, in reality, it is almost impossible to perform RSSI measurements thousands of times in each area (C ₁₁ , C ₁₂ , ... C ₄₃ , C ₄₄ ) of the target area. In order to solve this problem, it is desirable to augment the measured data through the data expansion unit 313 after performing the actual measurement dozens of times, and use the expanded data as learning data.

The data expansion method in the data expansion unit 313 is not particularly limited. For example, in consideration of the fact that the RSSI value changes every time a measurement is made under the same conditions, the data can be expanded in various ways, such as increasing or decreasing all or part of the measurement data within a certain range, or changing the value of some APs to the minimum I can.

The data expansion of the data expansion unit 313 is preferably performed on the secondary preprocessed data, but is not limited thereto. Therefore, data expansion may be performed on the measured data or the first preprocessed data.

On the other hand, when the positioning unit 330 of the positioning device 300 receives a positioning request for the positioning target terminal 100, each signal transmitter AP1, AP2, AP3, AP4, measured by the positioning target terminal 100 is received. .., APn)'s RSSI data is input into the machine learning algorithm that has been trained, and the output value is determined as the location (Cxy) of the positioning target terminal 100.

The storage unit 350 stores an OS program, a machine learning program, various parameters, and data for the operation of the positioning device 300.

The communication unit 370 includes a wired and/or wireless communication interface for connection with the communication network 200.

Hereinafter, an indoor positioning method according to an embodiment of the present invention will be described with reference to the flowchart of FIG. 6.

First, as illustrated in FIG. 2, a plurality of zones (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ) are set in the positioning target area, and the positioning device 300 is each signal transmitter measured in each zone. Receives and stores the RSSI value of (AP1, AP2, AP3, AP4, ..., APn). It is preferable that the RSSI measurement location is in the center of each zone (C ₁₁ , C ₁₂ , ..., C ₄₃ , C ₄₄ ). (ST11)

Subsequently, the machine learning unit 310 of the positioning device 300 performs primary data preprocessing on the RSSI measurement data. The primary data preprocessing is to convert an RSSI value expressed as a negative number into a positive number, specifically converting it into an absolute value's complement. However, the measured value marked as 0 is left unchanged without conversion. If the first pre-processing is performed in this way, positioning accuracy can be greatly improved. (ST12)

Subsequently, the machine learning unit 310 performs a second pre-processing of converting the unconverted value (0) into an arbitrary positive number in the first pre-processing process. At this time, it is converted to a value sufficiently smaller than the minimum value of the corresponding RSSI data set. Performing such secondary preprocessing can prevent the overfitting problem of machine learning. (ST13)

The machine learning unit 310 may perform data expansion in order to secure a more sufficient amount of training data after performing such a data preprocessing process, and store it in the training data DB. (ST14)

When an appropriate and sufficient amount of learning data is secured through the above process, the machine learning unit 310 executes the machine learning algorithm 314 to perform learning. At this time, the weights ([W]1, ..., [W]n) assigned to the machine learning algorithm can be continuously updated. (ST15)

When receiving a positioning request while sufficiently performing machine learning, the positioning unit 310 determines each signal transmitter (AP1, AP2, AP3, AP4, ..., measured by the positioning target terminal 100) at an arbitrary position. The RSSI data of APn) is acquired, inputted to the machine learning algorithm that has been trained, and an output value is determined as the location (Cxy) of the positioning target terminal 100. (ST16, ST17)

The positioning device 300 transmits the location (Cxy) information of the positioning target terminal 100 determined through the above process to the positioning target terminal 100 or a preset location-based service providing server. (ST18)

Meanwhile, the first data preprocessing unit 311, the second data preprocessing unit 312, the data expansion unit 313, and the positioning unit 330 constituting the positioning device 300 are software, hardware, or software and hardware. It can be implemented through a combination of. In addition, at least two of these components may be integrated and provided as one piece of software or one piece of hardware.

In the above, a preferred embodiment of the present invention has been described, but the present invention is not limited to the above-described embodiment and may be modified or modified in various forms.

For example, in the embodiment of the present invention, it has been described that the positioning target terminal 100 and the positioning device 300 are spaced apart from each other, but it is not limited thereto, so that some or all of the components of the positioning device 300 are positioned. It may be installed inside the target terminal 100.

As described above, the present invention may be variously modified or modified in a specific application process, and if the modified or modified embodiment includes the technical idea of the present invention disclosed in the claims to be described later, it is natural that it belongs to the scope of the present invention. Will do.

100: positioning target terminal 200: communication network
300: positioning device 310: machine learning unit
311: first data preprocessor 312: second data preprocessor
313: data extension 314: machine learning algorithm
330: positioning unit 350: storage unit
370: communication department

Claims (7)

Measuring received signal strength (RSSI) of a plurality of positioning signal transmitter signals for each of a plurality of areas defined indoors;
A data preprocessing step of converting a value excluding zero (0) among the RSSI measurement data of the entire area into an absolute value complement;
Learning by executing a machine learning algorithm using preprocessed RSSI data as training data;
Step of determining the location of the positioning target terminal by inputting RSSI information of a plurality of positioning signal transmitters measured by the positioning target terminal into the input node of the machine learning algorithm
Including,
The data pre-processing step includes converting the unconverted value (0) into a positive number smaller than the minimum value among the converted values after converting the RSSI measurement data to the complement of the absolute value. delete The method of claim 1,
After the data pre-processing step, indoor positioning method, characterized in that performing data augmentation on pre-processed RSSI data to provide sufficient learning data As a positioning system that determines the location of a positioning target terminal by using the received signal strength (RSSI) of wireless signals transmitted from multiple positioning signal transmitters installed indoors,
A first data preprocessing unit for converting a value excluding zero (0) from among the RSSI data of the plurality of positioning signal transmitters measured for each of a plurality of areas defined indoors into a complement of an absolute value;
A second data preprocessor for converting a value (0) that is not converted to an absolute value complement by the first data preprocessor into a positive number smaller than a minimum value among values converted to the absolute value complement;
A machine learning algorithm that performs machine learning by using preprocessed RSSI data as training data;
A positioning unit that determines the location of the positioning target terminal by inputting RSSI information of multiple positioning signal transmitters measured by the positioning target terminal into a machine learning algorithm.
Indoor positioning system including delete The method of claim 4,
Indoor positioning system comprising a data expansion unit that performs data augmentation on preprocessed RSSI data to provide sufficient learning data The method of claim 4,
The indoor positioning system, characterized in that the positioning signal transmitter is a Wi-Fi AP (Access Point)

Images