WO2012058960A1

WO2012058960A1 - Positioning method and device for mobile terminals

Info

Publication number: WO2012058960A1
Application number: PCT/CN2011/078057
Authority: WO
Inventors: 张晓勇
Original assignee: 中兴通讯股份有限公司
Priority date: 2010-11-05
Filing date: 2011-08-05
Publication date: 2012-05-10
Also published as: CN101986757B; CN101986757A

Abstract

The present invention relates to the field of mobile communications. Disclosed are a positioning method and a device for mobile terminals. The method can enhance the precision of the distance between a mobile terminal and a base station and enhance the positioning precision of a mobile terminal. According to the method: the base station in which a mobile terminal resides is identified; and, the cell ID of the base station and the current signal strength are multiplied with a forward deduction matrix of a training model to obtain the distance between the mobile terminal and the base station. When the cell ID and the signal strength are used as the input to multiply with the forward deduction matrix, the output value is the sampling distance.

Description

Positioning method and device for mobile terminal

The present invention relates to the field of mobile communications, and in particular, to a method and an apparatus for positioning a mobile terminal. Background technique

In the field of mobile communications, there are usually the following positioning technologies: A base station-based positioning technology, where the network side obtains the current base station information (Cell ID) of the mobile terminal to obtain the current location of the mobile terminal, and the accuracy depends on the mobile base station. The distribution and coverage size; the second is based on the advanced forward link trilateration (AFLT) positioning technology, AFLT is a unique technology of CDMA, in the positioning operation, the mobile terminal simultaneously monitors multiple The pilot information of the base station determines the distance to the nearby base station by using the chip delay, and finally calculates the specific location of the mobile terminal by using the triangulation method; the third is the positioning technology based on the wireless network assisted GPS positioning method (AGPS), and the AGPS will be the terminal The work is simplified, and the positioning server and the mobile terminal on the network side cooperate with each other to complete the positioning work, that is, the work of satellite scanning and positioning calculation is transferred from the mobile terminal to the positioning server.

The principle of the base station-based positioning technology is: Since each base station has a unique cell identifier (CID ID), the CID and signal strength of the base station can be moved each time the mobile terminal registers the network. The protocol stack of the terminal is parsed. The base station is used to locate the mobile terminal, and the following operation modes are as follows: First, local operation, obtaining a CID parameter in the mobile terminal protocol stack by a specific program, and correspondingly determining the CID and the area where the base station is located, but finally achieving the positioning, but The area information of the base station needs to be stored on the mobile terminal, and the database also needs to store the correspondence between the CID, the LAC (Location Area Code) and the specific base station location, and find the corresponding base station location in the database through the CID and the LAC; The mobile terminal reports the measurement information to the network side, and reports the measured neighboring cell, the camped cell, and the corresponding signal strength to the network side. The network side calculates the distance between the mobile terminal and the base station according to the reported measurement information, and determines the location of the mobile terminal according to the geographic information of the base station, and feeds back the operation result to the mobile terminal, and the mobile terminal displays the result on the screen.

However, in the above manner, the network side or terminal side programs must use CID, LAC, CH, TA, RxL, and TxPwr. First, the location information of the base station and the base station needs to be determined. Secondly, the distance between the mobile terminal and the base station needs to be calculated according to a certain model. There are two ways to calculate the distance between the mobile terminal and the base station. One is to use the signal attenuation formula to calculate, such as: distance L=TA*500+RxL*A+TxPwr*B, where A is the attenuation coefficient of the RF signal; B For the transmit power attenuation coefficient; TA parameter, the base station can know the TA value, establish a special call when the mobile station is in idle mode, and obtain the TA value; RxL is the signal receiving strength; TxPwr is the transmit power, or use other formula to obtain the mobile The distance between the terminal and the base station; the second is to establish a database of the correspondence between the signal strength and the distance of the base station, and obtain the distance according to the signal strength reported by the mobile terminal, and of course, according to the distance obtained from the signal strength reported several times. average value.

However, the method of obtaining the distance between the base station and the mobile terminal by using the above method is too theoretical, and the error caused by signal drift and interference cannot be avoided. The database model is used to find the distance, and a certain distance cannot be established in the database. For multiple signal strengths, a record must be kept by the averaging method or other strategies, so that the distance between the base station and the mobile terminal during positioning is not accurate. Summary of the invention

The invention provides a positioning method and device for a mobile terminal, which can improve the accuracy of the distance between the mobile terminal and the base station, and improve the positioning accuracy of the mobile terminal.

A method for locating a mobile terminal, the method comprising:

Determining a resident base station of the mobile terminal;

Multiplying the cell identity of the camping base station and the current signal strength by the forward derivation matrix of the training model to obtain the distance between the mobile terminal and the camping base station. A positioning device for a mobile terminal, comprising:

a base station determining module, configured to determine a camping base station of the mobile terminal;

The distance determining module is configured to multiply the cell identifier of the camping base station and the current signal strength by a forward derivation matrix of the training model to obtain a distance between the mobile terminal and the camping base station.

The positioning method of the mobile terminal provided by the embodiment of the present invention uses a neural network model to obtain a cell identifier of the base station and a correspondence matrix between the signal strength and the distance (ie, a forward derivation matrix). After the training is completed, the training model is used. The forward derivation matrix can obtain the distance between the mobile terminal and the base station according to the cell identity of the mobile station and the signal strength of the mobile station. In this method, since the forward derivation matrix of the training model is obtained by training according to the sampling data of each base station, the error caused by signal drift and interference can be weakened, thereby improving the accuracy of the distance between the mobile terminal and the base station, and improving The signal strength of the base station is used to locate the positioning accuracy of the mobile terminal. DRAWINGS

1 is a schematic flow chart of a positioning method of a mobile terminal according to the present invention;

2 is a flowchart of sampling data samples according to an embodiment of the present invention;

FIG. 3 is a flowchart of determining a forward derivation matrix by performing unified model training on each base station according to an embodiment of the present invention;

FIG. 4 is a flowchart of a method for locating a mobile terminal according to an embodiment of the present invention;

FIG. 5 is a structural diagram of a positioning apparatus of a mobile terminal according to an embodiment of the present invention. detailed description

Embodiments of the present invention provide a positioning method for a mobile terminal, which uses an artificial neural network model to determine a unified training model for each base station training, and collects a cell identifier of each base station, a signal strength of each sampling point, a sampling distance, and a training model. Training is performed to obtain a training model that characterizes the mapping relationship between cell identity, signal strength, and sampling distance. In the subsequent positioning of the mobile terminal, Simply input the cell identity and signal strength as inputs and multiply the forward derivation matrix of the training model to obtain the distance between the mobile terminal and the base station. Through the training method of the neural network, the forward derivation matrix of the training model can reflect the mapping relationship between cell identity, signal strength and distance, and can model the signal field of the base station, and the error caused by signal drift and interference. The weakening can be performed to improve the positioning accuracy of the mobile terminal by using the signal strength of the base station.

1 is a schematic flowchart of a positioning method of a mobile terminal according to the present invention. As shown in FIG. 1, the method includes:

Step 001, determining a camping base station of the mobile terminal;

Step 002: Multiply the cell identifier of the camping base station and the current signal strength by the forward derivation matrix of the training model to obtain a distance between the mobile terminal and the camping base station.

Specifically, before the mobile station residing the cell identifier of the base station and the current signal strength as an input, the method further includes: determining whether the resident base station passes the training of the training model, and if yes, moving The cell identity of the base station and the current signal strength are used as inputs, otherwise the model training is performed on the resident base station. The mobile terminal is located according to the distance between the mobile terminal and the camping base station and the location information of the camping base station.

Further, before determining the camping base station of the mobile terminal, the method further includes: performing unified model training on each base station in advance, and determining a forward derivation matrix of the training model.

Specifically, performing unified model training on each base station to determine a forward derivation matrix of the training model, specifically: performing unified model training by using data samples of each base station, and determining a forward derivation matrix of the training model, The data samples include: a cell identifier, a sampled signal strength, and a sampling distance corresponding to the sampled signal.

The foregoing positioning method of the mobile terminal provided by the embodiment of the present invention generally needs to include the following three parts: 1) sampling data samples; 2) performing unified model training on each base station, determining a forward derivation matrix; 3) determining a mobile terminal and a base station the distance.

FIG. 2 is a flowchart of sampling data samples according to an embodiment of the present invention, as shown in FIG. A flow chart of sampled data samples, including:

5101. Determine a sampling base station, and obtain a cell identifier of the sampling base station.

Specifically, the cell identifier of the sampling base station can be read by the mobile terminal from the mobile terminal protocol stack.

5102. Determine a sampling distance.

Specifically, the sampling distance, that is, the distance of the sampling point from the base station.

5103. Collect at least one sampled signal strength at each sampling distance;

Specifically, the signal strength received by the mobile terminal is collected at each sampling point, that is, the sampling distance, and the signal strength collected at each sampling point is at least one, and the collected sampling signal strength may be obtained by the mobile terminal from the protocol stack of the mobile terminal. Reading, the acquired signal strength ends with a signal change law that can reflect the point.

All the base stations that need to be trained are sampled by the above steps to obtain a two-dimensional table of "cell identification-sampling distance-signal strength". If the base station to be trained has established a database of correspondence between signal strength and distance, then it is not necessary to perform sampling.

The collected data samples, that is, the cell identifier, the sampling signal strength, and the sampling distance corresponding to the sampling signal are saved in the database, and the sampling process may be performed by the mobile terminal side and saved, or the sampling distance is recorded by the server side, and the mobile terminal The side reads the cell identifier of the base station and the sampled signal strength, and reports the signal strength to the server. The server saves the cell identifier, the sampled signal strength, and the sampling distance corresponding to the sampled signal.

Through the above steps, the data sample collection process is completed, and the collected data samples can be saved in the database of the training model. Next, the model training process is carried out. In order to realize the distance between the mobile terminal and the base station automatically according to the cell identifier and the sampled signal strength, it is necessary to train to determine the correspondence between the cell identifier, the sampled signal strength, and the sampling distance. The training model used in the embodiment of the present invention is a neural network model, and the correspondence between the input is the cell identifier, the signal strength, and the output is the distance through the training model.

FIG. 3 is a schematic diagram of performing unified model training on each base station according to an embodiment of the present invention. The flow chart of the guide matrix, as shown in FIG. 3, is a flow chart for determining a forward derivation matrix for unified model training for each base station, including:

S201. Determine a number of input parameters of the training model and a number of output values;

In the embodiment of the present invention, the input parameters are determined to be two (cell identifier, sampling signal strength), and the number of output values is one, that is, the sampling distance;

5202. Determine a number and a dimension of the forward derivation matrix.

Specifically, the forward derivation matrix may be composed of only one matrix or multiplied by several matrices. The dimension of the forward derivation matrix refers to the number of rows and the number of columns of the matrix; wherein, the forward derivation matrix is composed of at least one matrix; and, the number and dimension of the forward derivation matrix can be based on a database of the training model The scale and the actual training results of the matrix are adjusted, for example: Can be set to [ 2, M ], [ M, N ], [ N, 1 ], where [ 2, M ] is a matrix of 2 rows and M columns, [ M, N ] is a matrix of M rows and N columns; the values of the matrix, that is, the respective values (ie, weights) in the matrix, can be initialized to a random number between [1, 1];

5203. The cell identifier is read from the database, and the sampled signal strength is input as an input, and multiplied by the forward derivation matrix to obtain a test distance.

Specifically, the cell identifier and the sampled signal strength are read from the database as input, and the test distance is obtained.

5204. Obtain a distance error according to the obtained test distance and the recorded sampling distance. Specifically, the distance calculated according to the calculated test distance is compared with the sampling distance corresponding to the set of data in the database. All the cell identifiers and signal strengths in the database are calculated as above to obtain the test distance and the distance error.

5205. Determine an average distance error of each test distance error;

5206, determining whether the average distance error meets the set accuracy condition, and if so, executing step S207, if no, executing step S208;

S207. Determine a current forward derivation matrix as a forward derivation moment of the trained training model. Array

S208. Reversely adjust values in the forward derivation matrix until the average distance error satisfies the set precision condition, and determine the adjusted forward derivation matrix as the forward derivation matrix of the trained training model;

S209. If the average distance error still fails to meet the set precision condition after the reverse adjustment, return to step S202, re-determine the number and dimension of the forward derivation matrix, and retrain until the average distance error satisfies the set Accuracy conditions.

The set precision condition can be set as needed. For example, the average error can be set to be smaller than a certain fixed value. In the embodiment of the present invention, the setting of the precision condition is not limited.

Through the above steps S201~S209, the unified model training is performed on each base station to determine the forward derivation matrix. When the mobile terminal is subsequently located, only the cell identifier and the signal strength need to be determined as input, and multiplied by the forward derivation matrix of the training model. , the distance between the mobile terminal and the base station can be obtained.

The process of determining the forward derivation may be completed by the mobile terminal and sent to the positioning server, or may be completed by the positioning server on the network side and sent to the mobile terminal, and the forward derivation matrix of the determined training model may be saved in the mobile terminal, so that the mobile terminal can Locating according to the received signal strength and the cell identity of the camping base station, may also be stored in the positioning server, so that the positioning server can locate the mobile terminal according to the signal strength received by the mobile terminal and the resident base station where the mobile terminal is located. And sent to the mobile terminal.

When the forward derivation matrix of the training model is stored in the mobile terminal, the forward derivation matrix is recorded in the configuration file to prevent the mobile terminal from losing these parameters when the power is lost.

4 is a flowchart of a method for locating a mobile terminal according to an embodiment of the present invention. As shown in FIG. 4, a flowchart of a method for locating a mobile terminal according to an embodiment of the present invention includes:

S301. Determine a current camping base station of the mobile terminal.

S302. Determine whether the current camping base station of the mobile terminal passes the training of the training model, if If yes, go to step S303, if no, go to step S304;

Specifically, the training of the base station through the training model refers to data samples in the database of the training model in the process of performing unified training on each base station to determine the forward derivation matrix (ie, cell identification, sampling signal strength, and sampling in the database) The data collected by the base station is included in the sampling distance corresponding to the signal.

S303, taking the current cell identity of the camping base station and the signal strength received by the current mobile terminal as an input, multiplying the forward derivation matrix of the training model to obtain the distance between the mobile terminal and the camping base station, and executing step S304;

5304, using a formula method or a database method to obtain the distance between the mobile terminal and the base station, and executing step S305;

Specifically, the formula method or the database method is a method for locating the distance between the mobile terminal and the base station mentioned in the background art, and will not be described in detail herein.

S305. The mobile terminal is located according to the distance between the mobile terminal and the camping base station and the location information of the camping base station.

Specifically, the mobile terminal is located according to the obtained distance between the mobile terminal and the resident base station and the location information of the resident base station, and displayed;

Or further obtaining the distance between the mobile terminal and the neighboring base station, and the location information of the neighboring base station, and determining the location of the mobile terminal according to the triangulation method. The method for determining the distance between the mobile terminal and the neighboring base station may adopt a chip delay method in the CDMA network, or other methods, and will not be described herein.

The above positioning process can also be completed by the mobile terminal or the positioning server on the network side. In the foregoing embodiment, the forward derivation matrix is determined by collecting the cell identifier, the signal strength, and the sampling distance. In the embodiment of the present invention, the cell identifier, the signal strength, and the sampling coordinates are used as sampling data, and the forward derivation matrix is determined by training. The method is similar to the above method.

Wherein, in step S102, S103, it is necessary to determine sampling coordinates, and in each sampling coordinate At least one sampled signal strength is acquired, and a two-dimensional table of cell identification, sampling coordinates, and signal strength is stored in a database of the training model.

In steps S201~S209, the input parameter is still the cell identifier and the sampled signal strength, the output parameter is the calculated test coordinate, and the average coordinate error is determined according to the error of each test coordinate and the recorded sampling coordinate (ie, the expected value), according to the average The coordinate error inversely adjusts the forward derivation matrix until the error satisfies the coordinate error accuracy.

In steps S301-S305, the cell identifier of the current camping base station and the signal strength received by the current mobile terminal are used as inputs, multiplied by the forward derivation matrix of the training model, and the coordinates of the mobile terminal are obtained, and the mobile terminal is located.

The above method of the embodiment of the present invention may be completed by the mobile terminal on the mobile terminal side or by the positioning server on the network side.

FIG. 5 is a structural diagram of a positioning apparatus of a mobile terminal according to an embodiment of the present invention. As shown in FIG. 5, the apparatus includes: a base station determining module 41 and a distance determining module 42;

a base station determining module 41, configured to determine a camping base station of the mobile terminal;

The distance determining module 42 is configured to multiply the cell identifier of the mobile station resident base station and the current signal strength by a forward derivation matrix of the training model to obtain a distance between the mobile terminal and the camping base station.

The training model is a neural network model. When the cell identifier and the sampled signal strength are input parameters and multiplied by the forward derivation matrix, the output value is the sampling distance.

Preferably, the above apparatus further includes:

The training module 43 is configured to perform unified model training on each base station in advance, and determine a forward derivation matrix of the training model.

The training module 43 performs unified model training on each base station in advance, and determines a forward derivation matrix of the training model, including: performing, for each base station, respectively: determining a cell identifier and a sampling distance of the base station, at each sampling distance. , collecting at least one sampled signal strength, will Set the cell identifier, the collected sampled signal strength, and the corresponding sampling distance to save; set the number and dimension of the forward derivation matrix, take each acquired cell identifier, sampled signal strength as input, and forward derivation matrix Matrix multiplication operation, obtaining test distance; determining average distance error according to error of each test distance and sampling distance; determining whether the average distance error satisfies the set precision condition, and if so, directly determining the forward derivation matrix as training Training the forward derivation matrix of the model, if not, inversely adjusting the values in the forward derivation matrix until the average distance error satisfies the set precision condition, then determining the adjusted forward derivation matrix as the trained training model Forward derivation matrix.

The positioning method of the mobile terminal provided by the embodiment of the present invention uses an artificial neural network model to determine a unified training model for each base station training, and collects the cell identifier of each base station, the signal strength of each sampling point, and the sampling distance, and performs the training model on the training model. Training, obtaining a training model that characterizes the mapping relationship between cell identity, signal strength, and sampling distance. In the subsequent positioning of the mobile terminal, the cell identifier and the signal strength are simply input, and multiplied by the forward derivation matrix of the training model to obtain the distance between the mobile terminal and the base station. Through the training method of the neural network, the forward derivation matrix of the training model can reflect the mapping relationship between cell identity, signal strength and distance, and can model the signal field of the base station, and the error caused by signal drift and interference. The weakening can be performed to improve the positioning accuracy of the mobile terminal by using the signal strength of the base station.

The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included. Within the scope of protection of the present invention.

Claims

Claim

A method for locating a mobile terminal, the method comprising:

Determining a resident base station of the mobile terminal;

Multiplying the cell identity of the camping base station and the current signal strength by the forward derivation matrix of the training model to obtain the distance between the mobile terminal and the camping base station.

The method according to claim 1, wherein before the determining the camping base station of the mobile terminal, the method further comprises: performing unified model training on each base station in advance, and determining a forward derivation of the training model. matrix.

The method according to claim 2, wherein the performing unified model training on each base station determines a forward derivation matrix of the training model, which is:

The data samples of each base station are used for unified model training to determine a forward derivation matrix of the training model. The data samples include: a cell identifier, a sampled signal strength, and a sampling distance corresponding to the sampled signal.

The method according to claim 3, wherein the performing the unified model training by using the data samples of the respective base stations to determine the forward derivation matrix of the training model comprises: performing respectively for each base station: Decoding the cell identifier of the base station and the required sampling distance, collecting at least one sampling signal strength at each sampling distance, and storing the collected cell identifier, the collected sampling signal strength, and the sampling distance corresponding to the sampling signal;

Set the number and dimension of the forward derivation matrix, take each cell identifier and sampled signal strength as input, and perform matrix multiplication on the forward derivation matrix to obtain the test distance; according to the error of each test distance and sampling distance, Determine the average distance error;

Determine whether the average distance error satisfies the set precision condition, and if so, directly determine the forward derivation matrix as the forward derivation matrix of the training model, and if not, inversely adjust the value in the forward derivation matrix until the average distance error When the set precision condition is satisfied, the adjusted forward derivation matrix is determined as the forward derivation matrix of the training model.

The method according to claim 4, wherein, after performing the reverse adjustment, the method further includes:

When the average distance error still fails to meet the set accuracy condition, it returns to reset the number and dimension of the forward derivation matrix.

The method according to any one of claims 1 to 3, wherein, before the mobile station residing the cell identifier of the base station and the current signal strength as an input, the method further includes:

Determining whether the camping base station is trained by the training model, and if so, taking the cell identity of the mobile station camped on the base station and the current signal strength as input, otherwise performing model training on the camping base station.

The method according to claim 1, wherein the method further comprises: locating the mobile terminal according to a distance between the mobile terminal and the resident base station and location information of the resident base station.

A positioning device for a mobile terminal, the device comprising: a base station determining module and a distance determining module; wherein

The device according to claim 8, wherein the device further comprises: a training module, configured to perform unified model training on each base station in advance, and determine a forward derivation matrix of the training model.

The apparatus according to claim 9, wherein the training module performs unified model training on each base station in advance, and determines a forward derivation matrix of the training model, including: performing, for each base station separately: Describe the cell identity of the base station and the required sampling distance, At each sampling distance, at least one sampling signal strength is acquired, and the collected cell identifier, the collected sampling signal strength, and the corresponding sampling distance are saved; the number and dimension of the forward derivation matrix are set, and each of the collected The cell identifier and the sampled signal strength are used as inputs, and the matrix is multiplied by the forward derivation matrix to obtain the test distance; the average distance error is determined according to the error of each test distance and the sampling distance; and whether the average distance error satisfies the set precision condition is determined. If yes, the forward derivation matrix is directly determined as the forward derivation matrix of the training model. If not, the value in the forward derivation matrix is inversely adjusted until the average distance error satisfies the set precision condition, then the adjusted The forward derivation matrix is determined as the forward derivation matrix of the training model.