CN110830939B - Positioning method based on improved CPN-WLAN fingerprint positioning database - Google Patents
Positioning method based on improved CPN-WLAN fingerprint positioning database Download PDFInfo
- Publication number
- CN110830939B CN110830939B CN201910810666.6A CN201910810666A CN110830939B CN 110830939 B CN110830939 B CN 110830939B CN 201910810666 A CN201910810666 A CN 201910810666A CN 110830939 B CN110830939 B CN 110830939B
- Authority
- CN
- China
- Prior art keywords
- data
- fingerprint
- cpn
- database
- positioning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
- G06F16/2282—Tablespace storage structures; Management thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management
- H04W64/006—Locating users or terminals or network equipment for network management purposes, e.g. mobility management with additional information processing, e.g. for direction or speed determination
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Databases & Information Systems (AREA)
- Collating Specific Patterns (AREA)
Abstract
The invention relates to a WLAN fingerprint indoor positioning technology, and provides a WLAN fingerprint positioning database construction algorithm giving an improved CPN (compact peripheral component network) by improving a transmission neural network, so as to reduce the workload of an offline stage of WLAN fingerprint positioning. Step 1: selecting a practical measurement reference point in an area needing to be positioned and carrying out data measurement; step 2: using an autoencoder to perform dimension reduction on the reference point fingerprint information; and step 3: constructing a suitable neural network structure; and 4, step 4: training a neural network by using data measured by the actual measuring points; and 5: and constructing a fingerprint positioning database by using the trained neural network. The invention is mainly applied to WLAN fingerprint indoor positioning occasions.
Description
Technical Field
The invention relates to a WLAN fingerprint indoor positioning technology, in particular to a WLAN fingerprint positioning database construction and positioning method based on an improved neural Network (CPN).
Background
With the rapid development of intelligent mobile terminals such as mobile phones and tablet computers, Location Based Services (LBS) has increased rapidly, and especially, the demand for indoor positioning has increased rapidly. At present, the method based on indoor positioning mainly includes Radio Frequency Identification (RFID), Bluetooth (BT), Wireless Local Area Network (WLAN), Zigbee, Ultra Wide Band (UWB), and other schemes. The WLAN fingerprint positioning technology has wide social acceptance due to the advantages of no need of additional equipment, low cost and the like. However, since a large number of Reference Points (RPs) need to be measured in the off-line stage, manpower and material resources are consumed, so that the popularization of the WLAN fingerprint positioning technology has certain difficulty.
For the establishment of the off-line stage database, the following solutions are mainly available at present:
unnecessary RPs are removed by counting the signal characteristics, and the calculation complexity of an online stage is reduced; by designing a smoothing filter and a tailing filter, a small amount of RP is used for expanding a fingerprint positioning database, but the application scene is limited; the signal acquisition time of the RP is shortened, and a fingerprint positioning database is constructed by using kriging interpolation, but the workload of an off-line stage is not substantially reduced; the neural network is applied to predicting RSS (Received Signal Strength) fingerprints of unknown RPs, workload in an off-line stage is reduced, and the method is only suitable for the condition with low precision requirement; the database is constructed by a method that a terminal holder actively submits fingerprint information, but the database still needs to be done as an additional task; by applying the concept of crowd sensing, a user can complete the collection of fingerprint information without paying special attention, but the risk of privacy information leakage exists; the neural network is utilized to reduce the measurement area in the off-line stage and lighten the workload, but the positioning effect is poor.
In order to solve the above problems and reduce the workload of the offline phase, new and higher requirements are put on the construction of the WLAN fingerprint positioning database. Based on the method, the research on the WLAN fingerprint positioning database construction algorithm is carried out.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a WLAN fingerprint positioning database construction algorithm giving an improved CPN (central processing Network) by improving the CPN, so as to reduce the workload of the off-line stage of WLAN fingerprint positioning. Therefore, the technical scheme adopted by the invention is that the positioning method based on the improved CPN-WLAN fingerprint positioning database comprises the following steps:
step 1: selecting a practical measurement reference point in an area needing positioning and measuring data
Actually measured reference points are approximately and uniformly selected in an area to be positioned, the position coordinates and RSS fingerprints of the actually measured reference points are measured, the number of the actually measured reference points is obviously less than the total number of the reference points, and the workload of an off-line stage is reduced;
step 2: dimensionality reduction of reference point fingerprint information using an auto-encoder
(1) Constructing a proper self-encoder, setting the neuron number of an input layer and an output layer of the self-encoder to be the same as the AP (Access Point) number of an area to be positioned, and setting the neuron number of a hidden layer to be 64;
(2) and inputting the reference point fingerprint data into an autoencoder for iterative training.
And step 3: constructing a suitable neural network structure
(1) Modifying an input layer of a CPN (coherent neural network), and setting the neuron number as the sum of a two-dimensional space coordinate and the neuron number of a hidden layer of a self-encoder;
(2) for a Kohonen layer part of the CPN, setting the maximum 2 neurons input in an activation layer, and obtaining corresponding activation weights by using a Softmax function;
and 4, step 4: training neural networks using measured data from real-world points
Inputting the actual measurement point fingerprint information into an autoencoder for dimensionality reduction, then inputting the actual measurement point fingerprint information and the reference point spatial coordinates into a CPN together, and performing iterative training;
and 5: construction of fingerprint location database using trained neural networks
(1) Predicting fingerprint information of remaining reference points
Inputting the two-dimensional position coordinates of the rest reference points into the trained neural network according to a specified form, wherein the corresponding part in the output value is the fingerprint information of the corresponding reference point;
(2) constructing a complete fingerprint location database
And combining the fingerprint information of the rest reference points with the fingerprint information of the actually measured reference points, thereby completing the construction of the WLAN fingerprint positioning database for positioning.
Improving the activation strategy of the neurons in the Kohonen layer by using a Softmax function, activating 2 input maximum neurons at a time, mapping a group of data into a (0,1) interval by using the Softmax function, adding up the obtained numerical values to be 1, wherein each data corresponds to the classification probability of the size, and for an array V, containing nElement of ViRepresenting the ith element in V, then the Softmax value for this element is:
further comprising a verification step: and adopting the WLAN fingerprint positioning data set UJIIndorLoc obtained by actual measurement. The data set mainly comprises a tracing data set and a validation data set, wherein the information is shown in table 1:
TABLE 1 UJIIndenorLoc data set content form Table
Data at BUILDINGID of 0 and FLOOR of 0 is used: 1059 groups of data at 54 RPs are shared in the training data set, data at 10 RPs are taken as a training set and are regarded as measured data at an offline stage, and the rest data are taken as a verification set; the validation dataset had 78 groups of data at 78 RPs in common, as the test set;
the training set, the verification set and the test set are in the same plane space;
after the neural network training is finished, inputting RP coordinates in a verification set into the neural network, wherein input data of the improved CPN and the traditional CPN are (0, 0, … …, 0, 0, LONGITUDE, LATITUDE), and the output item 520 is a predicted value of RSS; the BPNN inputs data (LONGITUDE, LATITUDE), outputs RSS prediction values, is combined with training set data to complete the construction of a fingerprint positioning database, a testing set RP is positioned on the constructed database by using KNN and WKNN, all experimental data are repeated for 10 times, and an average value is taken for recording; comparing database construction conditions when RSS fingerprints at all RPs, 31 RPs and 16 RPs in the verification set are predicted and verified;
after the improved CPN training is completed, coordinate values of RP in a verification set are input into a neural network, a predicted value corresponding to an RSS fingerprint is output, the predicted value is combined with training set data, a fingerprint positioning database is constructed, the RP of a test set is used as a to-be-positioned point, the constructed fingerprint positioning database is subjected to construction effect evaluation by using a Nearest neighbor KNN (K Nearest neighbors), a weighted K Nearest neighbor WKNN (weighted K Nearest neighbors) algorithm, the algorithm training time is obtained by a TensorFlow interaction interface, the experiment is repeated for 10 times, an average value is taken for recording, and the construction conditions of the following fingerprint databases are respectively compared: constructing RSS fingerprints at all RP positions in a verification set; constructing RSS fingerprints at 31 RPs in the verification set; RSS fingerprints at 16 RPs in the verification set were constructed, and the experimental data are compared and shown in Table 2:
TABLE 2 comparison of experimental data
In table 2, the data in the same dataset are obtained by Original, IPCPN, CPN and BPNN respectively representing the measured database and the database constructed by improved CPN, traditional CPN and BPNN, the number after the horizontal line represents the number of RP in the constructed database, KNN and WKNN represent the average positioning error of the two methods, K represents the value of K when the error is minimum, and time is the training time of the neural network.
The invention has the characteristics and beneficial effects that:
the invention provides an improved CPN-based WLAN fingerprint positioning database construction algorithm, which is characterized in that a sparse autoencoder is used for improving an RBM part, and then a sparsity limit is used for improving a BPNN part, so that the number of measured points needing actual measurement in an offline stage can be greatly reduced, and better positioning precision is obtained, thereby effectively reducing the workload of the WLAN fingerprint positioning offline stage.
Description of the drawings:
fig. 1 training set and validation set RP space distribution plots.
Fig. 2 test set RP spatial distribution plots.
Figure 3 is a flow chart of steps.
Detailed Description
The invention provides a WLAN fingerprint positioning database construction algorithm giving an improved CPN (central processing Network) by improving a CPN (CPN), and reduces the workload of an offline stage of WLAN fingerprint positioning. The specific technical scheme is as follows:
step 1: selecting a practical measurement reference point in an area needing positioning and measuring data
Substantially uniformly selecting actual measurement reference points in an area to be positioned, measuring position coordinates and RSS fingerprints of the actual measurement reference points, wherein the number of the actual measurement reference points is obviously less than the total number of the reference points, and reducing the workload of an off-line stage
Step 2: dimensionality reduction of reference point fingerprint information using an auto-encoder
(1) Constructing a proper self-encoder, setting the neuron number of an input layer and an output layer of the self-encoder to be the same as the Access Point (AP) number in a region to be positioned, and setting the neuron number of a hidden layer to be 64.
(2) And inputting the reference point fingerprint data into an autoencoder for iterative training.
And step 3: constructing a suitable neural network structure
(1) And modifying the input layer of the CPN, and setting the neuron number as the sum of the two-dimensional space coordinate and the neuron number of the self-encoder hidden layer.
(2) For the Kohonen layer part of CPN, the maximum 2 neurons input in the activation layer are set, and the corresponding activation weights are obtained using the Softmax function.
And 4, step 4: training neural networks using measured data from real-world points
Inputting the actual measurement point fingerprint information into an autoencoder for dimensionality reduction, then inputting the actual measurement point fingerprint information and the reference point space coordinates into the CPN together, and performing iterative training.
And 5: construction of fingerprint location database using trained neural networks
(1) Predicting fingerprint information of remaining reference points
Inputting the two-dimensional position coordinates of the rest reference points into the trained neural network according to a specified form, wherein the corresponding part in the output value is the fingerprint information of the corresponding reference point;
(2) constructing a complete fingerprint location database
And combining the fingerprint information of the rest reference points with the fingerprint information of the actually measured reference points, thereby completing the construction of the WLAN fingerprint positioning database.
The present invention will be described in further detail with reference to the accompanying drawings and specific examples.
The technical scheme of the invention is as follows:
1) and reducing the dimension of the reference point fingerprint information by using an auto-encoder. For the classification problem, it is very important to extract the input feature information. In the conventional CPN, the input layer directly inputs data into the CPN without any processing of the input data. The RSS fingerprints in the fingerprint positioning are composed of RSS values of different RPs, and under the condition that other conditions are not changed, the number of the RPs contained in the RSS fingerprints is larger, namely the dimension of the RSS fingerprints is higher, and the positioning effect is generally better. Obstacles such as walls exist in indoor positioning, the environment is complex, the RSS value of the RP can be greatly attenuated, the RSS value becomes smaller and even is attenuated to 0, and therefore, in the RSS fingerprint, not every dimension of data has an important effect on classification. If the feature information in the fingerprint value is extracted without processing the input RSS fingerprint data, unnecessary computation power loss will be caused, and the accuracy of classification will also be reduced.
In view of the above problems, the present invention employs an auto-encoder to process input data. An auto-encoder (AE) is an unsupervised learning algorithm that makes the input and output the same by a back-propagation algorithm[53]Namely, encoding itself by using its own high-order features to obtain the high-order feature information of the input data, and the structure comprises an encoder (encoder) and a decoder (decoder). As the name implies, the process of passing data from the input layer to the hidden layer can be seen as a process of encoding information, while the process of passing data from the hidden layer to the output layer can be seen as a process of decoding information. In general, the dimension of the AE hidden layer is smaller than that of the input layer, so that the compressed representation of the input data and the feature information after dimension reduction can be learned.
As shown in fig. 1, data of the input layer is compressed to obtain hidden layer data, and the hidden layer data is decompressed to obtain output data. In order to equalize the input and output, the hidden layer needs to learn the compressed representation of the input data, i.e. extract the feature information from the input data. Therefore, dimension reduction can be performed on the RSS fingerprint by using the AE, the RSS fingerprint is used as input data and is input into the AE, and after training, the hidden layer data is high-order characteristic information obtained after dimension reduction of the RSS fingerprint.
2) The Kohonen layer activation strategy is improved using the Softmax function. In the conventional CPN, the Kohonen layer activates only one node in the layer according to input data, so that the output is 1. This makes the Grossberg layer only predictable from the only node activated in the Kohonen layer. Once the Kohonen layer classification deviates, it will cause the Grossberg layer input to deviate. Even if the Kohonen layer is classified correctly, and a plurality of groups of data with similar characteristic information are classified into one group, the output data of the data are completely the same, and the groups of data cannot be distinguished. Even if the coordinates of different test points are divided into the same type of coordinates, the respective RSS fingerprints cannot be completely the same, so that the same RSS fingerprint predicted values obtained by different coordinates generate larger errors.
Aiming at the problems, the algorithm improves the activation strategy of the neurons in the Kohonen layer by using a Softmax function, and 2 neurons with the largest input are activated at each time. The Softmax function maps a set of data into a (0,1) interval, and the resulting values are summed to 1, so that it can be understood that each data corresponds to a classification probability of its size. Suppose there is an array V comprising n elements, ViRepresents the ith element in V, then the Softmax value of this element is
The formula shows that the size of the Softmax value is related to the size of the element, and compared with the competitive rule, the element with a smaller value can be obtained with a certain probability. Unlike activating only the neuron with the largest input, the Softmax function is applied to process the 2 neurons with the largest input, and the corresponding classification probability is obtained. Because different input data can activate different 2 neurons, each with a different effect on the CPN output, different inputs can achieve different outputs.
Step 1: selecting a practical measurement reference point in an area needing positioning and measuring data
Substantially uniformly selecting actual measurement reference points in an area to be positioned, measuring position coordinates and RSS fingerprints of the actual measurement reference points, wherein the number of the actual measurement reference points is obviously less than the total number of the reference points, and reducing the workload of an off-line stage
Step 2: dimensionality reduction of reference point fingerprint information using an auto-encoder
(1) Constructing a proper self-encoder, setting the neuron number of an input layer and an output layer of the self-encoder to be the same as the Access Point (AP) number in a region to be positioned, and setting the neuron number of a hidden layer to be 64.
(2) And inputting the reference point fingerprint data into an autoencoder for iterative training.
And step 3: constructing a suitable neural network structure
(1) And modifying the input layer of the CPN, and setting the neuron number as the sum of the two-dimensional space coordinate and the neuron number of the self-encoder hidden layer.
(2) For the Kohonen layer part of CPN, the maximum 2 neurons input in the activation layer are set, and the corresponding activation weights are obtained using the Softmax function.
And 4, step 4: training neural networks using measured data from real-world points
Inputting the actual measurement point fingerprint information into an autoencoder for dimensionality reduction, then inputting the actual measurement point fingerprint information and the reference point space coordinates into the CPN together, and performing iterative training.
And 5: construction of fingerprint location database using trained neural networks
(1) Predicting fingerprint information of remaining reference points
Inputting the two-dimensional position coordinates of the rest reference points into the trained neural network according to a specified form, wherein the corresponding part in the output value is the fingerprint information of the corresponding reference point;
(2) constructing a complete fingerprint location database
And combining the fingerprint information of the rest reference points with the fingerprint information of the actually measured reference points, thereby completing the construction of the WLAN fingerprint positioning database.
According to the improved CPN-based WLAN fingerprint positioning database construction algorithm, part of data of a WLAN fingerprint positioning data set UJIIndorLoc is selected for experiment, the experiment result is compared with other algorithms, and the effectiveness of the method is verified as follows:
in this experiment, the WLAN fingerprint positioning data set ujiindioorloc obtained by actual measurement is adopted. The data set mainly comprises a tracing data set and an identification data set, wherein the main information of the files is shown in table 1.
TABLE 1 UJIIndenorLoc data set content form Table
The algorithm uses data with BUILDINGID of 0 and FLOOR of 0: 1059 groups of data at 54 RPs are shared in the training data set, data at 10 RPs are taken as a training set and are regarded as measured data at an offline stage, and the rest data are taken as a verification set; the validation dataset had 78 sets of data in common at 78 RPs as the test set. The spatial distribution of the RPs in the training set and validation set is shown in fig. 1.
The white dots in fig. 1 are RPs in the training set. The spatial distribution of the RPs in the test set is shown in fig. 2. As can be seen from fig. 1 and 2, the training set, the verification set, and the test set of the experiment are substantially in the same plane space, and can be used in the experiment.
After the neural network training is finished, inputting RP coordinates in a verification set into the neural network, wherein input data of the improved CPN and the traditional CPN are (0, 0, … …, 0, 0, LONGITUDE, LATITUDE), and the output item 520 is a predicted value of RSS; the BPNN input data is (LONGITUDE, LATITUDE), and the output is an RSS prediction value. And combining the data with the training set data to complete the construction of a fingerprint positioning database, and positioning the test set RP on the constructed database by using KNN and WKNN. All experimental data were repeated 10 times and averaged for recording. This document compares database build cases when verifying RSS fingerprints at all RPs, 31 RPs, 16 RPs in a set. The experimental data are compared in table 2:
after the improved CPN training is completed, the coordinate values of the RP in the verification set are input into the neural network, the predicted values of the corresponding RSS fingerprints are output, and the predicted values are combined with the data of the training set to construct a fingerprint positioning database. And (3) taking the RP of the test set as a to-be-positioned point, and evaluating the construction effect by using KNN (K Nearest Neighbors) and WKNN (weighted K Nearest Neighbors) algorithms on the constructed fingerprint positioning database. The algorithm training time is obtained from the TensorFlow interaction interface. All experimental data were repeated 10 times and the average was recorded. The following fingerprint databases are respectively compared: constructing RSS fingerprints at all RP positions in a verification set; constructing RSS fingerprints at 31 RPs in the verification set; RSS fingerprints at 16 RPs in the authentication set are constructed. The experimental data are compared in table 2.
TABLE 2 comparison of experimental data
The data in table 2 all obtain origin, IPCPN, CPN and BPNN on the same data set to respectively represent the measured database and the database constructed by the improved CPN, the traditional CPN and the BPNN, and the number behind the horizontal line represents the number of RP in the constructed database. KNN and WKNN represent the average positioning error of the two methods, K represents the value of K when the error is minimum, and time is the training time of the neural network. Therefore, the following steps are carried out:
1) compared with the traditional CPN and BPNN, the improved CPN constructed database has the highest positioning precision, the error is only less than 0.1 meter larger than that of the original database, and the difference is only about 1%; the BPNN construction effect is inferior to that of the improved CPN, and the comparison errors with the original database exceed 2.5 meters; the traditional CPN has the worst construction effect, and the contrast error of the traditional CPN and the original database exceeds 4 meters.
2) The improved CPN training time is shorter than conventional BPNN and conventional CPN.
3) The positioning effect of the IPCPN-26 is stronger than that of the CPN-54 and the BPNN-54, so that the IPCPN can achieve better effect than that of the traditional CPN and BPNN by constructing less RPs, and the workload of constructing the database is reduced.
Claims (3)
1. A positioning method based on an improved CPN-WLAN fingerprint positioning database is characterized by comprising the following steps:
step 1: selecting a practical measurement reference point in an area needing positioning and measuring data
Actually measured reference points are approximately and uniformly selected in an area to be positioned, the position coordinates and RSS fingerprints of the actually measured reference points are measured, the number of the actually measured reference points is obviously less than the total number of the reference points, and the workload of an off-line stage is reduced;
step 2: dimensionality reduction of reference point fingerprint information using an auto-encoder
(1) Constructing a proper self-encoder, setting the neuron number of an input layer and an output layer of the self-encoder to be the same as the AP (Access Point) number of an area to be positioned, and setting the neuron number of a hidden layer to be 64;
(2) inputting the reference point fingerprint data into an autoencoder to carry out iterative training;
and step 3: constructing a suitable neural network structure
(1) Modifying an input layer of a CPN (coherent neural network), and setting the neuron number as the sum of a two-dimensional space coordinate and the neuron number of a hidden layer of a self-encoder;
(2) for a Kohonen layer part of the CPN, setting the maximum 2 neurons input in an activation layer, and obtaining corresponding activation weights by using a Softmax function;
and 4, step 4: training neural networks using measured data from real-world points
Inputting the actual measurement point fingerprint information into an autoencoder for dimensionality reduction, then inputting the actual measurement point fingerprint information and the reference point spatial coordinates into a CPN together, and performing iterative training;
and 5: construction of fingerprint location database using trained neural networks
(1) Predicting fingerprint information of remaining reference points
Inputting the two-dimensional position coordinates of the rest reference points into the trained neural network according to a specified form, wherein the corresponding part in the output value is the fingerprint information of the corresponding reference point;
(2) constructing a complete fingerprint location database
And combining the fingerprint information of the rest reference points with the fingerprint information of the actually measured reference points, thereby completing the construction of the WLAN fingerprint positioning database for positioning.
2. The improved CPN-WLAN fingerprint location database-based location method of claim 1, wherein the activation strategy of neurons in Kohonen layer is improved by Softmax function, each time activating 2 neurons with maximum input, Softmax function maps a group of data into (0,1) interval, the obtained numerical values are summed to 1, each data corresponds to the classification probability of its size, and for an array V, comprising n elements, ViRepresenting the ith element in V, then the Softmax value for this element is:
3. the improved CPN-WLAN fingerprint location database-based location method according to claim 1, further comprising the step of verifying: the method comprises the following steps of adopting a WLAN fingerprint positioning data set UJIIndorLoc obtained through actual measurement, wherein the data set mainly comprises a training data set and an identification data set which are 2 files, and the information is shown in a table 1:
TABLE 1 UJIIndenorLoc data set content form Table
Data at BUILDINGID of 0 and FLOOR of 0 is used: 1059 groups of data at 54 RPs are shared in the training data set, data at 10 RPs are taken as a training set and are regarded as measured data at an offline stage, and the rest data are taken as a verification set; the validation dataset had 78 groups of data at 78 RPs in common, as the test set;
the training set, the verification set and the test set are in the same plane space;
after the neural network training is finished, inputting RP coordinates in a verification set into the neural network, wherein input data of the improved CPN and the traditional CPN are (0, 0, … …, 0, 0, LONGITUDE, LATITUDE), and the output item 520 is a predicted value of RSS; the BPNN inputs data (LONGITUDE, LATITUDE), outputs RSS prediction values, is combined with training set data to complete the construction of a fingerprint positioning database, a testing set RP is positioned on the constructed database by using KNN and WKNN, all experimental data are repeated for 10 times, and an average value is taken for recording; comparing database construction conditions when RSS fingerprints at all RPs, 31 RPs and 16 RPs in the verification set are predicted and verified;
after the improved CPN training is completed, coordinate values of RP in a verification set are input into a neural network, a predicted value corresponding to an RSS fingerprint is output, the predicted value is combined with training set data, a fingerprint positioning database is constructed, the RP of a test set is used as a to-be-positioned point, the constructed fingerprint positioning database is subjected to construction effect evaluation by using a Nearest neighbor KNN (K Nearest neighbors), a weighted K Nearest neighbor WKNN (weighted K Nearest neighbors) algorithm, the algorithm training time is obtained by a TensorFlow interaction interface, the experiment is repeated for 10 times, an average value is taken for recording, and the construction conditions of the following fingerprint databases are respectively compared: constructing RSS fingerprints at all RP positions in a verification set; constructing RSS fingerprints at 31 RPs in the verification set; RSS fingerprints at 16 RPs in the verification set were constructed, and the experimental data are compared and shown in Table 2:
TABLE 2 comparison of experimental data
In table 2, the data in the same dataset are obtained by Original, IPCPN, CPN and BPNN respectively representing the measured database and the database constructed by improved CPN, traditional CPN and BPNN, the number after the horizontal line represents the number of RP in the constructed database, KNN and WKNN represent the average positioning error of the two methods, K represents the value of K when the error is minimum, and time is the training time of the neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910810666.6A CN110830939B (en) | 2019-08-29 | 2019-08-29 | Positioning method based on improved CPN-WLAN fingerprint positioning database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910810666.6A CN110830939B (en) | 2019-08-29 | 2019-08-29 | Positioning method based on improved CPN-WLAN fingerprint positioning database |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110830939A CN110830939A (en) | 2020-02-21 |
| CN110830939B true CN110830939B (en) | 2021-04-27 |
Family
ID=69547860
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910810666.6A Expired - Fee Related CN110830939B (en) | 2019-08-29 | 2019-08-29 | Positioning method based on improved CPN-WLAN fingerprint positioning database |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110830939B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111461251A (en) * | 2020-04-10 | 2020-07-28 | 桂林电子科技大学 | Indoor positioning method of WiFi fingerprint based on random forest and self-encoder |
| CN111491367B (en) * | 2020-04-20 | 2021-03-30 | 电子科技大学 | Indoor positioning method based on crowd sensing and multi-fusion technology |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107037399A (en) * | 2017-05-10 | 2017-08-11 | 重庆大学 | A kind of Wi Fi indoor orientation methods based on deep learning |
| CN108540929A (en) * | 2018-03-29 | 2018-09-14 | 马梓翔 | Indoor fingerprint location system based on the sequence of RSSI signal strengths |
| CN109151727A (en) * | 2018-07-28 | 2019-01-04 | 天津大学 | WLAN fingerprint location database construction method based on improved DBN |
| CN109633530A (en) * | 2018-11-30 | 2019-04-16 | 哈尔滨工业大学（深圳） | A kind of localization method and system, equipment, storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10902243B2 (en) * | 2016-10-25 | 2021-01-26 | Deep North, Inc. | Vision based target tracking that distinguishes facial feature targets |
| CN108696932B (en) * | 2018-04-09 | 2020-03-17 | 西安交通大学 | Outdoor fingerprint positioning method using CSI multipath and machine learning |
-
2019
- 2019-08-29 CN CN201910810666.6A patent/CN110830939B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107037399A (en) * | 2017-05-10 | 2017-08-11 | 重庆大学 | A kind of Wi Fi indoor orientation methods based on deep learning |
| CN108540929A (en) * | 2018-03-29 | 2018-09-14 | 马梓翔 | Indoor fingerprint location system based on the sequence of RSSI signal strengths |
| CN109151727A (en) * | 2018-07-28 | 2019-01-04 | 天津大学 | WLAN fingerprint location database construction method based on improved DBN |
| CN109633530A (en) * | 2018-11-30 | 2019-04-16 | 哈尔滨工业大学（深圳） | A kind of localization method and system, equipment, storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于ZigBee位置指纹匹配法的室内定位算法研究;干建勇;《中国优秀硕士学位论文全文数据库信息科技辑》;20181215;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110830939A (en) | 2020-02-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cui et al. | A new subspace clustering strategy for AI-based data analysis in IoT system | |
| CN109151727B (en) | WLAN fingerprint positioning database construction method based on improved DBN | |
| CN110636445B (en) | WIFI-based indoor positioning method, device, equipment and medium | |
| CN108540929B (en) | Indoor fingerprint positioning method based on RSSI signal strength sequencing | |
| CN111898730A (en) | Structure optimization design method for accelerating by using graph convolution neural network structure | |
| CN103927550B (en) | A kind of Handwritten Numeral Recognition Method and system | |
| CN112231592B (en) | Graph-based network community discovery method, device, equipment and storage medium | |
| CN111352965A (en) | Training method of sequence mining model, and processing method and equipment of sequence data | |
| CN116596095B (en) | Training method and device of carbon emission prediction model based on machine learning | |
| CN113095370A (en) | Image recognition method and device, electronic equipment and storage medium | |
| Liu et al. | Large-scale deep learning framework on FPGA for fingerprint-based indoor localization | |
| CN110830939B (en) | Positioning method based on improved CPN-WLAN fingerprint positioning database | |
| CN110224771B (en) | Spectrum sensing method and device based on BP neural network and information geometry | |
| CN115311502A (en) | Remote sensing image small sample scene classification method based on multi-scale double-flow architecture | |
| Yu et al. | Fingerprint extraction and classification of wireless channels based on deep convolutional neural networks | |
| CN114491263A (en) | Recommendation model training method and device, and recommendation method and device | |
| Torres-Sospedra et al. | Scalable and efficient clustering for fingerprint-based positioning | |
| Ayinla et al. | SALLoc: An Accurate Target Localization In Wifi-Enabled Indoor Environments Via Sae-Alstm | |
| Ren et al. | A novel high precision and low consumption indoor positioning algorithm for internet of things | |
| CN114998725B (en) | Hyperspectral image classification method based on self-adaptive spatial spectrum attention kernel generation network | |
| Pu et al. | PaCNN-LSTM: A Localization Scheme Based on Improved Contrastive Learning and Parallel Fusion Neural Network | |
| Yoon et al. | DeepRSSI: Generative Model for Fingerprint-Based Localization | |
| CN113055111B (en) | Channel modeling method and system based on Bayesian optimization | |
| CN112529035B (en) | Intelligent identification method for identifying individual types of different radio stations | |
| Quezada-Gaibor et al. | Lightweight Hybrid CNN-ELM Model for Multi-building and Multi-floor Classification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210427 Termination date: 20210829 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |