CN117320150B - Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature - Google Patents
Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature Download PDFInfo
- Publication number
- CN117320150B CN117320150B CN202311236598.XA CN202311236598A CN117320150B CN 117320150 B CN117320150 B CN 117320150B CN 202311236598 A CN202311236598 A CN 202311236598A CN 117320150 B CN117320150 B CN 117320150B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- feature
- positioning
- cellular network
- fingerprints
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 230000001413 cellular effect Effects 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000012216 screening Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 42
- 210000004027 cell Anatomy 0.000 claims description 20
- 210000002569 neuron Anatomy 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000010295 mobile communication Methods 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims 3
- 238000004364 calculation method Methods 0.000 abstract description 3
- 238000005259 measurement Methods 0.000 abstract description 3
- 238000004891 communication Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000010267 cellular communication Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 238000007493 shaping process Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- 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
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computer Networks & Wireless Communication (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention belongs to the technical field of radio positioning, and discloses an outdoor fingerprint positioning method, a computer and a storage medium based on multiple characteristics of a mobile cellular network, wherein the method specifically comprises the following steps: performing mobile cellular signal measurement in a region to be measured; extracting signal characteristics; performing grid division on the area to be measured; constructing an offline fingerprint database; building and training a FNN model; and carrying out position prediction on the fingerprints received in real time. The method creatively introduces a brand-new feature screening algorithm in the stage of constructing the fingerprint, takes the mobile cellular network parameters screened by the algorithm as the multi-feature composition of the fingerprint, and greatly improves the distinguishing degree and the robustness of the fingerprint on the premise of not introducing redundant interference; in the stage of real-time position prediction, a FNN and KNN double-layer calculation model is innovatively used, and the real-time positioning accuracy is greatly improved.
Description
Technical Field
The invention belongs to the field of radio positioning, and particularly relates to an outdoor fingerprint positioning method, a computer and a storage medium based on multiple characteristics of a mobile cellular network.
Background
The rapid development of wireless technology has led to the fact that the internet of things (Internet ofThings, ioT) is shaping the future of many mass market applications, with location technology as the core technology to acquire spatial internet of things data being one of the most attractive directions of development. Urban outdoor scenes are key application scenes of IoT nodes, and application scenes such as intelligent traffic, smart cities and the like are not separated from accurate position information. An excellent positioning method needs to have high levels in terms of accuracy, continuity, integrity, usability, etc., which is very challenging in complex urban outdoor environments. Therefore, it is particularly important to develop an urban outdoor positioning method with strong noise immunity, easy deployment and low cost.
Currently, most IoT nodes carry communication modules (e.g., bluetooth, cellular communication) while also requiring positioning using global navigation satellite system (Global Navigation Satellite System, GNSS) positioning sensors, which results in higher deployment costs for the individual IoT nodes. Furthermore, even if IoT nodes are equipped with GNSS receivers, GNSS may not provide satisfactory positioning results in the event of poor satellite signals. On the other hand, for communication modules, thanks to the rapid development of wireless communication technology, there are various communication signals in urban environments, which makes it a viable target to achieve positioning with receivable communication signals. Positioning with communication signals may result in IoT nodes saving all or part of the positioning sensors, further reducing IoT node deployment costs and energy consumption.
The fingerprint positioning method is one of the methods for outdoor positioning using communication signals. The fingerprint positioning method can be mainly classified into a bluetooth-based fingerprint positioning method, a Wi-Fi-based fingerprint positioning method, and a mobile cellular network-based fingerprint positioning method according to the type of the received signal. The transmission distance of the Bluetooth signals is too short, and the distribution range of the Bluetooth beacons is not large enough; wi-Fi signals are extremely easy to interfere in an outdoor environment, and signals are not stable enough; whereas the coverage of a mobile cellular network is very large and the signal quality is high. Therefore, the fingerprint positioning method based on the mobile cellular network has the advantages of both outdoor environment and wide application range.
Most of traditional fingerprint positioning methods rely on shallow layer sub-model algorithms such as K nearest neighbor (K-NearestNeighbor, KNN), support vector machine and the like to find fingerprints which are most matched with real-time fingerprints in a fingerprint library, but the shallow layer models have limited learning ability and low matching precision; in addition, the current fingerprint positioning method based on the mobile cellular network utilizes single or small quantity of characteristics such as received signal strength indication (Received Signal Strength Indication, RSSI), reference signal received power (Reference Signal Receiving Power, RSRP), channel state information (Channel State Information, CSI) and the like to construct fingerprints, and the small quantity of characteristic information enables the pressure of mass fingerprint matching in an outdoor scene to be large, so that the positioning accuracy is directly reduced. However, if multiple pieces of characteristic information are introduced blindly, redundant interference is brought, and the final positioning accuracy is also affected. At present, the positioning precision of different outdoor fingerprint positioning methods is about 15-45 m, and a large lifting space is still reserved. Therefore, selecting appropriate signal characteristics to improve positioning accuracy is one of the problems that need to be solved in the fingerprint positioning method.
Disclosure of Invention
In order to solve the problems and overcome the defects of the prior art, the invention provides a city outdoor fingerprint positioning method based on multiple characteristics of a mobile cellular network. According to the method, a brand new feature screening algorithm is introduced in a fingerprint construction stage, mobile cellular network parameters screened by the algorithm are used as multi-feature components of fingerprints, so that the degree of distinction between fingerprints can be effectively increased, the robustness of the fingerprints is increased, and meanwhile, redundant interference is not introduced; in the real-time position prediction stage, a double-layer calculation model consisting of a feedforward neural network (Feedforward Neural Network, FNN) and KNN is used, so that real-time fingerprints can be more accurately and rapidly matched, and the positioning accuracy is greatly improved.
In order to achieve the above purpose, a method for positioning urban outdoor fingerprints based on multiple characteristics of a cellular network comprises the following steps:
s1: collecting fingerprints in the region to be measured, wherein all collected fingerprints form an integral set S data The method comprises the steps of carrying out a first treatment on the surface of the The original features of all fingerprints form an original feature matrix F; dividing the region to be measured into a plurality of grids with the same shape and area, marking each grid with independent labels L which are arranged in sequence, wherein the maximum value L of the labels L represents the total number of the grids; each fingerprint is assigned to a different grid according to its location parameters and is given a label l representing the grid, and at the same time, the set of fingerprints S data Is divided into L fingerprint data subsets corresponding to grid L one by one
Further, fingerprint collection is performed in the area to be measured, specifically:
collecting all receivable mobile cellular network parameters from a current service cell and two adjacent cells in a region to be measured by utilizing a mobile communication module, and collecting position information of each point; the structure of the fingerprint generated after acquisition is shown as formula (1):
S i =(loc i ,f i ) (1)
wherein S is i Representing an i-th fingerprint acquired; loc i A location parameter representing an ith fingerprint; f (f) i The original feature combination representing the ith fingerprint, i.e. all the acquired cellular network parameters that can be received, can be further represented asAn nth feature representing an ith fingerprint; f (f) i The feature screening algorithm is then further optimized to reject the interference information currently contained.
The step S1: the structure of the fingerprints after belonging to different grids is shown as a formula (2):
S i =(loc i ,f i ,l) (2)
s2: screening out positioning fingerprint features from original features of the fingerprints by using a feature screening algorithm; the method comprises the following steps:
s2.1: original feature matrix F for fingerprint, f= (F 1 ,f 2 ,…,f n ) Wherein f n Represented in the set S data The original data vector of the nth feature in the set is counted to count the number Y, Y= (Y) of different values contained in each feature on the respective original data vector 1 ,y 2 ,…,y n ) Y is a one-dimensional vector; y is n Representing the nth feature in its original data vector f n The number of different values contained thereon;
s2.2: statistics on different subsets of fingerprint dataIn which each feature contains a different number X of values on the respective data vector l ,/>X l Is the same as Y in dimension; />Representing the number of different values that the nth feature contains on its data vector in the first fingerprint data subset;
s2.3: for all X l Summing according to formula (3), and calculating the average value according to the quantity L of the fingerprint data subsets;
wherein x= (X) 1 ,x 2 ,…,x n ),x n A mean value representing the number of different values of the nth feature over all fingerprint data subsets;
s2.4: comparing X with Y according to formula (4),
wherein z= (Z) 1 ,z 2 ,…z i …,z n ) The method comprises the steps of carrying out a first treatment on the surface of the Setting a threshold H and selecting z i Features smaller than H are used as locating fingerprint features. Because of z i The larger the mean value of the different value numbers contained in the ith feature is, the closer the mean value is to the different value numbers contained in the overall feature matrix F, the smaller the contribution degree of the feature to grid classification can be further considered, so that z is required to be selected according to the actual application requirement i The smaller features serve as fingerprint features for outdoor fingerprint positioning.
S3: building and training a FNN model, wherein the number of neurons of an input layer is the same as the number of fingerprint positioning features in the step S2, the number of neurons of an output layer is L, and the number of neurons of a hidden layer is determined by the sizes of the input layer and the output layer; the output layer uses a Softmax function as shown in equation (5),
wherein h is L A net input to the output layer neuron; p is the activity value of the output layer neurons.
Using a cross entropy loss function as shown in equation (6) to quantify the difference between the predicted output and the actual grid to which the fingerprint belongs, using L 2 Regularization avoids an overfitting of the model,
wherein t is an L-dimensional one-hot vector used for representing the belonged grid label of the fingerprint; lambda is the regularization coefficient.
Will be set S data The method comprises the steps of dividing a training set, a testing set and a verification set, and training a FNN model to obtain weight parameters of the FNN model;
s4: position prediction is carried out on fingerprints received in real time:
s4.1, extracting a positioning feature vector of a received fingerprint in real time from a mobile cellular network signal, inputting the positioning feature vector into the FNN model trained in the step S3 to predict grids to which the current fingerprint belongs, and outputting the FNN model as the prediction condition probability of different grids to which the current fingerprint belongs; using a grid corresponding to the maximum conditional probability as a prediction result of the grid to which the current fingerprint belongs; the positioning feature vectors are vector representations of all positioning fingerprint features screened in the step S2;
s4.2, calculating a positioning fingerprint feature vector of the current real-time received fingerprint and the positioning fingerprint feature vectors of all the reference fingerprints in the grid to which the current real-time received fingerprint belongs by using a KNN algorithm, and solving Euclidean distance D;
s4.3, K reference fingerprints with the D average value smaller than a threshold value M are selected, the position average value of the K reference fingerprints is calculated according to the formula (7),
wherein, (x' j ,y' j ) Namely, whenThe predicted location of the fingerprint is received in real time.
The beneficial effects of the invention are as follows:
1) In the fingerprint construction stage, a brand new feature screening algorithm is innovatively introduced, the cellular network signal features of the service cell and the adjacent cells are screened to be used as multi-feature composition of the fingerprints, the degree of distinction between the fingerprints is effectively increased on the premise of not introducing redundant interference, and the robustness of the fingerprints is greatly improved;
2) In the stage of real-time position prediction, a double-layer calculation model consisting of FNN and KNN is innovatively used, so that the speed and accuracy of real-time fingerprint matching are greatly improved;
3) By using the outdoor fingerprint positioning method, the average positioning precision of the 4G cellular network is about 8.75m, and compared with other existing outdoor fingerprint positioning methods, the positioning precision is greatly improved.
Drawings
FIG. 1 is a diagram of an urban outdoor fingerprint positioning method framework based on multiple features of a mobile cellular network;
FIG. 2 is a diagram of an actual scenario of an embodiment of the present invention;
FIG. 3 is a graph showing the effect of grid size on prediction accuracy and training time according to an embodiment of the present invention;
FIG. 4 is a graph comparing feature screening algorithms for each feature in accordance with an embodiment of the present invention;
FIG. 5 is a diagram of a FNN structure in accordance with an embodiment of the present invention;
FIG. 6 is a graph showing the effect of K-value selection according to an embodiment of the present invention.
Detailed Description
The technical scheme of the invention is further described in detail below with reference to the accompanying drawings and examples. The following examples are illustrative of the invention and are not intended to limit the scope of the invention.
Examples:
a city outdoor fingerprint positioning method based on cellular network multi-feature comprises the following steps:
s1: fingerprint acquisition is carried out in the area to be measured, and all acquired fingerprints form a wholeVolume set S data The method comprises the steps of carrying out a first treatment on the surface of the The original features of all fingerprints form an original feature matrix F; dividing the region to be measured into a plurality of grids with the same shape and area, marking each grid with independent labels L which are arranged in sequence, wherein the maximum value L of the labels L represents the total number of the grids; each fingerprint is assigned to a different grid according to its location parameters and is given a label l representing the grid, and at the same time, the set of fingerprints S data Is divided into L fingerprint data subsets corresponding to grid L one by one
Further, fingerprint collection is performed in the area to be measured, specifically:
collecting all receivable mobile cellular network parameters from a current service cell and two adjacent cells in a region to be measured by utilizing a mobile communication module, and collecting position information of each point; the structure of the fingerprint generated after acquisition is shown as formula (1):
S i =(loc i ,f i ) (1)
wherein S is i Representing an i-th fingerprint acquired; loc i A location parameter representing an ith fingerprint; f (f) i The original feature combination representing the ith fingerprint, i.e. all the acquired cellular network parameters that can be received, can be further represented asAn nth feature representing an ith fingerprint; f (f) i The feature screening algorithm is then further optimized to reject the interference information currently contained.
The step S1: the structure of the fingerprints after belonging to different grids is shown as a formula (2):
S i =(loc i ,f i ,l) (2)
s2: screening out positioning fingerprint features from original features of the fingerprints by using a feature screening algorithm; the method comprises the following steps:
s2.1: original feature matrix F for fingerprint, f= (F 1 ,f 2 ,…,f n ) Wherein f n Represented in the set S data The original data vector of the nth feature in the set is counted to count the number Y, Y= (Y) of different values contained in each feature on the respective original data vector 1 ,y 2 ,…,y n ) Y is a one-dimensional vector; y is n Representing the nth feature in its original data vector f n The number of different values contained thereon;
s2.2: statistics on different subsets of fingerprint dataIn which each feature contains a different number X of values on the respective data vector l ,/>X l Is the same as Y in dimension; />Representing the number of different values that the nth feature contains on its data vector in the first fingerprint data subset;
s2.3: for all X l Summing according to formula (3), and calculating the average value according to the quantity L of the fingerprint data subsets;
wherein x= (X) 1 ,x 2 ,…,x n ),x n A mean value representing the number of different values of the nth feature over all fingerprint data subsets;
s2.4: comparing Y with Y according to formula (4),
wherein z= (Z) 1 ,z 2 ,…z i …,z n ) The method comprises the steps of carrying out a first treatment on the surface of the Setting a threshold H and selecting z i Features smaller than H are used as locating fingerprint features. Because of z i The larger the mean value of the different value numbers contained in the ith feature is, the closer the mean value is to the different value numbers contained in the overall feature matrix F, the smaller the contribution degree of the feature to grid classification can be further considered, so that z is required to be selected according to the actual application requirement i The smaller features serve as fingerprint features for outdoor fingerprint positioning.
S3: building and training a FNN model, wherein the number of neurons of an input layer is the same as the number of fingerprint positioning features in the step S2, the number of neurons of an output layer is L, and the number of neurons of a hidden layer is determined by the sizes of the input layer and the output layer; the output layer uses a Softmax function as shown in equation (5),
wherein h is L A net input to the output layer neuron; p is the activity value of the output layer neurons.
Using a cross entropy loss function as shown in equation (6) to quantify the difference between the predicted output and the actual grid to which the fingerprint belongs, using L 2 Regularization avoids an overfitting of the model,
wherein t is an L-dimensional one-hot vector used for representing the belonged grid label of the fingerprint; lambda is the regularization coefficient.
Will be set S data The method comprises the steps of dividing a training set, a testing set and a verification set, and training a FNN model to obtain weight parameters of the FNN model;
s4: position prediction is carried out on fingerprints received in real time:
s4.1, extracting a positioning feature vector of a received fingerprint in real time from a mobile cellular network signal, inputting the positioning feature vector into the FNN model trained in the step S3 to predict grids to which the current fingerprint belongs, and outputting the FNN model as the prediction condition probability of different grids to which the current fingerprint belongs; using a grid corresponding to the maximum conditional probability as a prediction result of the grid to which the current fingerprint belongs; the positioning feature vectors are vector representations of all positioning fingerprint features screened in the step S2;
s4.2, calculating a positioning fingerprint feature vector of the current real-time received fingerprint and the positioning fingerprint feature vectors of all the reference fingerprints in the grid to which the current real-time received fingerprint belongs by using a KNN algorithm, and solving Euclidean distance D;
s4.3, K reference fingerprints with the D average value smaller than a threshold value M are selected, the position average value of the K reference fingerprints is calculated according to the formula (7),
wherein, (x' j ,y' j ) The predicted position of the current real-time received fingerprint is obtained.
FIG. 1 is a frame diagram of the present invention, as shown in FIG. 1, the method of the present invention comprises the steps of:
1) A 4G network is selected as the mobile cellular network to which the present embodiment applies. All acceptable 4G cellular network parameters of a service cell and two adjacent cells connected by the current equipment are collected, and meanwhile, the position information of each collecting point, namely longitude and latitude information, is required to be collected. Fig. 2 is a diagram of an actual scenario of the measurement of a 4G cellular signal according to the embodiment of the present invention, as shown in fig. 2, in order to increase the representativeness of the embodiment, the signal measurement area includes busy backbone streets, less dense non-backbone streets for vehicles and pedestrians, urban building group environments, and urban open areas, and basically covers various scenarios of urban outdoor positioning applications.
2) The area to be measured is divided into a plurality of grids of the same shape and area, and each grid is marked with an independent, sequentially arranged label. The area of the area to be measured in the embodiment of the invention is 2.25km 2 FIG. 3 shows the effect of grid size on the accuracy of the subsequent FNN prediction and training time, as can be seen from FIG. 3, when the grid side length is increased from 10m to 20m, the accuracy of the prediction is greatly improvedLifting; when the grid side length is increased from 20m to 30m, the prediction accuracy difference distance of the grid side length and the grid side length is small, and then the grid side length tends to be stable; but when the grid side is 20m, the required network training time is approximately 2.5 times that when the grid side is 30 m. Therefore, based on the comprehensive consideration of the FNN prediction accuracy and the network training time, the embodiment of the invention selects the side length of 30m as the optimal value of the grid size, and the corresponding total number of grids L is 411.
3) And (3) optimizing the characteristics acquired in the step (1) by using a characteristic screening algorithm. Fig. 4 is a comparison of each feature by a feature screening algorithm, where "S" and "N" represent the features of the serving cell and neighboring cells, respectively, and a larger percentage value on the x-axis represents a smaller contribution of the feature to the grid classification. According to fig. 4, by setting a threshold H, the present embodiment selects nine signal features of a Cell unique identification (E-TURAN Cell ID, ECI), a Physical Cell ID (PCI), a carrier Frequency point number (E-TURAN Absolute Radio Frequency Channel Number, EARFCN), a Frequency Band ID (Frequency Band ID, FBI), a tracking area code (Tracking Area Code, TAC), RSRP, a reference signal reception quality (Reference Signal Receiving Quality, RSRQ), RSSI, and a signal-to-interference plus noise ratio (Signal to Interference plus Noise Ratio, SINR); five signal features, EARFCN, PCI, RSRP, RSRQ and RSSI, of two neighboring cells are selected, 19 4G cellular network signal features in total are used to construct a fingerprint, and the acquired features are composed into an offline fingerprint database.
4) Building and training a neural network:
4-1, constructing a FNN model consisting of an input layer, five hidden layers and an output layer, wherein the FNN structure is shown in figure 4, the number of neurons of the input layer is 19, the number of neurons of the output layer is 411 which is the same as the number of signal features of a used honeycomb network, the number of neurons of the hidden layers is the same as the total number of grids, the number of neurons of the hidden layers is determined by the sizes of the input layer and the output layer, the output layer adopts a Softmax function shown in a formula (5), and a cross entropy loss function shown in a formula (6) is used as a loss function;
4-2 will set S data Is divided into 60%, 20% and 20%,and respectively serving as a training set, a testing set and a verification set, and training the FNN according to the allocation duty ratio to obtain the FNN weight parameters.
5) Position prediction is carried out on fingerprints received in real time:
5-1 extracting 19 required features from the 4G cellular network signal received in real time, and representing the extracted features as feature vectors;
5-2, inputting the feature vector into the trained FNN to obtain a prediction result of the grid to which the current fingerprint belongs;
5-3, calculating the current fingerprint feature vector and 5-2 by using a KNN algorithm to obtain feature vectors of all reference fingerprints in the belonging grid, and solving Euclidean distance D;
5-4 selecting K reference fingerprints with the smallest value in D, fig. 5 shows the influence of the K value selection in the embodiment of the present invention, and as can be seen from fig. 5, by setting the threshold M, the positioning accuracy when determining k=3 is optimal, so in the embodiment of the present invention, k=3 is selected as the determining parameter of the positioning method, and it should be noted that the proper K value selection is affected by the region to be measured and the number of reference fingerprints;
5-5 the position mean of the selected k=3 reference fingerprints is calculated as shown in equation (4), resulting in (x' j ,y' j ) That is, the predicted position of the current received fingerprint in real time, and table 1 shows the comparison result of the positioning accuracy of the embodiment of the present invention, as can be seen from table 1, the positioning accuracy of the embodiment of the present invention is about 8.75m, and the positioning accuracy of other comparison methods is about 15m to 45m, which is far higher than that of the existing other comparison methods.
Table 1 comparison of positioning accuracy results
The foregoing describes the embodiment of the method for positioning urban outdoor fingerprints based on multiple characteristics of a cellular network in detail, and it should be pointed out that, for those skilled in the art, according to the ideas of the embodiments of the present invention, the specific implementation and application range of the method are changed, and in summary, the disclosure should not be construed as limiting the invention.
Claims (6)
1. The urban outdoor fingerprint positioning method based on the multi-feature of the cellular network is characterized in that the cellular network is a 4G cellular network and comprises the following steps:
s1: collecting fingerprints in the region to be measured, wherein all collected fingerprints form an integral setThe method comprises the steps of carrying out a first treatment on the surface of the The original features of all fingerprints constitute an original feature matrix +.>The method comprises the steps of carrying out a first treatment on the surface of the Dividing the region to be measured into a plurality of grids with the same shape and area, marking each grid with independent labels L which are arranged in sequence, wherein the maximum value L of the labels L represents the total number of the grids; each fingerprint is assigned to a different grid according to its location parameters and is given a label l representing the grid, while the fingerprint set +.>Is divided into L and grids->One-to-one fingerprint data subset>;
S2: screening 19 positioning fingerprint features from original features of the fingerprints by using a feature screening algorithm; the method comprises the following steps: the method comprises nine signal characteristics including a cell unique identification mark of a service cell, a physical cell mark, a carrier frequency point number, a frequency band mark, a tracking area code, RSRP, reference signal receiving quality, RSSI and signal-to-interference-plus-noise ratio; and the EARFCN, PCI, RSRP, RSRQ and RSSI of two neighboring cells, five signal characteristics each;
s3: building and training a FNN model, wherein the number of neurons of an input layer is the same as the number of the positioning fingerprint features in the step S2, and the number of neurons of an output layer is L; will be assembledThe method comprises the steps of dividing a training set, a testing set and a verification set, and training a FNN model to obtain weight parameters of the FNN model;
s4: position prediction is carried out on fingerprints received in real time:
s4.1, extracting a positioning feature vector of a received fingerprint in real time from a mobile cellular network signal, inputting the positioning feature vector into the trained FNN model in the step S3 so as to predict grids to which the current fingerprint belongs, and outputting the FNN model as the prediction condition probability of different grids to which the current fingerprint belongs; using a grid corresponding to the maximum conditional probability as a prediction result of the grid to which the current fingerprint belongs; the positioning feature vectors are vector representations of all positioning fingerprint features screened in the step S2;
s4.2, calculating a positioning fingerprint feature vector of the current real-time received fingerprint and the positioning fingerprint feature vectors of all the reference fingerprints in the grid to which the current real-time received fingerprint belongs by using a KNN algorithm, and solving Euclidean distance D;
s4.3, K reference fingerprints with the D average value smaller than a threshold value M are selected, the position average value of the K reference fingerprints is calculated according to the formula (7),
(7)
wherein,the predicted position of the current real-time received fingerprint is obtained.
2. The method for positioning urban outdoor fingerprints based on multiple characteristics of a cellular network according to claim 1, wherein said step S1: fingerprint acquisition is carried out in the area to be measured, specifically:
collecting all receivable mobile cellular network parameters from a current service cell and two adjacent cells in a region to be measured by utilizing a mobile communication module, and collecting position information of each point;
the step S1: the structure of the fingerprint after meshing is shown as formula (2):
(2)
wherein,represents the%>A bar fingerprint; />Represents->The original feature combination of the fingerprint, i.e. all the cellular network parameters collected that can be received,/->Represents->Position parameters of the fingerprint.
3. The urban outdoor fingerprint positioning method based on the cellular network multi-feature according to claim 1, wherein the feature screening algorithm of step S2 specifically comprises:
s2.1: original feature matrix for fingerprint,/>Wherein->Representative is in the collection->Middle->Original data vectors of the individual features; counting the number of different values each feature contains on the respective original data vector>,/>,/>Is a one-dimensional vector; />Indicate->The individual features are in their original data vector +.>The number of different values contained thereon;
s2.2: statistics on different subsets of fingerprint dataIn which each feature contains the number of different values on the respective data vector +.>,/>,/>Dimension and->The same; />Indicate->The individual features are in->The number of different values contained in the respective fingerprint data subsets on their data vectors;
s2.3: for all ofSumming according to (3) and according to fingerprint dataNumber of subsets->Calculating the average value;
(3)
wherein,,/>indicate->A mean of the number of different values of the individual features across all fingerprint data subsets;
s2.4: will beAnd->The comparison is carried out according to the formula (4),
(4)
wherein,the method comprises the steps of carrying out a first treatment on the surface of the Setting threshold H, selecting->Features smaller than H are used as locating fingerprint features.
4. The urban outdoor fingerprint positioning method based on the cellular network multi-feature according to claim 1, wherein the 4G cellular network, the built FNN model consists of one input layer, five hidden layers and one output layer.
5. A computer comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the computer program, implements the urban outdoor fingerprint positioning method based on cellular network multi-feature according to any one of claims 1 to 4.
6. A storage medium having stored thereon a computer program which when executed by a processor implements the cellular network multi-feature based urban outdoor fingerprint positioning method according to any of claims 1 to 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311236598.XA CN117320150B (en) | 2023-09-22 | 2023-09-22 | Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311236598.XA CN117320150B (en) | 2023-09-22 | 2023-09-22 | Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117320150A CN117320150A (en) | 2023-12-29 |
CN117320150B true CN117320150B (en) | 2024-04-05 |
Family
ID=89236548
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311236598.XA Active CN117320150B (en) | 2023-09-22 | 2023-09-22 | Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117320150B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3396400A1 (en) * | 2017-04-27 | 2018-10-31 | Deutsche Telekom AG | A system and method for clustering wi-fi fingerprints for indoor-outdoor detection |
WO2023005814A1 (en) * | 2021-07-28 | 2023-02-02 | 华为技术有限公司 | Construction method for indoor fingerprint map, and related device |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107613559B (en) * | 2017-10-12 | 2019-06-21 | 东南大学 | A kind of DOA fingerprint base localization method based on 5G signal |
CN112989888B (en) * | 2019-12-17 | 2024-06-07 | 华为技术有限公司 | Fingerprint anti-counterfeiting method and electronic equipment |
-
2023
- 2023-09-22 CN CN202311236598.XA patent/CN117320150B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3396400A1 (en) * | 2017-04-27 | 2018-10-31 | Deutsche Telekom AG | A system and method for clustering wi-fi fingerprints for indoor-outdoor detection |
WO2023005814A1 (en) * | 2021-07-28 | 2023-02-02 | 华为技术有限公司 | Construction method for indoor fingerprint map, and related device |
Non-Patent Citations (1)
Title |
---|
基于MR指纹库的两步室内外综合定位方法;余艺;;信息通信;20170415(第04期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN117320150A (en) | 2023-12-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111294921B (en) | RSSI wireless sensor network three-dimensional cooperative positioning method | |
CN108696932A (en) | It is a kind of using CSI multipaths and the outdoor fingerprint positioning method of machine learning | |
CN108540929B (en) | Indoor fingerprint positioning method based on RSSI signal strength sequencing | |
CN109640269B (en) | Fingerprint positioning method based on CSI and time domain fusion algorithm | |
CN109068267B (en) | Indoor positioning method based on LoRa SX1280 | |
CN110909924B (en) | Urban expansion multi-scenario simulation cellular automaton method based on cross entropy optimizer | |
CN107027148B (en) | Radio Map classification positioning method based on UE speed | |
CN101820676B (en) | Sensor node positioning method | |
CN110636515B (en) | Network planning evaluation method of electric power wireless private network | |
CN111867049A (en) | Positioning method, positioning device and storage medium | |
CN113779247B (en) | Network fault diagnosis method and system based on intention driving | |
CN112801411B (en) | Network traffic prediction method based on generation of countermeasure network | |
CN109379780A (en) | Wireless sensor network locating method based on adaptive differential evolution algorithm | |
CN113365323A (en) | Low-energy-consumption multipath transmission method for wireless sensor network | |
Gao et al. | Deep learning based location prediction with multiple features in communication network | |
CN101526604B (en) | Device and method for converting signal intensity of wireless positioning system | |
CN117320150B (en) | Outdoor fingerprint positioning method, computer and storage medium based on mobile cellular network multi-feature | |
Ma et al. | TV white space spectrum analysis based on machine learning | |
CN116437365A (en) | Wireless sensor network link quality evaluation method and system | |
CN115099385A (en) | Spectrum map construction method based on sensor layout optimization and adaptive Kriging model | |
CN114783215A (en) | Unmanned aerial vehicle clustering method and device and electronic equipment | |
CN108462940A (en) | A kind of indoor orientation method based on Wi-Fi hotspot | |
CN113313191A (en) | Distributed aperture interactive intelligent evaluation method based on unsupervised learning | |
CN112188388A (en) | Hybrid indoor positioning method based on machine learning | |
Lemelson et al. | Automatic identification of fingerprint regions for quick and reliable location estimation |
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 |