CN117062001A - 5G NR indoor positioning method and system based on interpretable deep learning - Google Patents
5G NR indoor positioning method and system based on interpretable deep learning Download PDFInfo
- Publication number
- CN117062001A CN117062001A CN202310957627.5A CN202310957627A CN117062001A CN 117062001 A CN117062001 A CN 117062001A CN 202310957627 A CN202310957627 A CN 202310957627A CN 117062001 A CN117062001 A CN 117062001A
- Authority
- CN
- China
- Prior art keywords
- data
- deep learning
- sequence type
- dimensional
- adopting
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 93
- 238000013135 deep learning Methods 0.000 title claims abstract description 49
- 238000012545 processing Methods 0.000 claims abstract description 50
- 238000013507 mapping Methods 0.000 claims abstract description 38
- 238000006243 chemical reaction Methods 0.000 claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims description 26
- 238000003860 storage Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 7
- 238000007634 remodeling Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 description 7
- 230000004044 response Effects 0.000 description 5
- 101100311460 Schizosaccharomyces pombe (strain 972 / ATCC 24843) sum2 gene Proteins 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 102100037182 Cation-independent mannose-6-phosphate receptor Human genes 0.000 description 3
- 101001028831 Homo sapiens Cation-independent mannose-6-phosphate receptor Proteins 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 208000008918 voyeurism Diseases 0.000 description 1
Classifications
-
- 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
- H04W4/029—Location-based management or tracking services
-
- 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
- H04W4/025—Services making use of location information using location based information parameters
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
An interpretable deep learning-based 5G NR indoor positioning method and system relate to the technical field of computers, in particular to the technical field of deep learning in computers. The method solves the problems of low positioning precision and efficiency, difficult deployment and high cost of the existing indoor fingerprint positioning method. The method comprises the following steps: converting the three-dimensional CSI data into two-dimensional sequence type data by adopting a CSI and sequence type data conversion module; processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data; processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data; and adopting a position mapping module to position the final output data to obtain a mapping position. The method is suitable for the 5G NR indoor positioning method based on the interpretable deep learning in the computer.
Description
Technical Field
The invention relates to the technical field of computers, in particular to the technical field of deep learning in computers.
Background
Indoor fingerprint positioning based on channel state information (Channel State Information, CSI) is widely used in related fields such as indoor navigation and personnel tracking. The 5G utilizes the multiple-input multiple-output (multiple input multiple output, MIMO) technology, the high-frequency band and the large-scale antenna array, and brings new characteristics and challenges for the fingerprint positioning based on the CSI.
At present, the indoor fingerprint positioning methods applied in engineering practice can be mainly divided into the following two types:
(1) Machine learning-based methods such as patent literature published at 12/29/2020: CN112153569a discloses an optimization method based on indoor fingerprint positioning, which adopts low-resolution wireless CSI or received signal strength indication, and has the problem that the higher information richness and feature resolution provided by 5G CSI cannot be effectively utilized. With the advent of high resolution 5GCSI, deep learning-based methods are increasingly replacing machine learning-based methods.
(2) Deep learning-based methods such as patent literature published at 25/06/2021: CN113038595a discloses a rapid fingerprint positioning method based on PQ and CNN, the output obtained by adopting the rapid positioning algorithm based on CNN is the two-dimensional coordinates of the vector to be positioned, although the expansion of the modeling range and the improvement of the modeling effectiveness bring higher positioning precision, the introduction of these tip technologies also bring challenges of parameter cost and calculation cost in actual fingerprint positioning.
Disclosure of Invention
The invention solves the problems of lower positioning precision and efficiency, difficult deployment and higher cost of the existing indoor fingerprint positioning method.
In order to achieve the above object, the present invention provides the following solutions:
the invention provides a 5G NR indoor positioning method based on interpretable deep learning, which comprises the following steps:
s1, converting three-dimensional CSI data into two-dimensional sequence type data by adopting a CSI and sequence type data conversion module;
s2, processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
s3, processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
s4, adopting a position mapping module to position and locate the final output data to obtain a mapping position.
Further, in a preferred embodiment, the processing flow of the CSI and sequence type data conversion module in step S1 is as follows:
s11, performing remolding operation on the three-dimensional CSI data by adopting remolding operation Re to obtain two-dimensional data H in ；
S12, converting the two-dimensional data H in Performing linear mapping to obtain two-dimensional sequence type data H in 。
Further, in a preferred embodiment, the three-dimensional CSI data is expressed as:
H∈R C×W×L ；
the above-described remodeling operation Re is expressed as:
wherein C is the number of input channels, W is the input width, L is the input length, and P w For the width of the image block, P l Is the length of the image block.
Further, in a preferred embodiment, the processing flow of the modeling module of the subcarrier in the single channel in the step S2 is as follows:
s21, defining a parameter matrix W which can be learned;
s22, using LayerNorm to convert the two-dimensional sequence type data H in Performing layer normalization processing to obtain processed two-dimensional sequence type data;
s23, inputting the processed two-dimensional sequence type data into the parameter matrix W with the parameter W 1 In the full connection layer of (2), W is obtained 1 Outputting data by the full connection layer of the (a);
s24, the W is 1 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s25, inputting the regularized processing data into the parameter matrix W, wherein the parameter is W 1 In the full connection layer of (2), W is obtained 2 Outputting data by the full connection layer of the (a);
s26, the W is 2 And the two-dimensional sequence type data H in The final output data X' is obtained by addition.
Further, in a preferred embodiment, the above-mentioned learnable parameter matrix W is expressed as:
W＝[w 1 ,w 2 ,w 3 ,w 4 ] · 。
further, in a preferred embodiment, the processing flow of the subcarrier modeling module in the multi-channel in the step S3 is as follows:
s31, carrying out normalization processing on the process layer of the output data X 'by adopting LayerNorm to obtain processed data X';
s32, inputting the processed data X' into the parameter matrix W, wherein the parameter is W 3 In the full connection layer of (2), W is obtained 3 Outputting data by the full connection layer of the (a);
s33, the W is 3 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s34, inputting the regularized processing data into the parameter matrix W with the parameter W 4 In the full connection layer of (2), W is obtained 4 Outputting data by the full connection layer of the (a);
s35, the W is 2 And the two-dimensional sequence type data H in Addition ofThe final output data X is obtained.
Further, in a preferred embodiment, the location mapping module in step S4 is implemented by using a mapping method based on deep learning.
The 5G NR indoor positioning method based on the interpretable deep learning can be realized by adopting computer software, so that the invention also provides a 5G NR indoor positioning system based on the interpretable deep learning, which comprises the following steps:
the storage device is used for converting the three-dimensional CSI data into two-dimensional sequence type data by adopting the CSI and sequence type data conversion module;
the storage device is used for processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
the storage device is used for processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
and the storage device is used for carrying out position location on the final output data by adopting a position mapping module to obtain a mapping position.
The invention also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs an interpretable deep learning based 5G NR indoor positioning method according to any of the preceding claims.
The invention also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the 5G NR indoor positioning method based on the interpretable deep learning.
The beneficial effects of the invention are as follows:
the invention provides a 5G NR indoor positioning method based on interpretable deep learning, which adopts a CSI and sequence type data conversion module to convert modeling data of a single base station and a plurality of base stations from complex three-dimensional CSI data to two-dimensional sequence type data, so that the CSI data can adapt to the input of an attention mechanism, and is convenient for the subsequent attention calculation process. Meanwhile, a modeling module of subcarriers in a single channel and a modeling module of subcarriers in multiple channels are adopted, the two modules can completely simulate subcarrier data of a single base station and multiple base stations by modeling through different full connection layers, abundant characteristic data can be extracted through the two modules, and the accuracy and the efficiency of the indoor positioning method can be effectively improved. Finally, a position mapping module is used, and the module obtains position deduction by mapping the obtained CSI characteristic representation to a linear layer formed by locators of coordinates. The module can output the obtained data into more visual position data representation, so that subsequent observation and comparison are facilitated.
The method is suitable for the 5G NR indoor positioning method based on the interpretable deep learning in the computer.
Drawings
FIG. 1 is a flow chart of a 5G NR indoor positioning method based on interpretable deep learning according to an embodiment one;
fig. 2 is a process flow diagram of a CSI and sequence type data conversion module according to a second embodiment;
fig. 3 is a flowchart of a process of a modeling module of subcarriers in a single channel according to the fourth embodiment;
fig. 4 is a flowchart of a process of a subcarrier modeling module in a multi-channel according to a sixth embodiment;
fig. 5 (a) is a graph comparing the results of the different methods under SNR20 data set according to embodiment eleven;
fig. 5 (b) is a graph comparing the results of the different methods under SNR50 data set according to embodiment eleven.
Fig. 6 is a diagram showing comparison of the error between the 5G NR indoor positioning method based on the interpretable deep learning according to the eleventh embodiment and the existing positioning method.
Where SNR is the signal-to-noise ratio.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples. The following examples will assist those skilled in the art in further understanding the present invention, but are not intended to limit the invention in any way. It should be noted that variations and modifications can be made by those skilled in the art without departing from the spirit of the invention, which falls within the scope of the invention.
Referring to fig. 1, the present embodiment provides a 5G NR indoor positioning method based on interpretable deep learning, where the method is as follows:
s1, converting three-dimensional CSI data into two-dimensional sequence type data by adopting a CSI and sequence type data conversion module;
s2, processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
s3, processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
s4, adopting a position mapping module to position and locate the final output data to obtain a mapping position.
In practical application, the embodiment comprises a CSI and sequence type data conversion module, a modeling module of subcarriers in a single channel, a modeling module of subcarriers in multiple channels and a position mapping module. The received resource grid rxGrid is first obtained by performing frequency division multiplexing demodulation and time offset correction on the received signal by the user equipment, rxGrid being defined as the received resource grid and the known pilots being defined as refGrid, each antenna may be represented as a k-dimensional column vector, i.e. may be represented as:
H raw ＝(rxGrid) k×1 /(refGrid) k×1 ；
where k represents the number of subcarriers, H raw Representing the channel impulse response.
The subcarriers of a single antenna may be denoted as h i ∈R 1×k Whereas the overall subcarriers of a 5G base station with multiple inputs and multiple outputs may be denoted as H' er n×k Which is provided withWhere n represents the number of antennas. In a typical indoor positioning task, multiple base stations are involved, so the overall CSI data can be expressed as H e R b×n×k Where b represents the number of base stations.
The fingerprint-based indoor positioning overall framework consists of two stages, offline and online, i.e. training and testing, respectively. The goal of the framework is to use the training dataset C train ＝{(csi 1 ,loc 1 ),(csi 2 ,loc 2 ),...,(csi n ,loc n ) The test data C is enabled to be query ＝{(csi n+1 ,loc n+1 ),(csi n+2 ,loc n+2 ),...,(csi m ,loc m ) Optimal positioning performance is achieved on the map, so that a mapping method with good generalization performance is learned. Wherein the training data and the test data are each composed of a set of paired data in which csi i Representing CSI data, loc i Representing the position data.
In the training phase, a deep learning based mapping method is utilized to minimize the loss function, wherein the predicted locations are represented by the output of the mapping method and the actual locations are represented as ground truth.
The embodiment provides a 5GNR indoor positioning method based on interpretable deep learning, which adopts a CSI and sequence type data conversion module to convert modeling data of a single base station and a plurality of base stations from complex three-dimensional CSI data to two-dimensional sequence type data, so that the CSI data can adapt to the input of an attention mechanism, and is convenient for the subsequent attention calculation process. Meanwhile, a modeling module of subcarriers in a single channel and a modeling module of subcarriers in multiple channels are adopted, the two modules can completely simulate subcarrier data of a single base station and multiple base stations by modeling through different full connection layers, abundant characteristic data can be extracted through the two modules, and the accuracy and the efficiency of the indoor positioning method can be effectively improved. Finally, a position mapping module is used, and the module obtains position deduction by mapping the obtained CSI characteristic representation to a linear layer formed by locators of coordinates. The module can output the obtained data into more visual position data representation, so that subsequent observation and comparison are facilitated.
Referring to fig. 2, the second embodiment is an illustration of a processing flow of the CSI and sequence type data conversion module in step S1 in the 5G NR indoor positioning method based on the interpretable deep learning according to the first embodiment, where the processing flow of the CSI and sequence type data conversion module is as follows:
s11, performing remolding operation on the three-dimensional CSI data by adopting remolding operation Re to obtain two-dimensional data H in ；
S12, converting the two-dimensional data H in Performing linear mapping to obtain two-dimensional sequence type data H in 。
In practical application, as shown in fig. 2, the present embodiment inputs three-dimensional data H e R for a given CSI C×W×L Performing remolding operation by using remolding operation Re to obtain two-dimensional data H in ；
Consider, for example, 2 base stations, each with 2 antennas, each with 4 subcarriers, the two CSI matrices are as follows:
after rearrangement, a matrix H can be obtained in The following is shown:
in the vertical direction, the element values are all from the same base station's subcarrier characterization. In the horizontal direction, aggregation from different patches from different base stations is combined.
For the resulting output matrix H in Obtaining final output data H using linear mapping in The method comprises the steps of carrying out a first treatment on the surface of the Linear mapping is used to increase feature richness of the matrix, facilitating subsequent modeling processes.
In the third embodiment, the three-dimensional CSI data and the remodeling operation Re in the 5GNR indoor positioning method based on the interpretable deep learning according to the second embodiment are illustrated, where the three-dimensional CSI data is expressed as:
H∈R C×W×L ；
the remodeling operation Re is expressed as:
wherein C is the number of input channels, W is the input width, L is the input length, and P w For the width of the image block, P l Is the length of the image block.
In practical application, the present embodiment gives one piece of CSI input data H e R C×W×L Where C represents the number of input channels and W and L represent the input width and length, respectively. To obtain the expected input sequence type data H of S non-overlapping sub-slides in A remodeling operation Re is defined.
Referring to fig. 3, the present embodiment is an example of a processing flow of a modeling module of a subcarrier in a single channel in step S2 in the 5G NR indoor positioning method based on the interpretable deep learning according to the first embodiment, where the processing flow of the modeling module of the subcarrier in the single channel is as follows:
s21, defining a parameter matrix W which can be learned;
s22, using LayerNorm to convert the two-dimensional sequence type data H in Performing layer normalization processing to obtain processed two-dimensional sequence type data;
s23, inputting the processed two-dimensional sequence type data into the parameter matrix W with the parameter W 1 In the full connection layer of (2), W is obtained 1 Outputting data by the full connection layer of the (a);
s24, the W is 1 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s25, inputting the regularized processing data into the parameter matrix W, wherein the parameter is W 1 In the full connection layer of (2), W is obtained 2 Outputting data by the full connection layer of the (a);
s26, the W is 2 And the two-dimensional sequence type data H in The final output data X' is obtained by addition.
In practical application, in order to fully simulate subcarriers of a single base station, the present embodiment considers H in Is required to model its columns. The specific modeling process is as follows:
firstly, a learnable parameter matrix W is defined, which is in the form of:
W＝[w 1 ,w 2 ,w 3 ,w 4 ] · ；
from H 0:,: The subcarrier format of (a) is as follows:
x sum1 ＝x 11 +x 13 +x 21 +x 23
x sum2 ＝x 12 +x 14 +x 22 +x 24 ；
from H 1:,: The subcarrier format of (a) is as follows:
y sum1 ＝y 11 +y 13 +y 21 +y 23
y sum2 ＝y 12 +y 13 +y 22 +y 23 ；
matrix multiplication WH in Output H of (2) out The formula is as follows:
in the actual implementation process, the method is realized through a full connection layer with one-dimensional convolution in deep learning, and the specific process is shown as the following formula:
X'＝X :,i +W 2 (Drop(σ(W 1 (LayerNorm(X) :,i ))))
for i＝1,2,...,C；
wherein sigma represents element mode as nonlinearity, using Gaussian Error Linear Unit (GELU) activation function for activation, layerNorm represents layer normalization, W 1 And W is 2 Representing the parameters of the fully connected layers used, respectively, drop represents a regularization operation to prevent overfitting.
In a fifth embodiment, the present embodiment is an example of a learnable parameter matrix W in the 5GNR indoor positioning method based on the interpretable deep learning according to the fourth embodiment, where the learnable parameter matrix W is expressed as:
W＝[w 1 ,w 2 ,w 3 ,w 4 ] · 。
referring to fig. 4, the present embodiment is an illustration of a processing flow of a subcarrier modeling module in a multi-channel in step S3 in the 5G NR indoor positioning method based on the interpretable deep learning according to the fourth embodiment, where the processing flow of the subcarrier modeling module in the multi-channel is as follows:
s31, carrying out normalization processing on the process layer of the output data X 'by adopting LayerNorm to obtain processed data X';
s32, inputting the processed data X' into the parameter matrix W, wherein the parameter is W 3 In the full connection layer of (2), W is obtained 3 Outputting data by the full connection layer of the (a);
s33, the W is 3 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s34, inputting the regularized processing data into the parameter matrix W with the parameter W 4 In the full connection layer of (2), W is obtained 4 Outputting data by the full connection layer of the (a);
s35, the W is 2 And the two-dimensional sequence type data H in Adding to obtain final output data X.
In practical application, the purpose of this module is to model subcarriers between different channels. The specific method comprises the following steps:
first, the learnable matrices K and s (xy) are defined, the specific formulas are as follows:
K＝[k 1 ,k 2 ,k 3 ,k 4 ]
s(xy)＝x sum1 +y sum1 +x sum2 +y sum2 ；
from H out K output H' out The formula of (2) is as follows:
in an actual implementation, this is done by a linear layer, defining the trainable parameter set used in this process as W 3 And W is 4 The specific process is shown in the formula:
X”＝X' j,: +W 4 (Drop(σ(W 3 (LayerNorm(X') j,: ))))
for j＝1,2,...,S；
wherein W is 3 And W is 4 Parameters representing the two linear layers used, respectively, X 'being the output obtained by the modeling module of the subcarriers within a single channel (single base station), each individual element k of the obtained matrix X' i w i s (xy) contains various elements in the whole CSI data, which means that the subcarriers of CSI from different base stations have interaction functions after modeling, and feature matrices with rich features can be obtained as inputs of the location mapping module.
In the seventh embodiment, the location mapping module of step S4 in the 5GNR indoor positioning method based on the interpretable deep learning according to the first embodiment is illustrated, and the location mapping module of step S4 is implemented by using a mapping method based on the deep learning.
In practical application, the embodiment can map the obtained features into three-dimensional coordinates through the position mapping module, so as to perform position mapping. The specific description formula is shown as follows:
(X,Y,Z)＝Mapping(Mean(LayerNorm(X)))。
the eighth embodiment provides a 5G NR indoor positioning system based on interpretable deep learning, the system being:
the storage device is used for converting the three-dimensional CSI data into two-dimensional sequence type data by adopting the CSI and sequence type data conversion module;
the storage device is used for processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
the storage device is used for processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
and the storage device is used for carrying out position location on the final output data by adopting a position mapping module to obtain a mapping position.
The ninth embodiment provides a computer-readable storage medium having a computer program stored thereon, which when executed by a processor, performs a 5G NR indoor positioning method based on interpretable deep learning as described in any one of the first to seventh embodiments.
In a tenth aspect, the present embodiment provides a computer apparatus including a memory and a processor, the memory storing a computer program, the processor executing the 5G NR indoor positioning method according to any one of the first to seventh embodiments, when the processor runs the computer program stored in the memory.
An eleventh embodiment is described with reference to fig. 5, which is a verification description of a 5G NR indoor positioning method based on interpretable deep learning according to any one of the first to seventh embodiments,
three real 5G scene data sets collected in the indoor space of a new experiment building of the Beijing China academy of sciences are adopted as experimental data, and the whole indoor space is 20 m wide, 60 m long and 4 m high, so that the three-dimensional space is a large room suitable for factories, museums and the like. To obtain CSI, five 5G base stations were deployed at 3.5 ghz using integrated sensing and communication, with a bandwidth of 100 mhz and a power of 40 watts. These base stations are mounted on plastic supports at a height of 2.4 meters above the ground and introduce a random float height of 0.1 meters during simulation to prevent coplanarity. The user device acts as a receiver and is placed on a marked car at a height of 1.2 meters from the ground, simulating a person holding a mobile phone at a height of 1.8 meters. The obtained data sets comprise 4816 positioning samples, wherein three data sets correspond to different representations of CSI at SNR signal to noise ratio 10, SNR20 and SNR 50. To divide the dataset into training, validation and test sets, approximate 6 was used according to different signal-to-noise ratios: 2:2, 2888, 964 and 964 samples were obtained, respectively. The size of the single CSI matrix is 5×16×193, meaning that there are 5 base stations, each with 16 antennas, each with 193 subcarriers.
Initial learning rate of 1×10 -4 After 100 training cycles, the learning rate was halved every 25 cycles, training round epoch was 300, and batch size was 16.
The two evaluation criteria used during the experiment were root mean square error (Root Mean Square Error, RMSE), mean absolute error (MeanAbsolute Error, MAE), respectively:
wherein y is i The predicted value is represented by a value of the prediction,representing the true value.
The comparison method of the present embodiment includes: clnet method Complex input lightweight neural network designed for massive MIMO CSI feedback (complex input lightweight neural network designed for massive MIMO CSI feedback), KNN method: deep learning for massive MIMO CSI feedback (deep learning of massive MIMO CSI feedback), MIMOnet method: maMIMO CSI-BasedPositioning using CNNs: peeking inside the Black Box (CNN-based MaMIMO CSI positioning: peeping inside the black box), hiloc method: hybrid Indoor Localization via Enhanced 5G NR CSI (hybrid indoor positioning based on enhanced 5GNR CSI), SVM method: toward 5G NRhigh-Precision Indoor Positioning via Channel Frequency Response: ANew Paradigm and Dataset Generation Method (5 GNR high precision indoor positioning based on channel frequency response: a new paradigm and data set generation method) and MPRI method: toward 5G NR High-Precision Indoor Positioning via Channel Frequency Response: ANew Paradigm andDataset Generation Method (5G NR High precision indoor positioning based on channel frequency response: a new paradigm and data set generation method).
Comparing the above method with the error of the 5G NR indoor positioning method based on the interpretable deep learning according to the present embodiment, as shown in fig. 6, it can be seen from fig. 6 that, compared with the CLnet, KNN, CSInet, MIMOnet, hiloc, SVM, MPRI method, the error of the 5G NR indoor positioning method based on the interpretable deep learning according to the present embodiment is greatly reduced, and the prediction accuracy is effectively improved.
Meanwhile, by comparing the above method with the cumulative distribution function distribution of the 5G NR indoor positioning method based on the interpretable deep learning according to the present embodiment, as shown in fig. 5 (a) and 5 (b), it can be seen that the positioning accuracy of the 5G NR indoor positioning method based on the interpretable deep learning according to the present embodiment is greatly improved compared with that of the CLnet, KNN, CSInet, MIMOnet, hiloc, SVM, MPRI method.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
The above description is only an example of the present invention and is not limited to the present invention, but various modifications and changes will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (10)
1. An interpretable deep learning-based 5G NR indoor positioning method is characterized by comprising the following steps:
s1, converting three-dimensional CSI data into two-dimensional sequence type data by adopting a CSI and sequence type data conversion module;
s2, processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
s3, processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
s4, adopting a position mapping module to position and locate the final output data to obtain a mapping position.
2. The method for 5G NR indoor positioning based on interpretable deep learning according to claim 1, wherein the processing flow of the CSI and sequence type data conversion module in step S1 is as follows:
s11, performing remolding operation on the three-dimensional CSI data by adopting remolding operation Re to obtain two-dimensional data H in ；
S12, converting the two-dimensional data H in Performing linear mapping to obtain two-dimensional sequence type data H in 。
3. The 5G NR indoor positioning method based on interpretable deep learning of claim 2, wherein the three-dimensional CSI data is represented as:
H∈R C×W×L ；
the remodeling operation Re is expressed as:
wherein C is the number of input channels, W is the input width, L is the input length, and P w For the width of the image block, P l Is the length of the image block.
4. The method for 5GNR indoor positioning based on interpretable deep learning of claim 1, wherein the processing flow of the modeling module of the subcarrier in the single channel in step S2 is as follows:
s21, defining a parameter matrix W which can be learned;
s22, using LayerNorm to convert the two-dimensional sequence type data H in Performing layer normalization processing to obtain processed two-dimensional sequence type data;
s23, inputting the processed two-dimensional sequence type data into the parameter matrix W with the parameter W 1 In the full connection layer of (2), W is obtained 1 Outputting data by the full connection layer of the (a);
s24, the W is 1 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s25, inputting the regularized processing data into the parameter matrix W, wherein the parameter is W 1 In the full connection layer of (2), W is obtained 2 Outputting data by the full connection layer of the (a);
s26, the W is 2 And the two-dimensional sequence type data H in The final output data X' is obtained by addition.
5. The 5G NR indoor positioning method based on interpretable deep learning according to claim 4, wherein the learnable parameter matrix W is represented as:
W＝[w 1 ,w 2 ,w 3 ,w 4 ] · 。
6. the method for 5G NR indoor positioning based on interpretable deep learning according to claim 4, wherein the processing flow of the subcarrier modeling module in the multi-channel in step S3 is as follows:
s31, carrying out normalization processing on the process layer of the output data X 'by adopting LayerNorm to obtain processed data X';
s32, inputting the processed data X' into the parameter matrix W, wherein the parameter is W 3 In the full connection layer of (2), W is obtained 3 Outputting data by the full connection layer of the (a);
s33, the W is 3 The output data of the full connection layer of the data processing system is input into a Drop layer for regularization processing, so that regularized processed data are obtained;
s34, inputting the regularized processing data into the parameter matrix W with the parameter W 4 In the full connection layer of (2), W is obtained 4 Outputting data by the full connection layer of the (a);
s35, the W is 2 And the two-dimensional sequence type data H in Adding to obtain final output data X.
7. The method for 5G NR indoor positioning based on interpretable deep learning according to claim 1, wherein the location mapping module of step S4 is implemented using a mapping method based on deep learning.
8. An interpretable deep learning-based 5G NR indoor positioning system, the system being:
the storage device is used for converting the three-dimensional CSI data into two-dimensional sequence type data by adopting the CSI and sequence type data conversion module;
the storage device is used for processing the two-dimensional sequence type data by adopting a modeling module of subcarriers in a single channel to obtain output data;
the storage device is used for processing the output data by adopting a subcarrier modeling module in a multi-channel to obtain final output data;
and the storage device is used for carrying out position location on the final output data by adopting a position mapping module to obtain a mapping position.
9. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, performs a 5G NR indoor positioning method based on interpretable deep learning according to any of claims 1-7.
10. A computer device, characterized by: the apparatus comprising a memory and a processor, said memory having stored therein a computer program, said processor performing an interpretable deep learning based 5G NR indoor positioning method according to any one of claims 1-7 when said processor runs said computer program stored in said memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310957627.5A CN117062001A (en) | 2023-08-01 | 2023-08-01 | 5G NR indoor positioning method and system based on interpretable deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310957627.5A CN117062001A (en) | 2023-08-01 | 2023-08-01 | 5G NR indoor positioning method and system based on interpretable deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117062001A true CN117062001A (en) | 2023-11-14 |
Family
ID=88658163
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310957627.5A Pending CN117062001A (en) | 2023-08-01 | 2023-08-01 | 5G NR indoor positioning method and system based on interpretable deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117062001A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110381440A (en) * | 2019-06-16 | 2019-10-25 | 西安电子科技大学 | The fingerprint indoor orientation method of joint RSS and CSI based on deep learning |
| CN112188613A (en) * | 2020-09-09 | 2021-01-05 | 国网浙江海盐县供电有限公司 | Multi-antenna indoor positioning method and device based on deep learning |
| CN114531729A (en) * | 2022-04-24 | 2022-05-24 | 南昌大学 | Positioning method, system, storage medium and device based on channel state information |
| CN115549742A (en) * | 2022-09-01 | 2022-12-30 | 浙江大学 | CSI compression feedback method based on deep learning |
-
2023
- 2023-08-01 CN CN202310957627.5A patent/CN117062001A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110381440A (en) * | 2019-06-16 | 2019-10-25 | 西安电子科技大学 | The fingerprint indoor orientation method of joint RSS and CSI based on deep learning |
| CN112188613A (en) * | 2020-09-09 | 2021-01-05 | 国网浙江海盐县供电有限公司 | Multi-antenna indoor positioning method and device based on deep learning |
| CN114531729A (en) * | 2022-04-24 | 2022-05-24 | 南昌大学 | Positioning method, system, storage medium and device based on channel state information |
| CN115549742A (en) * | 2022-09-01 | 2022-12-30 | 浙江大学 | CSI compression feedback method based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110308417B (en) | Method and device for estimating direction of arrival under nested array element failure based on matrix filling | |
| CN107015190A (en) | Relatively prime array Wave arrival direction estimating method based on the sparse reconstruction of virtual array covariance matrix | |
| CN107817465A (en) | The DOA estimation method based on mesh free compressed sensing under super-Gaussian noise background | |
| CN103295198B (en) | Based on redundant dictionary and the sparse non-convex compressed sensing image reconstructing method of structure | |
| CN107192878A (en) | A kind of trend of harmonic detection method of power and device based on compressed sensing | |
| CN113238227B (en) | Improved least square phase unwrapping method and system combined with deep learning | |
| CN110346654A (en) | Electromagnetic spectrum map construction method based on common kriging interpolation | |
| CN110907923B (en) | Bistatic EMVS-MIMO radar angle estimation algorithm and device based on parallel factor algorithm | |
| CN111313943A (en) | Three-dimensional positioning method and device under deep learning assisted large-scale antenna array | |
| Chen et al. | A wifi indoor localization method based on dilated cnn and support vector regression | |
| Li et al. | Robust Low-Rank Tensor Completion Based on Tensor Ring Rank via $\ell _ {p,\epsilon} $-Norm | |
| CN103679715A (en) | Method for extracting characteristics of mobile phone image based on non-negative matrix factorization | |
| CN109143151B (en) | Uniform area array tensor reconstruction method and information source positioning method for partial array element damage | |
| CN110954860A (en) | DOA and polarization parameter estimation method | |
| CN107770104B (en) | Channel estimation pilot frequency optimization method and device based on compressed sensing | |
| CN107622476B (en) | Image Super-resolution processing method based on generative probabilistic model | |
| CN117062001A (en) | 5G NR indoor positioning method and system based on interpretable deep learning | |
| CN117062002B (en) | 5G NR indoor positioning method and system based on lightweight TRANSFORMER | |
| CN110401474A (en) | A kind of phased antenna vector modulator control voltage determines method and system | |
| CN116383656A (en) | Semi-supervised characterization contrast learning method for large-scale MIMO positioning | |
| CN105846826A (en) | Approximate smoothed L0 norm-base compressed sensing signal reconstruction method | |
| CN109727295A (en) | Electromagnetic image extracting method, device, computer equipment and storage medium | |
| Lv et al. | A deep learning-based end-to-end algorithm for 5g positioning | |
| CN103607768A (en) | Target device positioning method and related equipment in non-centralized scene | |
| CN111914400B (en) | HRRP (high resolution redundancy protocol) feature extraction method based on multi-task learning |
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 |