CN109922432B - Target localization method by optimizing the number of fingerprint elements in wireless communication environment - Google Patents
Target localization method by optimizing the number of fingerprint elements in wireless communication environment Download PDFInfo
- Publication number
- CN109922432B CN109922432B CN201910245311.7A CN201910245311A CN109922432B CN 109922432 B CN109922432 B CN 109922432B CN 201910245311 A CN201910245311 A CN 201910245311A CN 109922432 B CN109922432 B CN 109922432B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- positioning
- elements
- neural network
- information
- 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 70
- 238000004891 communication Methods 0.000 title claims abstract description 33
- 230000004807 localization Effects 0.000 title 1
- 238000013528 artificial neural network Methods 0.000 claims abstract description 52
- 230000008569 process Effects 0.000 claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000004364 calculation method Methods 0.000 claims abstract description 32
- 239000013598 vector Substances 0.000 claims abstract description 25
- 238000012360 testing method Methods 0.000 claims abstract description 23
- 238000000611 regression analysis Methods 0.000 claims abstract description 4
- 238000011160 research Methods 0.000 claims abstract description 3
- 230000006870 function Effects 0.000 claims description 26
- 210000002569 neuron Anatomy 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- QVFWZNCVPCJQOP-UHFFFAOYSA-N chloralodol Chemical compound CC(O)(C)CC(C)OC(O)C(Cl)(Cl)Cl QVFWZNCVPCJQOP-UHFFFAOYSA-N 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Landscapes
- Position Fixing By Use Of Radio Waves (AREA)
- Mobile Radio Communication Systems (AREA)
- Collating Specific Patterns (AREA)
Abstract
本发明公开了一种无线通信环境下通过优化指纹元素个数的目标定位方法，以短距离无线通信定位系统为基础，以定位精度和计算效率为目标，以神经网络为研究工具，结合训练过程反馈的准确性以及计算周期不断优化特征维数；在训练阶段根据定位目标的历史轨迹，以及在这些轨迹下所采集到的特征信息所组成的指纹信息进行训练，采取神经网络进行回归分析，使得位置信息和特征信息能够实现映射，同时根据测试过程结果的调整对应位置上的特征向量的元素个数，从而实现用最合理的元素个数进行定位；最后在测试阶段，通过已经完成室内各个位置的训练模型，定位目标在室内具体位置将会被检测出。本发明可以在定位过程中针对动态情况选取最优的特征向量的元素个数，从而提高特征采集的效率以及提供更加精确的位置信息。
The invention discloses a target positioning method by optimizing the number of fingerprint elements in a wireless communication environment, based on a short-distance wireless communication positioning system, aiming at positioning accuracy and calculation efficiency, taking neural network as a research tool, combined with a training process The accuracy of the feedback and the calculation cycle continuously optimize the feature dimension; in the training phase, the training is performed according to the historical trajectory of the positioning target and the fingerprint information composed of the feature information collected under these trajectories, and the neural network is used for regression analysis, so that The location information and feature information can be mapped, and the number of elements of the feature vector at the corresponding position is adjusted according to the results of the test process, so as to achieve positioning with the most reasonable number of elements; finally, in the test stage, through the completion of each indoor location The training model of the locating target will be detected at the specific location indoors. The present invention can select the optimal number of elements of the feature vector according to the dynamic situation in the positioning process, thereby improving the efficiency of feature collection and providing more accurate position information.
Description
Technical Field
The invention relates to an indoor positioning system based on deep learning visual angle and short-distance wireless communication, belonging to the technical field of wireless communication positioning.
Background
Fingerprint identification is an important method for solving the problem of indoor positioning. The fingerprint (fingerprint information) refers to a feature set formed by feature information about each broadcasting node collected by a positioning target in a conventional positioning algorithm, such as Received Signal Strength (RSSI), time of arrival (TOA), time difference of arrival (TDOA), and the like. Therefore, in the present invention, a "fingerprint" is often used to represent the collected features, i.e., the feature information from each anchor node is combined into a vector for calculation.
The traditional positioning method mainly uses the acquired characteristic information directly for calculation, because in a two-dimensional plane, three anchor nodes in space can determine the position of a positioning target by solving a ternary-quadratic equation system, and if the height condition of the positioning target is considered, at least four anchor nodes are arranged in the three-dimensional plane. However, each calculation in the conventional positioning method uses a set of feature information from the anchor node at the current time, and due to the multipath effect in the space, the feature information used for calculation is constantly changed at the same position, so that higher positioning accuracy cannot be provided.
The fingerprint-based indoor positioning method uses a feature vector composed of a plurality of groups of features to represent the attribute of a specific position, and uses a neural network (the invention is based on a BP neural network) to mine the correlation between fingerprint information and the corresponding position, and simultaneously, the invention also learns the optimal feature vector dimension through the neural network. In an indoor positioning system, a positioning target continuously acquires characteristic information from each anchor node, and a characteristic vector is formed and put into a neural network to train by taking the position of the positioning target as a target. Because the number of the elements of the selected fingerprint vector can be adjusted, if the number of the selected elements is low, the position information is difficult to reflect, and the positioning precision is low; on the contrary, if the fingerprint length corresponding to each position is too long, the number of positions participating in the calculation in the whole calculation process becomes small, thereby causing the efficiency of the calculation to be reduced. Therefore, an optimal characteristic dimension is established for the system according to the training condition of the neural network, and the positioning precision and the calculation efficiency can be considered at the same time.
With the increase of the indoor positioning service demand, the relevance between the characteristic information and the position is difficult to be excavated by using the traditional positioning method or only using a simple statistical method, so that the method using the neural network solves the imbalance of data in indoor positioning by utilizing the theory of deep learning and overcomes the inevitable trend of various interferences.
Disclosure of Invention
The purpose of the invention is as follows: in order to overcome the defects in the prior art, the invention provides a target positioning method by optimizing the number of fingerprint elements in a wireless communication environment, and the method is based on the theory of a neural network, so that the problem of optimizing the dimension of fingerprint information input into the neural network in a short-distance wireless communication positioning system is solved, the positioning accuracy is considered in the positioning process, and the higher calculation efficiency can be kept.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the technical scheme that:
a target positioning method for optimizing the number of fingerprint elements in a wireless communication environment is based on a short-distance wireless communication positioning system, aims at positioning accuracy and calculation efficiency, takes a neural network as a research tool, and continuously optimizes a characteristic dimension by combining the feedback accuracy of a training process and a calculation period; since the output value of the neural network is a specific coordinate, the positioning system finally feeds back a motion track formed by fitting a plurality of coordinates.
In the invention, the evaluation index of the short-distance wireless positioning system is divided into positioning precision and calculation efficiency. The positioning accuracy refers to the error between the fingerprint information collected by the positioning target and the actual position after calculation in the neural network. The calculation efficiency refers to the number of position information fed back in a limited time range, because the dimension of the fingerprint influences the calculation, the number of elements of a proper fingerprint vector is selected when the neural network is constructed, and because the mobility of an object, the number of coordinates in the positioning process is not preset, the neural network cannot be simply considered as a multi-classification problem in the construction process, but a network structure with a regression function is constructed.
Training according to the historical tracks of the positioning target and fingerprint information consisting of the characteristic information collected under the tracks in the training stage, adopting a neural network to perform regression analysis, enabling the position information and the characteristic information to be mapped, and adjusting the element number of the characteristic vector on the corresponding position according to the test process result so as to realize positioning by using the most reasonable element number; finally, in the testing stage, the specific indoor position of the positioning target is detected through the training model of each indoor position.
In view of the above-mentioned requirement of the Neural Network, since the fingerprint is in a one-dimensional vector form, the present invention uses a commonly used BP Neural Network (BPNN) as a basic architecture of the Neural Network, and uses a Linear rectification function (ReLU) as an activation function to realize the regression output of the Neural Network. The BP neural network comprises an input layer, a hidden layer and an output layer. The input layer receives data, the output layer outputs data, the neuron in the previous layer is connected to the neuron in the next layer, information transmitted by the neuron in the previous layer is collected, the value of the previous layer is transmitted to the next layer through a ReLu activation function, and nonlinear mapping is achieved. The BP neural network has a forward propagation mechanism and a backward propagation mechanism, network parameters are continuously optimized, and a regression model of coordinates calculated by fingerprint vectors is finally learned.
In the short-distance wireless communication indoor positioning environment, the invention realizes the selection of the number of elements of a proper fingerprint vector to perform positioning under a dynamic condition by utilizing the regression model of the neural network, and as shown in figure 5, the method comprises the following steps:
step 2, slicing the fingerprint information according to different numbers of fingerprint elements and corresponding to coordinates;
step 3, repeating the step 1 and the step 2, and acquiring a large amount of data under the same number of fingerprint elements;
step 4, training by using a BP neural network, recording the positioning precision and the calculation efficiency in the test process by using a ReLU activation function and depending on a training model;
step 5, increasing the number of the fingerprint elements, repeating the steps 1-4, and comparing the results of different fingerprint elements obtained in the step 4; repeating the step 5 for multiple times to obtain results under multiple fingerprint element numbers;
step 6, finding the optimal number of fingerprint elements according to the result of the step 5 to obtain the optimal training model;
and 7, testing according to the training model in the step 6, and completing the construction of the positioning system.
Preferably: recording random movement of a positioning target in a positioning range within T time, and simultaneously recording the movement position of each moment within the T time to obtain a group of fingerprint information A from all anchor nodes within the T timeGeneral assemblyExpressed as:
Ageneral assembly＝{a0,a1,a2,…aS}
Wherein S is the total number of the characteristic information;
a is to beGeneral assemblyDividing into a plurality of subsets in accepted order, the subsets being denoted CmWhere m ∈ {1,2, 3., l }, the number of a in each subset is N, and the subset C ism＝{am-0,am-1,...,am-n-1In which the time interval of the subsets is tau, the number of subsets
This is because the elements in the subset will have characteristic information from different anchor nodes, and the characteristic information from all anchor nodes in the current time interval τ should be contained in one subset;
when the subset length N is determined and the feature information is combined into a fingerprint, the input to the BP neural network training process will be determined, and the subset CmCorresponding coordinates
By locating the target at CmThe center point of the movement range within the corresponding time interval τ is determined.
Preferably: the BP neural network inputs and outputs in the training process comprise the following processing procedures:
inputting:
1. fingerprint information of each time interval tau and the identification of the corresponding broadcast node are distinguished by modifying the identification of the anchor node according to the characteristics of the short-distance wireless communication system;
2. in thatAfter the number of elements of the fingerprint vector is given, under the condition of a certain total amount, the number of the elements occupied by each broadcast section should be the same; when the feature dimension of the currently selected fingerprint is N, if M anchor nodes exist, the number of fingerprint elements occupied by each anchor node should be N
And (3) outputting:
1. the coordinates of the training process are derived from the motion trajectory of the object and the corresponding fingerprint data on the same time axis, for example, the feature dimension of the fingerprint that we currently choose is N, and then the corresponding fingerprint can be represented as a ═ b0,a1,a2,…aN-1Where a is a combination of features, i.e., a ═ RSSI, TDOA, toa.]While we can be based on a0And aN-1The corresponding time axis of the fingerprint identification device finds the corresponding coordinate position, and the coordinate of the corresponding point of the fingerprint can be identified by calculating the middle point of the two points (experiments show that the fingerprint acquisition rate is high, and the error between the coordinate selected based on the received sequence interval and the actual coordinate can be ignored).
Preferably: the input vector of a single training is Cm＝{am-0,am-1,...,am-n-1Is, then the corresponding output vector is
According to the structure of the BP neural network, the ReLU function is
Where λ is set to a number close to 0 or directly to 0; let Wij kIs the connection weight of the jth neuron of the k-1 layer and the kth layer, bi kFor the bias of the ith neuron in the kth layer, the following results are obtained:
hi k＝f(neti k)
and h is the input element of each layer, the input of the first layer is CmTherein neti kIs the sum of the weights from the previous layer, i.e.
The calculation process of forward propagation is completed, and W needs to be corrected through backward propagation in the BP neural networkij kAnd bi k(ii) a Determining the loss function is required in performing back propagation
Wherein beta is a weight coefficient and 0 < beta < 1, TcostIs the calculation of the time of day,
representing the output value of the training process in a certain iteration of the neural network.
Meanwhile, the loss function is used as an objective function of the test process, namely the test objective is as follows:
finally, W is processed according to the defined loss function in the following wayij kAnd bi kUpdating:
wherein alpha is the learning rate, the corresponding N of the loss function is recorded in the process, and the optimal element number of the fingerprint is judged according to the result of the test process.
Preferably: the short-distance wireless communication positioning system comprises an anchor node, a positioning node and an upper layer server, wherein the anchor node continuously sends various characteristic information to the positioning node according to a short-distance wireless communication protocol, and the positioning node analyzes the information to analyze the identifier and the characteristic information of the anchor node; the positioning node forwards the received anchor node identification and the corresponding characteristic information to an upper-layer server in a wireless communication mode; constructing the collected information into fingerprint data in an upper-layer server, searching the optimal number of fingerprint elements through the training process of a neural network, and realizing the optimization of positioning precision and calculation efficiency
Compared with the prior art, the invention has the following beneficial effects:
the invention is based on the visual angle of the neural network theory, and by training the fingerprint information in the positioning engineering, the positioning accuracy and the calculation efficiency under different fingerprint dimensions are fed back, so that the optimal fingerprint dimension under the current system is determined.
Drawings
Fig. 1 is a positioning system based on short-range wireless communication.
Fig. 2 is a fingerprint vector structure diagram.
Fig. 3 is a basic structure diagram of a BP neural network.
Fig. 4 shows a neuron operational model based on the ReLU activation function.
Fig. 5 is a flow chart of an indoor positioning optimization algorithm based on the optimal number of fingerprint elements in a short-distance wireless communication system.
Detailed Description
The present invention is further illustrated by the following description in conjunction with the accompanying drawings and the specific embodiments, it is to be understood that these examples are given solely for the purpose of illustration and are not intended as a definition of the limits of the invention, since various equivalent modifications will occur to those skilled in the art upon reading the present invention and fall within the limits of the appended claims.
A target positioning method by optimizing the number of fingerprint elements in a wireless communication environment is characterized in that short-distance wireless communication equipment such as Bluetooth, RFID, wifi and the like are arranged indoors to serve as positioning environments of anchor nodes, so that a positioning target can sense characteristics such as signal strength, arrival time difference, arrival time and the like of peripheral nodes, the characteristics are used as constituent elements of fingerprints, and then the positioning target can be positioned in a self-adaptive mode through training and testing processes of a neural network; in the training stage, training is carried out according to the historical tracks of the positioning target and the characteristic information acquired under the tracks, a neural network is mainly adopted for carrying out regression analysis, so that the position information and the characteristic information can be mapped, and meanwhile, the number of elements of the characteristic vector on the corresponding position is adjusted according to the quality of a test process result, so that the positioning is carried out by using the most reasonable number of elements, and the reliable positioning precision and the higher operation efficiency are considered; finally, in the testing stage, because the building of the training model of each indoor position is completed, the specific indoor position of the positioning target can be detected. The invention can select the optimal element number of the characteristic vector aiming at the dynamic condition in the positioning process, thereby improving the efficiency of characteristic acquisition and providing more accurate position information. The positioning precision refers to the error between the fingerprint information acquired by the positioning target and the actual position after calculation in the neural network; the calculation efficiency refers to the number of position information fed back in a limited time range; the fingerprint information refers to a feature set composed of feature information about each broadcast node collected about a positioning target such as received signal strength, arrival time difference, and the like.
The method is based on the visual angle of deep learning, and the number of fingerprint elements in the indoor positioning environment is researched by constructing a neural network model. The invention can improve the calculation efficiency based on the fingerprint positioning algorithm on the premise of ensuring the positioning accuracy.
As shown in fig. 1, which is a block diagram of a short-range wireless communication indoor positioning system, it can be seen that the whole system includes an anchor node (i.e., a broadcasting node), a positioning node, and an upper server. According to the protocol of short-distance wireless communication, the anchor node can continuously send various feature information to the positioning node, and the positioning node analyzes the information to analyze the identifier and the feature information of the anchor node. The positioning node forwards the received anchor node identification and the corresponding feature information to an upper-layer server through wireless communication modes such as lora and wifi. The algorithm provided by the invention is completed in the upper-layer server, the collected information is constructed into fingerprint data, and the optimal number of fingerprint elements is searched through the training process of the neural network, so that the optimization of the positioning precision and the calculation efficiency is realized.
As shown in fig. 2, which is a set of fingerprint structures with N elements, it can be seen that the fingerprint is divided into M units, because for locating nodes, information of M anchor nodes may be received simultaneously within the time interval τ. Since the information of these anchor nodes is sent to the server by the positioning node in a random order, the server needs to sort the feature information from the M anchor nodes in the time interval τ, which constitutes the form of fig. 2. Here, it should be noted that, since the number of signal strengths of each anchor node in the fingerprint vector is the same, if the total number of fingerprint elements is set, it is necessary to ensure that the number of signal strength information of each anchor node in the time interval τ is at least as large as
Through the analysis, the total data acquisition time is set to be T, namely the target is positioned to move randomly in the positioning range within the T time, and the movement position of each moment within the T time is recorded (the actual movement position can be obtained through other positioning methods such as camera positioning and sensor binding positioning in the process, and a reference object is provided for fingerprint positioning based on short-distance wireless communication). In this way we will get a set of fingerprint information from all anchor nodes within time T. Can be expressed as:
Ageneral assembly＝{a0,a1,a2,…aS}
Wherein S is the total number of the fingerprint information, and since the data can be stored in the server, the receiving time of the characteristics from the anchor node and the position time of the positioning target can be in one-to-one correspondence in the invention. According to the situation described in the summary of the invention section, it is necessary to assign AGeneral assemblyThe division into subsets in accepted order, a subset can be denoted as CmWhere m ∈ {1,2, 3., l }, the number of a in each subset is N, and the subset C ism＝{am-0,am-1,...,am-n-1In which the time interval of the subsets is tau, the number of subsets
This is because the different distances between the anchor nodes and the positioning target will result in different numbers of the broadcast information received by the anchor nodes in a unit time, but at the same time, the requirement for satisfying the requirement that the number of signal strengths of each anchor node in the time interval τ is at least as high as
According to the previous constraint, since the number of pieces of feature information of each anchor node in N is required to be the same, it may cause part of the feature information to be discarded. In summary, when the subset length N is determined and the ranking process in the server is completed, the input to the BP neural network training process is determined, and the subset C is determinedmCorresponding coordinates
Can be positioned at C by positioning the targetmThe center point of the movement range within the corresponding time interval τ is determined.
Fig. 3 and 4 show the structure of the BP neural network and the process of passing neurons through the excitation function in the present invention. The BP neural network uses an error reverse algorithm to train the feedforward neural network, is widely applied in practical application, and can realize multi-target classification or regression prediction. In the invention, the regression of the positioning coordinates is realized through a BP neural network after the number of the elements of the fingerprint is given, and the number N of the elements is adjusted through verifying the quality of a regression model.
The input vector of a single training is Cm＝{am-0,am-1,...,am-n-1Is, then the corresponding output vector is
According to the structure of BP neural network, let the ReLU function be
Where λ may be set to a number close to 0 or directly to 0. Let Wij kIs the connection weight of the jth neuron of the k-1 layer and the kth layer, bi kThe bias of the ith neuron of the k layer is obtained by the following steps:
hi k＝f(neti k)
and h is the input element of each layer, the input of the first layer in the present invention is CmTherein neti kIs the sum of the weights from the previous layer:
the forward propagation calculation process is completed, and W and b need to be corrected through backward propagation in the BP neural network. Determining the loss function is required in performing back propagation
Wherein beta is a weight coefficient and 0 < beta < 1, TcostThe calculation time is mainly from the fact that the server needs to arrange the received fingerprint information according to the identification of the anchor node in the whole positioning process. Meanwhile, the loss function can also be used as an objective function of the test process, namely the test objective is as follows:
finally, W and b may be updated according to the defined loss function in the following way:
where α is the learning rate. According to the above process, the method records the corresponding N of the loss function, and since the initial N is not too large, the positioning accuracy is not too high even though the calculation rate is low, so that N is continuously increased in the method as shown in fig. 5, and the optimal number of elements of the fingerprint needs to be judged according to the result of the test process to avoid overfitting. The fingerprint preparation process in the test process needs to be consistent with the corresponding training process, and it is worth noting that if the training model is over-fitted, the positioning accuracy in the test process is not ideal, and over-fitting can be avoided only by selecting a proper number of fingerprint elements. In the indoor positioning problem, the optimal N may be different due to the change of environment, but the method provided by the present invention is general in the fingerprint-based indoor positioning problem.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (3)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910245311.7A CN109922432B (en) | 2019-03-28 | 2019-03-28 | Target localization method by optimizing the number of fingerprint elements in wireless communication environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910245311.7A CN109922432B (en) | 2019-03-28 | 2019-03-28 | Target localization method by optimizing the number of fingerprint elements in wireless communication environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109922432A CN109922432A (en) | 2019-06-21 |
| CN109922432B true CN109922432B (en) | 2021-07-16 |
Family
ID=66967549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910245311.7A Active CN109922432B (en) | 2019-03-28 | 2019-03-28 | Target localization method by optimizing the number of fingerprint elements in wireless communication environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109922432B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110764084A (en) * | 2019-11-07 | 2020-02-07 | 西北工业大学 | A radar detection method for obscured targets in complex terrain |
| CN111405461B (en) * | 2020-03-16 | 2021-10-08 | 南京邮电大学 | Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number |
| CN115119158B (en) * | 2021-03-19 | 2025-05-13 | 南京航空航天大学 | A method for indoor passive positioning using LoRa signal fingerprint |
| CN116887396A (en) * | 2023-07-07 | 2023-10-13 | 中国电信股份有限公司技术创新中心 | Position prediction model training method, terminal positioning method and device |
| CN117930135B (en) * | 2024-03-21 | 2024-08-16 | 广东海洋大学 | TDOA moving target positioning method based on NRZND model |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106482738A (en) * | 2016-11-09 | 2017-03-08 | 江南大学 | Transfinited the indoor fingerprinting localization algorithm of learning machine based on online increment |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7260369B2 (en) * | 2005-08-03 | 2007-08-21 | Kamilo Feher | Location finder, tracker, communication and remote control system |
| CN101414350A (en) * | 2007-12-14 | 2009-04-22 | 浙江师范大学 | Fingerprint image matching method syncretizing multi-reference node integral justification and direction field |
| CN102932738A (en) * | 2012-10-31 | 2013-02-13 | 北京交通大学 | Improved positioning method of indoor fingerprint based on clustering neural network |
| JP6596516B2 (en) * | 2015-05-22 | 2019-10-23 | 華為技術有限公司 | Method and apparatus for improving the positioning performance of an artificial neural network |
| CN105101408B (en) * | 2015-07-23 | 2018-10-23 | 常熟理工学院 | Indoor orientation method based on distributed AP selection strategy |
| CN107070687A (en) * | 2016-12-26 | 2017-08-18 | 南京理工大学 | WIFI localization methods based on integrated neural network |
| CN108235248B (en) * | 2018-01-04 | 2020-07-31 | 东北师范大学 | Multi-mode intelligent sensing indoor positioning method |
-
2019
- 2019-03-28 CN CN201910245311.7A patent/CN109922432B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106482738A (en) * | 2016-11-09 | 2017-03-08 | 江南大学 | Transfinited the indoor fingerprinting localization algorithm of learning machine based on online increment |
Non-Patent Citations (1)
| Title |
|---|
| 基于电信数据的室外定位技术研究;廖山河;《基于电信数据的室外定位技术研究》;20180430;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109922432A (en) | 2019-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109922432B (en) | Target localization method by optimizing the number of fingerprint elements in wireless communication environment | |
| Song et al. | Cnnloc: Deep-learning based indoor localization with wifi fingerprinting | |
| Song et al. | A novel convolutional neural network based indoor localization framework with WiFi fingerprinting | |
| Ibrahim et al. | CNN based indoor localization using RSS time-series | |
| CN108696932B (en) | Outdoor fingerprint positioning method using CSI multipath and machine learning | |
| CN105223546B (en) | Indoor orientation method based on received signal strength and reference point locations double focusing class | |
| CN109756842B (en) | Wireless indoor positioning method and system based on attention mechanism | |
| CN106851571B (en) | Decision tree-based rapid KNN indoor WiFi positioning method | |
| CN103916820B (en) | Wireless indoor location method based on access point stability | |
| Elesawi et al. | Hierarchical multi-building and multi-floor indoor localization based on recurrent neural networks | |
| Khassanov et al. | Finer-level sequential wifi-based indoor localization | |
| Xu et al. | An efficient indoor localization method based on the long short-term memory recurrent neuron network | |
| Tang et al. | Multi-output Gaussian process-based data augmentation for multi-building and multi-floor indoor localization | |
| CN108627798B (en) | WLAN indoor positioning algorithm based on linear discriminant analysis and gradient lifting tree | |
| CN109511085B (en) | A UWB Fingerprint Location Method Based on MeanShift and Weighted k-Nearest Neighbor Algorithm | |
| CN110636437B (en) | An Assisted Location Method Based on Fingerprint Pairs | |
| CN119172011B (en) | Adaptive calibration method of RFID tag antenna based on complex environment recognition | |
| CN113310490A (en) | Indoor positioning method and system combining GRNN with genetic algorithm | |
| CN108810799A (en) | A kind of more floor indoor orientation methods and system based on linear discriminant analysis | |
| Liu et al. | CTSLoc: An indoor localization method based on CNN by using time-series RSSI | |
| CN111405461B (en) | Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number | |
| Torres-Sospedra et al. | Let’s Talk about k-NN for Indoor Positioning: Myths and Facts in RF-based Fingerprinting | |
| Liu et al. | Novel robust indoor device-free moving-object localization and tracking using machine learning with Kalman filter and smoother | |
| CN101341513A (en) | Automatic 3-D object detection | |
| CN114679683A (en) | An Indoor Intelligent Localization Method Based on Derivative Fingerprint Migration |
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 |