CN109922432A - Pass through the object localization method of optimization fingerprint elements number under wireless communications environment - Google Patents

Pass through the object localization method of optimization fingerprint elements number under wireless communications environment Download PDF

Info

Publication number
CN109922432A
CN109922432A CN201910245311.7A CN201910245311A CN109922432A CN 109922432 A CN109922432 A CN 109922432A CN 201910245311 A CN201910245311 A CN 201910245311A CN 109922432 A CN109922432 A CN 109922432A
Authority
CN
China
Prior art keywords
fingerprint
positioning
neural network
characteristic information
subset
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.)
Granted
Application number
CN201910245311.7A
Other languages
Chinese (zh)
Other versions
CN109922432B (en
Inventor
朱晓荣
徐波
王福展
朱洪波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Post and Telecommunication University
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing Post and Telecommunication University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Post and Telecommunication University filed Critical Nanjing Post and Telecommunication University
Priority to CN201910245311.7A priority Critical patent/CN109922432B/en
Publication of CN109922432A publication Critical patent/CN109922432A/en
Application granted granted Critical
Publication of CN109922432B publication Critical patent/CN109922432B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Position Fixing By Use Of Radio Waves (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Collating Specific Patterns (AREA)

Abstract

The invention discloses pass through the object localization method of optimization fingerprint elements number under a kind of wireless communications environment, based on short-distance wireless communication positioning system, using positioning accuracy and computational efficiency as target, using neural network as research tool, the accuracy and calculating cycle of combined training process feedback continue to optimize intrinsic dimensionality;In the training stage according to the historical track of positioning target, and finger print information composed by the collected characteristic information of institute is trained under these tracks, neural network is taken to carry out regression analysis, so that location information and characteristic information can be realized mapping, simultaneously according to the element number of the feature vector on the adjustment corresponding position of test process result, positioned to realize with most reasonable element number;Finally in test phase, by having completed the training pattern of indoor each position, specific location will be detected positioning target indoors.The present invention can choose the element number of optimal feature vector for current intelligence in position fixing process, to improve the efficiency of collection apparatus and provide more accurate location information.

Description

Pass through the object localization method of optimization fingerprint elements number under wireless communications environment
Technical field
The present invention relates to one kind to be based on deep learning visual angle, builds the indoor locating system based on short-distance wireless communication, Belong to wireless communication field of locating technology.
Background technique
Fingerprint recognition is the important method for solving the problems, such as indoor positioning.So-called fingerprint (finger print information), just referring to will be traditional Location algorithm in received signal strength (RSSI), arrival time (TOA), reaching time-difference (TDOA) etc. position target acquisition The characteristic set that the characteristic information about each broadcast node arrived forms.So in the present invention, commonly using " fingerprint " to indicate These characteristic informations from each anchor node are combined into a vector and calculated by collected feature.
Above-mentioned collected characteristic information is mainly directly used in calculating by traditional localization method, this is because flat in two dimension In face, three anchor nodes in space can determine the position of positioning target by solving the secondary equation group of ternary, such as Fruit then at least arranges four anchor nodes in view of the altitudes of positioning target in three-dimensional planar.But traditional positioning side It is the one group of characteristic information put successively using current time from anchor that method calculates all every time, due to the multipath effect in space, is used In the characteristic information of calculating in same position be all continually changing, so higher positioning accuracy can not be provided.
The feature vector that multiple groups feature forms then is used to indicate a certain specific location by the indoor orientation method based on fingerprint Attribute, the relevance of finger print information and corresponding position is excavated by neural network (present invention be based on BP neural network), together When the present invention also pass through neural network and go to learn optimal feature vector dimension.In an indoor locating system, target is positioned The characteristic information put successively from each anchor can be constantly collected, and in meeting composition characteristic vector investment neural network, with fixed The position of position target is trained as target.Since the element number of the fingerprint vector of selection is adjustable, if choosing It takes the number of element lower, is then difficult to reflect location information, positioning accuracy is lower;, whereas if the corresponding finger in each position Line length is too long, then the position number for participating in calculating in entire calculating process can tail off, and reduces so as to cause the efficiency of calculating. So according to the training of neural network be system formulate a set of optimal intrinsic dimensionality can combine positioning accuracy and Computational efficiency.
With the increase of indoor positioning business demand, using traditional localization method or only using only simple statistics side Method is difficult to excavate the relevance of these characteristic informations and position, so using neural network using the theory of deep learning Method is to solve data nonbalance in indoor positioning, overcomes the inexorable trend of various interference.
Summary of the invention
Goal of the invention: in order to overcome the deficiencies in the prior art, the present invention provides leads under a kind of wireless communications environment The object localization method of optimization fingerprint elements number is crossed, this method theory neural network based is solved in short-distance wireless To the optimization of the dimension of the finger print information of input neural network in communications localization system, realization combines fixed in position fixing process The accuracy of position, and it is able to maintain higher computational efficiency.
Technical solution: to achieve the above object, the technical solution adopted by the present invention are as follows:
By the object localization method of optimization fingerprint elements number under a kind of wireless communications environment, with short-distance wireless communication Based on positioning system, using positioning accuracy and computational efficiency as target, using neural network as research tool, combined training process is anti- The accuracy and calculating cycle of feedback continue to optimize intrinsic dimensionality;Since the output valve of neural network is specific coordinate, so What the positioning system was finally fed back is motion profile made of having multiple coordinate fittings.
In the present invention, the evaluation index of short-distance wireless positioning system is divided into positioning accuracy and computational efficiency.Positioning accuracy It is that specific bit target finger print information collected passes through the error after calculating in neural network with physical location.Computational efficiency Refer to the number in the location information of limited time range internal feedback, since the dimension of fingerprint will affect calculating often, institute To choose the element number of suitable fingerprint vector when constructing neural network, simultaneously because the motility of object, is positioning Number of coordinates in the process is no preset, so it is one that neural network cannot be simply considered that in building process More classification problems, but one should be constructed and have the network structure for returning function.
In the training stage according to the historical track of positioning target, and the collected characteristic information institute, institute under these tracks The finger print information of composition is trained, and is taken neural network to carry out regression analysis, is enabled location information and characteristic information real It now maps, while according to the element number of the feature vector on the adjustment corresponding position of test process result, to realize with most Reasonable element number is positioned;Finally in test phase, by having completed the training pattern of indoor each position, positioning Specific location will be detected target indoors.
For neural network demand set forth above, since fingerprint is one-dimensional vector form, so using normal in the present invention Basic framework of the BP neural network (BPNN, Back Propagation Neural Network) as neural network, makes The recurrence output of neural network is realized as activation primitive with line rectification function (ReLU, Rectified Linear Unit). BP neural network contains input layer, hidden layer, output layer three-decker.Input layer receives data, and output layer output data is preceding One layer of neuron is connected to next layer of neuron, collects the information that upper one layer of neuron transmitting comes, will by ReLu activation primitive Upper one layer of value passes to next layer, realizes nonlinear mapping.BP neural network has the machine of propagated forward and backpropagation System, constantly optimizes network parameter, eventually learns to a regression model by fingerprint vector coordinates computed.
The present invention, using the regression model of neural network, realizes in short-distance wireless communication indoor positioning environment The element number of suitable fingerprint vector is chosen under current intelligence to be positioned, as shown in Figure 5, comprising the following steps:
Step 1, it builds based on short-distance wireless communication indoor positioning environment, acquires fingerprint in the case where positioning target motion conditions Information initializes the element number of fingerprint;
Step 2, finger print information is sliced according to different fingerprint elements numbers, and respective coordinates;
Step 3, step 1 and step 2 are repeated, acquires mass data under same fingerprint elements number;
Step 4, it is trained using BP neural network, using ReLU activation primitive, by training pattern, in test process Record location precision and computational efficiency;
Step 5, increase the element number of fingerprint, repeat step 1-4, and step 4 is obtained into the knot of fingerprint elements not of the same race Fruit is compared;It repeats step 5 more times, obtains the result under a variety of fingerprint elements numbers;
Step 6, optimal fingerprint elements number is found according to the result of step 5, obtains optimal training pattern;
Step 7, it according to the training pattern of step 6, is tested, completes building for positioning system.
It is preferred: to be recorded in T time and position target voltuntary movement in orientation range, while being recorded in T time every The movement position at a moment obtains one group within the T moment from the finger print information A of all anchor nodesAlways, indicate are as follows:
AAlways={ a0,a1,a2,…aS}
Wherein, S is total characteristic information number;
By AAlwaysMultiple subsets are divided into according to receiving sequence, subset is expressed as Cm, wherein m ∈ { 1,2,3 ..., l }, each The number of a is N, subset C in subsetm={ am-0,am-1,...,am-n-1, wherein the time interval of subset is τ, subset numberThis is because the element in subset has the characteristic information from different anchor nodes, it should be comprising working as in a subset Characteristic information from all anchor nodes in preceding time interval τ;
After determining subset length N and characteristic information is combined into fingerprint, the input of BP neural network training process will It can be determined, and subset CmCorresponding coordinateBy positioning target in CmMoving range in corresponding time interval τ Central point is determined.
Preferred: BP neural network includes treatment process below in training process input and output:
Input:
1. the mark of the finger print information of each time interval τ and corresponding broadcast node, according to short-distance wireless communication system The feature of system, by modifying the mark of anchor node to realize differentiation;
2., in the case where total amount is certain, each broadcast saves shared element number after given fingerprint vector element number It should be identical;When the fingerprint characteristic dimension currently chosen is N, if there is M anchor node, then each anchor puts shared fingerprint successively Element number should be
Output:
1. the coordinate of training process movement locus of object and corresponding finger print data on same time shaft, such as I The fingerprint characteristic dimension currently chosen be N, then corresponding fingerprint can be expressed as A={ a0,a1,a2,…aN-1, wherein a It is the combination of each feature, i.e. a=[RSSI, TDOA, TOA...], while we can be according to a0And aN-1The corresponding time Axis finds corresponding coordinate position, and the coordinates of the fingerprint corresponding points can be identified by asking the midpoint of two o'clock, and (experiment shows to refer to Line acquisition rate is very fast, and the error based on coordinate selected by the train interval received and actual coordinate can be ignored).
Preferred: the input vector of single training is Cm={ am-0,am-1,...,am-n-1, then corresponding output vector isAccording to the structure of BP neural network, ReLU function isWherein λ is disposed proximate to 0 Count or be directly disposed as 0;If Wij kFor the connection weight of -1 layer of j-th of neuron and kth layer of kth, bi kIt is i-th of kth layer The biasing of neuron, then:
hi k=f (neti k)
And h is every layer of input element, the input of first layer is Cm, wherein neti kBe from the sum of upper one layer weight i.e.
Just be over the calculating process of forward-propagating above, needs to correct W by backpropagation in BP neural networkij kWith bi k;Backpropagation is being executed it needs to be determined that loss functionIts Middle β is weight coefficient, and 0 < β < 1, TcostIt is to calculate the time,Expression is instructed in certain iteration of neural network Practice the output valve of process.
Objective function of the loss function as test process, i.e. test target simultaneously are as follows:
Finally according to the loss function of definition in the following manner to Wij kAnd bi kIt is updated:
Wherein, α is learning rate, and the correspondence N of loss function is recorded in the above process, by the result of test process come Judge the optimal element number of fingerprint
Preferred: short-distance wireless communication positioning system includes anchor node, positioning node and top service device, according to short The agreement of distance wireless communication, anchor node constantly send various characteristic informations to positioning node, and positioning node is by parsing this A little information analyze the mark and characteristic information of anchor node;The positioning node anchor section that mode will receive by wireless communication Point identification and corresponding characteristic information are transmitted to the server on upper layer;The information structuring that will will be collected into top service device Optimal fingerprint elements number is found by the training process of neural network at finger print data, positioning accuracy is realized and calculates effect The optimization of rate
The present invention compared with prior art, has the advantages that
The present invention is based on the visual angles of neural network theory to be fed back by being trained to the finger print information in positioning engineering Positioning accuracy and computational efficiency under different fingerprint dimensions, so that it is determined that optimal fingerprint dimension under present systems.
Detailed description of the invention
Fig. 1 is the positioning system based on short-distance wireless communication.
Fig. 2 is fingerprint vector structure figure.
Fig. 3 is BP neural network basic block diagram.
Based on the neuron operational model of ReLU activation primitive when Fig. 4.
Fig. 5 is the indoor positioning optimization algorithm process under short distance radio communication system based on best fingerprint elements number.
Specific embodiment
In the following with reference to the drawings and specific embodiments, the present invention is furture elucidated, it should be understood that these examples are merely to illustrate this It invents rather than limits the scope of the invention, after the present invention has been read, those skilled in the art are to of the invention various The modification of equivalent form falls within the application range as defined in the appended claims.
By the object localization method of optimization fingerprint elements number under a kind of wireless communications environment, by arrange indoors with Such as localizing environment of the short-range wireless communication apparatus such as bluetooth, RFID, wifi as anchor node, so that positioning target apperception week The signal strength of mid-side node, reaching time-difference, the features such as arrival time, and using these features as the component of fingerprint, so Afterwards by the training of neural network and test process enable positioning target it is adaptive position;It is needed in the training stage According to the historical track of positioning target, and under these tracks, the collected characteristic information of institute is trained, and mainly takes mind Regression analysis is carried out through network, so that location information and characteristic information can be realized mapping, while according to test process result The element number of feature vector on quality adjustment corresponding position, is positioned to realize with most reasonable element number, simultaneous Care for reliable positioning accuracy and higher operation efficiency;Finally in test phase, due to having completed indoor each position Training pattern is built, and specific location will be detected positioning target indoors.The present invention can be directed in position fixing process Current intelligence chooses the element number of optimal feature vector, to improve the efficiency of collection apparatus and provide more accurate Location information.Positioning accuracy is specific bit target finger print information collected by after calculating in neural network and actual bit The error set;Computational efficiency refers to the number in the location information of limited time range internal feedback;Finger print information refer to by about Received signal strength, arrival time, reaching time-difference etc. position the collected characteristic information about each broadcast node of target The characteristic set of composition.
The present invention is based on the visual angles of deep learning, study fingerprint under indoor positioning environment by building neural network model The number of element.The present invention can improve the computational efficiency based on fingerprinting localization algorithm under the premise of guaranteeing positioning accuracy.
It is short-distance wireless communication indoor locating system structure chart as shown in Figure 1, it can be seen that whole system includes anchor section The server on point (i.e. broadcast node), positioning node and upper layer.According to the agreement of short-distance wireless communication, anchor node can not It is disconnected to send various characteristic informations to positioning node, positioning node by parsing these information, analyze the mark of anchor node with And characteristic information.Positioning node can for example, by lora, the communications such as wifi the anchor node received is identified and Corresponding characteristic information is transmitted to the server on upper layer.Algorithm proposed by the present invention will be completed in top service device, will be collected The information structuring arrived finds optimal fingerprint elements number by the training process of neural network at finger print data, realizes positioning The optimization of precision and computational efficiency.
It is one group of dactylotype with N number of element as shown in Figure 2, it can be seen that fingerprint is divided into M unit, this is Because the information of M anchor node may be will receive simultaneously in time interval τ for positioning node.Due to these anchor sections The information of point is sequentially random when being sent to server by positioning node, so needing in server to time interval τ It is interior that the characteristic information from M anchor node is ranked up, form the form of Fig. 2.Illustratively restrictive condition is needed exist for, by Be in the number of the signal strength of anchor node each in fingerprint vector it is identical, so always a if setting fingerprint elements Number, then needing to guarantee that the signal strength information number of each anchor node in time interval τ will at least reach
By analyzing above, we set data acquisition total time as T, i.e., target is positioned in T time in orientation range Voltuntary movement, while being recorded in the movement position at each moment in T time (this process can be by other localization methods, example If camera positions, the modes such as binding sensor positioning obtain the position of actual motion, for the finger based on short-distance wireless communication Line positioning provides object of reference).In this way, we will obtain one group of finger print information from all anchor nodes within the T moment.It can be with table It is shown as:
AAlways={ a0,a1,a2,…aS}
Wherein S is to generally refer to line information number, since data can be in server repository, so coming from the present invention The receiving time of the feature of anchor node and the position time of positioning target can correspond.According to described by Summary The case where, it needs AAlwaysMultiple subsets are divided into according to receiving sequence, subset can be expressed as Cm, wherein m ∈ 1,2,3 ..., L }, the number of a is N, subset C in each subsetm={ am-0,am-1,...,am-n-1, wherein the time interval of subset is τ, subset NumberThis is because distance of each anchor node from positioning target is different, it will lead to anchor node and broadcasted in the unit time The number that information is received is different, but needs to meet the signal strength for needing to meet each anchor node in time interval τ simultaneously Number will at least reachAccording to constraint before, due to requiring the characteristic information number of each anchor node in N identical, so The characteristic information that will lead to part is rejected.In conclusion when determining subset length N and completing collated in server Journey, then the input of BP neural network training process will be determined, and subset CmCorresponding coordinateIt can pass through Target is positioned in CmThe central point of moving range is determined in corresponding time interval τ.
It is the process that the structure of BP neural network and neuron pass through excitation function in the present invention as shown in Figure 3 and Figure 4. BP neural network is widely used in practical applications using error inverse algorithm training feedforward neural network, BP mind Classification or the regression forecasting of multiple target may be implemented through network.In the present invention mainly after the element number of given fingerprint The recurrence that positioning coordinate is realized by BP neural network adjusts element number N by verifying regression model quality.
The input vector of single training is Cm={ am-0,am-1,...,am-n-1, then corresponding output vector isAccording to the structure of BP neural network, if ReLU function isWherein λ can be set to 0 is counted or is directly disposed as close to 0.If Wij kFor the connection weight of -1 layer of j-th of neuron and kth layer of kth, bi kFor kth layer The biasing of i-th of neuron, then be easy to get:
hi k=f (neti k)
And h is every layer of input element, the input of first layer is exactly C in the present inventionm, wherein neti kIt is from upper one layer The sum of weight is i.e.:
Just be over the calculating process of forward-propagating above, needs to correct W and b by backpropagation in BP neural network. Backpropagation is being executed it needs to be determined that loss functionWherein β It is weight coefficient, and 0 < β < 1, TcostIt is to calculate the time, mostlys come from service in entire position fixing process evaluation time of falling into a trap Device needs arrange the finger print information received according to the mark of anchor node.The loss function can also be used as test simultaneously The objective function of process, i.e. test target are as follows:
Finally W and b can be updated in the following manner according to the loss function of definition:
Wherein α is learning rate.According to the above process, the correspondence N of lower loss function will record in the present invention, due to initial N will not take excessive, so while have lower computation rate, but positioning accuracy will not be too high, so such as Fig. 5 institute N can be continuously increased in the present invention by showing, to avoid over-fitting, need to judge the member that fingerprint is optimal by the result of test process Plain number.Fingerprint preparation process in test process needs consistent with the progress of corresponding training process, it should be noted that if The positioning accuracy that the case where over-fitting so test process occurs in training pattern will not be too ideal, only chooses suitable fingerprint Element number is just avoided that over-fitting.Indoors in orientation problem, due to the variation of environment, optimal N may be different, but It is method provided by the present invention is general in the indoor positioning problem based on fingerprint.
The above is only a preferred embodiment of the present invention, it should be pointed out that: for the ordinary skill people of the art For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered It is considered as protection scope of the present invention.

Claims (7)

1. passing through the object localization method of optimization fingerprint elements number under a kind of wireless communications environment, it is characterised in that: with short distance Based on wireless communication positioning system, using positioning accuracy and computational efficiency as target, using neural network as research tool, in conjunction with The accuracy and calculating cycle of training process feedback continue to optimize intrinsic dimensionality;In the training stage according to the history of positioning target Track, and finger print information composed by the collected characteristic information of institute is trained under these tracks, takes neural network Regression analysis is carried out, so that location information and characteristic information can be realized mapping, while according to the adjustment pair of test process result The element number of the feature vector on position is answered, is positioned to realize with most reasonable element number;Finally in test rank Section, by having completed the training pattern of indoor each position, specific location will be detected positioning target indoors.
2. special according to claim 1 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: the following steps are included:
Step 1, it builds based on short-distance wireless communication indoor positioning environment, fingerprint letter is acquired in the case where positioning target motion conditions Breath, initializes the element number of fingerprint;
Step 2, finger print information is sliced according to different fingerprint elements numbers, and respective coordinates;
Step 3, step 1 and step 2 are repeated, acquires mass data under same fingerprint elements number;
Step 4, it is trained using BP neural network, is recorded by training pattern in test process using ReLU activation primitive Positioning accuracy and computational efficiency;
Step 5, increase the element number of fingerprint, repeat step 1-4, and by step 4 obtain the results of fingerprint elements not of the same race into Row compares;It repeats step 5 more times, obtains the result under a variety of fingerprint elements numbers;
Step 6, optimal fingerprint elements number is found according to the result of step 5, obtains optimal training pattern;
Step 7, it according to the training pattern of step 6, is tested, completes building for positioning system.
3. special according to claim 2 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: being recorded in T time, positions target voltuntary movement in orientation range, while being recorded in each moment in T time Movement position obtains one group within the T moment from the characteristic information A of all anchor nodesAlways, indicate are as follows:
AAlways={ a0,a1,a2,…aS}
Wherein, S is total characteristic information number;
By AAlwaysMultiple subsets are divided into according to receiving sequence, subset is expressed as Cm, wherein m ∈ { 1,2,3 ..., l }, each subset The number of middle a is N, subset Cm={ am-0,am-1,...,am-n-1, wherein the time interval of subset is τ, subset number This is because the element in subset has the characteristic information from different anchor nodes, it should include current time in a subset Characteristic information from all anchor nodes in interval τ;
After determining subset length N and characteristic information is combined into fingerprint, the input of BP neural network training process will be by It determines, and subset CmCorresponding coordinateBy positioning target in CmIn corresponding time interval τ in moving range Heart point is determined.
4. special according to claim 3 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: BP neural network includes treatment process below in training process input and output:
Input:
1. the mark of the finger print information of each time interval τ and corresponding broadcast node, according to short distance radio communication system Feature, by modifying the mark of anchor node to realize differentiation;
2., in the case where total amount is certain, each broadcast saves shared element number should after given fingerprint vector element number It is identical;When the fingerprint characteristic dimension currently chosen is N, if there is M anchor node, then the element occupied that each anchor is put successively Number should be
Output:
1. the coordinate of training process movement locus of object and corresponding finger print data on same time shaft, currently choose Fingerprint characteristic dimension is N, and corresponding fingerprint representation is A={ a0,a1,a2,…aN-1, wherein a is the combination of each feature, i.e. a =[RSSI, TDOA, TOA...], while according to a0And aN-1Corresponding time shaft find corresponding coordinate position, by asking two The midpoint of point identifies the coordinates of the fingerprint corresponding points.
5. special according to claim 4 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: the input vector of single training is Cm={ am-0,am-1,...,am-n-1, then corresponding output vector is According to the structure of BP neural network, ReLU function isWherein λ is disposed proximate to 0 number or straight It connects and is set as 0;If Wij kFor the connection weight of -1 layer of j-th of neuron and kth layer of kth, bi kFor the inclined of i-th of neuron of kth layer It sets, then:
hi k=f (neti k)
And h is every layer of input element, the input of first layer is Cm, wherein neti kBe from the sum of upper one layer weight i.e.
Just be over the calculating process of forward-propagating above, needs to correct W by backpropagation in BP neural networkij kAnd bi k;? Backpropagation is executed it needs to be determined that loss functionWherein β is Weight coefficient, and 0 < β < 1, TcostIt is to calculate the time,Indicate the training process in certain iteration of neural network Output valve;
Objective function of the loss function as test process, i.e. test target simultaneously are as follows:
Finally according to the loss function of definition in the following manner to Wij kAnd bi kIt is updated:
Wherein, α is learning rate, and the correspondence N of loss function is recorded in the above process, is judged by the result of test process The optimal element number of fingerprint.
6. special according to claim 5 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: short-distance wireless communication positioning system includes anchor node, positioning node and top service device, according to short-distance wireless The agreement of communication, anchor node constantly send various characteristic informations to positioning node, and positioning node is divided by parsing these information The mark and characteristic information of anchor node is precipitated;Positioning node by wireless communication mode by the anchor node received identify and Corresponding characteristic information is transmitted to the server on upper layer;In top service device by by the information structuring being collected at fingerprint number According to by the training process of neural network, finding optimal fingerprint elements number, realize the excellent of positioning accuracy and computational efficiency Change.
7. special according to claim 6 by the object localization method of optimization fingerprint elements number under wireless communications environment Sign is: the anchor node is the combination of one or more of bluetooth, RFID or wifi.
CN201910245311.7A 2019-03-28 2019-03-28 Target positioning method by optimizing number of fingerprint elements in wireless communication environment Active CN109922432B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910245311.7A CN109922432B (en) 2019-03-28 2019-03-28 Target positioning method by optimizing 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 positioning method by optimizing number of fingerprint elements in wireless communication environment

Publications (2)

Publication Number Publication Date
CN109922432A true CN109922432A (en) 2019-06-21
CN109922432B 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 positioning method by optimizing number of fingerprint elements in wireless communication environment

Country Status (1)

Country Link
CN (1) CN109922432B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110764084A (en) * 2019-11-07 2020-02-07 西北工业大学 Radar detection method for shielding target under complex terrain
CN111405461A (en) * 2020-03-16 2020-07-10 南京邮电大学 Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number
CN117930135A (en) * 2024-03-21 2024-04-26 广东海洋大学 TDOA moving target positioning method based on NRZND model

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080205535A1 (en) * 2005-08-03 2008-08-28 Kamilo Feher Touch screen, location finder, GSM, EDGE, CDMA cellular and OFDM, Wi-Fi 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
CN105101408A (en) * 2015-07-23 2015-11-25 常熟理工学院 Indoor positioning method based on distributed AP selection strategy
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
CN107070687A (en) * 2016-12-26 2017-08-18 南京理工大学 WIFI localization methods based on integrated neural network
CN107113764A (en) * 2015-05-22 2017-08-29 华为技术有限公司 The method and apparatus for improving artificial neural network positioning performance
CN108235248A (en) * 2018-01-04 2018-06-29 东北师范大学 A kind of multi-modal Intellisense indoor orientation method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080205535A1 (en) * 2005-08-03 2008-08-28 Kamilo Feher Touch screen, location finder, GSM, EDGE, CDMA cellular and OFDM, Wi-Fi 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
CN107113764A (en) * 2015-05-22 2017-08-29 华为技术有限公司 The method and apparatus for improving artificial neural network positioning performance
CN105101408A (en) * 2015-07-23 2015-11-25 常熟理工学院 Indoor positioning method based on distributed AP selection strategy
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
CN107070687A (en) * 2016-12-26 2017-08-18 南京理工大学 WIFI localization methods based on integrated neural network
CN108235248A (en) * 2018-01-04 2018-06-29 东北师范大学 A kind of multi-modal Intellisense indoor orientation method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
廖山河: "基于电信数据的室外定位技术研究", 《基于电信数据的室外定位技术研究 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110764084A (en) * 2019-11-07 2020-02-07 西北工业大学 Radar detection method for shielding target under complex terrain
CN111405461A (en) * 2020-03-16 2020-07-10 南京邮电大学 Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number
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
CN117930135A (en) * 2024-03-21 2024-04-26 广东海洋大学 TDOA moving target positioning method based on NRZND model

Also Published As

Publication number Publication date
CN109922432B (en) 2021-07-16

Similar Documents

Publication Publication Date Title
CN108696932B (en) Outdoor fingerprint positioning method using CSI multipath and machine learning
CN109922432A (en) Pass through the object localization method of optimization fingerprint elements number under wireless communications environment
CN106842123B (en) A kind of indoor orientation method and indoor locating system
CN107071743A (en) WiFi localization methods in a kind of quick KNN rooms based on random forest
CN106851571A (en) WiFi localization methods in a kind of quick KNN rooms based on decision tree
CN107396322A (en) Indoor orientation method based on route matching Yu coding and decoding Recognition with Recurrent Neural Network
CN106970379B (en) Based on Taylor series expansion to the distance-measuring and positioning method of indoor objects
CN106612495B (en) A kind of indoor orientation method and system based on propagation loss study
CN106792540B (en) A kind of improvement DV-Hop localization method based on route matching
CN110321401A (en) Space-time data is associated with deep learning method
CN104703276A (en) Locating system and method in light-weight light weight chamber based on channel state information ranging
CN107516127A (en) Service robot independently obtains people and wears the method and system for taking article ownership semanteme
CN107727095A (en) 3D indoor orientation methods based on spectral clustering and weighting reverse transmittance nerve network
CN103885030A (en) Locating method of mobile node in wireless sensor network
CN109672973A (en) A kind of indoor positioning blending algorithm based on most strong AP method
D'Aloia et al. Iot indoor localization with ai technique
CN103714554A (en) Video tracking method based on spread fusion
CN110188780A (en) Method and device for constructing deep learning model for positioning multi-target feature points
Chen et al. An indoor location system based on neural network and genetic algorithm
CN113747360A (en) Indoor positioning method based on mixed visible light and BLE
CN111405461A (en) Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number
CN109708628A (en) A kind of earth magnetism indoor orientation method based on integrated study and BP neural network
CN109541537A (en) A kind of pervasive indoor orientation method based on ranging
CN109658777A (en) Method and system for the teaching control of limb action class
Guo et al. Indoor positioning optimization based on genetic algorithm and RBF neural network

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