CN101173980A - Indoor node locating algorithm based on ultra-broadband - Google Patents
Indoor node locating algorithm based on ultra-broadband Download PDFInfo
- Publication number
- CN101173980A CN101173980A CNA2007100361739A CN200710036173A CN101173980A CN 101173980 A CN101173980 A CN 101173980A CN A2007100361739 A CNA2007100361739 A CN A2007100361739A CN 200710036173 A CN200710036173 A CN 200710036173A CN 101173980 A CN101173980 A CN 101173980A
- Authority
- CN
- China
- Prior art keywords
- radio frequency
- indoor
- frequency features
- node
- confidence level
- 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
Abstract
The invention relates to an indoor node positioning algorithm based on ultra-wideband, which uses ultra-wideband radio as the communication physical layer. The algorithm has the steps that reference nodes are arranged indoor, different indoor positions are selected, and the radio frequency characteristics of the received reference nodes at the positions are measured; the characteristics are blended, radio frequency characteristic confidence level is induced to compensate the influence of indoor non-sight distance path, a relational database is established to the modified radio frequency characteristics and corresponding position coordinates to be used as the training data set; a support vector regression model is induced, and the functional relation between the radio frequency characteristics and the node position coordinates is estimated through the training data set; the radio frequency characteristics of the nodes to be positioned are substituted into the support vector regression model, in order to calculate the node position coordinates. Being combined with the advantages of strong ultra-wideband anti-multipath and anti-interference capability, the positioning algorithm has the advantages of high reliability, high precision by utilizing the data fusion and machine learning theories.
Description
Technical field
The present invention relates to ultra broadband (UWB) technology and machine learning techniques, specifically is a kind of indoor node locating algorithm based on ultra broadband.
Background technology
At present, have multiple wireless technology to carry out indoor positioning, comprise indoor GPS, RFID, IR, WLAN, Bluetooth, they utilize fixer network, by the signal parameter that receives, individual or object at a time residing position measurement are finished the location according to specific algorithm.Common indoor locating system comprises at present:
The infrared ray indoor locating system, barrier can not pass in this system, makes that infrared-ray only can line-of-sight propagation.The Active Badge system of Olivetti research laboratory adopts this technology at present.
The ultrasound wave indoor positioning needs a large amount of bottom hardware equipment, therefore has the higher shortcoming of cost.Cricket System and Active Bat adopt this technology at present.
The bluetooth indoor locating system, main deficiency is the price comparison costliness of bluetooth devices and equipment, and for the space environment of complexity, the stability of Bluetooth system is poor slightly.
RFID, this system be based on the signal strength analysis method, adopts the polymerization algorithm that three dimensions is positioned, by label detection to the distance of signal power between representing to identify.Adopt the representative of this technology that companies such as Spoton, Wavetrend and Bewator Cotag are arranged.
Yet because that the indoor radio road has a multipath is serious, there are characteristics such as the probability of visual route is low between the transceiver, therefore said method and system worsen its positioning performance because of the ability of anti-multipath decline and shadow fading (non line of sight transmission), thereby are difficult to adapt to the actual demand of indoor positioning.The super-broadband tech conduct is the emerging a kind of radiotelegraphy that grows up in recent years, because of its distinctive high distance accuracy, anti-multipath jamming performance, is highly suitable for the application of indoor locating system.But the accurate orientation problem of node in the time of still can't solving the non line of sight transmission.
Yet the indoor positioning problem based on ultra-broadband radio has its specific characteristic:
1) feature of ultra-broadband radio itself, i.e. time resolution height;
2) indoor environment and indoor channel feature, promptly orientation range is less relatively, and indoor planimetric map is known; Indoor wireless channels multipath complexity, especially object, body of wall, the personage waits barrier and causes the non line of sight transmission.
Summary of the invention
The technical problem to be solved in the present invention is, defective at the prior art existence, a kind of indoor node locating algorithm based on ultra broadband is proposed, it combines ultra broadband anti-multipath and the strong advantage of interference performance, utilize data fusion, machine Learning Theory makes up location model, has characteristics such as the location cost is low, precision height.
Technical scheme of the present invention is, the step of described indoor node locating algorithm based on ultra broadband is:
1, indoor layout reference mode is chosen indoor diverse location, measures the radio frequency features that this position receives reference mode;
2, with these Feature Fusion, and introduce of the influence of radio frequency features confidence level, with radio frequency features and the corresponding position coordinates opening relationships database of revising, as training dataset with obstructed path in the compensated chamber;
3, introduce the support vector regression model,, promptly estimate the funtcional relationship of radio frequency features and node location coordinate by training dataset estimation model parameter;
4,, calculate the position coordinates of node with the radio frequency features substitution support vector regression model of node to be positioned.
Below the present invention made further specify.
The present invention considers the singularity of indoor positioning, orientation problem is embedded into the machine learning framework, radio signal by diverse location in the extraction chamber is a radio frequency features, the dependence model of setting up between signal characteristic and the position is realized the location, and take all factors into consideration the non line of sight influence, adopt the thought of data fusion further to improve the location algorithm performance.That is the present invention's ultra-broadband radio is a communication physical layer, by extracting the radio frequency features of node, utilizes data fusion, the location that the funtcional relationship of machine Learning Theory structure radio frequency features and node location realizes node.
In step of the present invention:
1, indoor layout reference mode is chosen indoor diverse location, measures the radio frequency features that this position receives reference mode; Saying further, in indoor layout reference mode, is communication physical layer with ultra-broadband radio, on indoor different position (coordinate is known), the various radio frequency features of measurement and positioning node are as time of arrival (toa), received signal intensity etc. obtain the radio frequency features vector;
2, with these Feature Fusion, and introduce of the influence of radio frequency features confidence level, with radio frequency features and the corresponding position coordinates opening relationships database of revising, as training dataset with obstructed path in the compensated chamber; Say further, merge measure various radio frequency features, by the relational model of machine learning structural attitude and node location; Because indoor obstructed path influence, the radio frequency features of measurement is not reliably, introduces for this reason and measures radio frequency features confidence level Cred
i j, it represents the confidence level of j node to the measurement of i reference mode; Confidence level be multiply by radio frequency features, as the training dataset of revising.
3, introduce the support vector regression model,, promptly estimate the funtcional relationship of radio frequency features and node location coordinate by training dataset estimation model parameter;
4,, calculate the position coordinates of node with the radio frequency features substitution support vector regression model of node to be positioned; Say further, after obtaining estimating good model, just set up the funtcional relationship of radio frequency features and position (coordinate), the radio frequency features of node measurement to be positioned is brought in the model, just can obtain corresponding position coordinates.
Among the present invention, described relational model is taked Support Vector Machine, by with known location and merge to such an extent that radio frequency features is a training dataset accordingly, simultaneously because the influence of indoor obstructed path, the radio frequency features of measuring is not reliably, introduces for this reason and measures radio frequency features confidence level Cred
i j, it represents the confidence level of j node to the measurement of i reference mode.By training dataset training relational model, estimation model parameter.
Be provided with the reference mode (AP) of l location aware and the terminal of m location aware, its time delay matrix of dummy terminal to reference mode
τ
IjRepresent the signal propagation delay time of i terminal, energy matrix (RSS) to j reference mode
Rs
IjRepresent signal energy and the coordinate of i terminal to j reference mode
Be known.Then need estimation function to concern pos=f (D), D=A
T+ B
T
Location algorithm is exactly in the funtcional relationship of estimating under the above-mentioned known conditions between radio signal characteristics and the position coordinates, at first signal characteristic is mapped to high-dimensional feature space, promptly
, is a mapping function, and estimation function is linear function f (D)=w D+b in the supposition feature space, and w represents weights, and b represents side-play amount, and then Support Vector Machine can be expressed as optimization problem (position with the x axle is an example, in like manner can estimate the y axial coordinate):
C wherein
rThe maximal value of expression vector, ξ
i *, ξ
iBe slack variable, ε is an error of fitting.
Because indoor obstructed path influence, measuring is not reliably, defines the confidence level Cred that measures for this reason
i j, it represents the confidence level of j node to the measurement of i reference mode.Consider the confidence level of measurement, we obtain following optimization problem after Support Vector Machine is improved.
The location estimation function of x axle:
The antithesis of above-mentioned optimization problem is as follows:
Find the solution dual problem, then the function of being asked is:
Introduce kernel function ker (), the location estimation function of then being asked becomes:
Cardinal principle of the present invention is, ultra broadband is owing to have multi-path resolved ability, antijamming capability, characteristics that penetration capacity is strong, make its be fit to very much location under complex environment, but indoor environment complexity, and the probability of signal line-of-sight propagation is very low, so the present invention adopts the communication physical layer of ultra broadband as location algorithm.However through problem, still still can't solve the problem of non-line-of-sight propagation, so the present invention is embedded into orientation problem in the machine learning framework more than can solving.At first the radio frequency features of diverse location in the extraction chamber then with these Feature Fusion, is considered indoor non-line-of-sight propagation influence simultaneously, and the confidence level of radio frequency features, the influence that comes non-line-of-sight propagation in the compensated chamber are measured in definition.Introduce the support vector regression model, diverse location radio frequency features that will be by the confidence level correction and corresponding position (coordinate) are as training dataset, the estimation model parameter, so just set up the funtcional relationship between radio frequency features and the position, bring the radio frequency features of node to be positioned into model, can calculate its position coordinates.This location algorithm is divided into three phases: the radio frequency features of diverse location in the measuring chamber, radio frequency features is merged, and introduce the measurement confidence level radio frequency features is revised with the influence of compensation non-line-of-sight propagation, introduce the support vector regression model, the estimation model parameter, funtcional relationship, the application that can obtain between radio frequency features and the position estimate that good model carries out the indoor node location.
The present invention combines super-broadband tech and machine learning techniques etc., has characteristics such as low, the high precision of location cost, dirigibility are big.
Embodiment
Location algorithm is divided into three phases: the radio frequency features of diverse location in the measuring chamber, radio frequency features is merged, and introduce the measurement confidence level radio frequency features is revised with the influence of compensation non-line-of-sight propagation, introduce the support vector regression model, the estimation model parameter, funtcional relationship, the application that can obtain between radio frequency features and the position estimate that good model carries out the indoor node location.Its concrete steps comprise:
1) in indoor layout reference mode, chooses indoor diverse location, measure the radio frequency features (as time of arrival, received signal intensity) that this position receives reference mode;
2) with these Feature Fusion, and introduce of the influence of radio frequency features confidence level, with radio frequency features and the corresponding position coordinates opening relationships database of revising, as training dataset with obstructed path in the compensated chamber;
3) introduce the support vector regression model,, promptly estimate the funtcional relationship of radio frequency features and node location coordinate by training dataset estimation model parameter;
4), can calculate the position coordinates (x, y) of node with the radio frequency features substitution support vector regression model of node to be positioned.
Claims (4)
1. the indoor node locating algorithm based on ultra broadband is characterized in that, its step is:
A. indoor layout reference mode is chosen indoor diverse location, measures the radio frequency features that this position receives reference mode;
B. with these Feature Fusion, and introduce of the influence of radio frequency features confidence level, with radio frequency features and the corresponding position coordinates opening relationships database of revising, as training dataset with obstructed path in the compensated chamber;
C. introduce the support vector regression model, estimate the funtcional relationship of radio frequency features and node location coordinate by training dataset;
D. with the radio frequency features substitution support vector regression model of node to be positioned, calculate the position coordinates of node.
2. according to the described indoor node locating algorithm of claim 1, it is characterized in that described radio frequency features is time of arrival (toa), received signal intensity based on ultra broadband.
3. according to the described indoor node locating algorithm of claim 1, it is characterized in that described radio frequency features confidence level is the measurement radio frequency features confidence level Cred of j node to i reference mode based on ultra broadband
i j, this confidence level be multiply by radio frequency features, as the training dataset of revising.
4. according to the described indoor node locating algorithm of claim 1, it is characterized in that the function of described radio frequency features and node location coordinate is based on ultra broadband:
Wherein ker () is a kernel function, α
Ix, α
Ix *, b
xThe Lagrange multiplier and the side-play amount of the location estimation function of x axle, α are found the solution in expression respectively
Iy, α
Iy *, b
yThe Lagrange multiplier and the side-play amount of the location estimation function of y axle found the solution in expression respectively.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2007100361739A CN101173980A (en) | 2007-11-21 | 2007-11-21 | Indoor node locating algorithm based on ultra-broadband |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CNA2007100361739A CN101173980A (en) | 2007-11-21 | 2007-11-21 | Indoor node locating algorithm based on ultra-broadband |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101173980A true CN101173980A (en) | 2008-05-07 |
Family
ID=39422607
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CNA2007100361739A Pending CN101173980A (en) | 2007-11-21 | 2007-11-21 | Indoor node locating algorithm based on ultra-broadband |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101173980A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8102315B2 (en) | 2008-11-27 | 2012-01-24 | Industrial Technology Research Institute | Algorithm of collecting and constructing training location data in a positioning system and the positioning method therefor |
CN101754235B (en) * | 2008-12-11 | 2012-11-28 | 财团法人工业技术研究院 | Method for collecting, constructing and positioning training position data of positioning system |
CN103369584A (en) * | 2012-03-30 | 2013-10-23 | 索尼公司 | Terminal device, terminal control method, program and information processing system |
CN105303139A (en) * | 2015-11-04 | 2016-02-03 | 英业达科技有限公司 | Indoor object positioning system and method thereof |
CN105960013A (en) * | 2016-05-06 | 2016-09-21 | 华东交通大学 | AOA-based cooperative localization method under non line-of-sight environment |
CN111246386A (en) * | 2018-11-13 | 2020-06-05 | 中国移动通信集团河南有限公司 | Terminal positioning method and device |
-
2007
- 2007-11-21 CN CNA2007100361739A patent/CN101173980A/en active Pending
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8102315B2 (en) | 2008-11-27 | 2012-01-24 | Industrial Technology Research Institute | Algorithm of collecting and constructing training location data in a positioning system and the positioning method therefor |
CN101754235B (en) * | 2008-12-11 | 2012-11-28 | 财团法人工业技术研究院 | Method for collecting, constructing and positioning training position data of positioning system |
CN103369584A (en) * | 2012-03-30 | 2013-10-23 | 索尼公司 | Terminal device, terminal control method, program and information processing system |
CN105303139A (en) * | 2015-11-04 | 2016-02-03 | 英业达科技有限公司 | Indoor object positioning system and method thereof |
CN105303139B (en) * | 2015-11-04 | 2018-06-26 | 英业达科技有限公司 | Interior articles alignment system and its method |
CN105960013A (en) * | 2016-05-06 | 2016-09-21 | 华东交通大学 | AOA-based cooperative localization method under non line-of-sight environment |
CN111246386A (en) * | 2018-11-13 | 2020-06-05 | 中国移动通信集团河南有限公司 | Terminal positioning method and device |
CN111246386B (en) * | 2018-11-13 | 2021-06-08 | 中国移动通信集团河南有限公司 | Terminal positioning method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107817469B (en) | Indoor positioning method based on ultra-wideband ranging in non-line-of-sight environment | |
CN109548141B (en) | Indoor environment base station coordinate position calibration method based on Kalman filtering algorithm | |
CN103402258B (en) | Wi-Fi (Wireless Fidelity)-based indoor positioning system and method | |
CN101466145B (en) | Dual-base-station accurate orientation method based on neural network | |
CN103501538B (en) | Based on the indoor orientation method of multipath energy fingerprint | |
US9407317B2 (en) | Differential ultra-wideband indoor positioning method | |
CN102981144B (en) | Method for three-dimensional passive positioning of targets by air moving platform | |
CN103746757B (en) | A kind of single star interference source localization method based on satellite multi-beam antenna | |
US20140106776A1 (en) | Method and system for estimation of mobile station velocity in a cellular system based on geographical data | |
CN103363988A (en) | Method for realizing geomagnetic indoor positioning and navigation by utilization of smartphone sensors | |
CN103929807A (en) | Method for precisely positioning device coordinate based on low power consumption | |
CN105954712A (en) | Multi-target direct positioning method in communication with adio signal complex envelope and carrier phase information | |
CN101173980A (en) | Indoor node locating algorithm based on ultra-broadband | |
CN103813448A (en) | Indoor positioning method based on RSSI | |
CN105911521A (en) | Over-the-horizon target direct locating method through combining radio signal complex envelop and carrier phase information | |
CN102427602B (en) | Sparse-based direct position determination method | |
CN103338516A (en) | Two-step positioning method of wireless sensor network based on total least squares | |
CN109975749A (en) | A kind of shortwave list under calibration source existence condition, which is stood erectly, connects localization method | |
CN102325369A (en) | WLAN (Wireless Local Area Network) indoor single-source linear WKNN (Weighted K-Nearest Neighbor) locating method based on reference point position optimization | |
CN102752849A (en) | Single receiving machine location method based on signal detection probability and wave angle estimation | |
CN108872932A (en) | The direct positioning result method for correcting error of over-the-horizon target neural network based | |
CN104038901A (en) | Indoor positioning method for reducing fingerprint data acquisition workload | |
CN104820204A (en) | Weighted least square positioning method with reduced deviation | |
CN104640204A (en) | Wireless sensor network node positioning method in indirect wave environment | |
CN101986756A (en) | Time-reversal signal-based wireless positioning scheme |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20080507 |