CN105072581A - Indoor positioning method of path attenuation coefficient based database construction - Google Patents

Abstract

The invention discloses an indoor positioning method of path attenuation coefficient based database construction. The method is divided into a training stage and a positioning stage; the training stage is mainly to obtain a corresponding attenuation coefficient of each reference node in the environment and an average attenuation coefficient of signal propagation in the environment by utilizing the known received signal strength and distance information; the positioning stage is mainly to obtain a distance between a blind node and the corresponding reference node and a distance between the blind nodes by utilizing the corresponding attenuation coefficients of the reference nodes in the environment which are obtained in the training stage, thereby obtaining a relative coordinate position of each blind node. The indoor positioning method is high in positioning precision, simple in implementation, low in system power consumption and low in cost.

Description

A kind of indoor orientation method building storehouse based on path attenuation coefficient

Technical field

The invention belongs to wireless network field of locating technology, be specifically related to a kind of indoor orientation method building storehouse based on path attenuation coefficient.

Background technology

Along with the rise of Internet of Things concept, and mobile computing device developing rapidly and universal, wireless sensor network (WirelessSensingNetwork, WSN) starts the focus becoming industry research.The target of wireless sensor network be will originally dispersion and independently hardware node couple together, form a shared network, each node in network has the ability of receiving and transmitting signal, and provides required environmental information, to reach the interaction of user and instrument.So-called environmental information, wherein has the very important point to be exactly spatial positional information, if can obtain the current location of node, the personalized function of so many practicalities just can realize.Such as: arrange wireless sensor node at bulk supermarket, at each shelf and shopping cart configuration node, such client just can find required shelf according to the positional information of shopping cart fast, makes shopping process be tending towards automation; Arrange wireless sensor node in hospital or community, each patient carries with a set of physical signs monitor, and passes to Surveillance center in real time by wireless network, and Surveillance center can obtain patient location at any time, holds therapic opportunity in time; Arrange wireless sensor node in museum, visitor allots small-sized transceiver, and intelligent guide system according to the current location real-time broadcast different content of visitor, and can not be subject to the restriction of travelling route.

In wireless sensor network, conventional mapping methods is the distance first estimated between mobile radio station and base station, and then by some estimations based on the algorithm realization positional information of dimensional measurement, therefore, distance estimations is the basis of location algorithm.In measuring the time of advent (TimeofArrival, TOA), the distance between signal node is obtained by the product of measuring-signal propagation time and signal velocity; In measuring the time of advent poor (TimeDifferenceofArrival, TDOA), the time difference that the range difference between unlike signal node arrives multiple receiving node by measuring-signal obtains; Received signal strength (TimeDifferenceofArrival, TDOA) is measured, and the propagation loss model of basis signal derives the direct distance of signal node.Although it is simple that above-mentioned conventional mapping methods has algorithm, easy realization, to the feature such as System Hardware Requirement is lower, but because the estimation of adjusting the distance easily is subject to multipath effect, the impact of the factors such as the signal attenuation of nlos environment and particular surroundings, the propagation of signal is difficult to provide in the mode of model predict accurately, and distance estimations has comparatively big error, and the estimation that result in positional information is inaccurate.These class methods general only can be applicable to outdoor localizing environment, and the positioning precision in indoor will reduce greatly.

A class localization method fast-developing be now then utilize fingerprint identification technology build storehouse recognition methods, and generally have employed and build library information based on received signal strength.Build storehouse recognition methods and be divided into off-line and online two stages.Off-line phase, navigation system in each coordinate known point collection and storage signal intensity, as building storehouse parameter.On-line stage, the signal strength signal intensity that the signal strength signal intensity measured and off-line phase collect is mated by navigation system, thus provides positional information.Conventional matching algorithm has, probabilistic method (ProbabilisticMethods), field, rank method (k-Nearest-Neighbor, kNN), neural net (NeuralNetworks), SVMs (SupportVectorMachine, SVM), minimum polygon (SmallestM-vertexPolygon, SMP) method.Although said method has positioning result advantage more accurately, but it is a complicated process that its off-line data builds the storehouse stage, very high request is proposed to the computing capability of system hardware and memory capacity, adds system cost, and the location requirement of emergency case cannot be tackled.These class methods general only can be applicable to indoor localizing environment, at the location cost of this system of outdoor utility considerably beyond the raising of positioning precision.

Summary of the invention

The object of the present invention is to provide a kind of indoor orientation method of environmental adaptation channel model, solve the problem that in prior art, positioning precision is low, enforcement is complicated, cost is high.

In order to achieve the above object, the present invention adopts following technical scheme:

Localization method of the present invention is divided into training stage and positioning stage two parts.

In the present invention, sub-fraction node has a priori location information, is referred to as reference node, and the positional information of all the other nodes to be positioned is unknown, is referred to as blind node.All nodes, do not consider whether it has the co-ordinate position information of priori, all need to carry out received signal strength measurement to the node in its effective propagation path, and are converted to euclidean distance between node pair by maximum Likelihood.

In the training stage, the blind node of a coordinate the unknown moves along random walk in the environment of required location, reference node sends signal, blind node acknowledge(ment) signal tracer signal from transmitting node, object along random walk tracer signal intensity is the received signal strength fully gathering each point in environment, can the channel fading coefficient of complete describe environment can try to achieve, by the certain distortion to channel attenuation model, utilize distance known between the received signal strength of each node and each node, be deduced the channel fading coefficient portraying current environment, be out of shape by formula, the attenuation coefficient that each reference node is corresponding in the environment can be tried to achieve simultaneously, and the mean attenuation coefficient that in environment, signal is propagated,

Positioning stage, the quantity of reference node and position are without the need to changing, signal transmitting and receiving mode is identical with the training stage, add one or more blind node to be positioned in the environment simultaneously, the all nodes comprising blind node and reference node send signal simultaneously, each blind node accepts the signal strength signal intensity coming from all the other nodes, and tracer signal from transmitting node.When signal comes from reference node, the attenuation coefficient that the reference node utilizing the training stage to try to achieve is corresponding in the environment, in conjunction with maximum likelihood range estimating equation, try to achieve the distance of this blind node and corresponding reference node, when signal comes from another blind node, the mean attenuation coefficient that the signal then utilizing the training stage to try to achieve is propagated, in conjunction with maximum likelihood range estimating equation, try to achieve corresponding blind internodal distance, after each internodal distance is all accurately estimated in the environment, facility maximum likelihood coordinate estimation formulas tries to achieve the relative coordinate position of each blind node.

Location algorithm comprises the following steps:

Step 1: try to achieve the path attenuation coefficient of each signal reference node in applied environment by measured signal intensity data

Reference node and blind node all adopt wide band direct sequence spread spectrum transceiver, by battery-powered, launch narrow band power signal, the signal antenna of receiver remains on the height of 1.5 meters, and along random walk, the narrow band power signal that each signal transmitter sends is measured in doors, record the distance on random walk between each measurement point and signal transmitter, on definition random walk, the signal strength measurement of the jth signal transmitter that i-th measurement point receives is p simultaneously _r(i, j) [dB], the path loss measurement that on definition random walk, i-th measurement point is corresponding with a jth signal transmitter is signal transmission power P _tthe difference of (i, j) [dB] and signal strength measurement, and be expressed as

$\begin{array}{c}{\mathrm{PL}}_{mea}(i,j)\[dB\]=10{\log }_{10}\left(\frac{P_T(i,j)}{P_R(i,j)}\right)\\ =P_T(i,j)\[dB\]-P_R(i,j)\[dB\]\end{array}$

Channel propagation model under definition indoor environment is

${\mathrm{PL}}_{mol}\left(d\right)\[dB\]=PL\left(d_0\right)\[dB\]+10{\mathrm{nlog}}_{10}\left(\frac{d}{d_0}\right)$

Wherein PL _mold () [dB] is the path loss prediction value when signal receiver and transmitter distance are known quantity d, PL (d ₀) [dB] be at reference distance d ₀the path loss measurement at place, gets d in indoor short-distance transmission situation usually ₀=1m, now can also calculate PL (d by free path loss model ₀) [dB], n is channel fading coefficient, which depict the relation of the path attenuation velocity and distance of signal strength signal intensity, and all path loss values are all the free path loss value (FreeSpacePathLoss with 1 meter, FSPL) as a reference, its decibel of expression formula be

$FSPL\left(dB\right)=20{\log }_{10}\left(\frac{4\mathrm{\π}}{c}df\right)$

Wherein c represents the light velocity, and d is reference distance 1 meter, and f is emission signal frequency,

Definition path attenuation coefficient vector is

n＝[n ₁n ₂…n _m]

N _j(j=1,2 ..., m) the corresponding path attenuation coefficient of each signal transmitter in indoor environment, thus, can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\[dB\]=PL\left(d_0\right)\[dB\]+10n_1{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\[dB\]=PL\left(d_0\right)\[dB\]+10n_2{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\[dB\]=PL\left(d_0\right)\[dB\]+10n_{N_i}{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point, w is the interchannel noise of obeying zero-mean gaussian distribution, and Systems with Linear Observation equation can be written as further

And simplification is written as

x _i＝H _in+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector x _i(i=1,2 ..., L) synthesize as next dimension is vector

$x=\left[\begin{array}{c}{x}_1\\ x_2\\ .\\ .\\ .\\ x_L\end{array}\right]$

And it is as follows to define N × m observing matrix H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝Hn+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index $J\left(\hat{n}\right)={(x-H\hat{n})}^T(x-H\hat{n})$

Reach minimum, solve into

$\hat{n}={\left(H^{T}H\right)}^{-1}{H}^{T}x$

Step 2: tried to achieve the average path attenuation coefficient in applied environment by measured signal intensity data

Can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\[dB\]=PL\left(d_0\right)\[dB\]+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\[dB\]=PL\left(d_0\right)\[dB\]+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\[dB\]=PL\left(d_0\right)\[dB\]+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point, w is that to obey average be zero variance be σ ²the interchannel noise of Gaussian Profile, Systems with Linear Observation equation can be written as further

${\left[\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\[dB\]-PL\left(d_0\right)\[dB\]\\ {\mathrm{PL}}_{mea}(i,2)\[dB\]-PL\left(d_0\right)\[dB\]\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\[dB\]-PL\left(d_0\right)\[dB\]\end{array}\right]}_{N_i\×1}=n{\left[\begin{array}{c}10{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)\\ 10{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)\\ .\\ .\\ .\\ 10{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)\end{array}\right]}_{N_i\×1}$

And simplification is written as

x _i＝nH _i+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector x _i(i=1,2 ..., L) synthesize as next dimension is vector

$X=\left[\begin{array}{c}{x}_1\\ x_2\\ .\\ .\\ .\\ x_L\end{array}\right]$

And it is as follows to define N × 1 measurement vector H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝nH+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index

$J\left(\hat{n}\right)={(x-\hat{n}H)}^T(x-\hat{n}H)$

Reach minimum, solve into

$\tilde{n}=\frac{H^{T}x}{\left(H^{T}H\right)}$

Step 3: each node implements point-to-point signal strength measurement, and calculate phase mutual edge distance

Start location, node used in environment measures the received signal strength value in its effective propagation path mutually, and the signal launched for i node reception j node in applied environment, its signal strength signal intensity measuring gained is P _i,j, the maximum likelihood estimator of its distance is

${\tilde{d}}_{ij}={\tilde{d}}_{ji}=d_0{\left(\frac{p_0}{p_{ij}}\right)}^{\frac{1}{n}}$

Wherein d ₀reference distance, p ₀it is the received signal strength recorded under reference distance, and n is path attenuation coefficient, when institute's acknowledge(ment) signal comes from reference node, apply the path attenuation coefficient that this reference node is corresponding, if come from blind node, then apply average path attenuation coefficient, if certain two internodal Signal transmissions is in disarmed state, then its mutual distance is set to default value, generally gets the average distance that applied environment is long and wide.

Step 4: finally calculate blind node location

At each node measurement and record other nodes send received signal strength value after, adopt maximum Likelihood, estimate the position of each blind node, formula is as follows

$\widetilde{\mathrm{\&theta;}}=\underset{\mathrm{\&theta;}}{\arg \min }\underset{i=1}{\overset{m+n}{\mathrm{\&Sigma;}}}\underset{j<i}{\underset{j\&Element;H\left(i\right)}{\mathrm{\&Sigma;}}}{\left(ln\frac{{\tilde{d}}_{ij}^2}{d^2(z_i,z_j)}\right)}^2$

Wherein, z _i=(x _i, y _i), θ=[z ₁z ₂z _n] represent n blind node coordinate, its maximum likelihood estimator, can be in the hope of by steepest descent method.

Beneficial effect of the present invention:

First: the training stage of the present invention utilizes known received signal strength and range information, try to achieve the Confirming model of required localizing environment fading channel, compensate for attenuation coefficient in basic skills and can not reflect true environment and the distance estimations error caused, thus improve positioning precision;

Second: simultaneously the self-organizing network positioning system that the present invention builds estimates multiple blind node location, can realize in navigation system the requirement of real-time that each blind node location is estimated, simultaneously, in guarantee positioning precision situation, adding of new blind node, the range information needed for location can be increased, positioning precision is improved, and do not need to do any setting to the node newly added, given full play to self-organizing network characteristic;

3rd: the present invention is owing to only building storehouse to environmental attenuation coefficient, and required hardware resource is much smaller than building storehouse localization method.And compare and basic fixed position method, positioning precision improves 20% within the scope of two meters;

4th: in the present invention, most range information be converted to by received signal strength information comes from the measurement between each blind node, so the reference node quantity required in the present invention is effectively reduced, is not affecting in positioning precision situation, system power dissipation, cost is effectively controlled;

5th: the present invention only need carry out one-shot measurement to localizing environment, after obtaining channel parameter, just accurately online location in real time can be realized, off-line is needed to build storehouse with tradition or field intensity curve learning method is compared, there is method be easy to implement, the feature that required priori data amount is minimum, greatly reduce the requirement to positioning system hardware, low cost can be reached, the requirement of low-power consumption;

6th: the channel parameter that the present invention tries to achieve contains two classes, one is the dissemination channel parameter of each reference node in applied environment, and parameter is different from the difference of transmitter, has good pairing characteristic; Two is average channel parameters of applied environment, propagate mainly for blind internodal signal and measure, the application of two class parameters effectively can improve system accuracy, achieves the pairing for specific environment, in the present invention, mobile node location and ipc monitor function can be realized simultaneously.

Accompanying drawing explanation

Below in conjunction with the drawings and specific embodiments, the present invention is further detailed explanation.

Fig. 1 is the present invention and basic fixed position method and builds the cumulative errors function ratio of storehouse localization method comparatively;

Fig. 2 is position error and blind number of nodes graph of a relation.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.

A kind of embodiment of the present invention, in the present embodiment, sub-fraction node has a priori location information, is referred to as reference node, and the positional information of all the other nodes to be positioned is unknown, is referred to as blind node.All nodes, do not consider whether it has the co-ordinate position information of priori, all need to carry out received signal strength measurement to the node in its effective propagation path, and are converted to euclidean distance between node pair by maximum Likelihood.In cellular network location information estimating method and local positioning system, range information needed for location only comes between blind node and reference node, in the present embodiment, most range information be converted to by received signal strength information comes from the measurement between each blind node.

Experimental situation is an indoor office room environmental, room-size length and width are at about 30 meters, the average path attenuation coefficient of environment is 3.4, and it is 4 that reference node is counted, in a monolithic chamber environment do not split by body of wall, reference node is placed in each corner of ceiling, and provide its coordinate, in indoor positioning, coordinate is traditionally arranged to be relative coordinate, the positioning flow of the present embodiment, can be divided into training stage and positioning stage two parts.

In the training stage, the blind node of a coordinate the unknown moves along random walk in the environment of required location, and reference node sends signal, blind node acknowledge(ment) signal tracer signal from transmitting node.Object along random walk tracer signal intensity is the received signal strength fully gathering each point in environment, can the channel fading coefficient of complete describe environment can try to achieve.By the certain distortion to channel attenuation model, utilize distance known between the received signal strength of each node and each node, derive the channel fading coefficient portraying current environment.Be out of shape by formula, the attenuation coefficient that each reference node is corresponding in the environment can be tried to achieve simultaneously, and the mean attenuation coefficient that in environment, signal is propagated, thus fully describe current environment, and the positional parameter obtained is only one group of vector, arrange four reference nodes for single ventricle environment, the element number of this group vector only has 5, has been significantly smaller than the received signal strength memory space of each point in the environment built during storehouse identifies and store.And training process is consuming time also much smaller than building storehouse recognition positioning method.

Positioning stage, the quantity of reference node and position are without the need to changing, and signal transmitting and receiving mode is identical with the training stage.Add one or more blind node to be positioned in the environment, all nodes comprising blind node and reference node send signal simultaneously simultaneously, and each blind node accepts the signal strength signal intensity coming from all the other nodes, and tracer signal from transmitting node.When signal comes from reference node, the attenuation coefficient that the reference node utilizing the training stage to try to achieve is corresponding in the environment, in conjunction with maximum likelihood range estimating equation, try to achieve the distance of this blind node and corresponding reference node, when signal comes from another blind node, the mean attenuation coefficient that the signal then utilizing the training stage to try to achieve is propagated, in conjunction with maximum likelihood range estimating equation, tries to achieve corresponding blind internodal distance.After each internodal distance is all accurately estimated in the environment, facility maximum likelihood coordinate estimation formulas tries to achieve the relative coordinate position of each blind node.

Utilize location algorithm to measure blind spot position to comprise the following steps:

Step 1: try to achieve the path attenuation coefficient of each signal reference node in applied environment by measured signal intensity data

Reference node and blind node all adopt wide band direct sequence spread spectrum transceiver, by battery-powered, launch narrow band power signal, the signal antenna of receiver remains on the height of 1.5 meters, and along random walk, the narrow band power signal that each signal transmitter sends is measured in doors, record the distance on random walk between each measurement point and signal transmitter, on definition random walk, the signal strength measurement of the jth signal transmitter that i-th measurement point receives is p simultaneously _r(i, j) [dB], the path loss measurement that on definition random walk, i-th measurement point is corresponding with a jth signal transmitter is signal transmission power P _tthe difference of (i, j) [dB] and signal strength measurement, and be expressed as

$\begin{array}{c}{\mathrm{PL}}_{mea}(i,j)\&lsqb;dB\&rsqb;=10{\log }_{10}\left(\frac{P_T(i,j)}{P_R(i,j)}\right)\\ =P_T(i,j)\&lsqb;dB\&rsqb;-P_R(i,j)\&lsqb;dB\&rsqb;\end{array}$

Channel propagation model under definition indoor environment is

${\mathrm{PL}}_{mol}\left(d\right)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d}{d_0}\right)$

Wherein PL _mold () [dB] is the path loss prediction value when signal receiver and transmitter distance are known quantity d, PL (d ₀) [dB] be at reference distance d ₀the path loss measurement at place, gets d in indoor short-distance transmission situation usually ₀=1m, now can also calculate PL (d by free path loss model ₀) [dB], n is channel fading coefficient, which depict the relation of the path attenuation velocity and distance of signal strength signal intensity, and all path loss values are all the free path loss value (FreeSpacePathLoss with 1 meter, FSPL) as a reference, its decibel of expression formula be

$FSPL\left(dB\right)=20{\log }_{10}\left(\frac{4\mathrm{\&pi;}}{c}df\right)$

Wherein c represents the light velocity, and d is reference distance 1 meter, and f is emission signal frequency,

Definition path attenuation coefficient vector is

n＝[n ₁n ₂…n _m]

N _j(j=1,2 ..., m) the corresponding path attenuation coefficient of each signal transmitter in indoor environment, thus, can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_1{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_2{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_{N_i}{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point, w is the interchannel noise of obeying zero-mean gaussian distribution, and Systems with Linear Observation equation can be written as further

And simplification is written as

x _i＝H _in+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector x _i(i=1,2 ..., L) synthesize as next dimension is vector

$x=\left[\begin{array}{c}{x}_1\\ x_2\\ .\\ .\\ .\\ x_L\end{array}\right]$

And it is as follows to define N × m observing matrix H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝Hn+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index $J\left(\hat{n}\right)={(x-H\hat{n})}^T(x-H\hat{n})$

Reach minimum, solve into

$\hat{n}={\left(H^{T}H\right)}^{-1}{H}^{T}x.$

Step 2: tried to achieve the average path attenuation coefficient in applied environment by measured signal intensity data

Can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point, w is that to obey average be zero variance be σ ²the interchannel noise of Gaussian Profile, Systems with Linear Observation equation can be written as further

${\left[\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\end{array}\right]}_{N_i\&times;1}=n{\left[\begin{array}{c}10{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)\\ 10{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)\\ .\\ .\\ .\\ 10{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)\end{array}\right]}_{N_i\&times;1}$

And simplification is written as

x _i＝nH _i+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector x _i(i=1,2 ..., L) synthesize as next dimension is vector

$x=\left[\begin{array}{c}{x}_1\\ x_2\\ .\\ .\\ .\\ x_L\end{array}\right]$

And it is as follows to define N × 1 measurement vector H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝nH+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index

$J\left(\hat{n}\right)={(x-\hat{n}H)}^T(x-\hat{n}H)$

Reach minimum, solve into

$\tilde{n}=\frac{H^{T}x}{\left(H^{T}H\right)}.$

Step 3: each node implements point-to-point signal strength measurement, and calculate phase mutual edge distance

Start location, node used in environment measures the received signal strength value in its effective propagation path mutually, and the signal launched for i node reception j node in applied environment, its signal strength signal intensity measuring gained is P _i,j, the maximum likelihood estimator of its distance is

${\tilde{d}}_{ij}={\tilde{d}}_{ji}=d_0{\left(\frac{p_0}{p_{ij}}\right)}^{\frac{1}{n}}$

Wherein d ₀reference distance, p ₀it is the received signal strength recorded under reference distance, and n is path attenuation coefficient, when institute's acknowledge(ment) signal comes from reference node, apply the path attenuation coefficient that this reference node is corresponding, if come from blind node, then apply average path attenuation coefficient, if certain two internodal Signal transmissions is in disarmed state, then its mutual distance is set to default value, generally gets the average distance that applied environment is long and wide.

Step 4: finally calculate blind node location

At each node measurement and record other nodes send received signal strength value after, the present invention adopts maximum Likelihood, and estimate the position of each blind node, formula is as follows

$\widetilde{\mathrm{\&theta;}}=\underset{\mathrm{\&theta;}}{\arg \min }\underset{i=1}{\overset{m+n}{\mathrm{\&Sigma;}}}\underset{j<i}{\underset{j\&Element;H\left(i\right)}{\mathrm{\&Sigma;}}}{\left(ln\frac{{\tilde{d}}_{ij}^2}{d^2(z_i,z_j)}\right)}^2$

Wherein, z _i=(x _i, y _i), θ=[z ₁z ₂z _n] represent n blind node coordinate, its maximum likelihood estimator, can be in the hope of by steepest descent method.

As can be seen from Figure 1, the present invention has similar cumulative errors distribution function with building compared with the localization method of storehouse, but hardware resource required for the present invention is much smaller than building storehouse localization method, and compare and basic fixed position method, positioning precision improves 20% within the scope of two meters.

As can see from Figure 2, along with the increase of blind number of nodes, positioning precision is also in increase, and when blind number of nodes continues to be increased to 16 nodes, the improvement change of positioning precision tends towards stability.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (5)

1. build the indoor orientation method in storehouse based on path attenuation coefficient, it is characterized in that: comprise training stage and positioning stage,

The described training stage is specially: the blind node of a coordinate the unknown moves along random walk in the environment of required location, reference node sends signal, blind node acknowledge(ment) signal tracer signal from transmitting node, by the distortion to channel attenuation model, utilize distance known between the received signal strength of each node and each node, derive the channel fading coefficient of current environment, be out of shape by formula, the attenuation coefficient that each reference node is corresponding in the environment can be tried to achieve simultaneously, and the mean attenuation coefficient that in environment, signal is propagated;

Described positioning stage is specially: the quantity of reference node and position are without the need to changing, signal transmitting and receiving mode is identical with the training stage, add one or more blind node to be positioned in the environment simultaneously, the all nodes comprising blind node and reference node send signal simultaneously, each blind node accepts the signal strength signal intensity coming from all the other nodes, and tracer signal from transmitting node, when signal comes from reference node, the attenuation coefficient that the reference node utilizing the training stage to try to achieve is corresponding in the environment, in conjunction with maximum likelihood range estimating equation, try to achieve the distance of this blind node and corresponding reference node, when signal comes from another blind node, the mean attenuation coefficient that the signal then utilizing the training stage to try to achieve is propagated, in conjunction with maximum likelihood range estimating equation, try to achieve corresponding blind internodal distance, after each internodal distance is all accurately estimated in the environment, maximum likelihood coordinate estimation formulas is utilized to try to achieve the relative coordinate position of each blind node.

2. a kind of indoor orientation method building storehouse based on path attenuation coefficient according to claim 1, is characterized in that: the concrete grammar of being tried to achieve the path attenuation coefficient of each signal reference node in applied environment by measured signal intensity data is:

Reference node and blind node all adopt wide band direct sequence spread spectrum transceiver, by battery-powered, launch narrow band power signal, the signal antenna of receiver remains on the height of 1.5 meters, and along random walk, the narrow band power signal that each signal transmitter sends is measured in doors, record the distance on random walk between each measurement point and signal transmitter, on definition random walk, the signal strength measurement of the jth signal transmitter that i-th measurement point receives is p simultaneously _r(i, j) [dB], the path loss measurement that on definition random walk, i-th measurement point is corresponding with a jth signal transmitter is signal transmission power P _tthe difference of (i, j) [dB] and signal strength measurement, and be expressed as

$\begin{array}{c}{\mathrm{PL}}_{mea}(i,j)\&lsqb;dB\&rsqb;=10{\log }_{10}\left(\frac{P_T(i,j)}{P_R(i,j)}\right)\\ =P_T(i,j)\&lsqb;dB\&rsqb;-P_R(i,j)\&lsqb;dB\&rsqb;\end{array}$

Channel propagation model under definition indoor environment is

${\mathrm{PL}}_{mol}\left(d\right)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d}{d_0}\right)$

Wherein PL _mold () [dB] is the path loss prediction value when signal receiver and transmitter distance are known quantity d, PL (d ₀) [dB] be at reference distance d ₀the path loss measurement at place, gets d in indoor short-distance transmission situation usually ₀=1m, now can also calculate PL (d by free path loss model ₀) [dB], n is channel fading coefficient, which depict the relation of the path attenuation velocity and distance of signal strength signal intensity, and all path loss values are all with the free path loss value of 1 meter as a reference, and its decibel of expression formula is

$FSPL\left(dB\right)=20{\log }_{10}\left(\frac{4\mathrm{\&pi;}}{c}df\right)$

Wherein c represents the light velocity, and d is reference distance 1 meter, and f is emission signal frequency,

Definition path attenuation coefficient vector is

n＝[n ₁n ₂…n _m]

N _j(j=1,2 ..., m) the corresponding path attenuation coefficient of each signal transmitter in indoor environment,

Thus, can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_1{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_2{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10n_{N_i}{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point,

W is the interchannel noise of obeying zero-mean gaussian distribution, and Systems with Linear Observation equation can be written as further

And simplification is written as

x _i＝H _in+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector x _i(i=1,2 ..., L) synthesize as next dimension is vector

$x=\left[\begin{array}{c}{x}_1\\ x_2\\ .\\ .\\ .\\ x_L\end{array}\right]$

And it is as follows to define N × m observing matrix H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝Hn+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index

$J\left(\hat{n}\right)={(x-H\hat{n})}^T(x-H\hat{n})$

Reach minimum, solve into $\hat{n}={\left(H^{T}H\right)}^{-1}{H}^{T}x$

3. a kind of indoor orientation method building storehouse based on path attenuation coefficient according to claim 1, is characterized in that: the concrete grammar of the average path attenuation coefficient of being tried to achieve in applied environment by measured signal intensity data is:

Can be expressed as at the Systems with Linear Observation equation of i-th measurement point

$i\left\{\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,1}}{d_0}\right)+w_1\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,2}}{d_0}\right)+w_2\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;=PL\left(d_0\right)\&lsqb;dB\&rsqb;+10{\mathrm{nlog}}_{10}\left(\frac{d_{i,N_i}}{d_0}\right)+w_{N_i}\end{array}\right.$

Wherein N _i(N _i≤ m) illustrate and effectively measure number at the signal receiving strength of i-th measurement point, w is that to obey average be zero variance be σ ²the interchannel noise of Gaussian Profile, Systems with Linear Observation equation can be written as further

${\left[\begin{array}{c}{\mathrm{PL}}_{mea}(i,1)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\\ {\mathrm{PL}}_{mea}(i,2)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\\ .\\ .\\ .\\ {\mathrm{PL}}_{mea}(i,N_i)\&lsqb;dB\&rsqb;-PL\left(d_0\right)\&lsqb;dB\&rsqb;\end{array}\right]}_{N_i\&times;1}=n{\left[\begin{array}{c}10{\log }_{10}\left(\frac{d_{i,1}}{d_0}\right)\\ 10{\log }_{10}\left(\frac{d_{i,2}}{d_0}\right)\\ .\\ .\\ .\\ 10{\log }_{10}\left(\frac{d_{i,N_i}}{d_0}\right)\end{array}\right]}_{N_i\&times;1}$

And simplification is written as

x _i＝nH _i+w _i,i＝1,2,...,L

After constructing the Systems with Linear Observation equation of i-th measurement point, whole L measurement vector

And it is as follows to define N × 1 measurement vector H and N dimension observation noise vector w accordingly

$H=\left[\begin{array}{c}{H}_1\\ H_2\\ .\\ .\\ .\\ H_L\end{array}\right],w=\left[\begin{array}{c}{w}_1\\ w_2\\ .\\ .\\ .\\ w_L\end{array}\right]$

Thus, the total observational equation on random walk can be written as

x＝nH+w

For making the mean square error of estimator reach minimum, according to the formation rule of Linear least square estimation amount, structure estimator performance index

$J\left(\hat{n}\right)={(x-\hat{n}H)}^T(x-\hat{n}H)$

Reach minimum, solve into

$\tilde{n}=\frac{H^{T}x}{\left(H^{T}H\right)}$

4. a kind of indoor orientation method building storehouse based on path attenuation coefficient according to claim 1, is characterized in that: each node implements point-to-point signal strength measurement, and the concrete grammar calculating phase mutual edge distance is:

Start location, node used in environment measures the received signal strength value in its effective propagation path mutually, and the signal launched for i node reception j node in applied environment, its signal strength signal intensity measuring gained is P _i,j, the maximum likelihood estimator of its distance is

${\tilde{d}}_{ij}={\tilde{d}}_{ji}=d_0{\left(\frac{p_0}{p_{ij}}\right)}^{\frac{1}{n}}$

Wherein d ₀reference distance, p ₀it is the received signal strength recorded under reference distance, and n is path attenuation coefficient, when institute's acknowledge(ment) signal comes from reference node, apply the path attenuation coefficient that this reference node is corresponding, if come from blind node, then apply average path attenuation coefficient, if certain two internodal Signal transmissions is in disarmed state, then its mutual distance is set to default value, generally gets the average distance that applied environment is long and wide.

5. a kind of indoor orientation method building storehouse based on path attenuation coefficient according to claim 1, is characterized in that: finally calculate blind node location and be specially:

At each node measurement and record other nodes send received signal strength value after, adopt maximum Likelihood, estimate the position of each blind node, formula is as follows

$\widetilde{\mathrm{\&theta;}}=\underset{\mathrm{\&theta;}}{\arg \min }\underset{i=1}{\overset{m+n}{\mathrm{\&Sigma;}}}\underset{j<i}{\underset{j\&Element;H\left(i\right)}{\mathrm{\&Sigma;}}}{\left(ln\frac{{\tilde{d}}_{ij}^2}{d^2(z_i,z_j)}\right)}^2$

Wherein, z _i=(x _i, y _i), θ=[z ₁z ₂z _n] represent n blind node coordinate, its maximum likelihood estimator, can be in the hope of by steepest descent method.