CN105960017A - Ultra-wideband node network-based device-free localization method - Google Patents

Abstract

The present invention discloses an ultra-wideband node network-based device-free localization method, belonging to the field of wireless localization technologies in a wireless network. According to the method, an ultra-wideband signal is sent by using a node so as to realize device-free localization (DFL), a channel pulse response in a line of sight (LOS) is detected, and localization is performed by using change of an LOS path signal power. In addition, the ultra-wideband node has a precise distance measurement capability, and can achieve precise node self-localization. Node self-localization is achieved by using distance measurement provided by the ultra-wideband node according to a cooperation localization method, shadow fading caused by shielding by a target is obvious, and an imaging effect is further improved, so that the method is more applicable to an environment with low LOS power, can improve localization precision in a more complex environment, and greatly reduces manpower and time cost for node deployment.

Description

Based on ultra broadband meshed network exempt from Portable device localization method

Technical field

The present invention relates to one and exempt from Portable device localization method, be a kind of based on the estimation of direct path shadow fading Ultra broadband object localization method, belong to the wireless location technology field in wireless network.

Background technology

Exempt from Portable device location (device-free localization, DFL) be developing progressively recent years into A kind of emerging technology, it allows target itself not carry any positioner.The application of DFL widely, Because target is unwilling to carry positioner or target is noncooperative in a lot of situations.Therefore DFL exists Intrusion detection, disaster relief, health care, and military operation have a lot of potential application.Traditional DFL Method uses optics, infrared method, and penetrance is poor and have may invasion of privacy.Along with wireless Popularizing of communication, the research of DFL based on wireless signal obtains the biggest concern, and wireless signal has Preferably penetration capacity, it is possible to penetrate the shelters such as wall.

Wireless DFL can be rough be divided into two classes, a class is method based on shadow fading, and another kind of is base Method in other characteristics such as signaling reflex of wireless signal.The invention mainly relates to based on shadow fading DFL method.In essence, what DFL based on shadow fading utilized is that target is to wireless link sighting distance (line Of sight, LOS) path blocks and causes the decay of signal.In traditional DFL based on shadow fading, Node send be narrow band signal, and utilize link received signal strength (received signal strength, RSS) change positions target.But it is in actual environment, owing to environment also existing various reflector, many Footpath is propagated extremely serious, and method based on narrow band signal RSS is relatively big by multi-path influence, and positioning precision declines. Additionally, DFL needs to measure the position of radio node, the method for manual measurement is the most all used to obtain node Position, needs a large amount of manpower and time cost, is not suitable for the occasion promptly disposed.

Summary of the invention

In order to solve the problem of multi-path jamming and node self-localization, the present invention proposes to utilize node to send ultra broadband Signal realizes the method for DFL.Use ultra-broadband signal can bring two benefits: preferable multi-path resolved energy Power and range capability.Owing to the big bandwidth of ultra-broadband signal can differentiate multipath signal, ultra broadband node measurement Each peak value of the channel impulse response obtained correspond to the power of each multipath component, the therefore present invention couple Channel impulse response (channel impulse response, CIR) under LOS path detects, and utilizes The change of LOS path signal power positions.Ultra broadband node has precision ranging ability simultaneously, can realize Accurately node self-localization, the method that the range measurement that the present invention utilizes ultra broadband node to provide uses co-positioned Realize the self-align of node, greatly reduce manpower and the time cost of node deployment.

Of the present invention based on ultra broadband meshed network exempt from Portable device localization method, comprise the following steps:

Step one: node deployment

K random being deployed in around monitored area of ultra-wideband detection node, this K node is positioned at same water In plane, it is assumed that the coordinate of i-th node is x_i=(x_i,y_i)^T, i=1,2 ..., K；Each node can Send and receive wireless signal, constituting η=K (K-1) bar link altogether；Additionally there are a base-station node, It is responsible for receiving and measures the channel impulse response CIR data that record of node, and data are issued PC do subsequent treatment； Respectively measure the wireless signal that node sent and received and be narrow pulse signal；

Step 2: measure when monitored area does not has a target channel of link between i-th node and jth node Impulse response CIR, is designated asAnd extract LOS path signal merit RateWith LOS path signal TOA, comprise the steps:

Step 2.1: monitored area keeps spaciousness, measures the CIR of the link constituted between any two nodes, It is designated as

Step 2.2: extract LOS path signal power from the multipath signal power of each of the links:

Use Edge Detection, it is believed that first CIR peak value correspondence LOS path letter exceeding given threshold value Number, i.e.

${\stackrel{\‾}{c}}_{(i,j)}^{{\hat{k}}^0}=\arg \underset{n}{min}\{P_n>P_{threshold},P_n\∈P_{(i,j)}\}$

Wherein P_thresholdFor the threshold value set, be typically based between false-alarm and missing inspection compromise determines；P_(i,j)It is CIR measured valueThe set of middle peak value, n is set P_(i,j)The sequence number of middle peak value element,For LOS path CIR peak value is at set P_(i,j)In sequence number, then the LOS road of link between node i and j when driftlessness blocks The power of footpath signal is

Step 2.3: after obtaining the CIR peak value of LOS path, the time judgement that LOS path peak value is arrived The TOA time of advent for signal；Calculate distance d between transmitting-receiving node i and j_i,j；

Step 3: carry out node self-localization according to the calculated euclidean distance between node pair of step 2, including walking as follows Rapid:

Step 3.1: obtained following square distance matrix by the distance of every pair of node, size is K × K；

Construct following matrix:

$B=-\frac{1}{2}JDJ;$

WhereinE_KIt is K rank unit matrixs, 1_KBe element value be all 1 K × 1 big Little column vector.

The Eigenvalues Decomposition of B is

B=V Λ V^T；

Wherein Λ=diag (λ₁,λ₂,....,λ_K) be B eigenvalue constitute diagonal matrix, eigenvalue is arranged in descending order Row, V=(v₁,v₂,....,v_K)^TThe matrix constituted for individual features value characteristic of correspondence vector；

Step 3.2: the overall barycenter constituted by K node is set to initial point, i.e.X The matrix constituted for the coordinate of all nodes, i.e.Make V₂=(v₁,v₂), Λ₂=diag (λ₁,λ₂), then node location is estimated as

$\hat{X}=V_2{\mathrm{\Λ}}_2^{1/2};$

I.e.For the estimation of node coordinate matrix X, be size be the matrix of K*2, footmark 1/2 represents opens radical sign；

Step 4: to link between every pair of node i and j, obtains owing to barrier blocks the LOS path caused Power change values

In monitored area, node measurement has link cir value c when blocking_(i,j), and according to step 2.2 method Extract now LOS path CIR peak valueMeasurement to unobstructed environment in integrating step two, without mesh When mark blocks, between node i and j, the LOS path CIR peak value of link isThen cause due to target occlusion Node i and j between the changed power of link LOS path be:

${\mathrm{\Δr}}_{(i,j)}^{LRSS}=10\lg {\stackrel{\‾}{c}}_{(i,j)}^{{\hat{k}}^0}-10\lg c_{(i,j)}^{{\hat{k}}^0};$

Step 5: use Portable device localization method of exempting from based on compressed sensing to realize target imaging:

Step 5.1: according to distance d between every pair of node of the self-align measurement of step 3 interior joint_i,jAnd estimation Each node locationCalculate weight matrix: method is as follows:

The method using wireless tomography (radio tomography imaging, RTI) in the present invention realizes target Location.In RTI, monitored area is divided evenly as square net, makes Δ x^mFor signal through monitoring section The decay of m-th grid in territory, total M grid, with weightRepresent that the signal of m-th grid declines Subtract the contribution of the signal attenuation of link, n between node i and j_(i,j)For the noise of link between node i and j, So between node i and j, the signal attenuation of link can be expressed as with aggregate form:

${\mathrm{\Δr}}_{(i,j)}^{LRSS}=\underset{m=1}{\overset{M}{\Σ}}{w}_{(i,j)}^m{\mathrm{\Δx}}^m+n_{(i,j)}$

Model of ellipse is used to calculate weight, as follows

$w_{(i,j)}^m=\left\{\begin{array}{c}1,d_{(i,j)}^m\left(1\right)+d_{(i,j)}^m\left(2\right)<d_{(i,j)}+\mathrm{\&gamma;}\\ 0\end{array}\right.$

WhereinWithIt is respectively the m grid distance to transmitting-receiving node, whereinExpression is arrived Receiving node,Represent to sending node, d_(i,j)For the distance between transmitting-receiving node；γ is adjustable parameter, Controlling the size of elliptic region, value is generally the half of signal wavelength；The weight that all links are obtained It is written as matrix form W ∈ R^L×N, and it is called weight matrix；

Step 5.2: utilize compressed sensing regularization method to realize target and position:

Considering all η bar links, the matrix form of LOS path power variation is

Δ r=W Δ x+n；

Wherein Δ r ∈ R^η×1It is observation vector, is by the LOS path signal power variations of η bar linkGroup Become, W ∈ R^η×MIt is the weight matrix of step 5.1 acquisition, n ∈ R^η×1Being noise vector, its element is n_(i,j)； Δx∈R^M×1It is vector to be estimated, the signal attenuation value on M grid that i.e. monitored area divides.Consider Element Δ x in Δ x^mNon-negative, based onThe regularization object function that norm compressed sensing is rebuild can be written as:

$\begin{array}{c}\underset{\mathrm{\&Delta;}x}{\min }\left|\right|\mathrm{\&Delta;}r-W\mathrm{\&Delta;}x|{|}^2+\mathrm{\&mu;}\underset{m=1}{\overset{M}{\&Sigma;}}{\mathrm{\&Delta;x}}^m\\ \begin{array}{cc}s.t.& {\mathrm{\&Delta;x}}^m\&GreaterEqual;0,m=1,2,...,M\end{array}\end{array};$

Wherein μ is regularization parameter；The grid being estimated as decay power maximum of target's center positionCoordinate in monitored area.

As preferably, narrow pulse signal described in step one, the persistent period is ns level or sub-ns level.

As preferably, in step 5.2, solve regularization object function by convex optimum theory.

As preferably, in step 2.1, the link constituted between any two nodes is carried out bidirectional ranging successively, This process is automatically performed by predetermined communication protocol by each node；Further, bidirectional ranging is according to TDMA Agreement completes.

As preferably, in step 2.3, by the method for symmetrical bilateral two-path DME calculate transmitting-receiving node i and j it Between distance d_i,j。

As preferably, ultra broadband node described in step 1 is ultra-wide belt segment based on IEEE802.15.4a agreement Point, it can provide the CIR of link to measure.Further, in ultra broadband node work described in step 1 Frequency of heart is 6.489GHz, carries a width of 499MHz and each node to assemble omnidirectional antenna.

As preferably, described barrier is the people stood still；

As preferably, barrier uses cylinder model.

Contrast prior art, the present invention has the beneficial effects that, and original DFL side based on arrowband RSS signal Method is compared because employing ultra-broadband signal, has reasonable multi-path resolved ability.And adopt on this basis By the method for LOS path power draw, it is more suitable for the environment that LOS power is relatively low, it is possible to more complicated Environment in improve positioning precision.Signal center frequency is improved to 6.489GHz by the 2.4GHz of narrow band signal, Make the shadow fading caused by target occlusion more notable, further increase imaging effect.In addition ultra-wide is taken a message Number can provide accurate range measurement, it is achieved node self-align, with DFL based on arrowband RSS signal The node deployment of method is compared and is saved extensive work.

Accompanying drawing explanation

Fig. 1: structured flowchart based on the ultra broadband node positioning method that direct path shadow fading is estimated；

Fig. 2: the layout schematic diagram of ultra broadband node in experiment；

The distribution of Fig. 3: experiment scene meeting room environment lower node and test point arrange schematic diagram；

Fig. 4: node self-localization result schematic diagram；

Fig. 5: DFL locating effect schematic diagram.

Detailed description of the invention

Below in conjunction with drawings and Examples, the present invention is described in detail, also described skill of the present invention simultaneously Art scheme solves the technical problem that and beneficial effect.The present invention proposes a kind of based on ultra broadband meshed network Exempting from Portable device localization method, the structured flowchart of this method is as it is shown in figure 1, specifically include and implement step as follows:

Step one: node deployment

K random being deployed in around monitored area of ultra-wideband detection node, as in figure 2 it is shown, this K joint Point is positioned in same level, it is assumed that the coordinate of i-th node is x_i=(x_i,y_i)^T, i=1,2 ..., K；Often Individual node can send and receive wireless signal, and therefore network may be considered full-mesh, may be constructed η=K (K-1) bar link；Additionally there are a base-station node, be responsible for receiving and measure the channel that node records Impulse response CIR data, and data are issued PC do subsequent treatment；

Node sends narrow pulse signal in the present invention, and the persistent period is the shortest, usually ns level and sub-ns Level, therefore the bandwidth of signal is the biggest；

It should be noted that in the experiment of the present invention be DecaWave company exploitation based on The ultra broadband node of IEEE802.15.4a agreement, it can provide the CIR of link to measure.In an experiment, super The operating center frequency of broadband node is 6.489GHz, carries a width of 499MHz, and node assembling is omnidirectional antennas Line.Using 10 ultra broadband nodes in experiment altogether, wherein node is measured in 9 conducts, and 1 is saved as base station Point.Measurement node has been responsible for the CIR of link and has been measured and send measured data to base-station node.Base station is saved Put to be connected by USB with local PC and data are issued PC and do subsequent treatment.Measure node deployment in monitoring Area periphery is also placed on support the distance with ground and is about 0.8m.In order to compare with self-align algorithm, We measure the position of all nodes in advance；

Step 2: measure when monitored area does not has a target channel of link between i-th node and jth node Impulse response CIR, is designated asAnd extract LOS path signal merit RateWith LOS path signal TOA, comprise the steps:

Step 2.1: monitored area keeps spaciousness, measures the CIR of the link constituted between any two nodes, It is designated asIn order to ensure accuracy, bidirectional measurement can be used to take Average method；

Step 2.2: extract LOS path signal power from the multipath signal power of each of the links:

The time of advent in view of LOS path signal is the shortest, but under complex environment, LOS path signal May be blocked by other barriers in environment, this means that to differ in the path that Article 1 arrives and is set to merit The path that rate is the strongest.Therefore the present invention uses Edge Detection (to accept signal phase information not utilizing In the case of, Edge Detection is the scheme uniquely screening LOS path.), it is believed that first exceed to Determine the CIR peak value correspondence LOS path signal of threshold value, i.e.

${\stackrel{\&OverBar;}{c}}_{i,j}^{{\hat{k}}^0}=\arg \underset{n}{min}\{P_n>P_{threshold},P_n\&Element;P_{(i,j)}\}$

Wherein P_thresholdFor the threshold value set, be typically based between false-alarm and missing inspection compromise determines；P_(i,j)It is CIR measured valueThe set of middle peak value, n is set P_(i,j)The sequence number of middle peak value element,For LOS path CIR peak value is at set P_(i,j)In sequence number, then the LOS road of link between node i and j when driftlessness blocks The power of footpath signal is

Step 2.3: after obtaining the CIR peak value of LOS path, the time judgement that LOS path peak value is arrived The TOA time of advent for signal；Calculated between transmitting-receiving node i and j by the method for symmetrical bilateral two-path DME Distance d_i,j；The method of symmetrical bilateral two-path DME refers to " IEEE Standard 802.15.4a-2007 ", P125；

It should be noted that the present invention has carried out target location under a typical indoor meeting room environment Experiment, in addition to surroundings wall, indoor have a lot of object to include the objects such as desk, chair, cupboard, this A little things can reflected radio signal, produce new multipath signal.In experiment, monitored area Node distribution is such as Shown in Fig. 3 (UWB node 1～9 in figure), the area of whole monitored area is 4.8m*2.4m, Wo Men Have selected 17 test positions in monitored area, as shown in the cross in figure, (be designated as L1～L17)；

Step 3: carry out node self-localization according to the calculated euclidean distance between node pair of step 2, including walking as follows Rapid:

Step 3.1: obtained following square distance matrix by the distance of every pair of node, size is K × K；

Construct following matrix:

$B=-\frac{1}{2}JDJ;$

WhereinE_KIt is K rank unit matrixs, 1_KBe element value be all 1 K × 1 big Little column vector.

The Eigenvalues Decomposition of B is

B=V Λ V^T；

Wherein Λ=diag (λ₁,λ₂,....,λ_K) be B eigenvalue constitute diagonal matrix, eigenvalue is arranged in descending order Row, V=(v₁,v₂,....,v_K)^TThe matrix constituted for individual features value characteristic of correspondence vector；

Step 3.2: the overall barycenter constituted by K node is set to initial point, i.e.X The matrix constituted for the coordinate of all nodes, i.e.Make V₂=(v₁,v₂), Λ₂=diag (λ₁,λ₂), then node location is estimated as

$\hat{X}=V_2{\mathrm{\&Lambda;}}_2^{1/2};$

I.e.For the estimation of node coordinate matrix X, be size be the matrix of K*2, footmark 1/2 represents opens radical sign；

Step 4: to link between every pair of node i and j, obtains owing to barrier blocks the LOS path caused Power change values

In monitored area, barrier remains stationary as on each test position, the test point in experimental situation Position is as it is shown on figure 3, node measurement has link cir value c when blocking_(i,j), and carry according to step 2.2 method Take out now LOS path CIR peak valueMeasurement to unobstructed environment in integrating step two, driftlessness When blocking, between node i and j, the LOS path CIR peak value of link isThen cause due to target occlusion Between node i and j, the changed power of link LOS path is:

${\mathrm{\&Delta;r}}_{(i,j)}^{LRSS}=10\lg {\stackrel{\&OverBar;}{c}}_{(i,j)}^{{\hat{k}}^0}-10\lg c_{(i,j)}^{{\hat{k}}^0};$

Step 5: use Portable device localization method of exempting from based on compressed sensing to realize target imaging:

Step 5.1: according to distance d between every pair of node of the self-align measurement of step 3 interior joint_i,jAnd estimation Each node locationCalculate weight matrix: method is as follows:

The method using wireless tomography (radio tomography imaging, RTI) in the present invention realizes target Location.In RTI, monitored area is divided evenly as square net, makes Δ x^mFor signal through monitoring section The decay of m-th grid in territory, total M grid, with weightRepresent that the signal of m-th grid declines Subtract the contribution of the signal attenuation of link, n between node i and j_(i,j)For the noise of link between node i and j, So between node i and j, the signal attenuation of link can be expressed as with aggregate form:

${\mathrm{\&Delta;r}}_{(i,j)}^{LRSS}=\underset{m=1}{\overset{M}{\&Sigma;}}{w}_{(i,j)}^m{\mathrm{\&Delta;x}}^m+n_{(i,j)}$

Because signal is mainly propagated along LOS path, therefore the mesh Weight close to LOS path is more than LOS The mesh Weight that path is remote, calculates weight by model of ellipse in the present invention, as follows

$w_{(i,j)}^m=\left\{\begin{array}{c}1,d_{(i,j)}^m\left(1\right)+d_{(i,j)}^m\left(2\right)<d_{(i,j)}+\mathrm{\&gamma;}\\ 0\end{array}\right.$

WhereinWithIt is respectively the m grid distance to transmitting-receiving node, whereinExpression is arrived Receiving node,Represent to sending node, d_(i,j)For the distance between transmitting-receiving node；γ is adjustable parameter, Value is generally the half of signal wavelength, and γ controls the size of elliptic region.The weight that all links obtain is write For matrix form W ∈ R^η×M, and it is called weight matrix；Weight matrix W is the matrix of η × M, each Row represents the weight of a link, whole M pictures under M element representation nth bar link of such as line n The weight of vegetarian refreshments.

Step 5.2: utilize compressed sensing regularization method to realize target and position:

Considering all η bar links, the matrix form of LOS path power variation is

Δ r=W Δ x+n；

Wherein Δ r ∈ R^η×1It is observation vector, is by the LOS path signal power variations of η bar linkGroup Become, W ∈ R^η×MIt is the weight matrix of step 5.1 acquisition, n ∈ R^η×1It is noise vector, N=[n₁,n₂,n₃…,n_η]^TWherein element is the η respective n of bar link_(i,j)；Δx∈R^M×1It is vector to be estimated, The signal attenuation value on M grid that i.e. monitored area divides.In view of element non-negative in Δ x, based onModel The regularization object function that number compressed sensing is rebuild can be written as:

$\begin{array}{c}\underset{\mathrm{\&Delta;}x}{\min }\left|\right|\mathrm{\&Delta;}r-W\mathrm{\&Delta;}x|{|}^2+\mathrm{\&mu;}\underset{m=1}{\overset{M}{\&Sigma;}}{\mathrm{\&Delta;x}}^m\\ \begin{array}{cc}s.t.& {\mathrm{\&Delta;x}}^m\&GreaterEqual;0,m=1,2,...,M\end{array}\end{array};$

Wherein μ is regularization parameter.Object function can be asked by existing method effectively by convex optimum theory Solve, the grid being estimated as decay power maximum of target's center positionIn monitored area Coordinate.

Test result indicate that testing of ultra broadband object localization method based on the estimation of direct path shadow fading In position error be 0.21m.In experiment, the result of node self-localization as shown in Figure 4, based on ultra-wide belt segment The schematic diagram of the DFL locating effect of spot net is as shown in Figure 5.Based on narrow band signal RSS compared to tradition DFL method, positioning precision is greatly improved, and achieves the self-align of node, greatly reduces The workload of node deployment.

The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited to This, any be familiar with the people of this technology in the technical scope that disclosed herein, it will be appreciated that the conversion expected and Replacing, all should contain within the scope of the comprising of the present invention, therefore, protection scope of the present invention should be with power The protection domain of profit claim is as the criterion.

Claims (9)

1. based on ultra broadband meshed network exempt from Portable device localization method, it is characterised in that include walking as follows Rapid:

Step one: node deployment

K random being deployed in around monitored area of ultra-wideband detection node, this K node is positioned at same water In plane, it is assumed that the coordinate of i-th node is x_i=(x_i,y_i)^T, i=1,2 ..., K；Each node can Send and receive wireless signal, constituting η=K (K-1) bar link altogether；Additionally there are a base-station node, It is responsible for receiving and measures the channel impulse response CIR data that record of node, and data are issued PC do subsequent treatment； Respectively measure the wireless signal that node sent and received and be narrow pulse signal；

Step 2: measure when monitored area does not has a target channel of link between i-th node and jth node Impulse response CIR, is designated asI=1,2 ..., K；J=1,2 ..., K；J ≠ i, and extract LOS path signal merit RateWith LOS path signal TOA, comprise the steps:

Step 2.1: monitored area keeps spaciousness, measures the CIR of the link constituted between any two nodes, It is designated asI=1,2 ..., K；J=1,2 ..., K；j≠i；

Step 2.2: extract LOS path signal power from the multipath signal power of each of the links:

Use Edge Detection, it is believed that first CIR peak value correspondence LOS path letter exceeding given threshold value Number, i.e.

${\stackrel{\&OverBar;}{c}}_{(i,j)}^{{\hat{k}}^0}=\arg \underset{n}{min}\{P_n>P_{threshold},P_n\&Element;P_{(i,j)}\}$

Wherein P_thresholdFor the threshold value set, be typically based between false-alarm and missing inspection compromise determines；P_(i,j)It is CIR measured valueThe set of middle peak value, n is set P_(i,j)The sequence number of middle peak value element,For LOS road Footpath CIR peak value is at set P_(i,j)In sequence number, then the LOS of link between node i and j when driftlessness blocks The power of path signal is

Step 2.3: after obtaining the CIR peak value of LOS path, the time judgement that LOS path peak value is arrived The TOA time of advent for signal；Calculate distance d between transmitting-receiving node i and j_i,j；

Step 3: carry out node self-localization according to the calculated euclidean distance between node pair of step 2, including walking as follows Rapid:

Step 3.1: obtained following square distance matrix by the distance of every pair of node, size is K × K；

Construct following matrix:

$B=-\frac{1}{2}JDJ;$

WhereinE_KIt is K rank unit matrixs, 1_KBe element value be all 1 K × 1 big Little column vector；

The Eigenvalues Decomposition of B is

B=V Λ V^T；

Wherein Λ=diag (λ₁,λ₂,....,λ_K) be B eigenvalue constitute diagonal matrix, eigenvalue is arranged in descending order Row, V=(v₁,v₂,....,v_K)^TThe matrix constituted for individual features value characteristic of correspondence vector；

Step 3.2: the overall barycenter constituted by K node is set to initial point, i.e.X The matrix constituted for the coordinate of all nodes, i.e.Make V₂=(v₁,v₂), Λ₂=diag (λ₁,λ₂), then node location is estimated as

$\hat{X}=V_2{\mathrm{\&Lambda;}}_2^{1/2};$

I.e.For the estimation of node coordinate matrix X, be size be the matrix of K*2, footmark 1/2 represents opens radical sign；

Step 4: to link between every pair of node i and j, obtains owing to barrier blocks the LOS path caused Power change values

In monitored area, node measurement has link cir value c when blocking_(i,j), and according to step 2.2 method Extract now LOS path CIR peak valueMeasurement to unobstructed environment in integrating step two, without mesh When mark blocks, between node i and j, the LOS path CIR peak value of link isThen cause due to target occlusion Node i and j between the changed power of link LOS path be:

${\mathrm{\&Delta;r}}_{(i,j)}^{LRSS}=10\lg {\stackrel{\&OverBar;}{c}}_{(i,j)}^{{\hat{k}}^0}-10\lg c_{(i,j)}^{{\hat{k}}^0};$

Step 5: use Portable device localization method of exempting from based on compressed sensing to realize target imaging:

Step 5.1: according to distance d between every pair of node of the self-align measurement of step 3 interior joint_i,jAnd estimation Each node locationCalculate weight matrix: method is as follows:

The method using wireless tomography in the present invention realize target location, monitored area be divided evenly into Square net, makes Δ x^mFor signal through the decay of m-th grid in monitored area, have M grid, With weightRepresent that the signal attenuation of m-th grid is to the tribute of the signal attenuation of link between node i and j Offer, n_(i,j)For the noise of link between node i and j, then between node i and j, the signal attenuation of link is permissible Aggregate form is expressed as:

${\mathrm{\&Delta;r}}_{(i,j)}^{LRSS}=\underset{m=1}{\overset{M}{\&Sigma;}}{w}_{(i,j)}^m{\mathrm{\&Delta;x}}^m+n_{(i,j)}$

Model of ellipse is used to calculate weight, as follows

$w_{(i,j)}^m=\left\{\begin{array}{c}1,d_{(i,j)}^m\left(1\right)+d_{(i,j)}^m\left(2\right)<d_{(i,j)}+\mathrm{\&gamma;}\\ 0\end{array}\right.$

WhereinWithIt is respectively the m grid distance to transmitting-receiving node, whereinTable Show receiving node,Represent to sending node, d_(i,j)For the distance between transmitting-receiving node；γ is can Adjusting parameter, control the size of elliptic region, value is generally the half of signal wavelength；All links are obtained WeightIt is written as matrix form W ∈ R^η×M, and it is called weight matrix；

Step 5.2: utilize compressed sensing regularization method to realize target and position:

Considering all η bar links, the matrix form of LOS path power variation is

Δ r=W Δ x+n；

Wherein Δ r ∈ R^η×1It is observation vector, is by the LOS path signal power variations of η bar linkGroup Become, W ∈ R^η×MIt is the weight matrix of step 5.1 acquisition, n ∈ R^η×1It is noise vector, wherein element N=[n₁,n₂,n₃…,n_η]^TIt is η bar link respective noise n_(i,j)；Δx∈R^M×1It is vector to be estimated, i.e. supervises Survey the signal attenuation value on M the grid that region divides；In view of element Δ x in Δ x^mNon-negative, based onModel The regularization object function that number compressed sensing is rebuild can be written as:

$\underset{\mathrm{\&Delta;}x}{\min }\left|\right|\mathrm{\&Delta;}r-W\mathrm{\&Delta;}x|{|}^2+\mathrm{\&mu;}\underset{m=1}{\overset{M}{\&Sigma;}}{\mathrm{\&Delta;x}}^m;$

s.t.Δx^m>=0, m=1,2 ..., M

Wherein μ is regularization parameter；The grid being estimated as decay power maximum of target's center positionCoordinate in monitored area.

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature Being, narrow pulse signal described in step one, the persistent period is ns level or sub-ns level.

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature Being, measuring node described in step one is ultra broadband node based on IEEE802.15.4a agreement.

Based on ultra broadband meshed network the most according to claim 3 exempt from Portable device localization method, its feature Being, the operating center frequency of ultra broadband node is 6.489GHz, carries a width of 499MHz, each node Assembling omnidirectional antenna.

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature It is, in step 5.2, solves regularization object function by convex optimum theory.

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature It is, in step 2.1, the link constituted between any two nodes is carried out bidirectional ranging, this mistake successively Journey is automatically performed by predetermined communication protocol by each node.

Based on ultra broadband meshed network the most according to claim 6 exempt from Portable device localization method, its feature Being, bidirectional ranging completes according to TDMA agreement.

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature Be, in step 2.3, by the method for symmetrical bilateral two-path DME calculate between transmitting-receiving node i and j away from From d_i,j。

Based on ultra broadband meshed network the most according to claim 1 exempt from Portable device localization method, its feature Being, described barrier uses cylinder model.