CN105530702A - Wireless sensing network mobile node positioning method based on self-organizing mapping - Google Patents

Abstract

The invention discloses a wireless sensing network mobile node positioning method based on self-organizing mapping, comprising steps of collecting fixed nodes in the indoor environment, a corresponding RSSI value between two mobile nodes to be positioned and a distance d, using the collected RSSI value and the distance d to train a one-dimension self-organizing mapping network, establishing a relation model between the RSSI value and the distance d, inputting the received RSSI value into the function to obtain the optimization distance d between the fixed node and the mobile node in the following practical usage, and obtaining three strongest fixing nodes of the RSSI to obtain three distances d with the mobile nodes, and obtaining the mobile node coordinate through the classic three-edge positioning method. Compared with the prior art, the method training based on one-dimension self-organizing mapping is short in study period and good in robustness and can further improve the positioning accuracy of the wireless sensor network.

Description

A kind of radio sensing network mobile node positioning method based on Self-organizing Maps

Technical field

The invention belongs to technology of wireless sensing network field, be specifically related to a kind of radio sensing network mobile node positioning method based on Self-organizing Maps.

Background technology

The successful Application of global positioning system makes the demand of people to positioning service increasing, but under indoor environment, propagate due to satellite-signal and be subject to building severe jamming, the precision of location can not meet application requirement, so people start to find replacement scheme to meet the demand of wireless location under indoor environment.Wireless sensor network, because its flexibility, cost are low, be easy to the characteristics such as layout, can facilitate, gather various information in time, accurately, and be subject to applying more and more widely.According in wireless sensor network positioning process the need of the distance measured between actual node, node locating can be divided into the mode based on range finding (range-based) and non-ranging (range-free) to carry out.Based in location algorithm location, during measured node spacing, adopt TOA (timeofarrive), TDOA (timedifferenceofarrive), RSSI (receivedsignalstrengthindication) etc.Because the location algorithm of range finding can provide higher required precision, the location algorithm based on range finding will obtain better development space in node locating technique.

In the location algorithm based on range finding, RSSI, due to cheap, obtains extensive use.But RSSI algorithm propagates the distance of empirical model calculating usually according to signal, comparatively large with actual conditions deviation, and in different environments, the parameter of model all needs to adjust, poor to the adaptive capacity of environment.Under indoor positioning environment, it is cause the main cause of indoor positioning error that inaccurate RSSI and radio signal propagation calculate distance, causes the use of the location algorithm based on RSSI to have significant limitation.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, the radio sensing network node mobile node positioning method based on Self-organizing Maps is provided.

For achieving the above object, a kind of radio sensing network node mobile node positioning method based on Self-organizing Maps, comprises the following steps:

Based on a radio sensing network mobile node positioning method for Self-organizing Maps, specifically comprise the following steps:

Steps A: the stationary nodes number disposed in radio sensing network is more than 4, then in place, location, gather stationary nodes and the mobile node training sample data at desired location, training sample data comprise mobile node to the received signal strength RSSI value of stationary nodes and distance value accordingly;

Step B: set up self organizing maps model, using the input value of the RSSI value of the stationary nodes that obtains in steps A and mobile node as training sample, data processing centre calculates the output of the actual ranging data between the stationary nodes of described fixed position and each mobile node as training sample; Described self organizing maps model is trained; Obtain the relational model of RSSI value and distance;

Step C: after mobile node to be positioned enters the locating area of radio sensing network, send wireless signal to stationary nodes, the RSSI signal received is uploaded to data processing centre by stationary nodes;

Step D: data processing centre using the RSSI value that receives as sample, be input in the self organizing maps model that step B trains, obtain mobile node and the distance d of corresponding stationary nodes according to this, preserve the position coordinates of these distance d lower and corresponding received signal strength and stationary nodes;

Step e: the distance value obtained according to step D, arranges RSSI value, selects value maximum in 3 RSSI value, the distance of record move node to stationary nodes and the coordinate of stationary nodes, adopts node locating method to obtain the coordinate of mobile node, completes location.

Described self organizing maps model adopts One dimensional Self organization mapping algorithm.

The input neuron number of described One dimensional Self organization mapping algorithm is consistent with the stationary nodes quantity of output neuron number and actual location.

The self study speed of described One dimensional Self organization mapping algorithm is automatic renewal.

Described node locating method adopts classical three limit ranging localization algorithms, obtains the position of unknown node according to the ranging data after error correction.

The present invention has following beneficial effect:

Compared with prior art, the present invention adopts One dimensional Self organization mapping algorithm to carry out error correction to RSSI, utilize the ability of interneuronal interaction and the characteristic of vying each other and self-organizing and study in the network of Self-organizing Maps algorithm, to be gone by intensity RSSI and distance d to received signal to train after the self organizing maps model success of setting up for different localizing environments, effectively improve the node locating precision of radio sensing network.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of embodiment of the present invention;

Fig. 2 is One dimensional Self organization mapping structure schematic diagram.

Embodiment

As shown in Figure 1, the present invention, on the basis researching and analysing wireless signal propagation model and traditional localization method, proposes and maps based on One dimensional Self organization the distance optimizing indoor node to be positioned and stationary nodes, thus reaches the object improving positioning precision.Below in conjunction with accompanying drawing, technical scheme of the present invention is described in detail:

Gather training sample

Adopt One dimensional Self organization mapping algorithm to obtain the relational model of RSSI value and distance in the present invention, need first to gather training sample to train One dimensional Self organization mapping, in place, wireless sensor network location, first utilize stationary nodes and mobile node to obtain training sample data, comprise mobile node to the received signal strength RSSI value of stationary nodes and the two distance value.

Adopt N (N >=4) individual stationary nodes and 1 mobile node to obtain training sample data in the present embodiment, with 0.2 meter for distance increment measures the RSSI value that in real training place, mobile node is corresponding with stationary nodes 0 ~ 5 meter distance, the mode be simultaneously averaging by multi collect data in implementation process reduces the error of single distance, tests 10 average seek distances in the present embodiment.

As shown in Figure 2, One dimensional Self organization maps training sample

Set up One dimensional Self organization mapping structure network,

Training sample vector set is expressed as X=(X ₁, X ₂..., X _n), network has N number of input node, and competition layer has Q neuron, is W by input layer to the connection weights of competition layer _ij, i=[1 ..., Q], j=[1 ..., N], initialization is connected weights W _ijgive random value and be normalized, obtaining W _ij(0); Determine initial learning rate α (0), (0 < α (0) < 1); Determine neighborhood N _rthe initial value N of (t) _r(0), determine total study number of times T simultaneously.

1) from training set, trained vector X is selected _k, k ∈ [1 ..., N], and be normalized, each neuron of competition layer is input to by parallel mode.

2) X is calculated _kwith each neuron (i.e. W _ij) between Euclid distance d _i, select to have in the topological neighborhood of neuron g and g of minimum range, by the neuronic weights in the topological neighborhood of formula (2) adjustment neuron g and g, other neuron weights remain unchanged, that is:

$d_i=\underset{0<i<Q}{\min }|\underset{j=0}{\overset{N}{\&Sigma;}}\&lsqb;X_{kj}-W_{ij}\left(t\right)\&rsqb;|---\left(1\right)$

W _ij(t+1)＝W _ij(t)+α(t)[X _k-W _ij(t)](2)

Wherein i ∈ NE _j(t), NE _j(t) for the topological neighborhood of triumph neuron g, t be current iteration number of times.α (t) is the learning rate factor, generally elects as:

$\mathrm{\&alpha;}\left(t\right)={\mathrm{\&alpha;}}_0\&times;(1-\frac{t}{t_{\max }})---\left(3\right)$

Wherein, t _maxfor total iterations, α ₀get [0,1].

3) to all training input patterns, repeat step 2), 3), until algorithmic statement or reach the maximum iteration time of initial setting.

The RSSI value of collection by training One dimensional Self organization mapping network, is obtained the relational model of RSSI value and distance as output as input, corresponding distance value.

Mapped by One dimensional Self organization and obtain distance value

After mobile node to be positioned enters positioned radio sensor network, the RSSI value of the signal receiving mobile node is input to the One dimensional Self organization mapping network that second step has trained by stationary nodes, obtains corresponding mobile node and the distance of stationary nodes

Three location, limits

By arranging RSSI value, select 3 values that RSSI value is the strongest, the distance of record move node to stationary nodes and the coordinate of stationary nodes, if the coordinate of three stationary nodes A, B, C is respectively (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) to mobile, they know that the distance of node D is respectively: d ₁, d ₂, d ₃, the coordinate of unknown node D is set to (x, y).Can following equations be obtained:

$\left\{\begin{array}{c}{d_1}^2={(x-x_1)}^2+{(y-y_1)}^2\\ {d_2}^2={(x-x_2)}^2+{(y-y_2)}^2\\ {d_3}^2={(x-x_3)}^2+{(y-y_3)}^2\end{array}\right\}$

The equation in coordinates that can obtain unknown node D according to above formula is:

$\left[\begin{array}{c}x\\ y\end{array}\right]={\left[\begin{array}{cc}2(x_1-x_3)& 2(y_1-y_3)\\ 2(x_2-x_3)& 2(y_2-y_3)\end{array}\right]}^{-1}\left[\begin{array}{c}{x}_1^2-x_3^2+y_1^2-y_3^2+d_3^2-{d_1}^2\\ x_1^2-x_3^2+y_1^2-y_3^2+d_3^2-{d_1}^2\end{array}\right]$

Claims (5)

1., based on a radio sensing network node mobile node positioning method for Self-organizing Maps, it is characterized in that, the method specifically comprises the following steps:

Based on a radio sensing network mobile node positioning method for Self-organizing Maps, specifically comprise the following steps:

Steps A: the stationary nodes number disposed in radio sensing network is more than 4, then in place, location, gather stationary nodes and the training sample data of mobile node in fixed position, training sample data comprise mobile node to the received signal strength RSSI value of stationary nodes and corresponding distance value;

Step B: set up self organizing maps model, using the input value of the RSSI value of the stationary nodes that obtains in steps A and mobile node as training sample, data processing centre calculates the output of the actual ranging data between the stationary nodes of described fixed position and each mobile node as training sample; Described self organizing maps model is trained; Obtain the relational model of RSSI value and distance;

Step C: after mobile node to be positioned enters the locating area of radio sensing network, send wireless signal to stationary nodes, the RSSI signal received is uploaded to data processing centre by stationary nodes;

Step D: data processing centre using the RSSI value that receives as sample, be input in the self organizing maps model that step B trains, obtain mobile node and the distance d of corresponding stationary nodes according to this, preserve the position coordinates of these distance d lower and corresponding received signal strength and stationary nodes;

Step e: the distance value obtained according to step D, arranges RSSI value, selects value maximum in 3 RSSI value, the distance of record move node to stationary nodes and the coordinate of stationary nodes, adopts node locating method to obtain the coordinate of mobile node, completes location.

2. a kind of radio sensing network node mobile node positioning method based on Self-organizing Maps according to claim 1, is characterized in that: described self organizing maps model adopts One dimensional Self organization mapping algorithm.

3. a kind of radio sensing network node mobile node positioning method based on Self-organizing Maps according to claim 2, is characterized in that: the input neuron number of described One dimensional Self organization mapping algorithm is consistent with the stationary nodes quantity of output neuron number and actual location.

4. a kind of radio sensing network node mobile node positioning method based on Self-organizing Maps according to Claims 2 or 3, is characterized in that: the self study speed of described One dimensional Self organization mapping algorithm is automatic renewal.

5. a kind of radio sensing network node mobile node positioning method based on Self-organizing Maps according to claim 1, it is characterized in that: described node locating method adopts classical three limit ranging localization algorithms, obtains the coordinate position of unknown node according to the ranging data after error correction.