CN102938876A - Passive target tracking method based on received signal strength of wireless sensor network - Google Patents

Abstract

The invention relates to a passive target tracking method based on received signal strength of a wireless sensor network. The method has the following technical characteristics that: the method comprises the following steps that: a node in the wireless sensor network broadcasts data, and all the nodes receive and record the data; the node calculates the absolute value of the difference value between the received signal strength value and the static network data, and calculates the target position likelihood through a passive target tracking model; and the nodes track the target in real time in combination with the target position likelihood through a particle filter algorithm and a parameter estimation algorithm. In the method provided by the invention, the characteristic that the radio frequency signal sent by the wireless sensor network has certain penetrability is fully used, little barriers cannot block the mutual communication between the nodes, and the transmission of the wireless signals cannot be affected by illumination, so that the method can be widely used at night or used for the indoor scenes lack of light, such as a warehouse and a basement.

Description

Passive target tracking method based on the wireless sensor network received signal strength

Technical field

The invention belongs to the wireless senser field, especially a kind of passive target tracking method based on the wireless sensor network received signal strength.

Background technology

Along with the extensive use of wireless network, people also increase day by day to the demand of target localization and tracking technique.Except need under outdoor environment, carrying out real-time location and tracking to mobile personnel, vehicle, in the indoor environment of complexity, especially also often need the positional information of acquisition personnel or article in high-end residential residential quarter, exhibition room, warehouse, supermarket, the underground parking.Owing to be subject to the restriction of complexity and the condition such as uncertain of indoor environment, traditional locating and tracking system, for example GPS navigation system and video monitoring system all have self limitation.The GPS navigation system can be subject to following 2 constraints when indoor use: on the one hand, the GPS receiver must be avoided being hidden from the top, guarantees with satellite communication unimpeded; On the other hand, the multipath transmisstion of satellite-signal will exert an influence to the process of clock synchronous.And video capture device can only gather the image information in the sight line, so can not have blocking of barrier between camera and the target during monitoring; On the other hand, the imaging performance of video capture device can be subject to the impact of illumination, and for example under the environment of high light or dark, the performance of obtaining image all can reduce.Therefore traditional track algorithm real-time of under indoor environment, locating, accuracy and all can't be guaranteed to the adaptivity of environment.

Wireless sensor network is the current hot research field that receives much concern, and is a revolution of information gathering and perception, can be widely used in the security monitoring field of military affairs, environmental monitoring and forecast, Smart Home, urban transportation and corridor and factory.Wherein, the location is one of main direction of studying of wireless sensor network with following the tracks of, mainly be to carry out the target perception by means of the employed radiofrequency signal of communication in the wireless sensor network, be also referred to as the localization method based on radio frequency receiving signal intensity, it judges the residing position of target according to the signal strength values that known node is sent.Yet existing wireless sensor network locating method great majority based on received signal strength are active location, and namely the necessary carry sensors node of target object receives the data that known node is sent.But the passive target (passive target) that can not be used for of the method is followed the tracks of, and the characteristics of passive target are: target can initiatively not carried out data interaction with tracking system, even might tracked target itself not wish to be traced into by system.Therefore, prior art can't effectively be followed the tracks of passive target.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, a kind of passive target tracking method based on the wireless sensor network received signal strength reasonable in design, simple and effective is provided.

The present invention solves its technical problem and takes following technical scheme to realize:

A kind of passive target tracking method based on the wireless sensor network received signal strength may further comprise the steps:

Step 1: the node in the wireless sensor network carries out data broadcast, and all nodes are answered and record data;

Step 2: node calculates the absolute value of the difference of received signal strength value and static network data, carries out the target location likelihood score by passive target tracking model and calculates;

Step 3: with particle filter algorithm and parameter estimation algorithm and combining target position likelihood score, carry out the real-time tracing of target.

And the data that described node is answered and recorded comprise received signal strength, signal source and recipient information.

And, described target location likelihood score computational methods are: at first, the relative position relation of every link in the possible position of calculating target and the wireless sensor network, then, calculating is in the difference of the absolute value of the theory signal Strength Changes value of this relative position and actual signal Strength Changes value, thus the calculated target positions likelihood score.

And described step 3 comprises following processing procedure:

⑴ initialization: give parameter θ ₀To any initial value, counter b is made as 0, from the prior probability distribution p (x of state ₀) particle that does well of sampling

⑵ arrange counter k=1,2 ..., N, N are population;

⑶ executing state is upgraded:

$x_{k+1}^{\left(n\right)}=f\left(x_k^{\left(n\right)}\right)+v_k\·y_k^{\left(n\right)}=h\left(x_k^{\left(n\right)}\right)+s_k$

F is system model, v _kBe the k system noise in step, h is measurement model, s _kIt is the k measurement noise in step.

⑷ particle filter, the computing mode estimation particle

And weight

θ wherein _B-1Be the b-1 parameter value in step,

Be the posterior probability formula of the state in b-1 step, Z _kBe the k actual observed value in step, normalization ${\mathrm{\ρ}}_k^{\left(n\right)}=\frac{{\mathrm{\ρ}}_k^{\left(n\right)}}{\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\ρ}}_k^{\left(n\right)}};$

Again sampling To obtain new even weight

From

In sample out

Weight is after upgrading

Normalization ${\mathrm{\ρ}}_k^{\left(m\right)}=\frac{{\mathrm{\ρ}}_k^{\left(m\right)}}{\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\ρ}}_k^{\left(m\right)}};$

⑸ carry out and expect maximization steps (onl ine EM algorithm) on the line:

${\widehat{\mathrm{\Ω}}}_b=(1-{\mathrm{\α}}_b){\widehat{\mathrm{\Ω}}}_{b-1}+{\mathrm{\α}}_b\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{W}_b^{\left(m\right)}\mathrm{\Ψ}(X_b^{\left(m\right)},Z_b),$ In the formula:

Ψ is sufficient statistic, and Ω is the set of sufficient statistic, the sufficient statistic weight Satisfy

$W_b^{\left(m\right)}\∝\frac{p_{{\mathrm{\θ}}_{b-1}}\left(x_b^{\left(m\right)}\right|Z_b)}{q_{{\mathrm{\θ}}_{b-1}}\left(x_b^{\left(m\right)}\right|Z_b)},$ ${\mathrm{\α}}_b=\frac{1}{\sqrt{b}}$ With $\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{W}_b^{\left(m\right)}=1;$

⑹ arrange variable count=k mod L, and L is the Cycle Length of expectation maximization steps on the line; If count=0 then carries out maximization and processes:

Counter b adds 1;

⑺ circulation execution in step 2 to ⑹ is until all processing end of all particles.

Advantage of the present invention and good effect are:

1, the present invention takes full advantage of the characteristics that radiofrequency signal that wireless sensor network sends has certain penetrability, can not stop mutually intercommunication between the node for a small amount of barrier, and the propagation of wireless signal is not subjected to illumination effect.Therefore, so that wireless sensor network can pass through the stopping of object such as wall, dense smoke people and object are positioned and follow the tracks of, and can be applied to night or or the indoor scene of the shortage illumination such as warehouse, basement.

2, the present invention adopts wireless sensor network, does not need the tracked target carry sensors to cooperate, nor need to the shape of surveyed area be limited, and can carry out quickly and efficiently target tracking.

3, the present invention can carry out the self adaptation operation to environment, namely by method for parameter estimation such as expectation maximizations on the line (online EM), automatically estimates environmental parameter, obtains high accuracy under varying environment.

Description of drawings

Fig. 1 is process chart of the present invention;

Fig. 2 is wireless sensor network distribution schematic diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing the present invention is further described.

A kind of passive target tracking method based on the wireless sensor network received signal strength as shown in Figures 1 and 2, may further comprise the steps:

Step 1: the node in the wireless sensor network carries out data broadcast, and all nodes are answered and record data.

The data that this step node is collected comprise received signal strength and the information such as signal source and recipient's (namely broadcasting the numbering of node number and the receiving node of this signal).

Step 2: node calculates the absolute value of the difference of received signal strength value and static network data, carries out the target location likelihood score by passive target tracking model and calculates.

In this step, the use of absolute value can be offset indoor multi-path influence.Described target location likelihood score computational methods are: at first, the relative position relation of every link in the possible position of calculating target and the wireless sensor network, then, calculating is in the difference of the absolute value of the theory signal Strength Changes value of this relative position and actual signal Strength Changes value, thus the calculated target positions likelihood score.

Step 3: with particle filter algorithm and parameter estimation algorithm and combining target position likelihood score, carry out (in real time) tracking on the line of target.

The concrete processing procedure that this step comprises is:

⑴ initialization: give parameter θ ₀To any initial value, counter b is made as 0, from the prior probability distribution p (x of state ₀) particle that does well of sampling

⑵ arrange counter k=1,2 ..., N, N are population;

⑶ executing state is upgraded:

$x_{k+1}^{\left(n\right)}=f\left(x_k^{\left(n\right)}\right)+v_k\·y_k^{\left(n\right)}=h\left(x_k^{\left(n\right)}\right)+s_k$

F is system model, v _kBe the k system noise in step, h is measurement model, s _kIt is the k measurement noise in step.

⑷ particle filter, the computing mode estimation particle

And weight

θ wherein _B-1Be the b-1 parameter value in step,

Be the posterior probability formula of the state in b-1 step, Z _kBe the k actual observed value in step, normalization ${\mathrm{\ρ}}_k^{\left(n\right)}=\frac{{\mathrm{\ρ}}_k^{\left(n\right)}}{\underset{n=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\ρ}}_k^{\left(n\right)}};$

Again sampling

To obtain new even weight

From In sample out

Weight is after upgrading

Normalization ${\mathrm{\ρ}}_k^{\left(m\right)}=\frac{{\mathrm{\ρ}}_k^{\left(m\right)}}{\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\ρ}}_k^{\left(m\right)}};$

⑸ carry out and expect maximization steps (onl ine EM algorithm) on the line:

${\widehat{\mathrm{\Ω}}}_b=(1-{\mathrm{\α}}_b){\widehat{\mathrm{\Ω}}}_{b-1}+{\mathrm{\α}}_b\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{W}_b^{\left(m\right)}\mathrm{\Ψ}(X_b^{\left(m\right)},Z_b),$ In the formula:

Ψ is sufficient statistic, and Ω is the set of sufficient statistic, the sufficient statistic weight

Satisfy

$W_b^{\left(m\right)}\∝\frac{p_{{\mathrm{\θ}}_{b-1}}\left(x_b^{\left(m\right)}\right|Z_b)}{q_{{\mathrm{\θ}}_{b-1}}\left(x_b^{\left(m\right)}\right|Z_b)},$ ${\mathrm{\α}}_b=\frac{1}{\sqrt{b}}$ With $\underset{m=1}{\overset{N}{\mathrm{\Σ}}}{W}_b^{\left(m\right)}=1;$

⑹ arrange variable count=k mod L, and L is the Cycle Length of expectation maximization steps on the line.If count=0 then carries out maximization and processes:

Counter b adds 1;

⑺ circulation execution in step 2 to ⑹ is until all processing end of all particles.

It is emphasized that; embodiment of the present invention is illustrative; rather than determinate; therefore the present invention includes and be not limited to the embodiment described in the embodiment; every other execution modes that drawn by those skilled in the art's technical scheme according to the present invention belong to the scope of protection of the invention equally.

Claims (4)

1. passive target tracking method based on the wireless sensor network received signal strength is characterized in that: may further comprise the steps:

Step 1: the node in the wireless sensor network carries out data broadcast, and all nodes are answered and record data;

Step 2: node calculates the absolute value of the difference of received signal strength value and static network data, carries out the target location likelihood score by passive target tracking model and calculates;

Step 3: with particle filter algorithm and parameter estimation algorithm and combining target position likelihood score, carry out the real-time tracing of target.

2. the passive target tracking method based on the wireless sensor network received signal strength according to claim 1 is characterized in that: the data that described node is answered and recorded comprise received signal strength, signal source and recipient information.

3. the passive target tracking method based on the wireless sensor network received signal strength according to claim 1, it is characterized in that: described target location likelihood score computational methods are: at first, the relative position relation of every link in the possible position of calculating target and the wireless sensor network, then, calculating is in the difference of the absolute value of the theory signal Strength Changes value of this relative position and actual signal Strength Changes value, thus the calculated target positions likelihood score.

4. the passive target tracking method based on the wireless sensor network received signal strength according to claim 1, it is characterized in that: described step 3 comprises following processing procedure:

⑴ initialization: give parameter θ ₀To any initial value, counter b is made as 0, from the prior probability distribution p (x of state ₀) particle that does well of sampling

⑵ arrange counter k=1,2 ..., N, N are population;

⑶ executing state is upgraded:

F is system model, v _kBe the k system noise in step, h is measurement model, s _kIt is the k measurement noise in step.

⑷ particle filter, the computing mode estimation particle

And weight θ wherein _B-1Be the b-1 parameter value in step,

Be the posterior probability formula of the state in b-1 step, Z _kBe the k actual observed value in step, normalization

Again sampling

To obtain new even weight

From

In sample out

Weight is after upgrading Normalization

⑸ carry out and expect maximization steps (onl ine EM algorithm) on the line:

In the formula:

Ψ is sufficient statistic, and Ω is the set of sufficient statistic, the sufficient statistic weight

Satisfy

With

⑹ arrange variable count=k mod L, and L is the Cycle Length of expectation maximization steps on the line; If count=0 then carries out maximization and processes:

Counter b adds 1;

⑺ circulation execution in step 2 to ⑹ is until all processing end of all particles.

Images