CN102685772B - Tracking node selection method based on wireless all-around sensor network - Google Patents

Abstract

The invention relates to a tracking node selection method based on a wireless all-around sensor network. The tracking node selection method comprises the following steps of: acquiring a transcendental state probability on the basis of particle filtering; obtaining a longest distance and a shorted distance from each node in a detection range to a target according to position and speed estimation at a previous moment and estimation factors; calculating a corresponding weight value of each particle with combination of a measurement value of the node, and weighting to obtain a reliable set of the measurement values of the nodes in the detection range; and selecting a reliable measurement value from the set as the basis for tracking estimation. The method can avoid possible tracking mistakes or target loss caused by mis-measurement.

Description

A kind of tracking node selecting method based on wireless omnidirectional sensor network

Technical field

The present invention relates to wireless sensor network technology field, particularly relate to a kind of tracking node selecting method based on wireless omnidirectional sensor network, be applicable to utilize omnidirectional's transducer tracing process interior joint such as shock sensor, sound transducer, radio-frequency antenna to select.

Background technology

Internet of Things (Internet of Things) represents the following direction with development communication technologies that calculates, and is considered to after computer, Internet, the third time development tide in information industry field.Internet of Things be one can (Anytime), place (Anyplace) at any time, realize the dynamic network that any object (Anything) is interconnected, it include between PC, interpersonal, interconnected between thing and people, between thing and thing.Internet of Things comprises sensing layer, network layer and application layer 3 levels.Wireless sensor network (WSN) is one of key technology of sensing layer employing.

In the application study of WSN, target locating is an important research direction.The sensor model of target locating is based on sound transducer, shock sensor, radio-frequency antenna etc. mostly.In tracing process, in fact due to multi-path jamming, block, the factor such as environment, cause indivedual signal value obtained to depart from actual value comparatively large, become mistake and measure, and mistake to measure be not that certain noise causes.If do not consider that in target locating the mistake that may exist is measured, the mistake of target information is very easily caused to estimate even track rejection.

Summary of the invention

Technical problem to be solved by this invention is to provide a kind of tracking node selecting method based on wireless omnidirectional sensor network, the trail-and-error avoiding mistake measurement to bring or track rejection.

The technical solution adopted for the present invention to solve the technical problems is: provide a kind of tracking node selecting method based on wireless omnidirectional sensor network, comprise the following steps:

(1) on the basis of particle filter, prior state probabilities is obtained;

(2) utilize the position of previous moment, velocity estimation and the estimation factor, obtain each node in investigative range to the maximum distance of target and minimum distance;

(3) measured value in conjunction with node itself calculates the corresponding weights of each particle, and weighting obtains the set of investigative range interior nodes measurement reliability;

(4) from described set, select reliable measured value to estimate foundation as tracking.

The method of Bayesian Estimation is adopted to obtain prior state probabilities in described step (1).

Three measured values are selected to estimate foundation as tracking in described step (4).

Beneficial effect

Owing to have employed above-mentioned technical scheme, the present invention compared with prior art, there is following advantage and good effect: the present invention is on particle filter basis, adopt Bayesian Estimation, obtain prior probability, utilize the position of previous moment, velocity estimation and the estimation factor, obtain each node in investigative range to target farthest and minimum distance, measured value again in conjunction with node itself calculates the corresponding weights of each particle, then weighting obtains the set of scope interior nodes measurement reliability, from this set, select reliable measured value to estimate foundation as tracking, thus can when there is erroneous measurements, select reliable node measurement value as tracking foundation, thus the trail-and-error avoiding mistake measurement to bring or track rejection.

Accompanying drawing explanation

Fig. 1 is flow chart of the present invention;

Fig. 2 is measurement point and maneuvering target graph of a relation in embodiment of the present invention;

Fig. 3 is the potential range schematic diagram of measurement point and maneuvering target in embodiment of the present invention.

Embodiment

Below in conjunction with specific embodiment, set forth the present invention further.Should be understood that these embodiments are only not used in for illustration of the present invention to limit the scope of the invention.In addition should be understood that those skilled in the art can make various changes or modifications the present invention, and these equivalent form of values fall within the application's appended claims limited range equally after the content of having read the present invention's instruction.

Embodiments of the present invention relate to a kind of tracking node selecting method based on wireless omnidirectional sensor network, as shown in Figure 1, comprise the following steps:

(1) on the basis of particle filter, prior state probabilities is obtained;

(2) utilize the position of previous moment, velocity estimation and the estimation factor, obtain each node in investigative range to the maximum distance of target and minimum distance;

(3) measured value in conjunction with node itself calculates the corresponding weights of each particle, and weighting obtains the set of investigative range interior nodes measurement reliability;

(4) from described set, select reliable measured value to estimate foundation as tracking.

Specifically, k moment prior state probabilities p (x is first obtained _k| y _1:k-1), wherein, x _krepresent status switch, y _krepresent and measure sequence, y _1:krepresent the measurement sequence in 1 to k moment.

Under measurement functions is nonlinear situation, according to bayesian criterion, known by particle filter sequential sampling, posterior probability density function is

$p\left(x_{0:k}\right|y_{1:k})=\underset{i=1}{\overset{N_p}{\mathrm{\Σ}}}{\mathrm{\ω}}_k^i\mathrm{\δ}(x_{0:k}-x_{0:k}^{\left(i\right)})$

Wherein N _prepresent population, represent that k moment i-th particle is to x _ksample value, represent corresponding weighted value, δ () represents Dirac function.

Known by Bayes' theorem:

p(x _0：k|y _1：k-1)＝p(x _k|x _0：k-1)p(x _0：k-1|y _1：k-1)

Above formula can be expressed as:

$p\left(x_{0:k}\right|y_{1:k-1})$

$=p\left(x_k\right|x_{0:k-1})\underset{i-1}{\overset{N_p}{\mathrm{\Σ}}}{\mathrm{\ω}}_{k-1}\mathrm{\δ}(x_{0:k-1}-x_{0:k-1}^{\left(i\right)})$

$=\underset{i-1}{\overset{N_p}{\mathrm{\Σ}}}{\mathrm{\ω}}_{k-1}p\left(x_k^{\left(i\right)}\right|x_{0:k-1})\mathrm{\δ}(x_{0:k-1}-x_{0:k-1}^{\left(i\right)})$

Prior probability p (x _k| y _1:k-1) be p (x _0:k| y _1:k-1) marginal density, i.e. p (x _0:k| y _1:k-1)=∫ ∫ .. ∫ p (x _0:k| y _1:k-1) dx ₀dx ₁... dx _k-1, thus prior state probabilities p (x _k| y _1:k-1) can be expressed as

$p\left(x_k\right|y_{1:k-1})=\underset{i=1}{\overset{N_p}{\mathrm{\Σ}}}{\mathrm{\ω}}_{k-1}^i\left(x_k\right|x_{k-1}^{\left(i\right)})$

Do uniform variable motion for maneuvering target, the relation of k moment measured node and maneuvering target can represent with Fig. 2.It is trusted area that definition may detect maneuvering target region, and the node definition in trusted area is trusted node.In figure, dash area represents that maneuvering target is at the possible range of movement of subsequent time, with target speed and estimate factor-related.According to mathematical knowledge, by abstract for the relation of k moment measured node and maneuvering target for being r with radius ₁, r ₂be θ with angle ₁, θ ₂fan-shaped relevant, wherein r ₁, r ₂be respectively node to disembark the possible minimum and ultimate range of moving-target, θ ₁, θ ₂for node measurement is to the minimum and maximum deflection of maneuvering target.Omnidirectional's sensor signal value and distance dependent.If make θ ₁, θ ₂value meaningful, then need the set considering whole trusted node, and verify all combinations, amount of calculation can be caused so huge.If only consider radius r ₁, r ₂, then only need to consider individual node measured value z _k,m, only need less amount of calculation.Therefore, the relation of k moment measured node and maneuvering target can abstract be further Fig. 3, only relevant with distance radius, have nothing to do with angle.

The Position And Velocity of k-1 moment maneuvering target is known, and the scope that may occur for measured node k moment maneuvering target is then as shown in Fig. 3 dash area.R ₁and r ₂the sample value of value and each particle i and measured value and node m relevant, be defined as follows

$r_{1,k}^{i,m}=\left|\right|(x_{k-1}^i,y_{k-1}^i)-(\mathrm{locx}\left(m\right),\mathrm{locy}\left(m\right))\left|\right|-{\mathrm{\ζ}}_{\mathrm{ns}}\left|\right|({\stackrel{\·}{x}}_{k-1}^i,{\stackrel{\·}{y}}_{k-1}^i)\left|\right|$

$r_{2,k}^{i,m}=\left|\right|(x_{k-1}^i,y_{k-1}^i)-(\mathrm{locx}\left(m\right),\mathrm{locy}\left(m\right))\left|\right|-{\mathrm{\ζ}}_{\mathrm{ns}}\left|\right|({\stackrel{\·}{x}}_{k-1}^i,{\stackrel{\·}{y}}_{k-1}^i)\left|\right|$

Wherein the estimated coordinates that i-th particle k-1 phase carves target, be the estimating speed that i-th particle k-1 phase carves target, (locx (m), locy (m)) is the coordinate of measured node, ζ _nsestimate the factor, ζ _ns> 1.

The reliability definition of the signal received by above formula node m is

$p\left(m\right)=\underset{D_m}{\∫\∫}p\left(x_k\right|y_{1:k-1}){\mathrm{dx}}_k{\mathrm{dy}}_k$

$=\underset{i=1}{\overset{N_s}{\mathrm{\Σ}}}{\mathrm{\ω}}_{k-1}^i\mathrm{\Φ}(x_{k-1}^{\left(i\right)},D_m,z_{k,m})$

be defined as follows

From above formula, if think that the measured value that measured node obtains is reliable, then if think that measured value is clutter measured value, and the degree that measured value departs from is larger, less.

The k moment gathers C _k={ m|z _{k, m}> η _ns, 0 < m < N _s, wherein η _nsthreshold value, relevant with the investigative range of node.Know from three-point fix, the location that realize maneuvering target needs the measured value of 3 nodes, thus according to formula set of computations C _kin the p (m) of all nodes, choose three measured values as locating information foundation according to the size of p (m).

Be not difficult to find, the present invention is on particle filter basis, adopt Bayesian Estimation, obtain prior probability, utilize the position of previous moment, velocity estimation and the estimation factor, obtain each node in investigative range to target farthest and minimum distance, measured value again in conjunction with node itself calculates the corresponding weights of each particle, then weighting obtains the set of scope interior nodes measurement reliability, from this set, select reliable measured value to estimate foundation as tracking, thus can when there is erroneous measurements, select reliable node measurement value as tracking foundation, thus the trail-and-error avoiding mistake measurement to bring or track rejection.

Claims (1)

1., based on a tracking node selecting method for wireless omnidirectional sensor network, it is characterized in that, comprise the following steps:

(1) on the basis of particle filter, prior state probabilities is obtained;

First k moment prior state probabilities p (x is obtained _k| y _1:k-1), wherein, x _krepresent status switch, y _krepresent and measure sequence, y _1:krepresent the measurement sequence in 1 to k moment, under measurement functions is nonlinear situation, according to bayesian criterion, known by particle filter sequential sampling, posterior probability density function is wherein N _prepresent population, represent that k moment i-th particle is to x _ksample value, represent corresponding weighted value, δ () represents Dirac function, is known by Bayes' theorem: p (x _0:k| y _1:k-1)=p (x _k| x _0:k-1) p (x _0:k-1| y _1:k-1), above formula is expressed as:

p(x _0:k|y _1:k-1)

Prior probability p (x _k| y _1:k-1) be p (x _0:k| y _1:k-1) marginal density, i.e. p (x _0:k| y _1:k-1)=∫ ∫ .. ∫ p (x _0:k| y _1:k-1) dx ₀dx ₁... dx _k-1, thus prior state probabilities p (x _k| y _1:k-1) be expressed as

(2) utilize the position of previous moment, velocity estimation and the estimation factor, obtain each node in investigative range to the maximum distance of target and minimum distance;

Minimum distance

Maximum distance

Wherein the estimated coordinates of i-th particle k-1 moment target, be the estimating speed of i-th particle k-1 moment target, (locx (m), locy (m)) is the coordinate of measured node, ζ _nsestimate the factor, ζ _ns> 1;

(3) measured value in conjunction with node itself calculates the corresponding weights of each particle, and weighting obtains the set of investigative range interior nodes measurement reliability;

The reliability definition of the signal that node m receives is:

be defined as follows

From above formula, if think that the measured value that measured node obtains is reliable, then if think that measured value is clutter measured value, and the degree that measured value departs from is larger, less, the k moment gathers C _k={ m|z _k,m> η _ns, 0 < m < N _s, wherein η _nsthreshold value, relevant with the investigative range of node;

(4) from described set, select reliable measured value to estimate foundation as tracking; Know from three-point fix, the location that realize maneuvering target needs the measured value of 3 nodes, thus according to formula set of computations C _kin the p (m) of all nodes, choose three measured values as locating information foundation according to the size of p (m).