CN102685772B

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

Info

Publication number: CN102685772B
Application number: CN201210112292.9A
Authority: CN
Inventors: 林振华; 夏凌楠; 王静; 林兴华; 姜群; 尹达
Original assignee: Shanghai Institute of Microsystem and Information Technology of CAS
Current assignee: Shanghai Institute of Microsystem and Information Technology of CAS
Priority date: 2012-04-17
Filing date: 2012-04-17
Publication date: 2014-12-24
Anticipated expiration: 2032-04-17
Also published as: CN102685772A

Abstract

The invention relates to a tracking node selection method based on a wireless omnidirectional sensor network, comprising the following steps: obtaining a priori state probability on the basis of particle filtering; The farthest distance and the shortest distance from the nodes within the detection range to the target; combine the measured values of the nodes themselves to calculate the corresponding weight of each particle, and weight them to obtain the reliability set of the measured values of the nodes within the detection range; select a reliable set from the set The measured value is used as the tracking estimation basis. The invention can avoid tracking error or target loss which may be caused by wrong measurement.

Description

A Tracking Node Selection Method Based on Wireless Omnidirectional Sensor Networks

技术领域 technical field

本发明涉及无线传感器网络技术领域，特别是涉及一种基于无线全向传感器网络的跟踪节点选择方法，适用于利用震动传感器、声音传感器、射频天线等全向传感器跟踪过程中节点选择。The invention relates to the technical field of wireless sensor networks, in particular to a tracking node selection method based on a wireless omnidirectional sensor network, which is suitable for node selection in the tracking process of omnidirectional sensors such as vibration sensors, sound sensors, and radio frequency antennas.

背景技术 Background technique

物联网(Internet of Things)代表了未来计算与通信技术发展的方向，被认为是继计算机、Internet之后，信息产业领域的第三次发展浪潮。物联网是一个能够在任何时间(Anytime)、地点(Anyplace)，实现任何物体(Anything)互联的动态网络，它包括了PC之间、人与人之间、物与人之间、物与物之间的互联。物联网包括感知层、网络层和应用层3个层次。无线传感器网络(WSN)是感知层采用的关键技术之一。The Internet of Things (Internet of Things) represents the direction of future computing and communication technology development, and is considered to be the third wave of development in the field of information industry after computers and the Internet. The Internet of Things is a dynamic network that can realize the interconnection of any object (Anything) at any time (Anytime), anywhere (Anyplace), which includes PCs, people-to-people, things-to-people, things-to-things interconnection between. The Internet of Things includes three layers: perception layer, network layer and application layer. Wireless sensor network (WSN) is one of the key technologies adopted by the perception layer.

在WSN的应用研究中，目标定位跟踪是一个重要的研究方向。目标定位跟踪的传感器模型大多是基于声音传感器、震动传感器、射频天线等。在跟踪过程中，实际上由于多径干扰、遮挡、环境等因素，造成个别获取的信号值偏离真实值较大，成为错误测量，而错误测量不是某种噪声造成。在目标定位跟踪中如果不考虑可能存在的错误测量，极易导致目标信息的错误估计甚至目标丢失。In the applied research of WSN, target location tracking is an important research direction. Most of the sensor models for target positioning and tracking are based on sound sensors, vibration sensors, radio frequency antennas, etc. In the tracking process, in fact, due to factors such as multipath interference, occlusion, and environment, the individual acquired signal values deviate greatly from the real value, which becomes an error measurement, and the error measurement is not caused by some kind of noise. If the possible erroneous measurement is not taken into account in target location tracking, it will easily lead to wrong estimation of target information and even target loss.

发明内容 Contents of the invention

本发明所要解决的技术问题是提供一种基于无线全向传感器网络的跟踪节点选择方法，避免错误测量可能带来的跟踪错误或目标丢失。The technical problem to be solved by the present invention is to provide a tracking node selection method based on a wireless omnidirectional sensor network to avoid tracking errors or target loss that may be caused by wrong measurements.

本发明解决其技术问题所采用的技术方案是：提供一种基于无线全向传感器网络的跟踪节点选择方法，包括以下步骤：The technical solution adopted by the present invention to solve the technical problem is: provide a kind of tracking node selection method based on wireless omnidirectional sensor network, comprising the following steps:

(1)在粒子滤波的基础上获得先验状态概率；(1) Obtain prior state probability on the basis of particle filter;

(2)利用前一时刻的位置、速度估计和估计因子，得到每一个在探测范围内的节点到目标的最远距离和最近距离；(2) Using the position, velocity estimation and estimation factor at the previous moment, obtain the farthest distance and the shortest distance from each node within the detection range to the target;

(3)结合节点本身的测量值计算每一粒子相应的权值，并加权得到探测范围内节点测量值可靠性集合；(3) Combining the measured value of the node itself to calculate the corresponding weight of each particle, and weighting to obtain the reliability set of the measured value of the node within the detection range;

(4)从所述集合中选择可靠的测量值作为跟踪估计依据。(4) Select reliable measurement values from the set as the basis for tracking estimation.

所述步骤(1)中采用贝叶斯估计的方法获得先验状态概率。In the step (1), the method of Bayesian estimation is adopted to obtain the prior state probability.

所述步骤(4)中选择三个测量值作为跟踪估计依据。In the step (4), three measured values are selected as the tracking estimation basis.

有益效果Beneficial effect

由于采用了上述的技术方案，本发明与现有技术相比，具有以下的优点和积极效果：本发明在粒子滤波基础上，采用贝叶斯估计，获得先验概率，利用前一时刻的位置、速度估计和估计因子，得到每一个在探测范围内的节点到目标的最远和最近距离，再结合节点本身的测量值计算每一粒子相应的权值，然后加权得到范围内节点测量值可靠性集合，从该集合中选择可靠的测量值作为跟踪估计依据，从而可以在存在错误测量值的情况下，选择可靠的节点测量值作为跟踪依据，从而避免错误测量可能带来的跟踪错误或目标丢失。Due to the adoption of the above-mentioned technical solution, the present invention has the following advantages and positive effects compared with the prior art: on the basis of particle filtering, the present invention adopts Bayesian estimation to obtain prior probability, and utilizes the position at the previous moment , speed estimation and estimation factor, to get the farthest and shortest distance from each node within the detection range to the target, and then combine the measured value of the node itself to calculate the corresponding weight of each particle, and then weighted to obtain the reliability of the measured value of the node within the range From this set, reliable measurement values are selected as the basis for tracking estimation, so that in the case of erroneous measurement values, reliable node measurement values can be selected as the basis for tracking, thereby avoiding tracking errors or target errors that may be caused by erroneous measurements. lost.

附图说明 Description of drawings

图1是本发明的流程图；Fig. 1 is a flow chart of the present invention;

图2是本发明实施方式中测量点与机动目标关系图；Fig. 2 is a diagram of the relationship between a measurement point and a maneuvering target in an embodiment of the present invention;

图3是本发明实施方式中测量点与机动目标的可能距离示意图。Fig. 3 is a schematic diagram of possible distances between a measurement point and a maneuvering target in an embodiment of the present invention.

具体实施方式 Detailed ways

下面结合具体实施例，进一步阐述本发明。应理解，这些实施例仅用于说明本发明而不用于限制本发明的范围。此外应理解，在阅读了本发明讲授的内容之后，本领域技术人员可以对本发明作各种改动或修改，这些等价形式同样落于本申请所附权利要求书所限定的范围。Below in conjunction with specific embodiment, further illustrate the present invention. It should be understood that these examples are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the teachings of the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of the present application.

本发明的实施方式涉及一种基于无线全向传感器网络的跟踪节点选择方法，如图1所示，包括以下步骤：Embodiments of the present invention relate to a method for selecting a tracking node based on a wireless omnidirectional sensor network, as shown in FIG. 1 , comprising the following steps:

具体地说，首先获得k时刻先验状态概率p(x_k|y_1：k-1)，其中，x_k表示状态序列，y_k表示测量序列，y_1：k表示1到k时刻的测量序列。Specifically, first obtain the prior state probability p(x _k |y _{1: k-1} ) at time k, where x _k represents the state sequence, y _k represents the measurement sequence, and y _{1: k} represents the measurement from time 1 to k sequence.

在测量函数是非线性的情况下，根据贝叶斯准则，由粒子滤波序列采样知，后验概率密度函数为In the case where the measurement function is nonlinear, according to the Bayesian rule, the posterior probability density function is known from the particle filter sequence sampling as

$p p (({x x}_{00 : : k k} | | {y the y}_{11 : : k k})) = = {Σ Σ}_{i i = = 11}^{{N N}_{p p}} {ω ω}_{k k}^{i i} δ δ (({x x}_{00 : : k k} - - {x x}_{00 : : k k}^{((i i))}))$

其中N_p表示粒子数，表示k时刻第i个粒子对x_k的抽样值，表示相应的权重值，δ(·)表示狄拉克函数。where N _p represents the number of particles, Indicates the sampling value of the i-th particle for x _k at time k, Indicates the corresponding weight value, and δ(·) indicates the Dirac function.

由贝叶斯定理知：According to Bayes theorem:

上式可以表示成：The above formula can be expressed as:

$p p (({x x}_{00 : : k k} | | {y the y}_{11 : : k k - - 11}))$

$= = p p (({x x}_{k k} | | {x x}_{00 : : k k - - 11})) {Σ Σ}_{i i - - 11}^{{N N}_{p p}} {ω ω}_{k k - - 11} δ δ (({x x}_{00 : : k k - - 11} - - {x x}_{00 : : k k - - 11}^{((i i))}))$

$= = {Σ Σ}_{i i - - 11}^{{N N}_{p p}} {ω ω}_{k k - - 11} p p (({x x}_{k k}^{((i i))} | | {x x}_{00 : : k k - - 11})) δ δ (({x x}_{00 : : k k - - 11} - - {x x}_{00 : : k k - - 11}^{((i i))}))$

先验概率p(x_k|y_1：k-1)是p(x_0：k|y_1：k-1)的边缘密度，即p(x_0：k|y_1：k-1)＝∫∫..∫p(x_0：k|y_1：k-1)dx₀dx₁...dx_k-1，因而先验状态概率p(x_k|y_1：k-1)可以表示为The prior probability p(x _k |y _1:k-1 ) is the edge density of p(x _0:k |y _1:k-1 ), that is, p(x _0:k |y _1:k-1 )= ∫∫..∫p(x _{0: k} |y _{1: k-1} )dx ₀ dx ₁ ... dx _k-1 , so the prior state probability p(x _k |y _{1: k-1} ) can be expressed for

$p p (({x x}_{k k} | | {y the y}_{11 : : k k - - 11})) = = {Σ Σ}_{i i = = 11}^{{N N}_{p p}} {ω ω}_{k k - - 11}^{i i} (({x x}_{k k} | | {x x}_{k k - - 11}^{((i i))}))$

以机动目标作匀变速运动为例，k时刻测量节点和机动目标的关系可以用图2表示。定义可探测到机动目标区域为可信区域，可信区域内的节点定义为可信节点。图中阴影部分表示机动目标在下一时刻可能的运动范围，与目标运动速度及估计因子有关。根据数学知识，将k时刻测量节点和机动目标的关系抽象为与半径为r₁、r₂和角度为θ₁、θ₂的扇形有关，其中r₁、r₂分别为节点离机动目标的可能最小与最大距离，θ₁、θ₂为节点测量到机动目标的最小与最大方向角。全向传感器信号值与距离有关。如果要使θ₁、θ₂的值有意义，则需考虑整个可信节点的集合，并验证所有组合，这样会导致计算量巨大。如果只考虑半径r₁、r₂，则仅需考虑单个节点测量值z_k,m，只需较小的计算量。因此，k时刻测量节点和机动目标的关系可以进一步抽象为图3，仅与距离半径有关，与角度无关。Taking the maneuvering target moving at a uniform speed as an example, the relationship between the measurement node and the maneuvering target at time k can be shown in Figure 2. The area where the maneuvering target can be detected is defined as a trusted area, and the nodes in the trusted area are defined as trusted nodes. The shaded part in the figure indicates the possible movement range of the maneuvering target at the next moment, which is related to the target movement speed and estimation factors. According to the mathematical knowledge, the relationship between the measuring node and the maneuvering target at time k is abstracted as being related to the sectors with radii r ₁ , r ₂ and angles θ ₁ , θ ₂ , where r ₁ and r ₂ are respectively the possibility of the node leaving the maneuvering target The minimum and maximum distances, θ ₁ and θ ₂ are the minimum and maximum direction angles measured by the node to the maneuvering target. The signal value of the omnidirectional sensor is related to the distance. If the values of θ ₁ and θ ₂ are to be meaningful, it is necessary to consider the entire set of trusted nodes and verify all combinations, which will lead to a huge amount of calculation. If only the radii r ₁ and r ₂ are considered, then only the measurement value z _k,m of a single node needs to be considered, and only a small amount of calculation is required. Therefore, the relationship between the measurement node and the maneuvering target at time k can be further abstracted as shown in Figure 3, which is only related to the distance radius and has nothing to do with the angle.

k-1时刻机动目标的位置与速度是已知的，对于测量节点k时刻机动目标可能出现的范围则是如图3阴影部分所示。r₁和r₂的取值与每一个粒子i的抽样值和测量值及节点m有关，具体定义如下The position and velocity of the maneuvering target at time k-1 are known, and the possible range of the maneuvering target at the measurement node k time is shown in the shaded part of Figure 3. The values of r ₁ and r ₂ are related to the sampling value and measurement value of each particle i and the node m, and the specific definitions are as follows

${r r}_{11,, k k}^{i i,, m m} = = | | | | (({x x}_{k k - - 11}^{i i},, {y the y}_{k k - - 11}^{i i})) - - ((locx locx ((m m)),, locy locy ((m m)))) | | | | - - {ζ ζ}_{ns ns} | | | | (({\overset{\cdot \cdot}{x x}}_{k k - - 11}^{i i},, {\overset{\cdot \cdot}{y the y}}_{k k - - 11}^{i i})) | | | |$

${r r}_{22,, k k}^{i i,, m m} = = | | | | (({x x}_{k k - - 11}^{i i},, {y the y}_{k k - - 11}^{i i})) - - ((locx locx ((m m)),, locy locy ((m m)))) | | | | - - {ζ ζ}_{ns ns} | | | | (({\overset{\cdot \cdot}{x x}}_{k k - - 11}^{i i},, {\overset{\cdot \cdot}{y the y}}_{k k - - 11}^{i i})) | | | |$

其中是第i个粒子k-1相刻目标的估计坐标，是第i个粒子k-1相刻目标的估计速度，(locx(m)，locy(m))是测量节点的坐标，ζ_ns是估计因子，ζ_ns＞1。in is the estimated coordinates of the i-th particle k-1 phase engraved target, is the estimated velocity of the i-th particle k-1 relative to the target, (locx(m), locy(m)) is the coordinate of the measurement node, ζ _ns is the estimation factor, ζ _ns >1.

由上式节点m接收到的信号的可靠性定义为The reliability of the signal received by node m in the above formula is defined as

$p p ((m m)) = = \underset{{D D.}_{m m}}{&Integral; &Integral; &Integral; &Integral;} p p (({x x}_{k k} | | {y the y}_{11 : : k k - - 11})) {dx dx}_{k k} {dy dy}_{k k}$

$= = {Σ Σ}_{i i = = 11}^{{N N}_{s the s}} {ω ω}_{k k - - 11}^{i i} Φ Φ (({x x}_{k k - - 11}^{((i i))},, {D D.}_{m m},, {z z}_{k k,, m m}))$

定义如下 defined as follows

由上式可知，如果认为测量节点得到的测量值是可靠的，则如果认为测量值是clutter测量值，并且测量值偏离的程度越大，越小。It can be seen from the above formula that if the measurement value obtained by the measurement node is considered to be reliable, then If the measurement is considered to be a clutter measurement, And the greater the degree of deviation of the measured value, smaller.

k时刻集合C_k＝{m|z_k，m＞η_ns，0＜m＜N_s}，其中η_ns是门限值，与节点的探测范围有关。从三点定位知，要实现机动目标的定位需要3个节点的测量值，因而根据式计算集合C_k中所有节点的p(m)，按照p(m)的大小选取三个测量值作为定位信息依据。Set C k at time _k = {m|z _{k, m} >η _ns , 0<m<N _s }, where η _ns is the threshold value, which is related to the detection range of the node. From the three-point positioning, the measurement values of three nodes are needed to realize the positioning of the maneuvering target, so the p(m) of all nodes in the set C _k is calculated according to the formula, and the three measurement values are selected as the positioning according to the size of p(m) Information basis.

不难发现，本发明在粒子滤波基础上，采用贝叶斯估计，获得先验概率，利用前一时刻的位置、速度估计和估计因子，得到每一个在探测范围内的节点到目标的最远和最近距离，再结合节点本身的测量值计算每一粒子相应的权值，然后加权得到范围内节点测量值可靠性集合，从该集合中选择可靠的测量值作为跟踪估计依据，从而可以在存在错误测量值的情况下，选择可靠的节点测量值作为跟踪依据，从而避免错误测量可能带来的跟踪错误或目标丢失。It is not difficult to find that on the basis of particle filtering, the present invention adopts Bayesian estimation to obtain prior probability, and uses the position, velocity estimation and estimation factors at the previous moment to obtain the farthest distance from each node within the detection range to the target. and the closest distance, combined with the measured value of the node itself to calculate the corresponding weight of each particle, and then weighted to obtain the reliability set of the measured value of the node within the range, from which the reliable measured value is selected as the basis for tracking estimation, so that in the existence In the case of erroneous measurement values, reliable node measurement values are selected as the tracking basis, so as to avoid tracking errors or target loss that may be caused by erroneous measurements.

Claims

1. A tracking node selection method based on wireless omnidirectional sensor network, is characterized in that, comprises the following steps:

(1) Obtain prior state probability on the basis of particle filter;

First, obtain the prior state probability p(x _k |y _1:k-1 ) at time k, where x _k represents the state sequence, y _k represents the measurement sequence, and y _1:k represents the measurement sequence from 1 to k time. When the function is nonlinear, according to the Bayesian criterion, the particle filter sequence sampling is known, and the posterior probability density function is where N _p represents the number of particles, Indicates the sampling value of the i-th particle for x _k at time k, Represents the corresponding weight value, δ( ) represents the Dirac function, known from Bayesian 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 ), the above formula is expressed as:

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

(2) Using the position, velocity estimation and estimation factors at the previous moment, get the farthest distance and the shortest distance from each node within the detection range to the target;

closest distance

furthest distance

in is the estimated coordinates of the target at time k-1 of the i-th particle, is the estimated speed of the target at the time of the i-th particle k-1, (locx(m), locy(m)) is the coordinates of the measuring node, ζ _ns is the estimation factor, ζ _ns >1;

(3) Combine the measured value of the node itself to calculate the corresponding weight of each particle, and weight it to obtain the reliability set of the measured value of the node within the detection range;

The reliability of the signal received by node m is defined as:

defined as follows

It can be seen from the above formula that if the measurement value obtained by the measurement node is considered to be reliable, then If the measurement is considered to be a clutter measurement, And the greater the degree of deviation of the measured value, The smaller the k time set C _k ={m|z _k,m >η _ns ,0<m<N _s }, where η _ns is the threshold value, which is related to the detection range of the node;

(4) Select reliable measured values from the set as the basis for tracking estimation; from the three-point positioning, the measured values of three nodes are needed to realize the positioning of the maneuvering target, so the p of all nodes in the set C _k is calculated according to the formula (m), select three measured values according to the size of p(m) as the basis of positioning information.