JP2008128726A - Positioning system, device and method using particle filter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a positioning method and a system having high accuracy by reducing an influence of a multipath. <P>SOLUTION: In one execution form, a terminal 40 radiates a prescribed signal, and a plurality of nodes 41-44 whose position information is known receive the signal, and record each reception time. A positioning signal processing part 46 determines a reception time difference between each node from each reception time, and corrects the reception time difference by using the first particle filters 81-1 to 6. Then, a terminal position is estimated from the corrected reception time difference and the position information of each node by using the second particle filter in a position estimation part 82. An influence of the multipath in a non-viewable environment is reduced by using the particle filters, thereby enable highly accurate positioning. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to terminal position detection in a network. More specifically, the present invention relates to a technique for estimating the position of a terminal using a hierarchical particle filter in a sensor network.

  Conventionally, the most common method for determining the position on the ground is GPS (Global Positioning System). This is because, as shown in FIG. 1, a receiver (for example, a vehicle 10) receives microwaves radiated from three or more satellites 11 to 14 in an orbit around the earth, and a pseudorange with each satellite. To determine the three-dimensional position of the receiver. Such a GPS positioning device is disclosed in Patent Document 1, for example. The GPS is intended to determine the position in a wide place because of its nature, and cannot be used in a room (for example, the terminal 15) where radio waves from the satellite do not reach.

  Patent Document 2 describes a positioning system that uses a time difference of arrival (TDOA) without using a satellite system such as GPS. As shown in FIG. 2, this is because a receiver (for example, base stations 21 to 24) whose position is known receives radio waves transmitted from a transmitter (for example, mobile terminal 20) whose position is unknown. This is a technique for estimating the position of a transmitter by measuring the difference in reception time at each receiver. This technology should also be used in indoor environments (for example, medical, office, and home environments) where the radio wave from the satellite does not reach and GPS technology cannot be used, such as the ubiquitous sensor network shown in FIG. Can do.

In positioning using TDOA, as shown in FIG. 2, a plurality of base stations 21 to 24 receive signals from the terminal 20 located at unknown coordinates [x _p , y _p , z _p ]. Assume that the coordinates [x _i , y _i , z _i ] of each base station are known (in FIG. 2, i is an integer from 1 to 4), and the time between the base station and the terminal cannot be synchronized. In the cellular system of FIG. 2 and the ubiquitous network of FIG. 3, an asynchronous environment is often assumed. In such an asynchronous environment, positioning is performed using a reception time difference (TDOA) t _i −t _j at each base station. Specifically, the relationship between the coordinates of two base stations [x _i , y _i , z _i ], [x _j , y _j , z _j ] and the terminal [x _p , y _p , z _p ] is TDOA information. Is represented by the following formula.
(1) c · (t _i −t _j ) = d _i −d _j
Here, c is the speed of light, d _i represents the distance from the terminal to the base station, and is defined as the following equation.

Since the relational expressions represented by the equations (1) and (2) are established by the number of pairs of base stations, the coordinates [x _p , y _p , z _p ] of the terminal are obtained by solving these nonlinear simultaneous equations. Can be estimated.

  However, as an actual problem, since the observed value includes noise, an approximate value is estimated for positioning. As noise sources, in addition to thermal noise and clock offset at the receiver of the base station, there are effects due to multipath in which radio waves arrive from the terminal through a non-line-of-sight (NLOS) propagation path. . As an approximate value derivation method, Non-Patent Documents 1 and 2 introduce the Newton method (also called Taylor-Series-Algorithm), but this method is vulnerable to the multipath problem caused by the NLOS propagation path. There is a problem that.

  As a method for deriving an approximate solution in TDOA type positioning, Non-Patent Document 3 introduces a positioning method that models a multipath propagation path and estimates an optimal arrival time. This method is known to be effective when the propagation path can be accurately modeled.

Japanese Patent No. 3634874 US Pat. No. 5,327,144 W. H. Foy, “Position-Location Solutions by Taylor-Series Estimation,” IEEE Trans. Aerosp. Electron. Syst., Vol. AES-12, no. 2, pp. 187-194, Mar., 1976. H. Kong, Y. Kwon and T. Sung, “Comparisons of TDOA Triangulation Solutions for Indoor Positioning,” in Proc. International Symposium on GNSS, December, 2004. Manabu Kawabata, Masato Asao, Takahiko Saikawa, Takeshi Hattori, “Position detection performance evaluation of TDOA system based on maximum likelihood estimation method in cellular system,” IEICE Transactions (B), vol. J87-B, no. 2, pp. 285-291, February, 2004.

  However, the conventional technique has the following problems. In other words, the conventional technology often depends on an external system such as GPS, which not only increases the cost of the apparatus, but also cannot be used in an indoor environment where GPS signals do not reach. Further, in the related art, when the propagation path between the transmitter and the receiver is out of line of sight, there is a problem that the positioning accuracy is deteriorated due to the influence of multipath in which radio waves arrive through a plurality of propagation paths. In order to compensate for the degradation of positioning accuracy due to multipath, the conventional technique performs estimation by accurately modeling a multipath propagation path. However, as a practical problem, it is difficult to accurately model the propagation path, and the amount of computation is increased. In the prior art, positioning is performed using a number of receivers in order to improve accuracy.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a positioning method and system with high accuracy by reducing the influence of multipath.

  In order to achieve such an object, the present invention provides a system for estimating a position of a terminal, wherein the terminal radiates a predetermined signal and a plurality of position information are known. A plurality of nodes that receive the predetermined signal and measure the reception time, receive the reception time from each node, obtain a reception time difference between the nodes, and determine the reception time difference and A signal processing unit that estimates the position of the terminal from position information of each node, wherein the signal processing unit corrects the reception time difference using a first particle filter.

  The invention according to claim 2 is the system according to claim 1, wherein the signal processing unit is configured to calculate the reception time difference corrected by using a second particle filter and the position information of each node. The terminal position is estimated.

  The invention according to claim 3 is the system according to claim 1 or 2, wherein the first particle filter is modeled as a time change of the reception time difference is constant. It is characterized by.

  According to a fourth aspect of the present invention, there is provided an apparatus for estimating a position of a terminal in a network including a plurality of nodes whose position information is known, and means for obtaining a signal reception time difference between the terminal and each node; And means for correcting the reception time difference using the first particle filter, and means for estimating the position of the terminal from the corrected reception time difference and position information of each node.

  The invention according to claim 5 is the apparatus according to claim 4, wherein the means for estimating is based on the reception time difference corrected using the second particle filter and the position information of each node. The terminal position is estimated.

  The invention according to claim 6 is the apparatus according to claim 4 or 5, wherein the first particle filter is modeled on the assumption that the time change of the reception time difference is constant. It is characterized by.

  The invention according to claim 7 is a method for estimating the position of a terminal, wherein the terminal radiates a predetermined signal, and a plurality of nodes whose position information is known receive the predetermined signal. Measuring the reception time, obtaining the reception time difference between the nodes from the reception time of each node, correcting the reception time difference using a first particle filter, and correcting the reception time Estimating the position of the terminal from the difference and position information of each node.

  The invention according to claim 8 is the method according to claim 7, wherein estimating the position of the terminal is performed by using the second particle filter to correct the received reception time difference and each node. The position of the terminal is estimated from position information.

  The invention according to claim 9 is a method for estimating a position of a terminal in a network including a plurality of nodes whose position information is known, and obtains a reception time difference of a signal between the terminal and each node. And correcting the reception time difference using a first particle filter, and estimating the position of the terminal from the corrected reception time difference and position information of each node.

  The invention according to claim 10 is the method according to claim 9, wherein the estimation is performed based on the reception time difference corrected using the second particle filter and the position information of each node. The terminal position is estimated.

  According to the present invention, it is possible to realize a positioning system that does not depend on GPS. Specifically, the position estimation of the mobile terminal can be realized without depending on GPS under the condition that the coordinate information of each node (base station) is known.

  Further, according to the present invention, resistance to an out-of-sight environment, particularly a multipath problem can be obtained. The present invention can be applied without any change in both the line-of-sight environment and the non-line-of-sight environment, and in particular, exhibits excellent positioning accuracy in the non-line-of-sight environment as compared with the conventional method.

  In addition, according to the present invention, an accurate positioning system is possible without requiring a propagation path model. That is, positioning is possible even in situations where the propagation path characteristics represented by the mathematical model / probability model are unknown. In one embodiment of the present invention, the effect of multipath can be reduced by setting a very simple precondition that “the time change of observation information is constant”.

  Moreover, according to this invention, there also exists an effect that the cost of a positioning system can be reduced. Since the positioning accuracy is improved, the number of nodes can be reduced. Further, the positioning signal processing can be distributed by adopting a hierarchical configuration. That is, it is possible to realize distributed processing in which the processing of the first layer is performed at each node and the processing of the second layer is performed by the positioning server. Such a reduction in signal processing burden is particularly effective in a sensor network or the like that has large hardware restrictions.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In this embodiment, positioning in the sensor network will be mainly described. However, the present invention is not limited to the sensor network, but can be applied to cellular and other networks.

(System model and positioning process)
FIG. 4 shows a system model of a positioning system in the sensor network. This system model includes a tag 40 whose position is unknown and nodes 41 to 44 whose positions are known. The positioning signal processing unit 46 is mounted on a positioning server of a sensor network, for example, obtains TDOA information based on the reception time information t _{1 to} t ₄ from the nodes 41 to 44, and estimates the position of the tag 40.

  FIG. 5 shows an example of a conventional positioning process based on the system model of FIG. In this positioning process, the tag 40 transmits a predetermined signal (step 502), each of the nodes 41 to 44 receives the signal transmitted from the tag 40, and records the reception time (step 504). Information regarding this reception time is transferred from each node to the positioning signal processing unit 46, where a reception time difference (TDOA) between the nodes is calculated (step 506). The positioning signal processing unit 46 estimates the position of the tag 40 by the Newton method using the calculated TDOA information (step 508).

  In contrast, FIG. 6 shows an example of the positioning process of the present invention based on the system model of FIG. In this positioning process, the tag 40 transmits a predetermined signal (step 602), each of the nodes 41 to 44 receives the signal transmitted from the tag 40, and records the reception time (step 604). Information regarding this reception time is transferred from each node to the positioning signal processing unit 46, where a reception time difference between the nodes is calculated (step 606). The positioning signal processing unit 46 processes the calculated TDOA information with a particle filter (PF) and corrects an error due to multipath in the propagation path (step 608). The positioning signal processing unit 46 estimates the position of the tag 40 using a particle filter using the corrected TDOA information (step 610).

FIG. 7 shows a configuration example of a positioning signal processing unit using a conventional Newton method. This positioning signal processing unit 46 includes buffers 47-1 to 4-4 that store reception time information (t _{1 to} t ₄ ) from each node, and TDOA calculation units 48-1 to 48-1 that calculate reception time differences between the nodes. 4, a memory 49 that stores the position information of each node, and a Newton method position estimation unit 70 that performs position estimation of the tag 40 by the Newton method from the calculated TDOA information and the position information of each node. In general, the TDOA information obtained by the TDOA calculation units 48-1 to 48-4 includes an observation error. In the positioning using the conventional Newton method, there is a problem that the estimation accuracy deteriorates because position estimation is performed without any correction for this error. In particular, the Newton method is an algorithm that performs a linear search that minimizes the gradient of the measurement error, and is therefore susceptible to observations with large errors.

Next, FIG. 8 shows a configuration example of a positioning signal processing unit according to an embodiment of the present invention. This positioning signal processing unit 46 includes buffers 47-1 to 4-4 that store reception time information (t _{1 to} t ₄ ) from each node, and TDOA calculation units 48-1 to 48-1 that calculate reception time differences between the nodes. 4, a memory 49 that stores position information of each node, a particle filter 81-1 to 6 that corrects the calculated TDOA information, and the position of the tag 40 by the particle filter from the corrected TDOA information and the position information of each node. The PF position estimation unit 82 executes estimation. In the positioning according to the embodiment of the present invention, a digital filter algorithm called a particle filter is used to correct the observation error of the TDOA information obtained by the TDOA calculation units 48-1 to 48-4 and perform position estimation. Therefore, the estimation accuracy is improved as compared with the prior art. In the configuration of FIG. 8, the positioning accuracy in a multipath environment is improved by adopting a two-stage particle filter. Hereinafter, a positioning system using a particle filter will be described in detail.

(Positioning system using particle filter)
The tag 40 according to an embodiment of the present invention periodically radiates radio waves to which its own ID is added. Nodes 41-44 receive the signal from the tag and identify the ID of the tag. Each node transfers the tag ID and the signal reception time to the positioning signal processing unit 46. Further, the node may transfer its own position information to the positioning signal processing unit 46 as necessary. The positioning signal processing unit 46 calculates the TDOA between the nodes from the signal reception time. Here, if the reception time at node #i is t _i and the reception time at node #j is t _j , the TDOA between the nodes is t _i -t _j . When this TDOA is multiplied by a high speed c, the distance difference Δd between the nodes #i and #j and the tag is obtained from the following equation.

Here, d _i represents the distance from the tag to the node #i as defined by the expression (2), and d _j represents the distance from the tag to the node #j in the same manner. Expression (3) is a relational expression in an ideal environment. In the actual environment, an error e due to multipath is added as in the following expression.

  Therefore, the actual observed value Δd ′ has a relationship of Δd ′ = Δd + e with respect to the true distance difference Δd. Therefore, in the present invention, the true value Δd is estimated from the observed value Δd ′ by the particle filter, the error is reduced from the observed value Δd ′, and a correction value Δd ″ close to the true value Δd is derived.

(First-level particle filter)
The particle filter is a signal processing algorithm for estimating a state vector modeled by a state space model. The estimation process is shown in FIG. In this specification, this processing is referred to as a “first layer particle filter”.

  In step 902 of FIG. 9, N particles, that is, N random vectors, are generated based on the initial distribution. In the particle filter, a state vector to be estimated is referred to as “particle”, and a final estimated value is obtained by analyzing its operation. Here, N can be set arbitrarily. In general, the larger N is, the better the characteristic is, but the amount of calculation increases.

  N particles according to initial distribution

Is represented in the following vector format.

Here, x ⁽ⁱ⁾ (0) is a random variable according to the initial distribution. As an initial distribution, a Gaussian distribution with an average Δd ′ (1) and a variance ν ² is given. Note that Δd ′ (1) is an observed value actually obtained at the initial time t = 1.

In general, the two elements included in the particle state vector X _t ⁽ⁱ⁾ at time t are the observed value Δd ′ (t) at the present time (time t) and the correction before the temporary point (time t−1). Represents the value Δd ″ (t−1). In the initial state (t = 1), since the time t−1 does not exist, the two elements average the observation values Δd ′ (1) at the time t = 1. Is used.

In step 904, a prediction is made one period ahead from the current distribution. Specifically, the observation value at the temporary point is corrected using the current observation value Δd ′ (t) and the correction value Δd ″ (t−1) before the temporary point. Here, the time of the observation value is corrected. As a prior model for predicting the change, it is assumed that “the time change of the observed distance information Δd ′ (t) is constant”. This assumption is valid when the movement of the tag, which is a mobile terminal, is slow, and has the effect of removing a sudden abnormal value due to multipath. This assumption is defined by the following equation.
(6) Δd ′ (t + 1) −Δd ′ (t) = Δd ′ (t) −Δd ′ (t−1)
Based on this assumption, the state vector

To make a new particle group

Get. The state transition equation is as follows.

Here, the system noise vector V _t mainly represents thermal noise, clock offset, etc., and gives a Gaussian distribution with an average of 0 and a variance ν ² . In addition, the elements of the state vector X _t ⁽ⁱ ) in Expression (7) are the observation value Δd ′ (t) at time t and the estimation result Δd ″ (t−1) before the temporary point. It is expressed in

Next, in step 906, using the observed value [Delta] d '(t) and observation noise distribution, calculates the likelihood of the individual particles P t (i) α t a (i).

Here, the observed noise distribution r (x; 0, η ² ) mainly represents the influence of multipath, and gives a Cauchy distribution defined by the following equation.

The random variable Δd ′ (t) − [1 0] · P _t ⁽ⁱ ) in Expression (9) indicates an error between the predicted value based on the one-term prediction and the actual observed value. A particle with a larger error has a lower likelihood, and a particle with a smaller error has a higher likelihood.

Next, in step 908, the filter distribution is calculated. Specifically, the particles P _t ⁽ⁱ⁾ are resampled based on the following equation.

As a result, a group of particles expressing the state of one period ahead

Get.

  In step 910, the state value of each particle after resampling is averaged to obtain a correction value for the observed value at that time. In step 904, the correction value is used for the prediction of one period ahead.

  In this way, by repeating Step 904 to Step 908, it is possible to obtain a correction value in which the influence of the multipath component included in the observation value is reduced.

(Second-level particle filter)
Next, the position of the tag that is the mobile terminal is estimated using the observation value corrected by the above-described process using the particle filter in the first layer. This estimation process is shown in FIG. In this specification, this processing is referred to as “second-level particle filter”.

In step 1002 of FIG. 10, initial estimated coordinates are set. In position estimation using a particle filter, a plurality of two-dimensional coordinates are set as initial estimation candidates, and these are called “particles”. We narrow down to true coordinates while examining the reliability of coordinates of individual particles. Here, the number of particles is N ′, and the initial state is X _n ⁽ⁱ⁾ = {x _i (n), y _i (n)}, i = 1,. . . , N ′. The two-dimensional coordinates are given by N ′ random two-dimensional variables that are uniformly distributed in the positioning area. The parameter n is a variable for counting the number of repetitions of step 1016 described later, and n = 0 at the initial setting stage.

In step 1004, the difference in distance from each node is calculated with respect to the initial coordinates of the particles set in step 1002. For example, the distance difference h _kj ⁽ⁱ⁾ between the particle {x _i (n), y _i (n)} and the node #k [x _k , y _k ] and the node #j [x _j , y _j ]. Is calculated by the following equation.

Such a distance difference is obtained for all node pairs. When the number of nodes is 4, the number of node pairs is ₄ C ₂ = 6, and thus 6 distance differences h _kj ⁽ⁱ⁾ are obtained. For simplicity, this distance difference is expressed as h ₁ ⁽ⁱ⁾ , h ₂ ^(i),. . . , H ₆ ⁽ⁱ⁾ .

Next, in step 1006, the likelihood of each particle is calculated. Specifically, the distance information y ₁ ,. . . , Y ₆ (in this embodiment, the correction value obtained in the first hierarchy) is obtained, and the square error E ⁽ⁱ⁾ is obtained.

Based on this error information, the likelihood α _n ⁽ⁱ⁾ of each particle is calculated.

Here, r (x; 0, μ ² ) gives a Gaussian distribution with an average of 0 and a variance of μ ² . An error E ⁽ⁱ⁾ is given as a random variable, and particles having a smaller E ⁽ⁱ⁾ have a higher likelihood.

  Next, in step 1008, the likelihood obtained in step 1006 is normalized by the following equation.

Next, in step 1010, the estimated value of the tag coordinates _{[X n} 'Y _n'] is calculated by the following equation.

This equation is a linear sum of state quantities (= two-dimensional coordinates) of all particles with likelihood as a weight.

  In step 1012, based on the probability expressed by

Is resampled and used for the next iteration.

次に、ステップ１０１４において、リサンプリングの後、粒子の状態量が初期分布で与えられた座標にのみ収束してしまうのを避けるため、粒子の状態量を次式に従って限定的に拡散させる。

Next, in step 1014, after resampling, in order to avoid that the particle state quantity converges only on the coordinates given by the initial distribution, the particle state quantity is diffused in a limited manner according to the following equation.

Here, the vector W = {w _x , w _y } is a two-dimensional Gaussian random number with an average of 0 and a variance μ ² / n.

  In step 1016, until the estimated coordinates converge, loop back to step 1004 and repeat steps 1004 to 1014 to obtain the final positioning result.

(Evaluation results)
Next, the result of computer simulation is shown about the positioning characteristic by one Embodiment of this invention, and it compares with the conventional system. In this simulation, a 100 m square two-dimensional plane is considered as a positioning area, and the number of base station nodes is four. The coordinates of the nodes are known, and they are set at four coordinates [x, y] = [25, 25], [25, 75], [75, 25], [75, 75]. The propagation environment was set to be a mixture of observation noise due to Gaussian noise and multipath delay according to the exponential distribution. Assume that Gaussian noise with an average of 0 and variance σ ² and an exponential distribution delay with an average λ are added to the distance information observed at each node.

As the movement trajectory of the tag, a trajectory that moves 160 m at a constant speed (V = 1 m / s) while giving a straight or 90 degree turn along the corridor in an indoor environment is given. The locus of the tag is shown as a solid line in FIGS. 11 (a) and 11 (b). The propagation environment is σ ² = 5 and λ = 2 in any case.

  FIG. 11A shows the positioning result by the conventional Newton method, and it can be seen that the positioning accuracy is greatly deteriorated due to the measurement error caused by the propagation environment. In general, in the Newton method, an estimated value does not converge correctly unless an initial estimated value is given appropriately. In FIG. 11A, since the center point [x, y] = [50, 50] surrounded by the nodes is given as the initial estimated value, the positioning accuracy is particularly poor in the area not surrounded by the nodes. The tag cannot be tracked.

FIG. 11B shows a positioning result obtained by the hierarchical particle filter of the present invention. The number of particles was N = 1000 for the first layer, N ′ = 50 for the second layer, and the initial parameters of the particle filter were ν ² = 3, η ² = 0.5, and μ ² = 10. It can be confirmed that the positioning accuracy is greatly improved compared to the conventional method.

In FIG. 12, as an evaluation showing the effect of error correction in the first hierarchy, from node # 1: [x, y] = [25, 25] and node # 2: [x, y] = [25, 75]. It shows how the error included in the observation distance to the tag changes during the tracking process. The propagation environment is an NLOS environment with σ ² = 5 and λ = 2. The conventional method indicated by the dotted line frequently generates large abnormal values because NLOS compensation is not performed. On the other hand, in the proposed method shown by the solid line, it can be seen that the error is suppressed to be small due to the effect of NLOS compensation in the first layer.

  FIG. 13 shows the root mean square error (RMSE) characteristics when λ representing the influence of NLOS propagation is changed among the two parameters σ and λ that characterize the propagation path. As λ becomes larger, the propagation environment becomes worse, and the positioning accuracy is deteriorated in any method. However, it can be seen that the hierarchical particle filter according to the present invention has an extremely low degree of degradation compared to the conventional method, and is an algorithm that is very robust against environmental changes.

(Other embodiments)
While the present invention has been described with respect to specific embodiments, the embodiments described herein are merely illustrative in view of the many possible embodiments to which the principles of the present invention can be applied. It does not limit the range. For example, in the above embodiment, the mobile terminal is configured to estimate the position of the mobile terminal, but the mobile terminal is configured to receive the signal from the network side and estimate the position of the mobile terminal. May be. In this case, coordinate information and time offset information can be included in the signal from the node.

  Further, in a network having a large number of nodes, for example, nodes may be selectively used, such as not using observation values of nodes in a poor environment. In addition, the number of particles can be changed adaptively, and in a static and good environment, the number of particles can be reduced to reduce the amount of computation, and when multipath is significant, the number of particles can be increased to increase the positioning accuracy. it can.

  The principle of the present invention can be applied not only to a TDOA type positioning system but also to a TOA (Time Of Arrival) type or RSS (Received Signal Strength) type positioning system. In this case, it is necessary to change the model used in Equation (6), but the configuration of a hierarchical particle filter can be applied as it is.

  As described above, the configuration and details of the embodiment exemplified here can be changed without departing from the gist of the present invention. Further, the illustrative components and procedures may be changed, supplemented, or changed in order without departing from the spirit of the invention.

  Fields in which the present invention can be used include, for example, home networks, office automation, shopping centers, medical environments, port facilities, ITS, and the like.

  In home networks, as mentioned above, it is considered that networking of digital home appliances and mobile devices in the home will progress with the penetration of the ubiquitous society, and the present invention can be applied to terminal position detection in such home networks. It is.

  As for office automation, there are many cases where a wired / wireless LAN network is built in an office environment, and the position of a terminal can be detected by incorporating the positioning system of the present invention into such a LAN base station network. Become.

  In shopping centers such as supermarkets, there is a need for position information recognition for product management. In addition, grasping the rate of attracting customers based on the grasp of the location of the shopping cart in the store, application to product placement, and the like are also potential needs. The present invention can be applied to an application in which several base stations are provided in a store and a signal is received from a terminal installed in a product or cart to perform positioning.

  In medical facilities such as hospitals, there is a high need for drug management and patient position determination, and the present invention can be applied to such fields. Also, in today's aging society, care is often performed in the home, and there is a need for a home network and security network that can determine the position of the elderly and the sick. is there.

  In harbor facilities, there is a high need for grasping the location of containers and the like. This is a facility where it is easy to construct a network using a wireless LAN, a power line, etc. For example, the present invention can be applied using a container as a mobile terminal.

  As a part of ITS (Intelligent Transport System), a technology for identifying position information of a vehicle has attracted a great deal of attention, but at present, only a technology for each vehicle to recognize its own position by GPS. However, in the future, construction of an ITS infrastructure for networking roadside belts, traffic lights, and vehicles is also being studied. The present invention can be applied as a tool for recognizing positions of an ITS network and individual vehicles, for example.

It is a figure which shows an example of the positioning system by the conventional GPS. It is a figure which shows an example of the positioning system by general TDOA. It is a figure which shows an example of the ubiquitous sensor network in an indoor environment. It is a figure which shows the system model of the positioning system in a sensor network. It is a flowchart which shows an example of the conventional positioning process based on the system model of FIG. It is a flowchart which shows an example of the positioning process of this invention based on the system model of FIG. It is a figure which shows the structural example of the positioning signal processing part using the conventional Newton method. It is a figure which shows the structural example of the positioning signal process part by one Embodiment of this invention. 4 is a flowchart illustrating an example of a first-tier particle filter estimation process according to an embodiment of the present invention. 3 is a flowchart illustrating an example of a second-tier particle filter estimation process according to an embodiment of the present invention. It is a graph which shows the positioning result by computer simulation, Fig.11 (a) is based on the conventional Newton method, FIG.11 (b) is based on the hierarchical particle filter of this invention. It is a graph which shows the effect of error correction by the particle filter of the 1st hierarchy in the tracking process of a mobile terminal. It is a graph which shows the characteristic of the root mean square error when changing the average delay time (lambda) in a line-of-sight environment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Vehicle 11-14 Satellite 15 Terminal 21-21 Base station 26 Signal processing part 40 Terminal 41-44 Node 46 Signal processing part 47-1-4 Buffer 48-1-6 TDOA calculation part 49 Memory 70 Newton method position estimation part 81 -1-6 1st particle filter 82 position estimation part

Claims (10)

A system for estimating the position of a terminal,
A terminal that emits a predetermined signal;
A plurality of nodes having known position information, receiving the predetermined signal, and measuring a reception time; and
A signal processing unit that receives the reception time from each node, obtains a reception time difference between the nodes, and estimates the position of the terminal from the reception time difference and position information of each node; and
The said signal processing part correct | amends the said reception time difference using a 1st particle filter, The system characterized by the above-mentioned. The system of claim 1, comprising:
The signal processing unit estimates the position of the terminal from the corrected reception time difference and position information of each node using a second particle filter. The system according to claim 1 or 2, wherein
The first particle filter is modeled on the assumption that the time change of the reception time difference is constant. An apparatus for estimating a position of a terminal in a network including a plurality of nodes whose position information is known,
Means for obtaining a signal reception time difference between the terminal and each node;
Means for correcting the reception time difference using a first particle filter;
An apparatus comprising: means for estimating the position of the terminal from the corrected reception time difference and position information of each node. The apparatus according to claim 4, comprising:
The estimation means estimates the position of the terminal from the corrected reception time difference and position information of each node using a second particle filter. An apparatus according to claim 4 or 5, wherein
The first particle filter is modeled on the assumption that the time change of the reception time difference is constant. A method for estimating the position of a terminal,
The terminal emits a predetermined signal;
A plurality of nodes whose position information is known receive the predetermined signal and measure a reception time;
Obtaining the reception time difference between the nodes from the reception time of each node;
Correcting the reception time difference using a first particle filter;
Estimating the position of the terminal from the corrected reception time difference and position information of each node. The method of claim 7, comprising:
The method of estimating the location of the terminal includes estimating the location of the terminal from the corrected reception time difference and location information of each node using a second particle filter. A method for estimating a position of a terminal in a network including a plurality of nodes whose position information is known,
Obtaining the signal reception time difference between the terminal and each node;
Correcting the reception time difference using a first particle filter;
Estimating the position of the terminal from the corrected reception time difference and position information of each node. The method of claim 9, comprising:
The estimating includes estimating the position of the terminal from the corrected reception time difference and position information of each node using a second particle filter.

Images