KR101628154B1 - Multiple target tracking method using received signal strengths - Google Patents

Abstract

The present invention relates to a multiple target tracking method using a received signal strength indicator (RSSI), and more specifically, to a multiple target tracking method using an RSSI, which generates RSSI data from a plurality of tracking signals detected by a radar, extracts only a tracking signal present within a certain effective distance among tracking signals in which the RSSI data is generated, generates the data related probability of the extracted tracking signal by using the RSSI data, and obtains a state vector computation value and an error covariance value for a target by using a tracking filter to which the data related probability is applied, thus updating the track of the target. According to the present invention, different RSSI data by target is applied to Joint Probability Data-Association (JPDA) or Cheap Joint Probability Data-Association (CJPDA), thus improving the performance of a data related technique.

Description

[0001] MULTIPLE TARGET TRACKING METHOD USING RECEIVED SIGNAL STRENGTHS [0002]

The present invention relates to a multi-target tracking method using received signal strengths, and more particularly, to a multi-target tracking method using received signal strengths to generate received signal strength data from a plurality of target signals detected by a radar, The method includes generating a data association probability of the extracted target signal using the received signal strength data after extracting only a target signal existing within an effective distance, performing a state vector calculation on the target using the tracking filter to which the data association probability is applied And a multi-target tracking method using the received signal strength to update the trajectory of the target by obtaining a covariance and an error covariance.

Multi-target tracking technology is a technique that simultaneously tracks multiple targets detected by a radar. In the multi-target tracking technique, high-precision tracking technology is used to reduce the rate of trafic perturbation for a large number of moving targets and increase the retention rate of a large number of moving targets, thereby enabling accurate tracking of a large number of targets in real time need.

1 is a flowchart illustrating a conventional multi-target tracking method.

Referring to FIG. 1, the detected results for each of the targets detected by the radar can be calculated through information such as distance, speed, and acceleration of each target. The wakes in which each of the targets are moving include a data associating technique for defining a relationship between a state vector measurement of each target and a wake based on the detection result for each of the targets and improving the wake of each of the targets over time Outgoing filter technique.

Thus, the performance of the multiple target tracking technique may vary depending on the data association technique. The data association technique includes a nearest neighbor technique (NN) capable of fast operation, a joint probability data association technique (JPDA) having a high complexity but showing accurate tracking performance, and the joint probability data association technique (JJPDA: Cheap Joint Probability Data-Association (CJPDA)) that considers only partial events can be used to reduce the complexity of JPDA.

On the other hand, in order to improve the performance of the multi-target tracking technique, a technique of applying the signal intensity information of the target obtained from each target, not only the distance information of each target, to the data association technique has been developed. However, if the signal strength information of a target that changes according to the type of target and posture characteristics of the target is not accurately reflected, a technique using the signal strength information of the target in the data association technique may deteriorate the performance of the multi-target tracking technique have. That is, when the received signal strength data of the target is applied to the multi-target tracking method, since the target's wake-up rate has been updated under the assumption that the average signal-to-noise ratio per target is the same, The performance may deteriorate.

Therefore, when the signal strength information of the target is used to improve the performance of the multi-target tracking technique, adaptive signal strength information according to the target needs to be applied.

KR 1402206 B1 KR 1001559 B1

1. Donald Lerro and Yaakov Bar-Shalom, "Automated Tracking with Target Amplitude Information," 2875-2880

The present invention has been proposed in order to solve the above problems. By improving the performance of a data association technique by applying different received signal strength data for each target to a joint probability data association technique or a simple joint probability data association technique, And to provide a multi-target tracking method using signal strength.

In order to achieve the above object, a multi-target tracking method using the received signal strength according to the present invention includes: a target position extracting step of extracting target position data from a target signal detected by a radar; Generating a trajectory of the target and a state vector prediction for the target from the extracted positional data of the target using a tracking filter; A target data generation step of generating target data including the distance of the target, the speed of the target, and the received signal strength data from the target, from the extracted positional data of the target; A gating step of extracting a target signal existing within a predetermined effective distance from a final navigation position of a target generated using the received signal strength data of the target signals from which the target data is generated; A data associating step of generating a data associating probability between the final navigation position of the generated target and the position of the extracted target signal using the received signal strength data from the target; And a navigation updating step of inputting a data association probability of the target into a tracking filter to update the generated map of the generated target; . ≪ / RTI >

At this time, in the target data generation step, the received signal strength data from the target may vary depending on the type of target and the posture of the target.

The data associating step may further include a step of calculating a state vector measurement value (x _j (k + 1)) calculated from a final trajectory position of the generated target and a position of the extracted target signal and a state vector predicted value computing a x ^t (k + 1)) of Innovation values _{(v j, t (k +} 1) differ) and effective to produce an effective matrix (Ω) from the innovation value _{(v j, t (k +} 1) ) A matrix generation step; And means for deriving a generator matrix from the valid matrix Ω and using the generator matrix to calculate the computed state vector measurements x _j (k + 1) An association probability generation step of generating a probability (? _{J, t} ); . ≪ / RTI >

Here, in the association probability generation step, the derived probability matrix? Is applied to a joint probability data association (JPDA) to calculate a state vector measurement value x _j (k + 1) ( _{J, t} ) to be associated with the target. At this time, the data association probability ( _{j, t} ) to which the joint probability data association technique (JPDA) is applied can be calculated by the following equation.

Also, in the association probability generation step, the derived generated matrix? Is applied to a simple joint probability data association (CJPDA), and the calculated state vector Z _j (k + 1)) may generate a data association probability (? _{J, t} ) to be associated with the target. At this time, the data association probability ( _{j, t} ) to which the simple association probability data association technique (CJPDA) is applied can be calculated by the following equation.

And

In particular, in the association probability generation step, the received signal strength (p ₁ ^? (A _j ^t )) can be calculated by the following equation.

및

And

(X (k + 1)) and an error covariance matrix (x (k)) by inputting the data association probability of the target as a filter gain weight of the tracking filter, P (k + 1)), thereby updating the wake of the generated target. At this time, the state vector calculation value (x ^t (k + 1)) and the error covariance matrix (P ^t (k + 1)) of the target can be calculated by the following equation.

및

And

 According to the multi-target tracking method using the received signal strength according to the present invention, not only the distance data of the target but also the received signal strength data of each target are applied to the combined probability data association technique, The performance of the data association technique used in the multi-target estimation method can be improved.

In addition, according to the multi-target tracking method using the received signal strength according to the present invention, by using the average of the plurality of accumulated received signal strength data as the received signal strength data for the target, data The error of the association probability can be reduced.

BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a flow chart illustrating a conventional multi-target tracking method.
2 is a block diagram illustrating a multi-target tracking apparatus using received signal strength according to the present invention.
3 is a flowchart illustrating a multi-target tracking method using received signal strength according to the present invention.
4 is an example of an effective matrix (Ω) generated in the effective matrix generation step among the multiple target tracking methods using the received signal strength according to the present invention.
FIG. 5 illustrates examples of a generable matrix θ derived from an association probability generation step among multiple target tracking methods using received signal strengths according to the present invention.

It is noted that the technical terms used herein are used only to describe specific embodiments and are not intended to limit the invention. Also, the technical terms used herein should be interpreted in a sense that is generally understood by those skilled in the art to which the present disclosure relates, unless otherwise specifically defined in the present specification, Or shall not be construed to mean excessively reduced. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are intended to provide further explanation of the invention as claimed. But is to be understood as including all modifications, equivalents, and alternatives falling within the scope of the appended claims.

Hereinafter, an embodiment of a multi-target tracking method using received signal strength according to the present invention will be described in detail with reference to the accompanying drawings.

First, a multi-target tracking device used in a multi-target tracking method using received signal strength according to the present invention will be briefly described.

2 is a block diagram illustrating a multi-target tracking apparatus using received signal strength according to the present invention.

2, an embodiment of the multi-target tracking apparatus 1 using the received signal strength according to the present invention includes a target detection unit 1000, a gate unit 2000, a data association unit 3000, And may include a management unit 4000. The target detection unit 1000 acquires the target signal 100 detected by the radar 1100 and extracts the positional data 110 of the target from the target signal 100. The target detection unit 1000 further includes target data 310 including target distance 310, target speed 320 and received signal strength 330 data from the target from the target position data 110 300). The gate unit 2000 extracts a target signal 100 within a predetermined effective distance from the target signal 100 in which the target data 300 is generated. Here, the extraction of the target signal 100 existing within the predetermined effective distance among the target signals 100 in which the target data 300 is generated can be expressed as gating. The data associating unit 3000 uses the received signal strength data 330 from the target 10 to calculate the data between the final navigation position 211 of the generated target and the position 340 of the extracted target signal And generates an association probability 700. The trace manager 4000 uses the trace filter 4100 to generate or update the trajectory 200 of the target. More specifically, the trace management unit 4000 generates a movement trajectory 210 in which the target moved from the extracted positional data 110 of the target using a tracking filter, The input tracing filter 4100 can be used to update the generated trajectory of the target.

FIG. 3 is a flowchart illustrating a multi-target tracking method using the received signal strength according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating a method of tracking an effective matrix (Ω) FIG. 5 is a graph showing examples of a generable matrix (?) Derived from an association probability generation step among multiple target tracking methods using received signal strength according to the present invention.

Referring to FIG. 3, an embodiment of a multi-target tracking method using received signal strength according to the present invention includes a target position extracting step S310, a snapshot generating step S320, a target data generating step S321, Step S330, a data association step S350, and a navigation update step S340, S360, a wake removal and management step S370, and the like.

In the target position extracting step S310, the position data 110 of the target is extracted from the target signal 100 detected by the radar 1100. More specifically, in the target location extraction step (S310), the target detection unit (1000) searches for a plurality of moving targets (10) from a radar (1100) The target signal 100 can be obtained. The target detection unit 1000 may then convert the target signal 100 to extract the target location data 110. That is, in the target position extraction step S310, the target detection unit 1000 detects the signal 100 for the moving plurality of targets 10 and outputs the detected target signal 100 to the target azimuth angle Data or target Doppler data, and the azimuth data of the transformed target or the Doppler data of the target can be converted into the target position data 110 again. Here, the target signal 100 obtained from the radar 1100 may be a target signal for a plurality of targets 10. Of course, prior to the target position extraction step (S310), initialization of the wake is performed.

In the trail generating step S320, a trail filter 4100 is used to generate the trajectory 200 of the target and the state vector predictor for the target from the extracted position data 110 of the target. More specifically, in the generation step (S320), the tracing filter (4100) of the navigation controller (4000) is used to move the extracted target from the position data (110) It is possible to generate the state vector predictions for the target, which means the wake 210 and the wake to move.

That is, in the trail generation step S320, the state vector 221 for the target t is calculated from the extracted positional data 110 of the target to the kth time, and the trajectory 210, Can be generated. In the trail generating step S320, the tracing filter 4100 is used to calculate, from the target position data 110, the trajectory of the t-th target in the (k + 1) A state vector prediction value and an error covariance matrix.

Here, the set of distances up to the t-th target during the k-th time to obtain the state vector predictions for the t-th target in the (k + 1) -th time can be calculated from the position data 110 of the t- . Here, the state vector prediction value for the target may be a state vector calculated from the position of the t-th target in the (k + 1) -th time and the position of the t-th target in the k-th time. The tracking filter 4100, A Kalman filter that can derive a state vector calculation value and an error covariance reflecting the target state vector prediction value and the error covariance matrix and the target received signal strength data 330.

In this case, when the set of distances for the t-th target from the target position data 110 to the k-th time is calculated, the predicted value x ^t (k + 1) for the state vector for the ^t- 1)) and the error covariance matrix P ^t (k + 1) for the state vector of the t-th target of the (K + 1)

Lt; / RTI >

Here, x ^t (k | k) is a state vector of the t-th target of the k-th time, which is expressed as x ^t (k) = [z _{X, k} , z _{Y, k} ] Location information. (K) is an error covariance matrix of the target at the k-th time, and F '(k) is a state transformation matrix of the target dynamic motion model, (K) is a covariance matrix representing the process noise in the target dynamic motion model.

That is, through the above-mentioned [Expression 1] and [Equation 2], in the trail generation step S320, the state vector of the t-th target from the extracted positional data 110 of the target up to the K- (x ^t (k)) to calculate a k + 1 th of the time t the second target to the state vector ^{(x t (k + 1)} ) predicted value and the error covariance matrix ^{(P t (k + 1)} ) by using the have.

In the target data generation step S321, the target data including the distance from the position data 110 of the extracted target to the target, the speed of the target, and the received signal strength data from the target through the radar sensor 1100 300). More specifically, in the target data generation step (S321), the generated wakefulness (200) of the generated target from the position data (110) of the target of the target signal (100) The distance data 310 to the target used to update, the velocity data 320 of the target, and the received signal strength data 330 from the target.

That is, in the target data generation step S321, the distance 310 from the extracted target position data 110 to each target of the moving target 10, the velocity 320 of each target, And the received signal strength data 330 from each target. Here, the distance data 310 to the target and the distance to the t-th target may be the same. Here, the received signal strength data 330 from the target may vary depending on the type of the target and the posture of the target.

In the gatewaying step S330, the target signal 100 existing within a predetermined effective distance from the final navigation position 211 of the generated target among the generated target signals 100 is extracted from the target data 300 do. More specifically, in the gateing step S330, in order to estimate the position of the t-th target in the (K + 1) -th time, the target data 300 is generated at the k- It is possible to extract only the target signal 100 existing within a predetermined radius centering on the final trajectory position 211 of the t-th target in FIG.

That is, in the gatewaying step S330, in order to estimate the next position at which the target 10 moves from the past warnings 210 of the target detected and generated by the radar sensor 1100, Only the target signal 100 existing within a predetermined radius around the final position 211 of the target 10 in the past trajectory 210 of the target 100 is extracted. At this time, the target signal 100 detected by the radar may be larger than the actual number of the target 10, for example, ten target signals 100 for two moving targets 10 may be radar Can be detected.

More specifically, in the gating step S330, each target moves from the wake of one or more targets 210 generated from one or more target signals 100 until the kth time obtained by the radar sensor 1100 In order to estimate the next position, the final position 211 of each target 10 in the past trail 210 of each target among the plurality of target signals 100 detected by the radar at the (k + 1) , and extracts only the target signal 100 existing within a predetermined radius around the position 211 of each target in the k-th time.

Here, the extraction of the target signal 100 existing within the predetermined effective distance among the signals 100 of the target in which the target data 300 is generated can be expressed as gating. Here, the predetermined effective distance may be a constant that limits a position at which the target can move in a (k + 1) -th time.

In this case, when the number of target signals 100 existing within a predetermined effective distance among the plurality of target signals 100 detected by the radar at the (k + 1) th time is m, The state vector (x ^t (k + 1)) 222 of the t-th target existing within the distance is given by the following equation,

. ≪ / RTI >

Here, the Z ^t _m (k + 1) is the m-th in the target signal position information for the k + 1 t-th sign of the second time (z _{X, k + 1,} z _{Y, k + 1)} and speed information (z _{'X, k + 1, z} ' Y, wherein Z ^t means the _{k + 1),} and _a (k + 1) is the signal strength data received from the k + 1 t-th sign of the second time in an m-th target signal ( 330). At this time, the type of the t-th target, the size of the target, and the attitude information of the target can be deduced from the received signal strength data (Z ^t _a (k + 1)) 330, The reliability of the data association probability of the target is obtained through the distribution of the received signal strengths and the average signal-to-noise ratio, which are different between the different targets 10, using the received signal strength information Z ^t _a (k + .

In the data associating step S350, the target signal position (211) existing within the predetermined effective distance from the final navigation position (211) of the generated target using the received signal strength data 330 from the target (10) 340). ≪ / RTI > That is, in the data associating step S350, it is determined whether or not there is a difference between the final navigation position 211 of each generated target using the received signal strength data 330 from each of the targets 10, m < / RTI > More specifically, the data association step S350 may include an effective matrix generation step (not shown) and an association probability generation step (not shown).

In the valid matrix generation step, the state vector measurement value x _j (k + 1) calculated from the final navigation position 211 of the generated target and the position 340 of the m extracted target signals, The innovation value v _{j, t} (k + 1)) is calculated by computing an innovation value v _{j, t} (k + 1) which is the difference between the state vector prediction value x ^t (?). That is, in the effective matrix generation step (S510), the state vector (x _j (k + 1)) calculated from the final navigation position 211 of each generated target and the position 340 of the m extracted target signals, )) and the generated status for each target vector predicted values (Innovation value indicating a difference between x ^t (k + 1)), (v _{j, t} (k + 1)) for operation and the innovation value (v _{j , t} (k + 1)).

At this time, the effective matrix (?) (500)

Lt; / RTI >

Here, w _{j, t} denotes "1" if the position of the j-th extracted target signal as an element of the valid matrix is within a predetermined effective range centered on the t-th target in the k-th time, 0 ", where m is the number of target signals 100 present within the predetermined effective range, T is the number of targets that generated the trajectory, and r is a predetermined effective range value.

510, 520, 530, 540, 550, 560, 570 that can be generated from the effective matrix (?) 500 in the association probability generation step (not shown) (? _{J, t} ) (700) to be associated with the target by using the generated state vector (x _j (k + 1)) using the generator matrix do.

4 and 5, in the association probability generation step (not shown), when two targets t and three target signals j are within a predetermined effective range, 5 500, 530, 550, 560, 570, 530, 530, 530, 530, 530, 530, 580 may be derived as shown in FIG. At this time, the method for deriving the generable matrix is such that only one target signal (j) exists in one target (t) and one target signal (j) exists only in one target To derive a matrix.

510, 520, 530, 540, 550, 560, 570, and 580 of the derived probability matrix (not shown) (? _{J, t} ) (700) to be associated with the target by applying the computed state vector (x _j (k + 1)) (400) to the Joint Probability Data- Can be generated.

At this time, the data association probability (β _{j, t} ) (700), to which the joint probability data association technique is applied,

Lt; / RTI >

Here, c is a normalized coefficient, V is a valid region interval, and πr | S (k) | ^1/2 , where r is the gateing threshold constant for determining the gate region, and φ is the number of signals in the current state, m-1, where N [v _{j, t} (k + 1); 0; S (k + 1) ] is a v _j, which is a Gaussian distribution is zero mean and variance of the _t S (k + 1), wherein τ _j (θ), δ _t from all generated available matrix (θ) Is a variable having a value of 1 only when the t-th target (track) of the t-th target (track) is associated with the j-th extracted target signal and P ^t _D is a previously defined system variable defined as the detection probability of the t-th target (track). In addition, the p ₁ ^η of (a ^t _j) is the j-th and the received signal strength of the extracted target signal, wherein a ^t _j is a PDF (Probability density function), the p ₀ ^η (a ^t _j) is the noise signal Means the received signal strength.

At this time, the received signal strength ^(η ₁ p (a ^t _j)) and the received signal strength of the noise signal _{^{(p 0 η (a t j}} )) of the j-th extracted the target signal is obtained using the equation,

Lt; / RTI >

Here, d represents the average signal-to-noise ratio of the target signal and has a value proportional to the received signal strength received from the FMCW (Frequency Modulated Continuous Wave) radar.

Since the present invention applies different target signal strengths, applying Equation (10) and Equation (11) to Equation (8) results in different signal-to-noise ratios (λ _i ) β _{j, t} ) 700 to obtain a data association probability β _{j, t} 700 reflecting the received signal strength data from the target.

510, 520, 530, 540, 550, 560, 570, and 580 may be combined with a simple joint probability data association technique ( _{J j, t} ) to be associated with the target by applying the computed state vector x _j (k + 1) to the CJPDA (Cheap Joint Probability Data Association).

At this time, the data association probability (? _{J, t} ) to which the simple association probability data association technique is applied is expressed by the following equation

Lt; / RTI >

이고, 상기 p₁ ^η(a_j ^t)는 j 번째 추출된 표적 신호의 수신 신호 세기이며, 상기 p₀ ^η(a^t _j) 는 노이즈 신호의 수신 신호 세기이다.Here, the above-mentioned? _J is an amplitude likelihood ratio

And wherein ^η ₁ p (a _j ^t) is the received signal strength of the j-th extracted the target signal, the ^η ₀ p (a ^t _j) is the received signal strength of the noise signal.

Here, the received signal strength p ₁ ^? (A _j ^{? T} )

Lt; / RTI >

Here, d represents the average signal-to-noise ratio of the target signal and is proportional to the received signal strength received from the FMCW radar.

The invention, because of applying a different target-specific signal strength, wherein if the [Equation 10] applied to the above [Equation 9] a different target-specific signal-to-noise ratio (λ _i) is the data associated probability (β _{j, t)} To obtain the data association probability (? _{J, t} ) in which the received signal strength data from the target is reflected.

Here, an average value (d _t ) of the received signal strength accumulated for a specific time to derive the signal-to-noise ratio (λ _i )

Lt; / RTI >

Here, L denotes an accumulated time for calculating the signal intensity average.

The average signal-to-noise ratio d expressed in Equation (8) for obtaining the association probability in the joint probability data association technique as well as Equation (9) for obtaining the association probability in the simple joint probability data association technique may be different from each other (11) can be applied.

In the navigation updating step S360, the data association probability of the target is input to the tracking filter 4100 to update the generated target wake 200. [ More specifically, in the navigation update step (S360), the data association probability generated by the joint probability data association technique or the simple association probability data association technique is input to the tracking filter 4100, the generated trajectory 200 of the generated target can be updated by obtaining the state vector operation value x (k + 1) for the t-th target and the error covariance matrix P (k + 1) of the state vector .

The state vector calculation value x (k + 1) for the t-th target in the (k + 1) th time and the error covariance matrix P (k + 1)

Lt; / RTI >

Here, the S (k + 1) is the error covariance matrix of the innovation, is the inverse of the S ^-1 (k + 1) is the S (k + 1), wherein W (k + 1) is a filter of the tracking filter (K + 1) is an error covariance matrix of the measurement noise, H (k + 1) is a state transformation matrix indicating a relationship between a measured value and a state variable, H (k + 1) is an inverse matrix of H (k + 1), v _t (k) is a weighted sum of Innovation, β ₀ is a data association probability and I is a unit matrix.

In the navigation update step S360, the target navigation update of the target is performed based on the filter gain weight W (K + 1) of the tracking filter 4100 by referring to Equation (12) The generated target wakefull 200 can be updated by calculating the target state vector calculation value x (k + 1) and the error covariance matrix P (k + 1).

The navigation is updated based on the tracking filter system constants so generated in the navigation update step (S360). In the trajectory elimination and management step (S370), trajectories are maintained and / or eliminated for the updated trajectory results. Of course, a step S340 of creating a new track may be configured before the track removal and management step S370.

It will be apparent to those skilled in the art that many other modifications and variations are possible in light of the above teachings and the scope of the present invention should be construed on the basis of the appended claims something to do.

1: Multi-target tracking device using received signal strength according to the present invention
10: Target
100: signal of the target
110: Position data of the target 111: Distance to the t-th target
200: Target wake
210: Movement trajectory to which the target moved
211: the final trajectory position of the target in the trajectory to which the target moved
220: Wand to move the target
300: target data
310: distance data to the target 320: velocity data of the target
330: received signal strength data from the target
331: Signal strength of the j-th extracted target signal
332: Signal strength of noise signal
340: Target signal position existing within a predetermined effective distance
500: effective matrix (Ω)
510, 520, 530, 540, 550, 560, 570, 580: generating matrix (?) That can be generated from the effective matrix (500)
700: Data association probability (? _{J, t} )
1000: target detection unit 1100: radar sensor
2000: Gating section
3000:
4000:
4100: Trace filter

Claims (10)

In a multiple target tracking method,
A target position extracting step of extracting target position data from the target signal detected by the radar sensor;
Generating a trajectory of the target and a state vector prediction for the target from the extracted positional data of the target using a tracking filter;
A target data generation step of generating target data including the distance of the target, the speed of the target, and the received signal strength data from the target, from the extracted positional data of the target;
A gating step of extracting a target signal existing within a predetermined effective distance from a final navigation position of a target generated using the received signal strength data of the target signals from which the target data is generated;
A data associating step of generating a data associating probability between the final navigation position of the generated target and the position of the extracted target signal using the received signal strength data from the target; And
A navigation update step of inputting the data association probability of the target into the tracking filter and updating the generated map of the generated target;
Wherein the multi-target tracking method is based on received signal strength.
The method according to claim 1,
In the target data generation step,
Wherein the received signal strength data from the target is dependent on the type of target and the attitude of the target.
The method according to claim 1,
Wherein the data associating step comprises:
A state vector prediction value x ^t (k + 1) for the generated target and a state vector measurement value x _j (k + 1) calculated from the position of the extracted target signal of the generated target, difference of the effective value Innovation matrix generation step of computing a _{(v j, t (k +} 1)) and to generate an effective matrix (Ω) from the innovation value _{(v j, t (k +} 1)); And
Deriving a generator matrix from the effective matrix Ω and using the generator matrix to calculate the calculated state vector measurements x _j (k + 1) (? _{j, t} );
Wherein the multi-target tracking method is based on received signal strength. The method according to claim 1,
In the navigation update step,
Wake update of the target is the filter gain state of the target desired to be updated by entering a weight vector computed value ^{(x t (k + 1)} ) and the error covariance matrix (P ^t (k of the tracking filter the data associated probability of the target +1)), thereby updating the generated trajectory of the target.
The method of claim 3,
In the association probability generation step,
Applying the derived generated matrix? To a joint probability data association (JPDA) to determine whether the computed state vector measure x _j (k + 1) And generating an association probability (? _{J, t} ) based on the received signal strength.
The method of claim 3,
In the association probability generation step,
Applying the derived generated matrix? To a simple Joint Probability Data-Association (CJPDA) so that the computed state vector measure x _j (k + 1) (? _{J, t} ) to be transmitted to the mobile station based on the received signal strength.
5. The method of claim 4,
The state vector computed value (x ^t (k + 1)) and the error covariance matrix (P ^t (k + 1)

And Equation

(Here, the S (k + 1) is the error covariance matrix of the innovation, the S ^-1 (k + 1) is of an inverse of the S (k + 1), wherein W (k + 1) is a tracking filter In the present invention, H (k + 1) is a state transformation matrix indicating a relationship between a measured value and a state variable, and R (k + 1) is an error covariance matrix of the measured noise. (K + 1) is an inverse matrix of H (k + 1), v _t (k) is a weighted sum of innovation, and β ₀ is a data association probability , Where I is a system matrix). ≪ / RTI >
6. The method of claim 5,
The data association probability (β _{j, t} ) applying the joint probability data association technique (JPDA)

And Equation

And Equation

Where c is a normalized coefficient, V is defined as a valid region period, πr S S (k) ^{1/2 1/2} , r is a gate threshold value for determining a gate region, wherein φ is here, m-1 in number of signals in the current state, the _{N [v j, t (k} + 1); 0; S (k + 1)] is a v _j, zero mean of _t and variance S (k) is a Gaussian distribution having a value of (k + 1), and τ _j (θ) and δ _t have a value of 1 only when the t th target (track) in all possible matrices Variable,

Is a t detection probability of the first targets (track), and the system variables are defined in advance, wherein p is ₁ ^η (a ^t _j) is the signal strength of the j-th measurement, wherein a ^t _j is a PDF (Probability density function), and , D is a mean signal-to-noise ratio of the target signal, and is a value proportional to a received signal strength received from the FMCW radar).
The method according to claim 6,
The data association probability (β _{j, t} ) applying the simple joint probability data association technique (CJPDA)

(Where, _i is the signal-to-noise ratio

And wherein p ₁ ^η (a _j ^t) is the received signal strength of the j-th extracted the target signal, wherein p ₀ ^η (a ^t _j) is being calculated by the received signal strength Im) of the noise signal Multi - target tracking method using received signal strength.
10. The method according to claim 8 or 9,
The received signal strength p ₁ ^? (A _j ^t )

And Equation

(Where d _t is an average value of the received signal strengths accumulated during a specific time in order to derive a signal-to-noise ratio (? _I ) for each target, and L is an accumulated time for calculating a signal intensity average) Wherein the multi-target tracking method is based on received signal strength.

Images