WO2020209053A1

WO2020209053A1 - Position estimating method, position estimating system, position estimating server, and position estimating program

Info

Publication number: WO2020209053A1
Application number: PCT/JP2020/013140
Authority: WO
Inventors: 阿部　直人; 瀬下　仁志; 宏志小西; 真道細田; 寛坂本; 智明小川
Original assignee: 日本電信電話株式会社
Priority date: 2019-04-08
Filing date: 2020-03-24
Publication date: 2020-10-15
Also published as: JP2020173122A

Abstract

A position estimating method for transmitting and receiving a measurement signal and a response signal between a wireless station provided with a plurality of (n) known antennas installed at known positions different from each other and a wireless terminal station provided with a terminal antenna, and measuring a round-trip time [RTT] or a received signal strength indication [RSSI] to estimate the position of the wireless terminal station, wherein the method comprises, for RTT/RSSI measurement values measured between each known antenna and the wireless terminal station: a step 1 for performing statistical processing for each antenna to obtain statistically processed RTT/RSSI; a step 2 for calculating instantaneous value coordinates of the wireless terminal station from the statistically processed RTT/RSSI; a step 3 for obtaining stabilized coordinates from which the vibration components below the threshold for the instantaneous value coordinates have been removed; and a step 4 for performing a moving average process on the stabilized coordinates and outputting statistically processed stabilized coordinates.

Description

Position estimation method, position estimation system, position estimation server and position estimation program

The present invention relates to a position estimation method, a position estimation system, a position estimation server, and a position estimation program for estimating the position of a wireless terminal station based on the known positions of a plurality of antennas.

Using a distributed antenna, the position of the wireless terminal station is estimated from the round-trip delay time RTT (Round Trip Time) and received signal strength RSSI (Received Signal Strength Indicator) between multiple antennas whose positions are known and the terminal. There is a method (Non-Patent Document 1).

In Non-Patent Document 1, the position estimation formula by RTT and the position estimation formula by RSSI are completely independent, and each is a method of obtaining coordinates, and two types of coordinates with different accuracy can be obtained. Furthermore, the calculation process was completely independent, and the position estimation calculation could not be integrated in consideration of the accuracy of both. Further, in order to obtain one position estimation result, it was necessary to obtain the two coordinates separately and then integrate the coordinates.

Non-Patent Document 1 does not have a specific formula or calculation method for estimating the position in two or three dimensions. When estimating the 3D position using RTT, it is easy to think of the position estimation by applying the GPS calculation method, but the specific formula and calculation method for estimating the 3D position using RSSI Is not shown. Further, neither RTT / RSSI nor a specific formula or calculation method for estimating the two-dimensional position is shown.

In the position estimation formula by RSSI of Non-Patent Document 1, it is necessary to determine the propagation loss coefficient α of the space by measuring it in advance according to the actual space.

Non-Patent Document 1 shows a method using a locus in addition to using a GPS calculation method, and states that when estimating a two-dimensional position by RTT, the intersection of two hyperbolas may be obtained. However, when the number of antennas is 3 or more, the number of hyperbolas is 2 or more, and there may be a case where there is no intersection where all the hyperbolas intersect. In that case, the position cannot be estimated. Similarly, when estimating the three-dimensional position using the locus in RTT, there may be cases where the number of antennas is 4 or more, the number of hyperboloids rotating in two leaves is 3 or more, and there is no point where all the hyperboloids intersect. In that case, the position could not be estimated. Similarly, in the case of RSSI, if the number of antennas is large, there may be a case where all the circles or spheres do not intersect, and in that case, the position cannot be estimated.

If there is an error in the RTT / RSSI measurement value, the coordinates of the position estimation result may deviate significantly, and the coordinates outside the range in which the terminal can move (exist) may be output.

Since Non-Patent Document 1 uses a distributed antenna, a cost for laying a coaxial cable or the like between the known position antenna and the AP is required.

An object of the present invention is to provide a position estimation method, a position estimation system, a position estimation server, and a position estimation program capable of accurately estimating the position of a wireless terminal station using RTT and / and RSSI.

The first invention transmits and receives a measurement signal and a response signal between a radio station having a plurality of known antennas installed at different known positions and a radio terminal station having a terminal antenna, and delays the round trip. In the position estimation method for estimating the position of the wireless terminal station by measuring the time RTT, the RTT measurement value measured between the known antenna i (i = 1, 2, ..., N) and the wireless terminal station. Step 1 to obtain the statistical processing RTT by performing statistical processing for each antenna, step 2 to calculate the coordinates of the wireless terminal station from the statistical processing RTT and output as instantaneous value coordinates, and vibration below the threshold value of the instantaneous value coordinates. Step 3 of removing minutes and converting instantaneous value coordinates to stabilized coordinates within a range that does not exceed the threshold for vibrations above the threshold, and round-trip averaging of the stabilized coordinates to output statistical processing stabilized coordinates. Step 4 and the like.

In the second invention, a measurement signal and a response signal are transmitted and received between a radio station having a plurality of known antennas installed at different known positions and a radio terminal station having a terminal antenna, and the received radio wave thereof. In the position estimation method for estimating the position of the wireless terminal station by measuring the intensity RSSI, the RSSI measurement value measured between the known antenna i (i = 1, 2, ..., N) and the wireless terminal station. Step 1 to obtain statistical processing RSSI by performing statistical processing for each antenna, step 2 to calculate the coordinates of the wireless terminal station from statistical processing RSSI and output as instantaneous value coordinates, and vibration below the threshold value of instantaneous value coordinates. Step 3 of removing minutes and converting instantaneous value coordinates to stabilized coordinates within a range that does not exceed the threshold for vibrations above the threshold, and moving averaging processing of the stabilized coordinates to output statistical processing stabilized coordinates. It has step 4 and the like.

A third invention is a position estimation system that estimates the position of a radio terminal station by the position estimation methods of the first and second inventions, in which RTT and / and are measured by a radio station having a plurality of n known antennas, respectively. The RSSI is transferred to the position estimation server, and the position estimation server calculates the statistical processing stabilization coordinates to obtain the position of the wireless terminal station.

A fourth aspect of the present invention is a position estimation system for estimating the position of a wireless terminal station by the position estimation method of the first to second inventions, wherein a plurality of n known antennas are cables of a known length from a wireless base station. It is a distributed antenna distributed via the radio base station, and the RTT and / and RSSI for each distributed antenna measured by the radio base station are transferred to the position estimation server, and the statistical processing stabilization coordinates are calculated by the position estimation server. The configuration is to find the position of the wireless terminal station.

A fifth invention is a position estimation server that estimates the position of a wireless terminal station by the position estimation method of the first to second inventions, and RTT and / and are measured by a radio station having a plurality of n known antennas, respectively. The RSSI is transferred, the statistical processing stabilization coordinates are calculated, and the position of the wireless terminal station is obtained.

The position estimation program of the sixth invention causes the computer to execute the process executed by the position estimation server of the fifth invention, calculates the statistical processing stabilization coordinates, and obtains the position of the wireless terminal station.

The present invention can accurately estimate the position of a wireless terminal station using RTT and / and RSSI. In addition, concrete position estimation can be performed in two dimensions and three dimensions.

Further, even if there is an error in the measured values of RTT or / and RSSI, the coordinates of the position estimation result will not be significantly deviated by taking the moving average of the stabilized coordinates by the coordinate adjustment unit.

It is also possible to adopt a configuration that estimates the position without using a distributed antenna, in which case it is not necessary to lay a cable between the known position antenna and the AP, and the cost can be reduced.

It is a figure which shows the whole structure example of the position estimation system of this invention. It is a figure which shows the configuration example of a radio station 10. It is a figure which shows the configuration example of a wireless terminal station 20. It is a figure which shows the example 1 of the sequence of the measurement signal and the response signal. It is a figure which shows the example 2 of the sequence of the measurement signal and the response signal. It is a figure which shows the other overall configuration example of the position estimation system of this invention. It is a figure which shows the configuration example of the position estimation server 30. It is a figure which shows the example of the positioning result calculated using the data measured in the test environment.

FIG. 1 shows an overall configuration example of the position estimation system of the present invention.
In FIG. 1, a measurement signal and a response signal are exchanged between two or more known antennas 11 of a radio station 10 whose position is known and a terminal antenna 21 of a radio terminal station 20 whose position is unknown, and RTT or Measure RSSI or both. The measured values are collected from each radio station 10 to the position estimation server 30 via the network. The position estimation server 30 performs position estimation calculation, and calculates and estimates the positions of the wireless terminal station 20 and the terminal antenna 21.

FIG. 2 shows a configuration example of the radio station 10 in the present invention.
In FIG. 2, the radio station 10 includes a signal transmitting unit 12 and a signal receiving unit 13 connected to a known antenna 11, an RTT measuring unit 14 and an RSSI measuring unit 15 connected to the signal receiving unit 13, and a clock 16. The position estimation server 30 is connected to the measurement unit 14 and the RSSI measurement unit 15.

FIG. 3 shows a configuration example of the wireless terminal station 20 in the present invention.
In FIG. 3, the wireless terminal station 20 includes a signal receiving unit 22, a signal transmitting unit 23, a control unit 24, and a clock 25, if necessary, connected to the terminal antenna 21.

(1) RTT / RSSI measurement
(1.1) RTT / RSSI measurement example 1
FIG. 4 shows Example 1 of the sequence of the measurement signal and the response signal.
In FIG. 4, a measurement signal is transmitted from the radio station 10 to the radio terminal station 20, and the radio terminal station 20 returns a response signal to the radio station.

(1.1.1) Measurement signal and response signal A wireless LAN action frame or management frame can be used for the measurement signal. Since the wireless terminal station 20 that has received them returns ACK, this can be used as a response signal. Of course, other types of frames such as data frames may be used for the measurement signal and the response signal. In addition to the wireless LAN, other wireless communication methods capable of exchanging the measurement signal and the response signal may be used.

(1.1.2) RTT measurement When RTT is measured, the radio station 10 records the time t1 using the clock 16 when transmitting the measurement signal from the RTT measurement unit 14 via the signal transmission unit 12. When the signal receiving unit 22 receives the measurement signal by the control unit 24, the wireless terminal station 20 transmits a response signal from the signal transmitting unit 23. At this time, the clock 25 may be used to record the reception time t2 of the measurement signal and the transmission time t3 of the response signal. When the radio station 10 receives the response signal to the signal receiving unit 13, the RTT measuring unit 14 records the time t4 using the clock 16.

The RTT measurement unit 14 can obtain the RTT including the terminal delay time (t3-t2), which is the delay time of the wireless terminal station 20, by calculating (t4-t1). This is called rawRTT. When measuring only rawRTT, the clock 25 of the wireless terminal station 20 is unnecessary.

If the wireless terminal station 20 is equipped with a clock 25 and can record t2 and t3, a method such as placing the terminal delay time (t3-t2) or t2 and t3 itself on the response signal or on another signal. May be transmitted to the radio station 10. Then, by calculating ((t4-t1)-(t3-t2)), the RTT excluding the terminal delay time can be obtained. This is referred to as a no-delay RTT.

(1.1.3) RSSI measurement When measuring RSSI, the radio station 10 acquires RSSI when the signal receiving unit 13 receives the response signal by the RSSI measuring unit 15.

(1.2) RTT / RSSI measurement example 2
FIG. 5 shows Example 2 of the sequence of the measurement signal and the response signal. Here, FTM (Fine Timing Measurement) of IEEE Std 802.11-2016 is used.

(1.2.1) In the RTT measurement radio station 10, the RTT measurement unit 14 transmits an Initial FTM Request from the signal transmission unit 12 to the radio terminal station 20 as an initiator, and transmits the measurement signal to the radio terminal station 20 a specified number of times. Request. When the signal receiving unit 22 receives the Initial FTM Request, the wireless terminal station 20 starts the operation of the control unit 24 as a responder. The wireless terminal station 20 transmits FTM1 as the first measurement signal from the signal transmission unit 23, and the control unit 24 records the transmission time as t1 using the clock 25. When the radio station 10 receives FTM1, it returns ACK as a response signal. At this time, the RTT measurement unit 14 uses the clock 16 to record the reception time as t2 and the transmission time as t3. Upon receiving the ACK, the wireless terminal station 20 transmits the next measurement signal FTM2. At this time, the control unit 24 uses the clock 25 to record the reception time t4 and the transmission time t1'. The previous transmission time t1 and reception time t4 are included in the second and subsequent measurement signals.

Similarly, the round trip between the measurement signal and the response signal continues until the number of times specified in the first request is reached, and the radio station 10 and the radio terminal station 20 record the transmission / reception time and the measurement signal of the previous time (t1, t4). ) Grant. As a result, the radio station 10 can receive the measurement signal including the previous (t1, t4), and together with the previous reception time t2 and transmission time t3 recorded by the RTT measurement unit 14, the (((t1, t4) By calculating t4-t1)-(t3-t2)), it is possible to obtain a non-delayed RTT in the round trip between the previous measurement signal and the response signal.

Here, in this sequence, the directions of the measurement signal and the response signal of the sequence of FIG. 4 are opposite to each other, but the radio station 10 triggers the non-delayed RTT between the radio station 10 and the radio terminal station 20. Can be measured.

(1.2.2) RSSI measurement When measuring RSSI, the radio station 10 acquires RSSI when the signal receiving unit 13 receives the measurement signals FTM1, FTM2, etc. by the RSSI measuring unit 15.

(1.3) Configuration example using distributed antennas In the system configuration shown in FIG. 1, the position of the known antenna 11 of the radio station 10 is known, and the measurement signal and the response signal are exchanged with the terminal antenna 21 of the radio terminal station 20 to perform RTT. Alternatively, RSSI was measured, but other configurations may be used.

For example, if a distributed antenna has already been installed and no new cable laying cost is incurred, the plurality of sets of radio stations 10 and known antennas 11 in FIG. 1 are combined with one radio base station 40 and a cable as shown in FIG. It may be replaced with a plurality of distributed antennas 42 connected via 41.

In this case, by correcting each of the raw RTT, the non-delayed RTT, and the RSSI in consideration of the loss and the delay due to passing through the cable 41, the measurement value equivalent to the configuration as shown in FIG. 1 can be obtained, and the same position can be obtained. An estimation calculation method can be used.

(1.4) Summary of RTT / RSSI measurement With the above method, the radio station 10 or radio base station 40 can measure part or all of raw RTT, no-delay RTT, and RSSI, and this can be measured by the location estimation server via the network. Tell 30.

(2) Position estimation FIG. 7 shows a configuration example of the position estimation server 30 in the present invention.
In FIG. 7, the position estimation server 30 includes a measured value statistical processing unit 31, a coordinate calculation unit 32, a coordinate stabilization unit 33, and a coordinate adjustment unit 34 connected to the radio station 10.

(2.1) Measured value statistical processing unit 31
Since the wireless terminal station 20 not only processes the measurement signal and the response signal but also performs other processes at the same time, the terminal delay time varies depending on the status of the other processes. Therefore, in the case of a configuration in which only raw RTT can be measured, the measured value statistical processing unit 31 performs statistical processing for each known antenna 11 to obtain a statistical processing RTT from which variations have been removed by methods such as removal of outliers, moving average, and regression. You may. In the case of non-delayed RTT, since the terminal delay time is not included and the variation is small, it is optional to perform statistical processing. It is optional to perform statistical processing on RSSI as well.

The measured value statistical processing unit 31 may calculate the reliability of the RTT measured value and the RSSI measured value as an index of accuracy for each known antenna 11. For example, if the number of measurements in the past fixed period is large, the reliability of the measured value of the known antenna 11 is high, and if the variance or standard deviation of the measured values in the past fixed period is small, the reliability of the measured value of the known antenna 11 is high. , Can be calculated. In addition, the reliability can be corrected according to the characteristics, location, and surrounding conditions of each known antenna 11. For example, the number of measurements in the past fixed period is s, the standard deviation is σ, the correction parameter for each known antenna 11 is a _size , the correction offset is b _size , and the standard deviation is a _sigma. If the correction offset of the standard deviation is b _sigma , the reliability w is w = (a _size · s + b _size ) / (a _sigma · σ + b _sigma ).
Can be obtained at.

(2.2) Coordinate calculation unit 32
Since it is not necessary to distinguish between raw RTT, no-delay RTT, and statistical processing RTT, the coordinate calculation unit 32 treats any of them as RTT and calculates the coordinates indicating the position of the wireless terminal station 20 as follows.

(2.2.1) Premise of coordinate calculation The number of known antennas 11 or distributed antennas 42 is n, and they are collectively referred to as known antennas i (i = 1, 2, ..., N).

The coordinates of the known antenna i are (x _i , y _i ) in the case of two dimensions and (x _i , y _i , z _i ) in the case of three dimensions. These are known and the unit of coordinates is m. _{Let trti} [s] be the RTT measured with the known antenna i.
Let RSSI _i [dBm] be the RSSI measured by the known antenna i.

If there is something that affects the characteristics of the known antenna i, the cable length, or other measured values for each known antenna i, correct it in advance. For example, if the cable length to which the known antenna i is connected is longer than that of other known antennas, the cable delay and cable loss will be larger than those of other known antennas. Therefore, subtract the round trip of the cable delay from _trti to obtain another known antenna. corrects the known antenna conditions equivalent to t _rti is obtained by adding the cable loss in RSSI _i condition equivalent RSSI _i other known antenna previously performed correction obtained.

The coordinates of the terminal antenna 21 are (x _s , y _s ) in the case of two dimensions and (x _s , y _s , z _s ) in the case of three dimensions. These are unknown and the unit of coordinates is m.

_Let t _d [s] be the sum of the terminal delay time and other delay times. This is also unknown.
The speed of light c is 299792458 [m / s]. This is exactly the same constant in any space.

Let α be the propagation loss coefficient in space. This is α = 2 in the free space, but here it is measured in advance according to the actual space. Therefore it is known.

The wavelength of the radio wave used for wireless communication is λ [m]. This is known because the frequency is determined once the channel is determined.

In the case of two dimensions, assuming that the distance between the known antenna i and an arbitrary point (x, y) is _di (x, y),
d _i (x, y) = √ [(x _i − x) ² + (y _{i −} y) ² ]
Will be. Similarly, in the case of three dimensions, if the distance between the known antenna i and an arbitrary point (x, y, z) is _di (x, y, z),
_{d i (x, y, z} ) = √ [(x i -x) 2 + (y i -y) 2 + (z i -z) 2]
Will be.

Assuming that the true distance between the known antenna i and the terminal antenna 21 is l _i [m], in the case of two dimensions,
l _i = d (x _i , y _i )
In the case of 3D,
l _i = d (x _i , y _i , z _i )
Will be.

(2.2.2) RTT evaluation function _trti, which is the RTT measured by the known antenna i, is the round-trip time, but the distance equivalent to this one-way distance is defined as the pseudo distance l _pi [m].
l _pi = ct _rti / 2
And. This is a value that can be calculated from the measured value, but since it is calculated based on _tr ti including the delay time t _d , it is longer than the true distance l _i between the known antenna i and the terminal antenna 21. ing.

If the difference between l _pi and l _i is l _d [m], this is equivalent to the one-way distance of the delay time t _d .
l _d = ct _d / 2
And
l _i = l _pi -l _d
Is.

Here, the unknowns to be obtained are three of the coordinates x _s , y _s and l _d of the terminal antenna 21 in the case of two dimensions, and the coordinates x _s , y _s , z _s and l _d of the terminal antenna 21 in the case of three dimensions. There will be four. Consider searching for these unknowns to obtain the coordinates of the terminal antenna 21. As the estimated value during the search, three variables x, y, l in the case of two dimensions and four variables x, y, z, l in the case of three dimensions are used. The distance between the coordinates and the known antenna i estimates, in the case of 2-dimensional d _i (x, y)
In the case of three dimensions, _di (x, y, z)
Will be.

Further, the estimated value of the true distance, which can be obtained from the estimated value l of the difference between the pseudo distance and the true distance, is
l _pi -l
Will be. If each estimate is correct, they match.

Therefore, the difference between them is defined as the RTT evaluation function f _ti for evaluating the estimated value as follows.
For _{2-dimensional: f ti (x, y,} l) = d i (x, y) - (l pi -l)
For _{3-D: f ti (x, y,} z, l) = d i (x, y, z) - (l pi -l)

(2.2.3) RSSI evaluation function If the RSSI measured by bringing the terminal antenna 21 close to the known antenna i is B [dBm], the model in which the measured RSSI _i is as follows is adopted.
RSSI _i = B-10α log ₁₀ (4πl _i / λ)

Here, if RSSI _i is converted into mW notation as _PRESIi [mW] and B is converted into mW notation as P _B [mW].
P _RSSIi = P _B {λ / (4πl _i )} ^ α
Will be. If you solve this for l _i ,
l _i = {λ / (4π)} P _B ^ (1 / α) ・_PRESIi ^ (-1 / α)

here,
r _B = {λ / (4π)} P _B ^ (1 / α)
R _i = _PRESIi ^ (-1 / α)
And. r _B is unknown, but R _i is known because it can be calculated from the measured values. Then
l _{_i} = r _B R _i
Will be.

As a result, in the case of RSSI, when the unknown number is two-dimensional, the coordinates x _s , y _s and r _B of the terminal antenna 21 are three, and when three-dimensional, the coordinates x _s , y _s , z of the terminal antenna 21 are obtained. _There are four, _s and r _B. Consider searching for these unknowns to find the coordinates of the terminal antenna. As the estimated value during the search, three variables x, y, r in the case of two dimensions and four variables x, y, z, r in the case of three dimensions are used. The RSSI evaluation function _fr , which evaluates the estimated value in the same way as RTT, is defined as follows.
For _{2-dimensional: f ri (x, y,} r) = d i (x, y) -rR i
For _{3-D: f ri (x, y,} z, r) = d i (x, y, z) -rR i

(2.2.4) Range evaluation function Define a range evaluation function to give a penalty to the evaluation value when the estimated value such as coordinates is out of the movable (existing) range of the terminal. The range evaluation function is zero when the estimated value is within the range, and when it is out of the range, the value increases according to the degree of deviation.

For example, in the case of two dimensions, the estimated value is x _min <x <x _max.
y _min <y <y _max
l _min <l <l _max
r _min <r <r _max
If it must be within the range indicated by, the coordinate range evaluation function f _L can be defined as follows.
f _L (x, y, l, r) = f _Lx (x) + f _Ly (y) + f _Ll (l) + f _Lr (r)

However,
f _Lx (x) = x _min -x (x <x _min )
= 0 (x _min <x <x _max )
= X-x _max (x> x _max )
f _Ly (y) = y _min −y (y <y _min )
= 0 (y _min <y <y _max )
= Y-y _max (y> y _max )
f _Ll (l) = l _min -l (l <l _min )
= 0 (l _min <l <l _max )
= L-l _max (l> l _max )
f _Lr (r) = r _min -r (r <r _min )
= 0 (r _min <r <r _max )
= R-r _max (r> r _max )

(2.2.5) Simultaneous equations An example of a method of obtaining coordinates by making each evaluation function a simultaneous equation and finding the solution is shown.

(2.2.5.1) Example of calculation method for RTT only (similar to GPS) To calculate coordinate estimates using the same method as GPS with four known antennas, combine the three-dimensional RTT evaluation functions as follows. Create a system of equations.
f _t1 (x, y, z, l) = 0
f _t2 (x, y, z, l) = 0
f _t3 (x, y, z, l) = 0
f _t4 (x, y, z, l) = 0

By solving this, the solutions x, y, z, l can be obtained, and the position can be estimated. This system of equations is non-linear and difficult to solve analytically, but it can be solved using Newton's method. When there are four or more known antennas, a similar simultaneous equation is established, but a solution cannot be obtained because it is a dominant determination system in which four or more equations are used for four unknowns. In that case, the method of least squares is used together to obtain a solution. In that case, more accurate position estimation is possible by using the weighted least squares method in which the reliability w of the RTT measurement value for each known antenna is weighted and weighted for each corresponding equation. ..

Hereinafter, when the same method as GPS is applied to two dimensions, the following simultaneous equations are created as an example of calculating the two-dimensional coordinate estimates.
f _t1 (x, y, l) = 0
_ft2 (x, y, l) = 0
f _t3 (x, y, l) = 0
f _t4 (x, y, l) = 0

Since it is a superior determination system with 4 equations for 3 unknowns, the solution can be obtained by Newton's method which also uses the least squares method. In that case, more accurate position estimation is possible by using the weighted least squares method in which the reliability w of the RTT measurement value for each known antenna is weighted and weighted for each corresponding equation. ..

(2.2.5.2) Example of calculation method when RTT evaluation function and range evaluation function are combined As an example of calculating two-dimensional coordinate estimation value by combining range evaluation functions using RTT measurement values of four known antennas, the following Make a system of equations like this.
f _t1 (x, y, l) = 0
_ft2 (x, y, l) = 0
f _t3 (x, y, l) = 0
f _t4 (x, y, l) = 0
f _L (x, y, l) = 0

However, in this case, since there is no unknown number r peculiar to RSSI, the range evaluation function is
f _L (x, y, l) = f _Lx (x) + f _Ly (y) + f _Ll (l)
Do not use r as such.

Since there are four superior determination systems with only the equation of the RTT evaluation function for unknown number 3, the solution can be obtained by Newton's method which also uses the least squares method. At this time, due to the effect of the range evaluation function, a force that keeps the unknown within the specified range works, making it easier to select a solution within the range and not greatly deviating. As in the previous section, for the equation of the RTT evaluation function, more accurate position estimation is possible by using the weighted least squares method, which uses the reliability w of the RTT measurement value of the corresponding known antenna as a weight. Furthermore, weights can also be used for the equation of the range evaluation function, and the strength of the force that tries to keep the coordinates to be the solution within the range can be set by the relative magnitude relation with other weights.

(2.2.5.3) Example of calculation method for RSSI only As an example of calculating three-dimensional coordinate estimates using RSSI measurement values of four known antennas, the following simultaneous equations are created.
f _r1 (x, y, z, r) = 0
f _r2 (x, y, z, r) = 0
f _r3 (x, y, z, r) = 0
f _r4 (x, y, z, r) = 0

Similarly, the following simultaneous equations are created as an example of calculating a two-dimensional coordinate estimate value using RSSI measurement values of four known antennas.
f _r1 (x, y, r) = 0
f _r2 (x, y, r) = 0
f _r3 (x, y, r) = 0
f _r4 (x, y, r) = 0

As in the case of RTT, the simultaneous equations are non-linear, so it is difficult to solve them analytically, and the solution can be obtained using Newton's method. If it becomes a dominant decision system, a solution can be obtained by using the least squares method together. Further, by using the weighted least squares method weighted by the reliability w of the RSSI measurement value for each known antenna, more accurate position estimation becomes possible.

(2.2.5.4) Example of calculation method when RSSI evaluation function and range evaluation function are combined As an example of calculating two-dimensional coordinate estimates by combining range evaluation functions using RSSI measurement values of four known antennas, the following Make a system of equations like.
f _r1 (x, y, r) = 0
f _r2 (x, y, r) = 0
f _r3 (x, y, r) = 0
f _r4 (x, y, r) = 0
f _L (x, y, r) = 0

However, in this case, since there is no unknown number l peculiar to RTT, the range evaluation function is
f _L (x, y, r) = f _Lx (x) + f _Ly (y) + f _Lr (r)
Do not use l as such.

Since there are four superior determination systems with only the equations of the RSSI evaluation function for unknown number 3, the solution can be obtained by Newton's method which also uses the least squares method. At this time, due to the effect of the range evaluation function, a force is exerted so that the unknown is within the specified range, and the solution within the range is easily selected and does not deviate significantly. As in the previous section, for the equation of the RSSI evaluation function, more accurate position estimation is possible by using the weighted least squares method, which uses the reliability w of the RSSI measurement value of the corresponding known antenna as the weight. Furthermore, weights can also be used for the equation of the range evaluation function, and the strength of the force that tries to keep the coordinates to be the solution within the range can be set by the relative magnitude relation with other weights.

(2.2.5.5) Example of calculation method combining RTT and RSSI As an example of calculating two-dimensional coordinate estimates using both RTT measurement values and RSSI measurement values of four known antennas, the following simultaneous equations are used. create.
f _t1 (x, y, l) = 0
_ft2 (x, y, l) = 0
f _t3 (x, y, l) = 0
f _t4 (x, y, l) = 0
f _r1 (x, y, r) = 0
f _r2 (x, y, r) = 0
f _r3 (x, y, r) = 0
f _r4 (x, y, r) = 0

Since this is also a dominant decision system, the solution can be found by Newton's method, which also uses the least squares method.
Furthermore, for the equation of the RTT evaluation function, the reliability w of the RTT measurement value of the corresponding known antenna is used as a weight, and for the equation of the RSSI evaluation function, the reliability w of the RSSI measurement value of the corresponding known antenna is used as a weight. By using the weighted least squares method, which is used, more accurate position estimation becomes possible. Further, by increasing or decreasing the weight of RTT and the weight of RSSI relatively, it is possible to adjust which measured value is emphasized.

(2.2.5.6) Example of calculation method using RTT and RSSI combined with range evaluation function Using both RTT measurement value and RSSI measurement value of 4 known antennas, combine range evaluation function and calculate 2D coordinate estimation value As an example of doing this, create the following simultaneous equations.
f _t1 (x, y, l) = 0
_ft2 (x, y, l) = 0
f _t3 (x, y, l) = 0
f _t4 (x, y, l) = 0
f _r1 (x, y, r) = 0
f _r2 (x, y, r) = 0
f _r3 (x, y, r) = 0
f _r4 (x, y, r) = 0
f _L (x, y, l, r) = 0

Due to the effect of the range evaluation function, a force that keeps the unknown within the specified range works, making it easier to select a solution within the range, and it does not deviate significantly. By using the weighted least squares method that uses the reliability w of the RTT / RSSI measurement value as the weight, more accurate position estimation is possible, and the weight is also used for the equation of the range evaluation function. It is also possible to set the strength of the force that tries to keep the coordinates to be the solution within the range by the relative magnitude relationship with the weight of.

(2.2.6) Minimal search Instead of solving simultaneous equations, each evaluation function is combined into one error evaluation function, and the estimated value that minimizes the value of the error evaluation function is searched and optimized. An example of how to obtain the coordinates is shown.

(2.2.6.1) Error evaluation function An error evaluation function f _error that combines all of the RTT evaluation function, RSSI evaluation function, and range evaluation function of n known antennas is defined.

However, the reliability of the RTT measurement value of the known antenna i is the weight w _ti , the reliability of the RSSI measurement value is the weight w _ri , and the weight of the range evaluation function is w _L.

If any of the RTT, RSSI, and range evaluation functions are not used, the error evaluation function with that term deleted should be used. There are a maximum of five parameters of the error evaluation function, x, y, z, l, and r, but in the case of two dimensions, z disappears, when RTT is not used, l disappears, and when RSSI is not used, r is. It disappears. Since either RTT or RSSI must be used, the minimum number of x, y, l combs is x, y, r.

(2.2.6.2) Search Search for the combination of parameters that minimizes the error evaluation function. The minimum may be searched by the Newton method, the search may be performed by a quasi-Newton method algorithm such as the L-BFGS method, or the search may be performed by Bayesian optimization. There is a simpler method of brute force search within a certain range at regular intervals, or a method of searching at rough intervals first and then searching around the area where the evaluation value is likely to be low at fine intervals. is there.

All parameters may be searched at the time of search, but there is also a method of not searching for l or r by finding l or r that minimizes the error evaluation function from the estimated values of the coordinates.
For example, when the estimated value of the coordinates is (x ^, y ^) in two dimensions, it is sufficient to take the partial differential of l or r of the error evaluation function and find l or r at which it becomes zero. Note that x ^ and y ^ are x hat and y hat. That is,

Find l that meets

Find r that satisfies, and use each as an estimate of l and r at the coordinates (x ^, y ^) to evaluate the error evaluation function. If this method is adopted, only the coordinates need to be searched, and the search space can be significantly reduced.

(2.3) Estimating the spatial propagation loss coefficient α Up to this point, the spatial propagation loss coefficient α required for the method of obtaining coordinates from RSSI has been assumed to be a known value that has been measured in advance. However, if many known antennas can be used, both RTT and RSSI can be measured, and the measurement error of each is small, not only the coordinates but also α can be estimated.

When adopting the method using simultaneous equations, create simultaneous equations with α as an unknown number as well as coordinates, l and r. Since there are many known antennas and both RTT / RSSI measurement results can be used, it becomes a dominant decision system, and the solution can be obtained by Newton's method using the least squares method. By using the weighted least squares method using the reliability w of the RTT / RSSI measurement value as a weight, more accurate position estimation and α estimation become possible. The range of α can be set in the equation of the range evaluation function, and the strength of the force to keep α within the range can be set by using the weight and the relative magnitude relationship with other weights. it can.

When adopting the method by the minimum search, create the RSSI evaluation function and the range evaluation function not only for the coordinates, l, and r but also for α, and assemble the error evaluation function. Then, it can be estimated by searching for α and performing optimization.

By estimating α by these methods, it is not necessary to measure α in advance. In addition, once the space α is obtained by performing high-precision measurement using a large number of known antennas, this α is used as a known value in actual operation, with a small number of known antennas and general accuracy. The method of estimating the position can also be adopted.

(2.4) Coordinate stabilization unit 33
Even if the terminal antenna 21 is stationary, the coordinates obtained by the coordinate calculation unit 32 (hereinafter, instantaneous value coordinates) may vibrate finely. Therefore, the coordinate stabilizing unit 33 obtains the coordinates (hereinafter referred to as the stabilizing coordinates) from which fine vibrations have been removed.

For example, if a certain threshold value is set as "play" and the difference or distance between the instantaneous value coordinates and the stabilization coordinates is smaller than the threshold value, the stabilization coordinates are maintained to absorb the fine vibration of the terminal antenna 21, and the threshold value is set. It can be stabilized by updating the stabilization coordinates by regarding it as a change in position only when it exceeds.

In addition, the instantaneous value coordinates may be passed through an LPF or a Kalman filter to remove fine vibrations as the stabilized coordinates.

(2.5) Coordinate adjustment unit 34
In the statistical processing (moving average processing) for the measured values of the measured value statistical processing unit 31 of the position estimation server, for example, the statistical processing RTT or the statistical processing RSSI obtained by moving average the measured values from N seconds before to the present is calculated for each antenna. However, when N seconds become longer, the movement is delayed from the actual movement, and the change in the measured value is averaged and becomes smaller. On the other hand, when N seconds are shortened, it reacts sensitively to changes in radio waves due to sudden movements such as turning around by a person who owns the wireless terminal station 20, and changes in measured values are regarded as changes in position, and the moving average. The processing is also affected.

Furthermore, even if the coordinate stabilization unit 33 converts the instantaneous value coordinates to the stabilized coordinates, the effect is unavoidable. For example, when the person who owns the wireless terminal station 20 turns around and starts moving as described above, the instantaneous value coordinates change greatly from the threshold value and the stabilization coordinates also change greatly, and it is estimated as a position that suddenly moves several meters. Will be done. Therefore, it is necessary to adjust the parameters of the LPF and the Kalman filter used as the coordinate stabilizing unit 33, and further, learning data is also required.

In order to deal with such a situation, the coordinate adjusting unit 34 performs moving average processing on the stabilized coordinates obtained by the coordinate stabilizing unit 33, and statistically processes the average value of the stabilized coordinates for the past M times. Calculate the stabilization coordinates.

FIG. 8 shows an example of the positioning result calculated using the data measured in the test environment.
In FIG. 8, the test environment is the vicinity of the point A on the return route when reciprocating between the point A as the starting point near the antenna A and the point B near the antenna B 6 m away from the point A. There is an obstacle in the route and it is a detour route.

(1) shows the change in the stabilized coordinates output by the coordinate stabilizing unit 33 when the moving average interval N of the measured value statistical processing unit 31 is N = 10 seconds. As described above, the stabilized coordinate values are delayed according to the length of the moving average interval, and the changes are averaged to reduce the accuracy of the position.

(2) shows the change in the stabilized coordinates output by the coordinate stabilizing unit 33 when the moving average interval N of the measured value statistical processing unit 31 is N = 2 seconds. As described above, the stabilized coordinate values respond to sudden changes in radio waves due to sudden movements such as starting to move or turning around, the intervals between the points in the graph are widened, and it is as if the wireless terminal station 20 has moved significantly. Looks like.

(3) shows the output of the coordinate adjustment unit 34 when the moving average section N = 2 seconds of the measured value statistical processing unit 31 and the moving average section M = 4 times of the coordinate adjustment unit 34. That is, the stabilized coordinates of (2) are moved and averaged by the coordinate adjusting unit 34. Therefore, the step-like change of the stabilization coordinates due to the width of the threshold value is averaged, and the sudden change of the radio wave due to the sudden movement such as the start of movement or turning around is absorbed, and the interval of the stabilization coordinates is interpolated. Coordinate values can be obtained.

The processing of each part of the position estimation server 30 described above can be realized by a computer program that operates the computer. The computer program can be stored on a computer-readable storage medium or provided over a network.

10 Radio station 11 Known antenna 12 Signal transmitter 13 Signal receiver 14 RTT measurement unit 15 RSSI measurement unit 16 Clock 20 Wireless terminal station 21 Terminal antenna 22 Signal receiver 23 Signal transmitter 24 Control unit 25 Clock 30 Position estimation server 31 Measurement Value statistics processing unit 32 Coordinate calculation unit 33 Coordinate stabilization unit 34 Coordinate adjustment unit 40 Radio base station 41 Cable 42 Distributed antenna

Claims

A measurement signal and a response signal are transmitted and received between a radio station having a plurality of known antennas installed at different known positions and a radio terminal station having a terminal antenna, and the round-trip delay time of the RTT is measured. In the position estimation method for estimating the position of the wireless terminal station,
Step 1 of obtaining statistical processing RTT by performing statistical processing for each antenna with respect to the RTT measurement value measured between the known antenna i (i = 1, 2, ..., N) and the wireless terminal station.
Step 2 in which the coordinates of the wireless terminal station are calculated from the statistical processing RTT and output as instantaneous value coordinates, and
Step 3 of removing vibration components below the threshold value of the instantaneous value coordinates and converting the instantaneous value coordinates into stabilizing coordinates within a range not exceeding the threshold value for vibration components above the threshold value.
A position estimation method comprising step 4 in which a moving average process is performed on the stabilized coordinates and a statistical process stabilized coordinates are output.
A measurement signal and a response signal are transmitted and received between a radio station having a plurality of known antennas installed at different known positions and a radio terminal station having a terminal antenna, and RSSI, which is the received radio wave intensity, is measured. In the position estimation method for estimating the position of the wireless terminal station,
Step 1 of obtaining statistical processing RSSI by performing statistical processing for each antenna with respect to the RSSI measurement value measured between the known antenna i (i = 1, 2, ..., N) and the wireless terminal station.
Step 2 in which the coordinates of the wireless terminal station are calculated from the statistical processing RSSI and output as instantaneous value coordinates, and
Step 3 of removing vibration components below the threshold value of the instantaneous value coordinates and converting the instantaneous value coordinates into stabilizing coordinates within a range not exceeding the threshold value for vibration components above the threshold value.
A position estimation method comprising step 4 in which a moving average process is performed on the stabilized coordinates and a statistical process stabilized coordinates are output.
In the position estimation system that estimates the position of the wireless terminal station by the position estimation method according to claim 1 or 2.
The RTT and / and the RSSI measured by the radio station having the plurality of n known antennas are transferred to the position estimation server, the position estimation server calculates the statistical processing stabilization coordinates, and the radio terminal station. A position estimation system characterized in that it has a configuration for finding the position of.
In the position estimation system that estimates the position of the wireless terminal station by the position estimation method according to claim 1 or 2.
The plurality of known antennas are distributed antennas distributed from the radio base station via a cable of a known length, and the RTT and / and the RSSI for each distributed antenna measured at the radio base station. Is transferred to a position estimation server, and the position estimation server calculates the statistical processing stabilization coordinates to obtain the position of the wireless terminal station.
In the position estimation server that estimates the position of the wireless terminal station by the position estimation method according to claim 1 or 2.
The configuration is such that the RTT and / and the RSSI measured by the radio station having the plurality of n known antennas are transferred, the statistical processing stabilization coordinates are calculated, and the position of the radio terminal station is obtained. A featured location estimation server.
A position estimation program characterized by causing a computer to execute the process executed by the position estimation server according to claim 5 and calculating the statistical processing stabilization coordinates to obtain the position of the wireless terminal station.