An error correction method for TOA non-line-of-sight scenarios

Abstract

The present invention discloses an error correction method for TOA non-line-of-sight scenarios, including: truncating the TOA measurement values of base stations with NLOS, where when the target is in motion, the truncation length is related to the real-time requirement; taking time as the horizontal axis and data values as the vertical axis, performing robust locally weighted regression fitting on the truncated data to obtain the corresponding TOA fitting curve; calculating the deviation between the measurement value and the corresponding fitting value at each moment within the preset data length range for this base station; finding the moment point with the largest deviation value, and translating the fitting curve as a whole along the vertical axis to make the measurement value at this moment point coincide with the fitting value, thereby obtaining the corrected TOA measurement data curve. The present invention does not require long-term observation, can directly fit the measurement data of base stations with NLOS, realizes dynamic correction of TOA measurement values, meets the requirement of real-time correction for base stations, and improves the positioning accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of positioning, and more particularly to an error correction method for TOA non-line-of-sight scenarios. Background Art

[0002] In positioning, there are many factors affecting positioning accuracy, but the most important factor comes from the impact of non-line-of-sight propagation errors. When there are obstacles blocking between the base station and the target, the signal between them will be propagated through the reflection or refraction of other obstacles such as walls, resulting in non-line-of-sight errors.

[0003] Existing technologies can measure data for a period of time for NLOS identification and NLOS mitigation. For example, the Wylie algorithm mentioned in "Research on LOS-NLOS Propagation Environment Identification and NLOS Error Elimination Technology". However, this type of technology cannot achieve dynamic correction, that is, it cannot meet the real-time requirements. Summary of the Invention

[0004] In view of the deficiencies in the prior art, the present invention provides an error correction method for TOA non-line-of-sight scenarios. When it is known that a certain base station has non-line-of-sight errors, it is not necessary to conduct long-term observations, and the measurement data of the base station with NLOS can be directly fitted to correct the errors.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] An error correction method for TOA non-line-of-sight scenarios, the error correction method comprising the following steps:

[0007] S1, truncate the TOA measurement values of the base station with NLOS. When the target is in motion, the truncation length is related to the real-time requirement;

[0008] S2, taking time as the horizontal axis and data values as the vertical axis, perform robust locally weighted regression fitting on the truncated data to obtain the corresponding TOA fitting curve;

[0009] S3, calculate the deviation between the measurement value and the corresponding fitting value at each moment within the preset data length range of the base station;

[0010] S4, find the moment point with the largest deviation value, and translate the fitting curve as a whole along the vertical axis direction so that the measurement value at this moment point coincides with the fitting value, obtaining the corrected TOA measurement data curve.

[0011] To optimize the above technical solutions, the specific measures taken further include:

[0012] Further, in step S1, when the target is in motion, obtain the minimum allowable delay duration τ, and calculate the maximum truncation length l according to the following formula:

[0013] l = τ * f

[0014] In the formula, f is the sampling frequency.

[0015] Further, in step S1, the error correction method further includes:

[0016] The minimum allowable delay durations in different time periods are relatively independent.

[0017] Further, in step S2, the process of performing robust locally weighted regression fitting on the truncated data includes the following steps:

[0018] S21, Let time be the horizontal axis and TOA measurement data be the vertical axis to construct a coordinate system;

[0019] S22, Assume any truncated data within the current time period is (x i , y i ), i = 1, 2,..., n, where n is the total number of truncated data within the current time period, and y i is the TOA measurement data corresponding to the time point x i ;

[0020] S23, With the time point x i of any truncated data as the center, along the horizontal axis direction, intercept a section of length frac of data. For this section of intercepted data, perform weighted linear regression processing using a weight function, and denote as the central value of the corresponding regression line, and i as the fitted value corresponding to the time point x

[0021] S24, Repeat step S23 until n regression lines corresponding to all truncated data are obtained;

[0022] S25, Obtain the TOA fitting curve corresponding to the current time period according to the n regression lines corresponding to all truncated data.

[0023] Further, in step S23, the process of performing weighted linear regression processing using a weight function includes the following steps:

[0024] S231, Assume the ranging values (TOA) of a single base station within a period of time are Y = {y 1 ,..., y n}, in units of cm, and the corresponding sampling time series is X = {x 1 ,..., x n}, in units of s. The ranging value after weighted regression is

[0025] S232, at any time x i Cut the data with length L+1 for the center And the sampling time series of this segment of data Normalize

[0026] S233, will Substitute into the weight function to solve the weight W = {w 1 , ..., w L+2}:

[0027]

[0028] S234, substituting the weight W into the weighted regression calculation to obtain the fitting value of the distance measurement value

[0029] S235, calculate the residual of the measured value and the fitted value of TOA Let the constant s be {|e 1 |, ..., |e L+2 |} median;

[0030] S236, substitute the weight correction function to calculate the weight coefficient Coef = {δ 1 , ..., δ L+ 2}:

[0031]

[0032] S237, calculate the corrected weight W′={w′ 1 , ..., w′ L+2} = Coef*W;

[0033] S238, taking W′ as the new weight, repeating steps S234 to S237 until convergence.

[0034] The beneficial effects of the present invention are:

[0035] When it is known that a base station has a non-line-of-sight error, there is no need to conduct long-term observations. The measurement data of the base station with NLOS can be directly fitted to achieve dynamic correction of the TOA measurement value, meet the requirements of real-time correction of the base station, and further improve the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of an error correction method for TOA non-line-of-sight scenarios according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The present invention will now be described in further detail with reference to the accompanying drawings.

[0038] It should be noted that the terms such as "upper", "lower", "left", "right", "front", and "rear" cited in the invention are only for the sake of clarity in description and are not used to limit the scope of implementation of the present invention. The change or adjustment of their relative relationship, without substantial change in the technical content, should also be regarded as the scope of implementation of the present invention.

[0039] Figure 1 is a flowchart of an error correction method for TOA non-line-of-sight scenarios in an embodiment of the present invention. Refer to Figure 1 and the error correction method includes the following steps:

[0040] S1. Truncate the TOA measurement values of the base stations with NLOS. When the target is in motion, the truncation length is related to the real-time requirement.

[0041] S2. With time as the horizontal axis and data values as the vertical axis, perform robust locally weighted regression fitting on the truncated data to obtain the corresponding TOA fitting curve.

[0042] S3. Calculate the deviation between the measurement value and the corresponding fitting value at each moment within the preset data length range of the base station.

[0043] S4. Find the moment point with the largest deviation value, and translate the fitting curve as a whole along the vertical axis so that the measurement value at this moment point coincides with the fitting value, obtaining the corrected TOA measurement data curve.

[0044] The premise for implementing the error correction method mentioned in this embodiment is that it is known that a certain base station has non-line-of-sight errors. That is, this embodiment only corrects non-line-of-sight errors and does not involve the error judgment process. In practical applications, it can be judged by the staff according to the scene layout whether a base station has non-line-of-sight errors (NLOS), or a more accurate judgment can be made by using the data testing method. It should be understood that for base stations without non-line-of-sight errors, this embodiment will not affect their accuracy, but only cause waste of computing resources.

[0045] The error correction method of this embodiment will be further described below. Specifically, the error correction method includes the following steps:

[0046] Step 1. Truncated data sampling

[0047] Truncate the TOA measurement values of the base stations with NLOS. When the target is in motion, the truncation length is related to the real-time requirement. Specifically, when the target is in motion, obtain the minimum allowable delay duration τ, and calculate the maximum truncation length l according to the following formula:

[0048] l = τ * f

[0049] Where f is the sampling frequency, and the sampling point interval duration is 1 / f.

[0050] Preferably, in this embodiment, the minimum allowable delay durations in different time periods are relatively independent. That is, in different time periods, different minimum allowable delay durations can be set according to the actual requirements of the scenario to further optimize the computing resources.

[0051] Step 2: Perform robust locally weighted regression fitting on the truncated data to obtain the corresponding TOA fitting curve

[0052] Step 2 specifically includes the following sub-steps:

[0053] S21: Use time as the horizontal axis and TOA measurement data as the vertical axis to construct a coordinate system.

[0054] S22: Let any truncated data in the current time period be (x i , y i ), where i = 1, 2,..., n, and n is the total number of truncated data in the current time period. y i is the TOA measurement data corresponding to the time point x i .

[0055] S23: With the time point x i of any truncated data as the center, intercept a section of data with a length of frac along the horizontal axis direction. For this section of intercepted data, perform weighted linear regression processing using a weight function. Denote as the central value of the corresponding regression line, as the fitting value corresponding to the time point x i .

[0056] S24: Repeat step S23 until n regression lines corresponding to all truncated data are obtained. The connection of the central values of each regression line is the Lowess curve of this section of data.

[0057] S25: Obtain the TOA fitting curve corresponding to the current time period according to the n regression lines corresponding to all truncated data.

[0058] In step S22, the selection of the intercepted length will affect the accuracy of the regression line. Through practical verification, for the positioning requirements in areas such as factories, 0.4 times the data length of the current time period can be intercepted as the intercepted data, which can improve the computing efficiency and the convergence efficiency of relevant values as much as possible on the basis of meeting the real-time requirements.

[0059] Exemplarily, in step S23, the process of performing weighted linear regression processing using a weight function includes the following steps:

[0060] S231. Assume that the ranging values (TOA) of a single base station within a period of time are \(Y = \{y\) 1 , \(\cdots, y\) n}\), with the unit of em, and the corresponding sampling time series is \(X=\{x\) 1 , \(\cdots,\) x _n\), with the unit of s. The ranging value after weighted regression is

[0061] S232. Intercept the data with a length of \(L + 1\) centered at any time \(x\) i and normalize the sampling time series of this segment of data

[0062] S233. Substitute into the weight function to solve the weights \(W=\{w\) 1 , \(\cdots, w\) L+2 \}:

[0063]

[0064] S234. Substitute the weights \(W\) into the weighted regression calculation to obtain the fitted value of the ranging value

[0065] S235. Calculate the residual between the measured value and the fitted value of TOA Denote the constant \(s\) as the median of \(\{|e\) 1 |, \(\cdots, |e\) L+2 |\}\).

[0066] S236. Substitute into the weight correction function to calculate the weight coefficient \(Coef = \{\delta\) 1 , \(\cdots, \delta\) L+2 \}:

[0067]

[0068] S237. Calculate the corrected weights \(W'=\{w'\) 1 , \(\cdots, w'\) L+2 \} = Coef * W\).

[0069] S238. Take \(W'\) as the new weights and repeat steps S234 to S237 until convergence. The convergence state can be basically reached when the number of iterations is 3.

[0070] Step 3. Calculate the deviation between the measured value and the corresponding fitted value at each moment within the preset data length range of this base station

[0071] ​​There is measurement data of NLOS base stations, and its error is a non - negative random number. Therefore, the deviation between the fitting value and the measurement value at each moment can be calculated.

[0072] Step Four: Obtain the corrected TOA measurement data curve

[0073] Within this data length, find the moment with the largest deviation, and then translate the entire fitting curve to the point at that moment, thus obtaining the corrected TOA measurement data. Assume that within this section of data, at a certain point, the difference between y and is the largest, which is α. Then subtract α from all y values to obtain the TOA curve after correcting the non - line - of - sight error.

[0074] The above is only the preferred implementation mode of the present invention. The protection scope of the present invention is not limited to the above - mentioned embodiments. All technical solutions within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and retouches should be regarded as within the protection scope of the present invention.

Claims

1. An error correction method for TOA non-line-of-sight scenarios, characterized in that, the error correction method comprises the following steps: S1. Truncate the TOA measurement values of the base stations with NLOS. When the target is in motion, the truncation length is related to the real-time requirement; S2. With time as the horizontal axis and data values as the vertical axis, perform robust locally weighted regression fitting on the truncated data to obtain the corresponding TOA fitting curve; S3. Calculate the deviation between the measurement value and the corresponding fitting value at each moment within the preset data length range of the base station; S4. Find the moment point with the largest deviation value, and translate the fitting curve as a whole along the vertical axis so that the measurement value at this moment point coincides with the fitting value, obtaining the corrected TOA measurement data curve.

2. The error correction method for TOA non-line-of-sight scenarios according to claim 1, characterized in that, in step S1, when the target is in motion, obtain the minimum allowable delay duration τ, and calculate the maximum truncation length l according to the following formula: l = τ * f where f is the sampling frequency.

3. The error correction method for TOA non-line-of-sight scenarios according to claim 2, characterized in that, in step S1, the error correction method further comprises: The minimum allowable delay durations in different time periods are relatively independent.

4. The error correction method for TOA non-line-of-sight scenarios according to claim 1, characterized in that, in step S2, the process of performing robust locally weighted regression fitting on the truncated data comprises the following steps: S21. Let time be the horizontal axis and TOA measurement data be the vertical axis to construct a coordinate system; S22. Let any truncated data within the current time period be (x i , y i ), where i = 1, 2, …, n and n is the total number of truncated data within the current time period. y i is the TOA measurement data corresponding to the time point x i ; S23, at any moment point x of the truncated data i as the center, along the horizontal axis direction, intercept a section of data with a length of frac, and for this section of intercepted data, perform weighted linear regression processing using a weight function, denoted as as the central value of the corresponding regression line, as the moment point x i corresponding fitted value; S24. Repeat step S23 until n regression lines corresponding to all truncated data are obtained; S25. Obtain the TOA fitting curve corresponding to the current time period according to the n regression lines corresponding to all truncated data.

5. The error correction method for TOA non-line-of-sight scenarios according to claim 4, characterized in that, in step S23, the process of performing weighted linear regression processing using a weight function comprises the following steps: S231, assume that the ranging value TOA of a single base station within a period of time is Y = {y 1 ,..., y n}, with the unit of cm, and the corresponding sampling time series is X = {x 1 ,..., x n}, with the unit of s. The ranging value after weighted regression is S232, at any moment x i Intercept data with a quantity of L + 1 centered around it And for the sampling time series of this segment of data Perform normalization processing S233, substitute into the weight function to solve for the weights W = {w 1 ,..., w L+1}: S234. Substitute the weight value W into the weighted regression calculation to obtain the fitted value of the ranging value. For S235, calculate the residuals between the measured values and the fitted values of TOA Denote the constant s as the median of {|e 1 |,…,|e L+1 |} S236, substitute into the weight correction function to calculate the weight coefficient Coef = {δ 1 ,..., δ L+1}: S237, calculate the corrected weight \(W'=\{w' 1 ,...,w' L+1 \}=Coef * W; S238. Take W′ as the new weight, and repeat steps S234 to S237 until convergence.

Images