CN1235429C - Method for estimating position - Google Patents

Abstract

The present invention discloses a method for estimating positions, which comprises: the time difference of arrival (TDOA) between a main base station and two adjacent base stations referring to a measuring and positioning request is firstly measured; the zero mean value correction of the errors of the non light of sight (NLOS) in the TDOA measurement quantity is carried out by making use of the NLOS error mean value calculated by the error distribution parameters of NLOS. Accordingly, the time of arrival (TOA) from a mobile station to the main base station is estimated, and TOA from the mobile station to the adjacent base stations is calculated. The zero mean value correction of the NLOS errors in TOA from the mobile station to the main base station and from the mobile station to the adjacent base stations is carried out by making use of the NLOS error mean value. The self-adaptive adjustment of the weighting matrix in the TOA position estimation is carried out by making use of NLOS error variance; the weighting least square estimation is carried out by making use of the adjusted weighting matrix, and the positional estimate value of the mobile station is obtained. The present invention can greatly suppress the NLOS errors and the GDOP influence, and the position estimating accuracy is obviously increased.

Description

Position estimation method

Technical Field

The invention relates to a positioning technology in the field of wireless communication, in particular to a position estimation method for positioning a mobile station.

Background

In the field of wireless communications, cellular mobile station positioning consists of two basic elements, measurement and position estimation. In the measurement link, there are two basic measurement quantities, one is a time difference of arrival (TDOA) measurement quantity, and the other is a time of arrival (TOA) measurement quantity; in the location estimation step, there are two basic location estimation methods, one is a TDOA location estimation method, and the other is a TOA location estimation method. From the analysis of measurement quantity, although both TDOA and TOA measurement quantities are affected by non line of sight (NLOS) errors, generally, TOA measurement quantities are obtained by using the Round Trip Time (RTT) measurement function of the system, and besides the NLOS errors, TOA measurement quantities also include some errors introduced in the RTT measurement process, such as radio frequency channel delay errors of the base station and the mobile station, and receiving and transmitting delay errors of the mobile station, which will reduce the accuracy of TOA measurement quantities in high-precision positioning. TDOA measurements do not contain NLOS errors and errors introduced during RTT measurements, and therefore TDOA measurements are preferred over TOA measurements in mobile station location. From a position estimation perspective analysis, the TOA position estimate is more susceptible to geometric dilution of precision (GDOP) than the TOA position estimate due to the nonlinear effects of the hyperbolas, in other words, the TOA position estimate suppresses the GDOP effects relative to the TDOA position estimate, and thus is more desirable than the TDOA position estimate.

Among existing TDOA position estimation methods, the chen algorithm described in "a simple and effective hyperbolic positioning position estimator" article ("a simple and effective positioning estimator for hyperbaric location", IEEE Trans Signal processing, vol.42, No.8, aug.1994, pp.1905-1915), published by y.t. chen (y.t. chen) in the Signal processing branch of the institute of electrical and electronics engineers, volume 42, 8, 1994), is a typical representative.

In the existing TOA position estimation method, an article "TOA positioning algorithm considering non-line-of-sight propagation effect" in "journal of communications" published in 3.2001, volume 22, phase 3 is a typical representative, and the discussion of the TOA position estimation algorithm is divided into two cases, namely a position estimation algorithm in a line-of-sight (LOS) channel environment and a position estimation algorithm in an NLOS channel environment, where the position estimation algorithms in the LOS environment and the NLOS environment discussed in the document are respectively referred to as LOS-TOA chen algorithm and NLOS-TOA relaxation algorithm.

In the existing TOA-TDOA hybrid location estimation algorithm, the "method and apparatus for location estimation using a main base station TOA and a TDOA relative to the main base station" patent application is one representative.

Among the three position estimation methods, LOS-TOA Chen's algorithm and TOA-TDOA mixed position estimation algorithm are constructed by taking the solution idea of the Chen's algorithm as a reference. The common features of these two position estimation methods are: both adopt twice weighted linear least square estimation to approximate and realize the maximum likelihood estimation; the sensor array has wide universality, can be suitable for the condition that the number of the measured quantities is equal to the number of the radiation source coordinates, is also suitable for the condition that the number of the measured quantities is greater than the number of the radiation source coordinates, and is also suitable for the condition that the sensors are linearly arranged and the condition that the sensors are randomly arranged. The common disadvantage of both position estimation methods is: because the existing algorithms are all constructed on the premise of assuming that a direct path exists between a radiation source and a sensor or a positioning receiver, under the assumption, an error signal input to a position estimation algorithm only comes from a delay estimation error of the sensor, and the delay estimation error is zero mean Gaussian distribution, so the position estimation algorithms do not have NLOS error suppression capability.

Among the three position estimation methods, the basic idea of the NLOS-TOA relaxation algorithm is to introduce a relaxation variable into TOA measurement and to use a search method to solve a reasonable position solution. The geometric nature of this method is to adjust the radius of the TOA circle containing the NLOS error in a certain way, and it is expected that this method will eliminate some of the NLOS error influence and obtain a more accurate position estimation. The search criteria of this NLOS location estimation method are: and taking the point which is searched and is closest to the position obtained when the LOS algorithm is used for the NLOS channel as the output of the NLOS algorithm. The NLOS-TOA relaxation algorithm has the following disadvantages: no method for determining the relaxation variable with objective basis is provided, so that the relaxation variable is difficult to select in practical application and has no practicability.

In the patent application "a method for improving positioning accuracy by using TDOA/TOA smoothing and reconstruction", a method for estimating a position by comprehensively using TOA and TDOA measurement quantities and having NLOS error suppression capability is provided, which is characterized in that: 1) the distance d between the mobile station and the base station, obtained by rough position estimation, is determined by means of the delay spread tau_rmsMedian value T at one thousand meters₁(empirical value) and formula τ_rms＝T₁d^εξ to obtainDelay spread tau of power delay profile_rmsIn the formula, ζ is a random variable of lognormal distribution; epsilon is an exponential factor with the value of 0.5-1. Extending the delay by tau_rmsAs a distribution parameter for NLOS errors. 2) NLOS identification and estimation of NLOS error variance are performed using smoothing of TOA measurements.

However, this method has certain disadvantages, that is: 1) delay spread tau_rmsThere is no logical connection to the distribution parameters of NLOS errors. 2) Formula τ_rms＝T₁d^εξ is an abstracted relationship for channel modeling and the delay spread τ calculated therefrom_rmsτ from the actual position of the mobile station_rmsThe anastomosis is also difficult. 3) The method for performing NLOS identification and NLOS error variance estimation by utilizing the smoothing of the TOA measured value has no real-time performance and poor accuracy. In view of the above disadvantages, this method is difficult to be effectively applied to actual measurement and calculation.

In summary, in the existing cellular mobile station positioning technology, a position estimation algorithm with NLOS error suppression performance, GDOP suppression performance and practicability at the same time is currently lacking.

Disclosure of Invention

In view of the above, an object of the present invention is to provide a position estimation method that can simultaneously achieve NLOS error suppression performance, GDOP suppression performance, and practicality.

The purpose of the invention is realized by the following technical scheme:

a position estimation method measures a time difference of arrival (TDOA) between a main base station and two neighbor base stations involved in a positioning request, and then:

a. performing zero-mean correction on NLOS errors in TDOA measurement quantity by using NLOS error mean values obtained by non-line-of-sight (NLOS) error distribution parameters, estimating the time of arrival (TOA) from a mobile station to a main base station according to the corrected TDOA measurement quantity, and calculating the TOA from the mobile station to an adjacent base station by using the TOA from the mobile station to the main base station and the uncorrected TDOA from the mobile station to the main base station and the adjacent base station obtained by system measurement;

b. performing zero mean correction on NLOS errors in TOAs from the mobile station to the main base station and from the mobile station to the adjacent base station by using NLOS error mean values calculated by the NLOS error distribution parameters;

c. and carrying out self-adaptive adjustment on a weighting matrix in the TOA position estimation by using the NLOS error variance obtained by the NLOS error distribution parameters, and carrying out weighted least square estimation on the position of the mobile station by using the adjusted weighting matrix to obtain an estimated value of the position of the mobile station.

In the above position estimation method, steps a to c may be repeated to obtain more than 1 mobile station position estimation values and average them to obtain a final mobile station position estimation value.

In the above position estimation method, step a may further include:

a1. TDOA measurement quantity needing zero mean value correction is determined through NLOS identification, and mean value of NLOS errors in the TDOA measurement quantity is determined through NLOS error distribution parameters;

a2. performing zero-mean correction on the NLOS error in the TDOA measurement according to the NLOS error mean value determined in the step al;

a3. calculating the TOA from the mobile station to the main base station according to the TDOA measured quantity obtained after correction in the step a2, and calculating the TOA from the mobile station to each adjacent base station according to the calculated TOA value from the mobile station to the main base station and the uncorrected TDOA measured by the system between the mobile station to the main base station and the adjacent base station.

In the position estimation method, the step a3 may further include the following steps: and after the TOA of the main base station is calculated, calculating more than 1 time, and calculating the TOA value of each adjacent base station according to the average value of the obtained TOA values more than 1 and the uncorrected TDOA measured by the system from the mobile station to the main base station and the adjacent base station.

In the above position estimation method, step b may further include:

b1. determining the TOA values of the main base station and the adjacent base stations needing zero-mean correction, which are obtained in the step a3, by NLOS identification, and determining the mean value of NLOS errors in the TOA values of the main base station and the adjacent base stations needing zero-mean correction, which are obtained in the step a3, by using NLOS error distribution parameters;

b2. and c, performing zero-mean correction on the mean value of the NLOS errors in the TOA values of the main base station and the adjacent base stations which need to be subjected to zero-mean correction and are obtained in the step a3.

In the above position estimation method, NLOS identification may be performed by selecting a size of a set of sample discrete coefficients of the strongest path from the power delay profile obtained from the radiation source, or may be performed by using an inter-path power difference or an inter-path amplitude difference on the power delay profile obtained from the radiation source.

In the above position estimation method, step c may further include:

c1. determining a weighting matrix form by setting a main diagonal element as a sum of a covariance of a TOA delay estimation error and a covariance of a corrected residual of the TOA in a line-of-sight (LOS) environment;

c2. setting the covariance value of correction residual errors of the TOA corresponding to each channel containing NLOS errors in the channels for calculating the TOA to be zero;

c3. the TOA measurement estimation is carried out on the system under the LOS channel environment, or the covariance of the TOA time delay estimation error under the LOS environment needing to be adjusted in the weighting matrix of the step c1 is determined through system simulation or through the statistics of the TDOA measurement error; if the TOA contains the NLOS error in a discrete form, determining the covariance of the TOA correction residual needing to be adjusted in the weighting matrix of the step c1 according to the NLOS error distribution parameter and the probability density function of the discrete NLOS error in the TOA; if the NLOS error contained in the TOA is in a continuous form, the probability density function of the continuous NLOS error of the TDOA is obtained through the NLOS error distribution parameters, and then the covariance of the TOA correction residual to be adjusted in the weighting matrix of step c1 is determined through the probability density function of the continuous NLOS error of the TDOA.

In the above position estimation method, the TDOA measurement may be determined by GDOP minimum criterion or LOS channel criterion in step a.

According to the technical scheme, in the TOA position estimation process, the precision influence of NLOS errors on TOA estimated values is greatly inhibited through the steps of conducting zero-mean correction on NLOS errors in TDOA measured quantity, conducting zero-mean correction on NLOS errors in TOA, conducting adaptive adjustment on a weighting matrix of TOA through NLOS error variance, solving the TOA position estimated values according to the weighting matrix and the like. Meanwhile, the TOA position estimation method is adopted, so that the influence of GDOP can be effectively inhibited. According to the technical scheme, the steps are simple and convenient to implement, and the method has good practicability in practical application.

Drawings

FIG. 1 is a flow chart of a position estimation method of the present invention for suppressing NLOS errors and GDOP effects;

FIG. 2 is a flowchart of a zero-mean correction method for NLOS errors in TDOA measurements according to the present invention;

FIG. 3 is a flowchart of a zero-mean correction method for NLOS errors in TOA values calculated by the present invention;

fig. 4 is a flow chart of a method of adaptively adjusting a weighting matrix according to the present invention.

Detailed Description

The invention is further described with reference to the following figures and specific examples.

The invention provides a method for converting TDOA measurement quantity into a TOA value and then utilizing the TOA value to carry out position estimation. Specifically, the TOA from the main base station to the mobile station is first determined using two TDOA measurements obtained from the main base station and any two adjacent base stations adjacent to the main base station₁(ii) a Then using TOA₁And TOA_i＝TDOA_i，1+TOA₁TDOA measured by mobile station_i，1Conversion of measurement into TOA_iWherein TDOA_i，1Indicating a TDOA value between an ith base station other than the master base station and the master base station, i being 2, 3. Finally using TOA_iAnd the constructed TOA position estimation algorithm estimates the position of the mobile station.

NLOS error in TDOA measurements both reduces TOA derived from TDOA measurements₁The accuracy of the subsequent position estimation based on the TOA value is also reduced, and in order to inhibit the influence of NLOS errors, the basic thought adopted by the invention is as follows: under the premise of obtaining the mean value and the variance of the NLOS error, the mean value of the NLOS error is used for correcting the nonnegative NLOS error into a random variable of a zero mean value, then the variance of the NLOS error is used for constructing a weighting matrix in weighted least square estimation to preliminarily inhibit the influence of the NLOS error on position estimation, and finally the NLOS error is further inhibited through multiple averaging of position estimation results according to the zero mean characteristic of the corrected NLOS error.

The basic idea of the present invention is explained in detail below with reference to fig. 1.

Fig. 1 is a flow chart of the position estimation method for suppressing NLOS error and GDOP influence according to the present invention, and as can be seen from fig. 1, the method of the present invention is composed of the following six basic steps:

and step 101, performing zero-mean correction on the NLOS error contained in the TDOA measurement quantity by using the mean value of the NLOS error calculated by the NLOS error distribution parameters. Two TDOA measurement quantities and corresponding base stations and pseudo random codes are selected from two or more TDOA measurement quantities according to a certain criterion, such as a GDOP minimum criterion or a LOS channel existence criterion, and then zero-mean correction of NLOS errors is carried out on the two TDOA measurement quantities.

As shown in fig. 2, the method for zero-mean correction of NLOS error in TDOA measurement specifically includes substeps 201, 202, and 203:

substep 201 of step 101 determines the amount of TDOA measurements that require zero-mean correction by NLOS identification. First, TDOA measurements containing NLOS errors are identified_(m)i，j，TDOA_(m)i，jRepresents the raw TDOA measurement without zero mean correction of the NLOS error, the subscripts i, j indicate that this TDOA measurement is the time difference of arrival between the ith and jth radiation sources, such as base stations in a cellular network; ${\mathrm{TDOA}}_{\left(m\right)i,j}={\mathrm{TDOA}}_{\left(\mathrm{los}\right)i,j}^{\left(0\right)}+n_{\left(n\right)i,j}+{\mathrm{\μ}}_{\left(n\right)i,j}+n_{i,j}.$ in the above formula, TDOA_(los)i，j ⁽⁰⁾Is an ideal TDOA value that does not contain any error; n is_(n)i，jThe residual error of the NLOS error after zero-mean correction, wherein the mean value is zero; mu.s_(n)i，jThe mean value of the NLOS error is calculated according to the NLOS error distribution parameters; n is_i，jThe TDOA time delay estimation error of the system in the LOS environment is obtained without considering the influence of clock drift among radiation sources on TDOA measurement, so that the TDOA time delay estimation error is a random variable with zero-mean normal distribution. Determining TDOA_(m)i，jThe method for judging whether the NLOS error is included is to perform NLOS identification on the ith radiation source and the jth radiation source respectively, the identification method here is various, and one of the methods can be arbitrarily selected, for example, the NLOS identification can be realized by using the magnitude of a group of sample discrete coefficients of the strongest path selected from a group of power delay profiles acquired from the radiation source, or the NLOS identification can be realized by using the inter-path power difference or amplitude difference on a single or multiple power delay profiles acquired from the radiation source.

Substep 202 of step 101 determines the mean of the NLOS errors in the TDOA measurements. According to the NLOS recognition result of step 201, first, a distribution parameter of the NLOS error is obtained from a set of power delay distributions of the channel corresponding to the TDOA measurement containing the NLOS error. In concrete implementation, the NLOS error distribution parameter p may be implemented by using the following formula (1) or formula (2)_iAnd p_jIs estimated.

$p_i=\frac{(m_1+m_2+\·\·\·+m_N)\×a}{W\×N}---\left(1\right)$

In the formula (1), p_iIs the NLOS error distribution parameter; m is_kThe number of the paths detected in the kth (k belongs to 1, 2.., N) scatterer statistical window, wherein the scatterer statistical window is intercepted from the kth power delay distribution, and the starting point of the scatterer statistical window can be a certain position behind the first path and can also comprise the position of the first path; w is the width of a scatterer statistical window, the unit is a chip, and the value of W is usually between 1-10 chips; n is to obtain a p_iThe number of power time delay distributions adopted by the estimated value is usually 1-10, and the N power time delay distributions are obtained by performing multipath search for N times in a certain time interval; a is the number of sampling performed in one chip, that is, the number of sampling points included in one path, and the value range is usually 1 to 32.

$p_i=\frac{s_1+s_2+\·\·\·{+s}_N}{W\×N}---\left(2\right)$

In the formula (2), p_iIs NLOS error distribution parameter; s_kThe number of detected sampling points exceeding a detection threshold in a kth (k belongs to 1, 2.. multidot.n) scatterer statistical window, wherein the scatterer statistical window is intercepted from the kth power delay distribution, and the starting point of the scatterer statistical window can be a certain position behind the first path and can also comprise the position of the first path; w is the width of a scatterer statistical window, the unit is a sampling point, the value of W is usually within 40 sampling points, and the typical value is 20 sampling points; n is to obtain a p_iThe number of power delay distributions adopted by the estimated value is usually 1-10, and the N power delay distributions are obtained by performing multipath search for N times in a certain time interval.

After obtaining the distribution parameter of the NLOS error, the NLOS error distribution parameter p is utilized_iAnd p_jAnd calculating the average value of the NLOS errors in the distribution form of the NLOS errors.

The mean of the discrete form NLOS error can be taken as p_i、p_jAnd NLOS error δ of TDOA in discrete form represented by equation (3)_(s)i，jThe probability density function of (a) is obtained directly.

$f_{{\mathrm{\&delta;}}_{\left(s\right)i,j}}\left({\mathrm{\&delta;}}_{\left(s\right)i,j}\right)=\left\{\begin{array}{cc}{p}_i{p}_j\frac{(1-p_i)(1-p_j)}{1-(1-p_i)(1-p_j)};{\mathrm{\&delta;}}_{\left(s\right)i,j}=0& \left(3a\right)\\ p_i{p}_j\frac{(1-p_i)(1-p_j)}{1-(1-p_i)(1-p_j)}(1-p_j{)}^{{\mathrm{\&delta;}}_{\left(s\right)i,j}};{\mathrm{\&delta;}}_{\left(s\right)i,j}>0& \left(3b\right)\\ p_i{p}_j\frac{(1-p_i)(1-p_j)}{1-(1-p_i)(1-p_j)}{(1-p_i)}^{{\mathrm{\&delta;}}_{\left(s\right)i,j}};{\mathrm{\&delta;}}_{\left(s\right)i,j}<0& \left(3c\right)\end{array}\right.$

In the formula (3), δ_(s)i，jIs the NLOS error amount of TDOA corresponding to the ith and jth base stations in the unit of the number of sampling points, delta_(s)i，jE (· -3, -2, -1, 0, 1, 2, 3.) and the product of the number of sampling points and the sampling point interval is the NLOS error, and the dimension is time; p is a radical of_iAnd p_jNLOS error amount delta measured by TOA of ith and jth base stations, respectively_(s)i、δ_(s)jThe distribution parameter of (2); delta_(s)i，j＝δ_(s)i-δ_(s)j。

The mean of the continuous form of the NLOS error can be taken as p_i、p_jAnd equation (4) to determine the distribution parameter θ of the NLOS error in the continuous TDOA measurements_iAnd theta_jThen using theta_i、θ_jAnd NLOS error δ of TDOA in continuous form represented by equation (5)_i，jThe probability density function of (2) averages the NLOS errors.

${\mathrm{\&theta;}}_i=T\frac{-1}{\ln (1-p_i)}---\left(4\right)$

In equation (4), T is the system sampling interval time in microseconds.

$f_{{\mathrm{\&delta;}}_{i,j}}\left({\mathrm{\&delta;}}_{i,j}\right)=\left\{\begin{array}{cc}\frac{1}{{\mathrm{\&theta;}}_i+{\mathrm{\&theta;}}_j}{e}^{\frac{{\mathrm{\&delta;}}_{i,j}}{{\mathrm{\&theta;}}_j}};{\mathrm{\&delta;}}_{i,j}<0& \left(5a\right)\\ \frac{1}{{\mathrm{\&theta;}}_i+{\mathrm{\&theta;}}_j};{\mathrm{\&delta;}}_{i,j}=0& \left(5b\right)\\ \frac{1}{{\mathrm{\&theta;}}_i+{\mathrm{\&theta;}}_j}{e}^{-\frac{{\mathrm{\&delta;}}_{i,j}}{{\mathrm{\&theta;}}_i}};{\mathrm{\&delta;}}_{i,j}>0& \left(5c\right)\end{array}\right.$

In the formula (5), θ_iAnd theta_jIs simply delta_i，jThe distribution parameter of (2).

Substep 203 of step 101, using the mean μ of the NLOS errors obtained in step 202_(n)i，jZero-mean correction of NLOS errors is performed according to equation (6).

${\mathrm{TDOA}}_{i,j}^{(\mathrm{nlos}\_\mathrm{miti})}={\mathrm{TDOA}}_{\left(m\right)i,j}-{\mathrm{\&mu;}}_{\left(n\right)i,j}={\mathrm{TDOA}}_{\left(\mathrm{los}\right)i,j}^{\left(0\right)}+n_{\left(s\right)i,j}+n_{i,j}---\left(6\right)$

In equation (6), TDOA_i，j ^(nlos_miti)Represents the TDOA measurements after NLOS error zero-mean correction.

102, calculating the TOA of the main base station according to the TDOA measurement quantity after zero mean correction₁TOA with neighbor base stations_i。

First, the TOA of the master base station is calculated₁. The specific implementation method comprises the following steps: TDOA obtained from equation (6)_2，1 ^(nlos_miti)And TDOA_3，1 ^(nlos_miti)Multiplying by the speed of light instead of r in equation (7)_2，1And r_3，1Obtaining the position coordinates x, y of the mobile station containing the intermediate variable r₁Is shown in (a).

$\left[\begin{array}{c}x\\ y\end{array}\right]=-{\left[\begin{array}{cc}{x}_{\mathrm{2,1}}& y_{\mathrm{2,1}}\\ x_{\mathrm{3,1}}& y_{\mathrm{3,1}}\end{array}\right]}^{-1}\&times;\{\left[\begin{array}{c}{r}_{2,1}\\ r_{\mathrm{3,1}}\end{array}\right]r_1+\frac{1}{2}\left[\begin{array}{c}{r_{\mathrm{2,1}}}^2-K_2+K_1\\ {r_{\mathrm{3,1}}}^2-K_3+K_1\end{array}\right]\}---\left(7\right)$

In the formula (7), the first and second groups, $K_i=x_i^2+y_i^2(i=\mathrm{1,2,3}),$ (x_i，y_i) Is the location coordinates of the base station.

The result of equation (7), i.e., x, y, is then taken to contain the intermediate variable r₁Substituting the value of (c) into the formula (8) when i is equal to 1, and obtaining r₁Of the square form r₁The positive value of the solution is the distance from the mobile station to the reference base station, where the reference base station is numbered 1, and the distance divided by the speed of light is the desired TOA₁. In very special cases, r₁There are two positive roots, and it is necessary to determine a correct solution by using a priori knowledge and combining partial measurement data, such as RTT measurement amount and sector information where the mobile station is located, and abandon the other solution. To increase the TOA₁Can obtain a plurality of TOAs for a plurality of TDOA measurements₁The averaging is performed. Due to TOA₁Using corrected TDOA measurements_i，1 ^(nlos_miti)Obtained, TOA₁The average of the NLOS error of (1) is zero.

$r_i^2={(x_i-x)}^2+{(y_i-y)}^2=K_i-2x_{i}x-2y_{i}y+x^2+y^2,i=\mathrm{1,2},...,M---\left(8\right)$

In the formula (8), the first and second groups, $K_i=x_i^2+y_i^2,$ (x_i，y_i) Is the position coordinate of the base station and is a known parameter; (x, y) are the mobile station position coordinates and are unknown parameters, i.e., parameters to be solved for.

In calculating the TOA of the main base station₁Then, TOA of each adjacent base station is calculated_iTOA here_iThe method comprises NLOS error, and is implemented by calculation according to formula (9).

TOA_i＝-TDOA_(m)i，1+TOA₁ (9)

In formula (9), TDOA_(m)i，1Representing the amount of TDOA measurements between the ith base station and the 1 st base station that contain NLOS errors.

Step 103, performing zero-mean correction on the NLOS error included in the TOA measurement quantity using the mean of the NLOS error calculated from the NLOS error distribution parameters.

The process of zero-mean correction of the NLOS error in the TOA measurement, as shown in fig. 3, includes the following sub-steps 301, 302, 303:

substep 301 of step 103 determines, by NLOS identification, the amount of TOA measurements that require zero-mean correction. Identifying TOAs containing NLOS errors_iThe subscript i denotes this TOA_iIs the time of arrival or pseudorange between the i-th radiation source and the receiver, such as a base station in a cellular network; ${\mathrm{TOA}}_i={\mathrm{TOA}}_{\left(\mathrm{los}\right)i}^{\left(0\right)}+n_{\left(n\right)i}+{\mathrm{\&mu;}}_{\left(n\right)i}+n_j.$ wherein, TOA_(los)i ⁽⁰⁾TOA value which is an ideal state containing no error; n is_(n)iThe residual error of the NLOS error after zero-mean correction, wherein the mean value is zero; mu.s_(n)iThe mean value of the NLOS error is calculated according to the NLOS error distribution parameters; n is_iThe TOA time delay estimation error of the system in the LOS environment is a random variable with zero-mean normal distribution. Determine TOA_iThe method of determining whether to include the NLOS error is to perform NLOS identification on the ith radiation source, where the identification method is a well-known technique, and may be any one of a plurality of identification methods, for example, NLOS identification may be implemented by using the magnitude of a set of sample discrete coefficients of the strongest path selected from a set of power delay profiles acquired from the radiation source, or NLOS identification may be implemented by using the inter-path power difference or amplitude difference on a single or multiple power delay profiles acquired from the ith radiation source.

Substep 302 of step 103, conducting TOA_iEstimation of the mean of the medium NLOS error.

First, from the result of NLOS recognition in step 301, TOA containing NLOS error_iObtaining the distribution parameter of the NLOS error from a group of power delay distributions of the corresponding channel, and when the NLOS error distribution parameter p is specifically implemented, the NLOS error distribution parameter p may be implemented by using either formula (1) or formula (2)_iIs estimated.

Then, using the NLOS error distribution parameter p_iAnd TOA_iAnd calculating the average value of the NLOS errors in the distribution form of the NLOS errors.

Discrete form of TOA_iThe medium NLOS error is a geometric distribution, and the mean value can be p_iAnd a discrete form of TOA represented by equation (10)_iMedium NLOS error delta_(s)iThe probability density function of (a) is obtained directly.

$f_{\mathrm{\&delta;}}\left({\mathrm{\&delta;}}_{\left(s\right)i}\right)=\left\{\begin{array}{cc}{p}_i{(1-p_i)}^{{\mathrm{\&delta;}}_{\left(s\right)i}},& {\mathrm{\&delta;}}_{\left(s\right)i}\&Element;(\mathrm{0,1,2}...)\\ 0,& {\mathrm{\&delta;}}_{\left(s\right)i}\&NotElement;(\mathrm{0,1,2}...)\end{array}\right.---\left(10\right)$

δ in equation (10)_(s)iThe number of sampling points of the system sample is represented, the product of the number of sampling points and the sampling point interval is NLOS error, and the dimension of the NLOS error is time and p_iIs a distribution parameter of the geometric distribution.

Continuous form of TOA_iThe NLOS error of (1) follows a single-sided exponential distribution, the mean of which can be represented by p_iAnd equation (4) to find TOA in continuous form_iDistribution parameter theta of NLOS error in (1)_iThen using theta_iAnd TOA of continuous form represented by formula (11)_iNLOS error δ of_iThe probability density function of (2) averages the NLOS errors.

$f_{\mathrm{\&delta;}}\left({\mathrm{\&delta;}}_i\right)=\left\{\begin{array}{cc}\frac{1}{{\mathrm{\&theta;}}_i}{e}^{-\frac{{\mathrm{\&delta;}}_i}{{\mathrm{\&theta;}}_i}},& {\mathrm{\&delta;}}_i>0\\ 0,& {\mathrm{\&delta;}}_i\&le;0\end{array}\right.---\left(11\right)$

Delta in the formula (11)_iRepresenting NLOS error, θ, of successive values_iIs a distribution parameter.

Substep 303 of step 103, using the mean μ of the NLOS errors obtained in step 302_(n)iTOA according to equation (12)_iZero mean correction of medium NLOS errors.

${\mathrm{TOA}}_i^{(\mathrm{nlos}\_\mathrm{miti})}={\mathrm{TOA}}_i-{\mathrm{\&mu;}}_{\left(n\right)i}={\mathrm{TOA}}_{\left(\mathrm{los}\right)i}^{\left(0\right)}+n_{\left(n\right)i}+n_i---\left(12\right)$

In the formula (12), TOA_i ^(nlos_miti)The TOA measurement quantity after NLOS error zero-mean correction is carried out is represented; n is_(n)iIs TOA_iThe corrected residual error of (2); n is_iAn error is estimated for the delay.

And step 104, utilizing the variance of the NLOS error obtained by the NLOS error distribution parameters to carry out adaptive adjustment on elements in a weighting matrix in weighted least square estimation.

The adaptive adjustment of the weighting matrix is shown in fig. 4, and further includes sub-steps 401, 402, and 403:

in sub-step 401 of step 104, a form of weighting matrix is determined, and the present invention uses equation (13) as the weighting matrix for the TOA least squares estimation.

$Q_{\left(r\right)}=Q_{\left(l\right)}+Q_{\left(n\right)}=\left.-\begin{array}{ccccc}{{\mathrm{\&sigma;}}_1}^2& 0& 0& \&CenterDot;& 0\\ 0& {{\mathrm{\&sigma;}}_2}^2& 0& \&CenterDot;& 0\\ 0& 0& {{\mathrm{\&sigma;}}_3}^2& \&CenterDot;& 0\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;\\ 0& 0& 0& \&CenterDot;& {{\mathrm{\&sigma;}}_M}^2\end{array}\right]+\left[\begin{array}{ccccc}{{\mathrm{\&sigma;}}_{\left(n\right)1}}^2& 0& 0& \&CenterDot;& 0\\ 0& {{\mathrm{\&sigma;}}_{\left(n\right)2}}^2& 0& \&CenterDot;& 0\\ 0& 0& {{\mathrm{\&sigma;}}_{\left(n\right)3}}^2& \&CenterDot;& 0\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;\\ 0& 0& 0& \&CenterDot;& {{\mathrm{\&sigma;}}_{\left(n\right)M}}^2\end{array}\right]---\left(13\right)$ (13)

Wherein,

$Q_{\left(l\right)}=\left[\begin{array}{ccccc}{{\mathrm{\&sigma;}}_1}^2& 0& 0& \&CenterDot;& 0\\ 0& {{\mathrm{\&sigma;}}_2}^2& 0& \&CenterDot;& 0\\ 0& 0& {{\mathrm{\&sigma;}}_3}^2& \&CenterDot;& 0\\ \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;& \&CenterDot;\\ 0& 0& 0& \&CenterDot;& {{\mathrm{\&sigma;}}_M}^2\end{array}\right]---\left(14\right)$

Q_(l)time delay estimation error N ═ N of TOA under LOS environment₁n₂...n_M]^TN is an M-dimensional vector, the mean value of N is zero, and the covariance matrix Q of N_(l)Is an M × M dimensional symmetric matrix. Sigma_i ²The covariance of the TOA delay estimation error in the LOS environment.

$Q_{\left(n\right)}=\left[\begin{array}{ccccc}{{\mathrm{\&sigma;}}_{\left(n\right)1}}^2& 0& 0& \&CenterDot;& 0\\ 0& {{\mathrm{\&sigma;}}_{\left(n\right)2}}^2& 0& \&CenterDot;& 0\\ 0& 0& {{\mathrm{\&sigma;}}_{\left(n\right)3}}^2& \&CenterDot;& 0\\ \&CenterDot;& 0& 0& \&CenterDot;& {{\mathrm{\&sigma;}}_{\left(n\right)M}}^2\\ 0& & & & \end{array}\right]---\left(15\right)$

Q_(n)Corrected NLOS error n for zero mean_(n)iConstructed M-dimensional vector N_(n)＝[n_(n)1 n_(n)2... n_(n)M]^TCovariance matrix of N_(n)Has a mean value of zero, Q_(n)Is an M × M dimensional symmetric matrix.

Sub-step 402 of step 104 determines the elements of the weighting matrix that need to be adjusted according to the NLOS recognition result. This step directly utilizes the results of step 401, so long as the TOA is constructed_iThe corresponding channel of the ith radiation source(s) contains NLOS errors, the corresponding element in the weighting matrix, i.e. the weighting coefficient, is adjusted. For example, if the channel corresponding to the 1 st radiation source is determined to be an LOS channel through NLOS identification, Q in formula (13)_(n)Containing sigma_(n)1 ²The element of (a) needs to be adjusted, i.e. sigma_(n)1 ²Take the value zero.

In sub-step 403 of step 104, the values of the elements to be adjusted in the weighting matrix are determined.

First, a matrix Q is determined_(l)The value of each element in (1), i.e. sigma in formula (14)_i ²The value of (a). Sigma can be approximated by carrying out TOA measurement error statistics on the system in an LOS channel environment_i ²σ can also be approximated by system simulation_i ²σ can also be approximated by statistics of TDOA measurement errors_i，j ²Then use ${\mathrm{\&sigma;}}_i^2=\frac{1}{2}{{\mathrm{\&sigma;}}_{i,j}}^2$ To determine sigma_i ²。

Then, a matrix Q is determined_(n)The value of each element in the list. P found in step 103 can be used here_iAnd TOA represented by the formula (10)_iNLOS error δ of_(s)iIs calculated as a function of the probability density_(n)i ²Step 103 may also be utilizedFound p_iAnd equation (4) to determine θ_iThen, the obtained theta is reused_iAnd TOA represented by formula (11)_iMedium NLOS error delta_iIs calculated as a function of the probability density_(n)i ²。TOA₁The variance of the NLOS error of (1) adopts the distribution parameter p obtained from the main base station₁Performing calculation, the specific steps and TOA_iThe variance of (c) is calculated in the same way.

And 105, performing weighted least square estimation to obtain a group of estimated values of the position of the mobile station which preliminarily inhibits the influence of NLOS errors.

When TDOAs of a plurality of base stations greater than 3 are measured, one TOA acquired in step 102 is used₁And a plurality of TOAs_i ^{(i＝2，3，...，M M≥4)}Substituting equation (8) to construct a nonlinear equation system by introducing an intermediate variable d₁＝(x-x₁)²+(y-y₁)²And solving the system of nonlinear equations using weighted least squares estimation, the position coordinates (x, y) of the mobile station are obtained. In order to provide NLOS error suppression capability to the solution of the system of equations (8) obtained when M is 4 or greater, the Q determined in step 104 is used in solving this system of nonlinear equations using weighted least squares estimation_(r)The covariance matrix of the estimation error of the substitution equation set (8). The position estimation value obtained in this way suppresses the GDOP error and the NLOS error well.

Step 106, further suppressing the influence of the NLOS error on the position estimation accuracy by averaging the set of mobile station position values acquired in step 105. That is, first, a plurality of position estimation values output on the basis of independent TOA measurements are acquired in a short time interval such as several seconds, and then these position estimation values (coordinate components) are averaged to obtain one position coordinate, by step 105.

In this step, by using the characteristic that the mean value of the NLOS error of the TOA measurement value after zero-mean correction according to the formula (12) is zero, the plurality of position estimation results obtained in step 105 are averaged, thereby further suppressing the influence of the NLOS error correction residual on the position estimation accuracy and reducing the variance of the position estimation error.

The above is merely a detailed description of one embodiment of the present invention and is not intended to limit the scope of the invention.

Claims (9)

1. A method for position estimation, wherein a time difference of arrival TDOA between a main base station and two neighboring base stations involved in a positioning request is measured, the method further comprising:

a. performing zero-mean correction on NLOS errors in TDOA measurement quantity by using NLOS error mean values obtained by non-line-of-sight NLOS error distribution parameters, estimating the TOA of the mobile station to a main base station according to the corrected TDOA measurement quantity, and calculating the TOA of the mobile station to an adjacent base station by using the TOA of the mobile station to the main base station and the uncorrected TDOA between the mobile station to the main base station and the adjacent base station obtained by system measurement;

b. performing zero mean correction on NLOS errors in TOAs from the mobile station to the main base station and from the mobile station to the adjacent base station by using NLOS error mean values calculated by the NLOS error distribution parameters;

c. and carrying out self-adaptive adjustment on a weighting matrix in the TOA position estimation by using the NLOS error variance obtained by the NLOS error distribution parameters, and carrying out weighted least square estimation on the position of the mobile station by using the adjusted weighting matrix to obtain an estimated value of the position of the mobile station.

2. The method of claim 1, wherein the steps a to c are repeated to obtain more than 1 mobile station position estimation values, and the more than 1 mobile station position estimation values are averaged to obtain a final mobile station position estimation value.

3. The position estimation method according to claim 1 or 2, characterized in that step a further comprises:

a1. TDOA measurement quantity needing zero mean value correction is determined through NLOS identification, and mean value of NLOS errors in the TDOA measurement quantity is determined through NLOS error distribution parameters;

a2. performing zero-mean correction on the NLOS error in the TDOA measurement quantity according to the NLOS error mean value determined in the step a 1;

a3. calculating the TOA from the mobile station to the main base station according to the TDOA measured quantity obtained after correction in the step a2, and calculating the TOA from the mobile station to each adjacent base station according to the calculated TOA value from the mobile station to the main base station and the uncorrected TDOA measured by the system between the mobile station to the main base station and the adjacent base station.

4. The position estimation method according to claim 3, further comprising the step of, in step a 3: and after the TOA of the main base station is calculated, calculating more than 1 time, and calculating the TOA value of each adjacent base station according to the average value of the obtained TOA values more than 1 and the uncorrected TDOA measured by the system from the mobile station to the main base station and the adjacent base station.

5. The position estimation method according to claim 3, wherein the step b further comprises:

b1. determining the TOA values of the main base station and the adjacent base stations needing zero-mean correction, which are obtained in the step a3, by NLOS identification, and determining the mean value of NLOS errors in the TOA values of the main base station and the adjacent base stations needing zero-mean correction, which are obtained in the step a3, by using NLOS error distribution parameters;

b2. and c, performing zero-mean correction on the mean value of the NLOS errors in the TOA values of the main base station and the adjacent base stations which need to be subjected to zero-mean correction and are obtained in the step a3.

6. The method of claim 5, wherein the NLOS identification is performed by selecting a set of magnitudes of sample discrete coefficients of the strongest path from a power delay profile obtained from a radiation source.

7. The position estimation method of claim 5, wherein the NLOS identification is performed using an inter-path power difference or an inter-path amplitude difference on a power delay profile obtained from a radiation source.

8. The position estimation method according to claim 1 or 2, characterized in that step c further comprises:

c1. determining a weighting matrix form by setting main diagonal elements of the matrix as the sum of the covariance of TOA time delay estimation errors and the covariance of correction residuals of TOA in a line-of-sight (LOS) environment;

c2. setting the covariance value of correction residual errors of the TOA corresponding to each channel containing NLOS errors in the channels for calculating the TOA to be zero;

c3. determining the covariance of the TOA delay estimation error in the LOS environment to be adjusted in the weighting matrix of step c1 by performing TOA measurement estimation on the system in an LOS channel environment, or by system simulation, or by means of TDOA measurement error statistics; judging whether the NLOS error contained in the TOA is in a discrete form or a continuous form, if the NLOS error is in the discrete form, determining the covariance of TOA correction residuals needing to be adjusted in the weighting matrix of the step c1 according to the NLOS error distribution parameters and the probability density function of the discrete NLOS error in the TOA; if the correction residual error is in a continuous form, the probability density function of the continuous NLOS error of the TDOA is obtained through the NLOS error distribution parameters, and then the covariance of the correction residual error of the TOA to be adjusted in the weighting matrix of step c1 is determined through the probability density function of the continuous NLOS error of the TDOA.

9. A method of position estimation according to claim 1 or 2, characterized in that in step a TDOA measurements are determined by GDOP minimum criterion or by the presence of LOS channel criterion.

Images