Method for locating a transmitter

Abstract

The invention relates to a method for locating a transmitter, wherein the following steps are carried out in the method: Emitting a wave-based signal by the transmitter, so that a wave field emanates from the transmitter, Receiving the wave-based signal by at least one, in particular several, receiver, which has at least one receiving antenna, preferably an antenna array with several spatially offset receiving antennas, and Forming a measurement signal in each of the at least one receiver for several, preferably each receiving antenna, wherein the respective measurement signal depends on the spatial distribution of the wave field and its measured phase response is characteristically influenced by a signal propagation time from the transmitter to the respective receiving antenna, and The measured phase response of the respective measurement signal is used both to locate the transmitter and to estimate a receiver-specific measurement error power at at least one receiver.

Description

[0001] The present invention relates to a method for locating a transmitter, in particular by a PDOA (phase-difference-of-arrival)-based determination of the transmitter's position.

[0002] There are already several providers of angle-based positioning systems, particularly using the Bluetooth standard, e.g., Quuppa, infsoft, Bluelot, u-blox. These systems use PDOA measurements from multiple receiver arrays distributed throughout the room, the positions of which are known in advance, to first estimate the angle to the transmitter at each receiver array and then to locate the transmitter using multiangulation. Such a system is in Fig. 1 illustrated.

[0003] It is known from the prior art that the localization accuracy of PDOA systems depends directly on the aperture size of the receiver arrays relative to the measurement distance; see Sippel, Erik; Geiss, Johanna; Brückner, Stefan; Gröschel, Patrick; Hehn, Markus; Vossiek, Martin: Exchanging Bandwidth With Aperture Size in Wireless Indoor Localization - Or Why 5G / 6G Systems With Antenna Arrays Can Outperform UWB Solutions. In: IEEE Open Journal of Vehicular Technology Vol. 2 (2021), pp. 207-217.

[0004] Known angle-based PDO localization approaches do not work optimally for large receiver arrays because the transmitter to be located is typically in the receiver's near field. This contradicts the assumption of plane received waves used by angle estimators. Due to the spherical waves emitted by the transmitter, this assumption only holds true for limited spatial regions (namely the far field) and can therefore only be considered valid for small receiver arrays. The near field is often referred to in the literature as the Fresnel region, where the Fresnel approximation applies, but not the Fraunhofer approximation. The far field, on the other hand, is often referred to as the Fraunhofer region, where the Fraunhofer approximation applies.

[0005] Furthermore, according to current technology, receiver arrays for angle estimation are typically fully populated with receiving antennas, so that the entire area is covered with antennas, usually spaced half a wavelength apart. This becomes very complex and expensive for large receiving arrays. Therefore, sparsely populated receiver arrays are often used for positioning with large arrays; see Pavlenko, T.; Reustle, C.; Dobrev, Y.; Gottinger, M.; Jassoume, L.; Vossiek, M.: Design and Optimization of Sparse Planar Antenna Arrays for Wireless 3-D Local Positioning Systems. In: IEEE Transactions on Antennas and Propagation Vol. 65 (2017), No. 12, pp. 7288-7297.

[0006] Consequently, for precise and cost-efficient PDOA-based positioning systems, the evaluation of raw data (for example, complex amplitudes of the received signals) without angle estimation is necessary.

[0007] It is known, for example, from DE102019110512A1, that the position of the transmitter can be efficiently determined by recursively evaluating the phase differences between the spatially distributed receiving antennas of the receiving arrays. This is usually done by applying recursive filters, in particular the extended Kalman filter in the form of the holographic extended Kalman filter (HEKF), cf. BRÜCKNER, STEFAN; SIPPEL, ERIK; LIPKA, MELANIE; GEISS, JOHANNA; VOSSIEK, MARTIN: Phase Difference Based Precise Indoor Tracking of Common Mobile Devices Using an Iterative Holographic Extended Kalman Filter. In: IEEE Open Journal of Vehicular Technology Vol. 3 (2022), pp. 55-67.

[0008] The metric used here is always implicitly least squares. Calculating difference phases from the measured absolute phases generates correlated measurement noise, which is represented within the covariance matrix by entries outside the main diagonal. The correct implementation of this covariance matrix within the HEKF is essential, as it ensures that information from the relative phases of distant antenna pairs is implicitly included, even though this information is not directly calculated within the phase difference evaluation matrix. See, for example: SIPPEL, ERIK: Holographic 3D Indoor Localization, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 2022.

[0009] The position estimation, based on the least-squares metric, assumes normally distributed additive noise (AWGN - additive white Gaussian noise). The position estimation thus implicitly assumes constant measurement conditions from each receiver to the transmitter. However, this is not accurate in large, changing environments.

[0010] One reason for this is, for example, the fact that measurement conditions in real-world situations are constantly changing. In particular, the line of sight being evaluated is regularly blocked, which automatically leads to poorer location results, as the measurements are assumed to be correct and included in the evaluation.

[0011] This necessitates automatic detection of the receiver's current measurement conditions.

[0012] Systems that adaptively estimate the covariance matrices of measurement and system noise within Kalman filters are known from the prior art; see AKHLAGHI, SHAHROKH; ZHOU, NING; HUANG, ZHENYU: Adaptive adjustment of noise covariance in Kalman filter for dynamic state estimation. In: 2017 IEEE Power Energy Society General Meeting, 2017, pp. 1-5; M EHRA , R.: On the identification of variances and adaptive Kalman filtering. In: IEEE Transactions on Automatic Control Vol. 15 (1970), No. 2, pp. 175-184; MEHRA, R.: Approaches to adaptive filtering. In: IEEE Transactions on Automatic Control Vol. 17 (1972), No. 5, pp. 693-698; M OHAMED , AH; S CHWARZ , KP: Adaptive Kalman Filtering for INS / GPS. In: Journal of Geodesy Vol. 73 (1999), No. 4, pp. 193-203 or W ANG, JINLING : Stochastic Modeling for Real-Time Kinematic GPS / GLONASS Positioning. In: NAVIGATION Vol. 46 (1999), No. 4, pp. 297-305.

[0013] This implies that the estimation of the covariance matrices of the measurement noise is primarily relevant for the HEKF. This is always fully estimated in the proposed systems. However, the evaluation of phase differences results in a very large number of entries in the covariance matrix of the measurement noise, making it difficult to estimate.

[0014] If the covariance matrix of the measurement noise is estimated, it will have an arbitrary, indeterminate form. However, the evaluation of difference matrices within the HEKF necessarily requires a representation shaped according to the evaluation of the phase differences. Therefore, it follows that the complete estimation of the covariance matrices of the measurement noise known from the literature is not possible with the HEKF.

[0015] The present invention overcomes the disadvantages listed above or at least eliminates them.

[0016] According to the invention, a method for locating a transmitter is provided, wherein the following steps are carried out in the method: Emitting a wave-based signal by the transmitter, so that a wave field emanates from the transmitter, Receiving the wave-based signal by at least one, in particular several, receiver, which has at least two receiving antennas, preferably an antenna array with several spatially offset receiving antennas, and Forming a measurement signal in each of the at least one receiver for several, preferably each receiving antenna, wherein the respective measurement signal depends on the spatial distribution of the wave field and its measured phase response is characteristically influenced by a signal propagation time from the transmitter to the respective receiving antenna, and The measured phase response of the respective measurement signal is used both to locate the transmitter and to estimate a receiver-specific measurement error power at at least one receiver.

[0017] According to a further optional development of the present invention, it can be provided that an Extended Kalman Filter, EKF, in particular a Holographic Extended Kalman Filter, HEKF, is used to locate the transmitter.

[0018] Advantageously, it can be provided that a covariance matrix of a residual of the localization is estimated or generated, and that the receiver-specific measurement error powers at at least one receiver are deduced from the entries of the covariance matrix of the residual.

[0019] According to a further modification of the present invention, it can be provided that a receiver is used to locate the transmitter only if its currently estimated receiver-specific measurement error power is low, preferably below a predetermined threshold value.

[0020] Furthermore, according to an advantageous embodiment of the present invention, it can be provided that the estimated receiver-specific measurement error power of a respective receiver is used to weight the influence of the receiver on the localization of the transmitter, in particular wherein a high estimated receiver-specific measurement error power leads to a reduction in the weighting of the respective receiver.

[0021] According to a further optional embodiment of the present invention, it can be provided that at least one receiving array of the at least one receiver is so large that the transmitter to be localized is located in the near field of this receiver.

[0022] Furthermore, according to an advantageous modification of the present invention, it can be provided that a Doppler evaluation, a magnetic field-based position determination, an optical system, an ultrasound signal and / or additional sensor values, in particular sensor values ​​of an inertial sensor, are taken into account for the localization of the transmitter.

[0023] The method according to the invention can be further developed by the following refinement, in which subgroups of receiving antennas are formed at at least one receiver, the receiver-specific measurement error powers of which are estimated.

[0024] Advantageously, it may be provided that the estimated receiver-specific measurement error power of a subgroup is used to weight the influence of the subgroup on the localization of the transmitter, in particular where a higher estimated receiver-specific measurement error power leads to a reduction in the weighting of the subgroup.

[0025] The invention further relates to a system for carrying out a method according to one of the aspects discussed above.

[0026] Further features, details and advantages of the invention will become apparent from the following description of the concept according to the invention and a detailed embodiment. The following are shown: Fig. 1: A representation of the state of the art with a localization environment in which three receivers, each with six receiving antennas, locate a transmitter, Fig. 2: a sketch of a recipient from Fig. 1, where the dashed lines represent a “spanning tree” for evaluating the phase differences, Fig. 3: a representation of a HEKF with adaptive estimation of the measurement error power of the individual receivers, and Fig. 4: A representation of an ECF with adaptive estimation of the entire covariance matrix. Basic concept of the invention

[0027] According to the invention, a concept is provided which enables the adaptive detection of measurement error power at each array within a localization system, such as that described in Fig. As shown in Figure 1, the system can be structured from many receiving arrays. The HEKF is used as the localization algorithm, as already known from DE 10 2019 110 512 A1. The adaptive detection of the measurement error power enables the detection of challenging multipath conditions, blocked lines of sight, poorly calibrated receivers, etc. Kalman filters always assume that a system state x k The task is to estimate which changes over time steps k according to a system model and is to be estimated using a measurement model. The system model xk=ƒk(xk−1)+wk In HEKF, it consists of an arbitrarily selectable state transition model f. k and a system noise w kA typical system model for a localization algorithm like the HEKF is the constant-velocity model, which assumes that the target moves at approximately a constant speed. The system model can be extended to include any number of additional parameters to be estimated, which have little or nothing to do with the actual evaluation of the HEKF, such as inertial sensors. Furthermore, the influence of system noise can be eliminated by choosing the appropriate system model and system noise level. The changing system state x k is done using the measurement model zk=hk(xk)+vk estimated, where h k the measurement function and v k This represents the measurement noise. In the case of the HEKF, h k The phase differences between the spatially distributed antennas are evaluated at each receiver. This assumes that at time step k at the n A tenth antenna out of a total of N A,nR antennas of the nR ten recipients of a total of N R Recipients the phase φnR,nA,k=mod2π'(ω0τnR,nA,k+φin,nR,k) is measured, cf. BRÜCKNER, STEFAN; SIPPEL, ERIK; LIPKA, MELANIE; GEISS, JOHANNA; VOSSIEK, MARTIN: Phase Difference Based Precise Indoor Tracking of Common Mobile Devices Using an Iterative Holographic Extended Kalman Filter. In: IEEE Open Journal of Vehicular Technology Vol. 3 (2022), pp. 55-67 and SIPPEL, ERIK: Holographic 3D Indoor Localization, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 2022, where mod2π'(⋅) the ambiguous phases with mod2π'(θ)={mod2π(θ)if mod2π(θ)≤πmod2π(θ)−2πif mod2π(θ)>π maps to (-π, π], ω0 = 2πf0 corresponds to the angular frequency, φ in,nR,k represents the unknown incoherent phase of the transmit-receive process and the delay τ nR,nA,k from the position of the transmitter p TX,k and the receiving antenna p nR,nA to τnR,nA,k=‖pTX,k−pnR,nA‖c0 This results in... To use the measurements for position estimation, the phase measurements are summarized and phase differences are determined. For the n R ten receivers thus yields the measurement vector φnR,k=(φnR,1,k⋮φnR,NA,nR,k).

[0028] The corresponding noisy measurement vector is obtained by superimposing it with the noise vector v. nR,k to φnR,k,Mess=φnR,k+νnR,k.

[0029] The measurement errors v nR,k have the recipient-specific performance that changes over time σnR,k2. This is referred to below as receiver-specific measurement error power. Depending on the system and measurement model, the time-varying receiver-specific measurement error power can vary. σnR,k2 using a model from the previous receiver-specific measurement error power σnR,k−12 The measurement errors can arise from various effects such as noise, calibration errors, multipath propagation, or shadowing. Multipath propagation and shadowing are the dominant sources of error in indoor localization systems. The absolute phases are derived from the evaluation matrix A. nR the phase differences to ΔφnR,k=mod2π'(AnRφnR,k) and ΔφnR,k,Mess=mod2π'(AnRφnR,k,Mess) calculated. The evaluation matrix A nR Each row calculates the phase differences between two antennas and contains only one "1" and one "-1" per row, otherwise being filled with zeros. To evaluate all the information, it must have rank N. A,nR - 1, where one degree of freedom is lost due to the unknown incoherent phase. Since the evaluated phase difference becomes more ambiguous with increasing antenna distance, it is advisable to use the evaluation matrix AnR to design it in such a way that a "spanning tree" is created by evaluating the phase differences of neighboring antennas, which connects all antennas together, cf. BRÜCKNER, STEFAN ; SIPPEL, ERIK ; LIPKA, MELANIE ; GEISS, JOHANNA; VOSSIEK, MARTIN: Phase Difference Based Precise Indoor Tracking of Common Mobile Devices Using an Iterative Holographic Extended Kalman Filter. In: IEEE Open Journal of Vehicular Technology Vol. 3 (2022), pp. 55-67 and SIPPEL, ERIK: Holographic 3D Indoor Localization, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 2022. An example "spanning tree", in which the evaluation matrix evaluates the phase differences of the antenna pairs 1-2, 2-3, 3-4, 4-6, 6-5, is shown in Fig. 2 can be seen. Assuming that the measurement error contributions at the individual antennas were uncorrelated with each other, i.e., the covariance matrix of v nR , k the form σnR,k2I Evaluating the phase differences yields the following for Δφ nR,k the covariance matrix RnR,k=σnR,k2AnRAnRT for the measurement errors at n R ten recipients. This characteristic structure AnRANRT the resulting covariance matrix R nR This is essential for the functionality of the HEKF, as it allows the relative phase differences of antenna pairs to be implicitly evaluated within the HEKF, even if these are not directly within A. nR calculated, but only connected via the "spanning tree", such as the phase difference of antenna pair 1-5 in Fig. 2. At the same time, the assumption that the measurement error contributions at the individual antennas were uncorrelated is not true for multipath propagation. It follows directly from this that the complete estimates of covariance matrices within Kalman filters, established in the literature, cannot be applied to HEKF, since their estimation of the covariance matrices can generally be arbitrarily structured. This leads to the correlation of multipath propagation being reproduced by entries in the estimated covariance matrix. Thus, the estimation of the receiver-specific measurement error power is σnR,k2 This is necessary, and knowledge of the characteristic structure of the covariance matrix within the HEKF is taken into account. To combine the phase differences of all receivers, these are combined into a vector. hk(xk)=Δφk=(Δφ1,k⋮ΔφnR,k) In summary, the corresponding erroneous measured values ​​result from... zk=mod2π'(hk(xk)+νk)=(Δφ1,k,Meas⋮ΔφNR,k,Measure). where the covariance matrix becomes Rk=(R1,k0…00R2,k⋱⋮⋮⋱⋱00…0RNR,k) results.

[0030] In general, Kalman filters alternately use the system model in the prediction step to derive the estimated system state x. k-1|k-1 and its estimated covariance P k-1|k-1 from the previous time step k - 1 the prediction for the system state x k|k-1 and its estimated covariance P k|k-1 to create the current time step, and in the correction step (English: Update Step) the measurement model in order to derive the system state x from the prediction made and the available measurements. k|k of the current time step and its covariance P k|kto estimate. This invention disclosure describes the correction step by estimating the receiver-specific measurement error power at all receivers. ∑k=(σ1,k2⋮σNR,k2) expanded. ∑^k, σ^nR,k2 and R^k Here, denotes the estimated measurement error powers and their covariance matrices. In the kth step, only the measurement error power estimate from the (k - 1)th step can be used for localization using HEKF. This results in the following algorithm steps: (xk|k−1 , Pk|k−1)=Prediction(xk−1|k−1 , Pk−1|k−1) R^k−1=Combine(∑^k−1) (xk|k, Pk|k, ∑^k)=Correction(xk|k−1, Pk|k−1, zk, R^k−1) k=k+1 and go to step 1

[0031] These are in Fig. Figure 3 is shown as a sketch. Here, in step 1, the prediction of the current system state x is made. k|k-1 and its uncertainty P k|k-1To combine this information with the measurements, in step two, the previously estimated receiver-specific measurement error powers Σ̂ are first calculated for all receivers. k-1 The covariance matrix of the measurement errors of the phase difference evaluation at each receiver R̂ nR,k-1 calculated and used to calculate the total covariance matrix R̂ k-1 joined together. Since steps 1 and 2 are executed independently, their order of execution is irrelevant. The total covariance matrix R̂ created in step 2 k-1 is used, in combination with the prediction from step 1 and the measured phase differences in step 3, to determine the current system state x k|k , the covariance matrix P k|k and receiver-specific measurement error powers at all receivers Σ̂ kto estimate. The process then continues with the next time step k + 1. Depending on the currently estimated receiver-specific measurement error powers, only those receivers with currently good measurement conditions, i.e., low measurement error powers, can be used for position estimation. Furthermore, the presented algorithm can be implemented to estimate the measurement error powers of arbitrary subgroups of receiving antennas at any receiver. In comparison to the HEKF with adaptive estimation of the measurement error power of individual receivers, the established process of adaptive estimation of the covariance matrix in Kalman filters is Fig. Figure 4 shows the covariance matrix R̂. k estimated and evaluated unchanged in the next time step, which would prevent the evaluation of the difference bunnies in HEKF from working. Detailed example

[0032] The following is a detailed example demonstrating how adaptive measurement error estimation can be performed within the HEKF. The implementation of the HEKF is based on BRÜCKNER, STEFAN; SIPPEL, ERIK; LIPKA, MELANIE; GEISS, JOHANNA; VOSSIEK, MARTIN: Phase Difference Based Precise Indoor Tracking of Common Mobile Devices Using an Iterative Holographic Extended Kalman Filter. In: IEEE Open Journal of Vehicular Technology Vol. 3 (2022), pp. 55-67, although for the sake of simplicity, the iterative execution of the correction step is omitted. Steps 1-3 of the estimation process are sequentially examined.

[0033] Step 1: The widely used constant-velocity model is employed for prediction. Here, the system state is... xk=(px,kpy,kpz,kνx,kνy,kνz,k) from position p k and the speed v k of the sender. This allows for the prediction of the system state with xk|k−1=Fxk−1|k−1 calculate, where the matrix F=(100TS000100TS000100TS000100000010000001) This ensures a linear motion model. The corresponding covariance matrix is ​​calculated using Pk|k−1=FPk−1|k−1FT+Qk calculated where Q k The covariance matrix of the constant-velocity model is represented. Step 2: The covariance matrices used for the measurement errors of all receivers in the kth step are calculated based on the previously estimated receiver-specific measurement error powers and the preprocessing matrices used. R^nR,k=σ^nR,k2AnRAnRT calculated. Subsequently, the entire covariance matrix of the measurement errors of all receivers is derived from this. R^k−1=(R^1,k−10…00R^2,−1k⋱⋮⋮⋱⋱00…0R^NR,k−1) compound.

[0034] Step 3: The implementation of step 3 is carried out here by first performing the known correction of the HEKF in step 3.1 and then evaluating the residual in step 3.2 to determine the measurement error powers.

[0035] Step 3.1: First, phase differences of the measurement data are determined at all receivers. ΔφnR,k,Mess=mod2π'(AnRφnR,k,Mess) calculated and to the measurement vector zk=(Δφ1,k,Meas⋮ΔφnR,k,Measure) In summary, the hypothetical phase differences h are derived from the measurement vector. k (x k|k-1 ) subtracted, which is determined by the predicted system state x k|k-1 This results in the measurement difference (residuum) mapped to (-π,π]. d(xk|k−1)=mod2π'(zk−hk(xk|k−1)) The Kalman gain is calculated to determine the correction of the system state. Kk=Pk|k−1HkT(HkPk|k−1HkT+R^k−1)−1 determined, where H kthe Jacobian matrix of the measurement function h k (x k ) on the predicted system state x k|k-1 This results in the corrected system state. xk|k=xk|k−1+Kkd(xk|k−1) and the associated covariance matrix Pk|k=(I−KkHk)Pk|k−1.

[0036] Step 3.2: The previously estimated system state x k|k and the associated covariance matrix P k|k are now used to determine receiver-specific measurement error powers using the measurement residual. σ^nR,k2 to estimate. The measurement residue is calculated as follows: εk=mod2π'(zk−hk(xk|k)).

[0037] Evaluating the expected covariance matrix of the residual allows it to be transformed into, after some adjustments, Ck=E{εkεkT}≈Rk−HkPk|kHkT approximate. Consequently, for estimating the covariance matrix of the observed measurement noise, R^k' An estimation of the covariance matrix of the residual is necessary. This involves the covariance matrix of the observed measurement noise. R^k' only a calculation aid for estimating the measurement error power of the arrays σ^nR,k2 The covariance matrix of the residual is recursively calculated by C^k=(1−α)C^k−1+αεkεkT estimated, where α represents a "forgetting" factor that can be used to control the convergence rate. Ĉ k This represents another system state to be stored. From this, the corresponding covariance matrix of the measurement noise that occurred can be derived. R^k'=C^k+HkPk|kHkT to be calculated. This now has the structure R^k'=(R^1,k'0⋯00R^2,k'⋱⋮⋮⋱⋱00⋯0R^nR,k'), from which the covariance matrix of the observed measurement noise R^nR,k' for each recipient n R is extracted. Since the structure of R^nR,k' Since the phase differences within the HEKF are known from the calculation, this is used to determine the receiver-specific measurement error power of the n. R ten recipients from R^nR,k' to extract by first multiplying with (AnRAnRT)−1 A matrix is ​​created, with the receiver-specific measurement error power to be estimated on its main diagonal. The estimated receiver-specific measurement error power is then extracted by averaging this main diagonal. This yields σ^nR,k2=Trace((AnRAnRT)−1R^nR,k')NA,nR−1, where trace(·) evaluates the sum of the main diagonals of a matrix. These are then combined to Σ^k=(σ^1,k2 ⋮σ^NR,k2) combined. Since the formulas used in step 3.2 are linear transformations, the calculations can also be performed individually for each receiver.

[0038] The method presented in step 3.2 for estimating receiver-specific measurement error power represents both an efficient and high-performing algorithm. The very high performance of estimating receiver-specific measurement error power instead of entire covariance matrices is also directly evident.

[0039] Prior art methods typically determine angles from the measured phase profiles in order to use them for position estimation via a Kalman filter. If the covariance matrix of measurement errors is then adaptively estimated in such a method, this is done solely by directly comparing the angle measurement with the hypothetical angles that would result at the estimated transmitter position. Since the transmitter position is estimated exclusively based on the determined angles, this results in an inherently unstable estimation process, as the erroneous angle estimates directly manifest themselves in erroneous position estimates and thus, in turn, in incorrect estimates of the covariance matrix.

[0040] In contrast, the presented combination of the HEKF according to the invention with the adaptive estimation of the receiver-specific measurement error power compares all evaluated phase differences of the measurement with the hypothetical phase differences that arise at the estimated transmitter position. Thus, the phase differences measured at a receiver relative to each other are also implicitly used for the adaptive estimation of the receiver-specific measurement error power. This makes it possible to detect the deterioration of the measurement conditions even in cases where the position estimate as a whole fits the impaired measured values. In the detailed example shown, this is achieved by using all entries of the covariance matrix, including those beyond the main diagonal, of the measurement noise that has occurred. R^nR,k' for calculating σ^nR,k2. This makes the presented HEKF, with its adaptive estimation of receiver-specific measurement error power, far superior to all known methods. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] DE 102019110512A1 [0007, 0027] Cited non-patent literature

[0000] Sippel, Erik; Geiss, Johanna; Brückner, Stefan; Gröschel, Patrick; Hehn, Markus Vossiek, Martin: Exchanging Bandwidth With Aperture Size in Wireless Indoor Localization - Or Why 5G / 6G Systems With Antenna Arrays Can Outperform UWB Solutions. In: IEEE Open Journal of Vehicular Technology Bd. 2 (2021), S. 207-217

[0003] Pavlenko, T.; Reustle, C.; Dobrev, Y.; Gottinger, M.; Jassoume, L.; Vossiek, M.: Design and Optimization of Sparse Planar Antenna Arrays for Wireless 3-D Local Positioning Systems. In: IEEE Transactions on Antennas and Propagation Bd. 65 (2017), Nr. 12, S. 7288-7297

[0005] BRÜCKNER, STEFAN; SIPPEL, ERIK; LIPKA, MELANIE; GEISS, JOHANNA; VOSSIEK, MARTIN: Phase Difference Based Precise Indoor Tracking of Common Mobile Devices Using an Iterative Holographic Extended Kalman Filter. In: IEEE Open Journal of Vehicular Technology Bd. 3 (2022), S. 55-67 [0007, 0027, 0029, 0032] SIPPEL, ERIK: Holographic 3D Indoor Localization, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 2022

[0008] AKHLAGHI, SHAHROKH; ZHOU, NING; HUANG, ZHENYU: Adaptive adjustment of noise covariance in Kalman filter for dynamic state estimation. In: 2017 IEEE Power Energy Society General Meeting, 2017, S. 1-5

[0012] M EHRA , R.: On the identification of variances and adaptive Kalman filtering. In: IEEE Transactions on Automatic Control Bd. 15 (1970), Nr. 2, S. 175-184

[0012] MEHRA, R.: Approaches to adaptive filtering. In: IEEE Transactions on Automatic Control Bd. 17 (1972), Nr. 5, S. 693-698

[0012] M OHAMED , A. H.; S CHWARZ , K. P.: Adaptive Kalman Filtering for INS / GPS. In: Journal of Geodesy Bd. 73 (1999), Nr. 4, S. 193-203

[0012] W ANG, JINLING : Stochastic Modeling for Real-Time Kinematic GPS / GLONASS Positioning. In: NAVIGATION Bd. 46 (1999), Nr. 4, S. 297-305

[0012] SIPPEL, ERIK: Holographic 3D Indoor Localization, Friedrich-Alexander-University Erlangen-Nuremberg

[0027]

Claims

[1] Method for locating a transmitter, wherein the following steps are performed in the method: Emitting a wave-based signal by the transmitter, so that a wave field emanates from the transmitter, Receiving the wave-based signal by at least one, in particular several, receiver, which has at least two receiving antennas, preferably an antenna array with several spatially offset receiving antennas, and Forming a measurement signal in each of the at least one receiver for several, preferably each receiving antenna, wherein the respective measurement signal depends on the spatial distribution of the wave field and its measured phase response is characteristically influenced by a signal travel time from the transmitter to the respective receiving antenna, characterized by , that The measured phase response of the respective measurement signal is used both to locate the transmitter and to estimate a receiver-specific measurement error power at at least one receiver. [2] Method according to the preceding claim 1, wherein an Extended Kalman Filter, EKF, in particular a Holographic Extended Kalman Filter, HEKF, is used to locate the transmitter. [3] Method according to the preceding claim 2, wherein a covariance matrix of a residual of the localization is estimated or generated, and the receiver-specific measurement error performance at at least one receiver is inferred from the entries of the covariance matrix of the residual. [4] Method according to one of the preceding claims, wherein a receiver is used to locate the transmitter only if its currently estimated receiver-specific measurement error power is low, preferably below a predetermined threshold. [5] Method according to any of the preceding claims, wherein the estimated receiver-specific measurement error power of a respective receiver is used to weight the influence of the receiver on the localization of the transmitter, in particular wherein a high estimated receiver-specific measurement error power leads to a reduction in the weighting of the respective receiver. [6] Method according to one of the preceding claims, wherein at least one receiving array of the at least one receiver is such that the transmitter to be localized is located in the near field of this receiver. [7] Method according to one of the preceding claims, wherein a Doppler evaluation, a magnetic field-based position determination, an optical system, an ultrasound signal and / or additional sensor values, in particular sensor values ​​of an inertial sensor, are taken into account for the localization of the transmitter. [8] Method according to one of the preceding claims, wherein subgroups of receiving antennas are formed at at least one receiver, the receiver-specific measurement error powers of which are estimated. [9] Method according to the preceding claim 8, wherein the estimated receiver-specific measurement error power of a subgroup is used to weight the influence of the subgroup on the localization of the transmitter, in particular wherein a high estimated receiver-specific measurement error power leads to a reduction in the weighting of the subgroup. [10] System for carrying out a method according to any of the preceding claims.

Images