DE102019209119A1 - Method for classifying a position signal - Google Patents

Abstract

A method for classifying position signals transmitted from at least one satellite (60, 62, 64, 66) and received by a receiver (50), wherein both a multi-frequency transmission and a post-correlation analysis are performed so that a correlation form is received at two frequencies is used to classify the received signals and in this way to detect multipath errors and NLOS errors.

Description

The invention relates to a method and an arrangement for classifying a position signal. The invention also relates to a computer program for performing the method and a machine-readable storage medium.

State of the art

A global navigation satellite system (GNSS: Global Navigation Satellite System) is a system for position determination and navigation that works on the basis of signals received from navigation satellites. This represents an important position determination technology that is increasingly being used, for example, for vehicle and pedestrian navigation. Applications can be found in particular in the field of autonomous driving and the Internet of Things (loT: Internet of Things).

Since GNSS technology is based on the reception of satellite signals, it is available almost worldwide. The satellites of systems such as GPS (Global Positioning System) and Galileo are each assigned a code C (pseudo-random noise, PRN), which enables the receiver to distinguish them from one another. The receiver accomplishes this by modulating codes onto a carrier signal and correlating this signal with the received satellite signal. A three-dimensional search, namely for PRN, code delay and carrier frequency shift, is thus carried out until the correlation is sufficiently high. The satellite also transmits a message that contains information about when the code was sent, which allows the receiver to estimate the time the signal will travel. By measuring the transit time of at least four satellites, the receiver can solve the navigation equations and calculate its three-dimensional position.

1. 1. Liegt das Sichtliniensignal vor oder nicht,
2. 2. die Anzahl an Reflexionen N_R, wobei jede Reflexion k bezüglich des Sichtlinenpfads definiert ist durch:
   1. (a) Verzögerung τ_k [Sekunden],
   2. (b) Leistungsverlust α_k,
   3. (c) Dopplerverschiebung f_k [Hz],
   4. (d) Trägerphasenoffset ϕ_k [im Bogenmaß].

Errors occur in the transmission of signals from the satellites until a position is calculated. Most of these errors have been sufficiently minimized through research. The main remaining errors are the so-called multipath (MP) and NLOS errors (NLOS: non-light-of-sight: no line of sight) 2 referenced, which are caused by a signal reflection on objects, such as. Skyscrapers. These MP and NLOS states or conditions can be defined by the following parameters:

1. 1. Is the line-of-sight signal present or not,
2. 2. the number of reflections N _R , where each reflection k is defined with respect to the line of sight path by:
   1. (a) delay τ _k [seconds],
   2. (b) power loss α _k ,
   3. (c) Doppler shift f _k [Hz],
   4. (d) Carrier phase offset ϕ _k [in radians].

• Es wurden umfassende Untersuchungen vorgenommen, um MP- und NLOS-Fehler zu verringern. Die Verfahren können durch ihre Schritte in der GNSS-Signalverarbeitungskette kategorisiert werden:
  1. 1. antennenbasierte Verfahren,
  2. 2. empfängerbasierte Verfahren,
  3. 3. Verfahren unter Zuhilfenahme von externen Sensoren oder Signalen.

It should also be noted that all of these parameters change dynamically, depending on the movement of the satellite, the receiver and the obstacles between them. This dynamic can be modeled:

• Extensive research has been done to reduce MP and NLOS errors. The procedures can be categorized by their steps in the GNSS signal processing chain:
  1. 1. antenna-based methods,
  2. 2. recipient-based procedures,
  3. 3. Procedure with the aid of external sensors or signals.

However, the proposed solutions have some disadvantages. For example, it is necessary to include complex hardware. Furthermore, these are not sufficiently robust. In addition, these solutions are not suitable for reducing the errors sufficiently.

Disclosure of the invention

Against this background, a method according to claim 1 and an arrangement for performing the method according to claim 7 are presented. A computer program according to claim 8 and a machine-readable storage medium according to claim 9 are also presented. Embodiments emerge from the dependent claims and from the description.

The presented method is used to classify signals, in particular signals for navigation or for position determination, which are typically transmitted by satellites and received by a receiver whose position is to be determined and evaluated for position determination. Before this evaluation, the signals are classified, i. H. these are assigned to predetermined classes, the classes allowing a statement to be made as to the extent to which the assigned signals are suitable, conditionally suitable or unsuitable for determining the position.

A restriction is made here to methods based on individual antennas and individual receivers. External sensors or additional wireless signals can be used later can be added to further improve the results.

1. 1. Multifrequenzübertragung
2. 2. Postkorrelationsanalyse

Two recipient-based methods provide valuable and relevant information in particular to record MP and NLOS errors:

1. 1. Multi-frequency transmission
2. 2. Post correlation analysis

So far, these two methods have been used independently of each other, a combination of the two methods has not yet been used. Since the multi-frequency and post-correlation methods only involve relatively slight modifications in hardware and software, these represent an interesting option for improving the performance of mass market GNSS receivers, especially in urban canyons.

• Das erste Verfahren, die Multifrequenzübertragung, verwendet verschiedene Trägerfrequenzen, um die Satellitensignale zu übertragen. GPS stellte bspw. zuvor lediglich Signale auf einer einzelnen Trägerfrequenz FC für zivile Nutzer bereit, nämlich das L1-Signal. Ab 2019 wurde GPS mit den L2C- und L5-Signalen erweitert und ein weiteres Beispiel ist das Galileosystem, das im Wesentlichen drei Frequenzen bereitstellen wird: E1, E5A und E5B. Weiterhin werden Zweifrequenz-Empfänger für den Massenmarkt verfügbar. Diese sind in der Lage, Signale von bis zu zwei verschiedenen Trägerfrequenzen zu empfangen, wie bspw. L1CA und L2C. Das macht diese für viele Anwendungen interessant, wie bspw. für die Fahrzeug- und Fußgängernavigation. Da Signale auf verschiedenen Trägerfrequenzen unterschiedlich reagieren, wenn diese an Gebäuden reflektiert werden, können wichtige Informationen über einen Multipfad-Empfang von diesen Empfängern extrahiert werden.

The procedures mentioned are discussed below:

• The first method, multi-frequency transmission, uses different carrier frequencies to transmit the satellite signals. For example, GPS previously only provided signals on a single carrier frequency FC for civil users, namely the L1 signal. From 2019, GPS has been expanded with the L2C and L5 signals and another example is the Galileo system, which will essentially provide three frequencies: E1, E5A and E5B. Furthermore, two-frequency receivers are becoming available for the mass market. These are able to receive signals of up to two different carrier frequencies, such as L1CA and L2C. This makes them interesting for many applications, such as vehicle and pedestrian navigation. Since signals on different carrier frequencies react differently when they are reflected off buildings, important information can be extracted from these receivers via multipath reception.

The second method, post correlation analysis, uses information from the correlator output. Few receivers output the results of the correlator, but this information is relevant for MP and NLOS acquisition. For a binary phase modulation (BPSK: Binary Phase Shift Keying), which is used for example with the GPS-L1CA signal, an SPLOS reception generates a triangular shape, the peak or peak location of which indicates the delay. However, when reflected signals are also received, the triangle shape is affected by additional peaks, making it difficult to correctly identify the delay. The idea behind the post-correlation analysis is to identify these influences, for example by evaluating the symmetry of the correlator output. In particular, the early, phase and late correlator outputs can be analyzed, such as the early-late phase parameters (ELP: Early Phase Late) or the Fast Fourier transfer-based early-minus-late detector (Fast Fourier Transform-Based Early Minus Late Detector). ELPs are three points of the correlation shape that are used to detect multipath. The fast Fourier transmission based early-minus-late detector corresponds to ELP, only transformed into the frequency range. This is also used to recognize multipath.

Deconvolution-based methods are also known, such as, for example, the projection-onto-convex series. Other filtering methods include the linear adaptive filters, stationary uncorrelated scattering in the broad sense, or virtual MP mitigation techniques. Signal quality monitoring is the method of monitoring parameters related to signals in order to detect errors such as multipath errors. Signal quality monitoring (SQM: Signal Quality Monitoring) is the method of monitoring parameters that are related to the signal in order to detect errors such as MP. SQM can also be performed with the correlator output as an input. More recently, machine learning methods have also been applied to the correlator output, for example by using a software-defined receiver (SDR) or in combination with the correlator output.

The method presented provides a possibility of recognizing a multipath error and further distinguishing multipath errors from non-sight line situations or errors. The multipath cases can be further processed to extract a useful pseudorange measurement, while the non-line of sight cases should be discarded and not used for the navigation solution. With the method presented, this is achieved without placing high demands on additional hardware.

The method is characterized in that the correlation form is received at two frequencies in order to classify the received signals into one of three classes in an embodiment: normal (SPLOS), multipath (MP), or non-sight line (NLOS). The advantage here is that the classification problem can be solved and only simple hardware and software modifications are required.

A correlator performs the function of cross-correlation, so that the time offset between two signals can be determined.

Further advantages and configurations of the invention emerge from the description and the accompanying drawings.

It goes without saying that the features mentioned above and those yet to be explained below can be used not only in the respectively specified combination, but also in other combinations or on their own, without departing from the scope of the present invention.

Figure list

• 1 shows an embodiment of the described method in a flow chart.
• 2 shows a possible scenario for performing the method in a schematic representation.
• 3 shows triangular shapes in four graphs that are evaluated for carrying out the method.

Embodiments of the invention

The steps to compute the Dual-Frequency Post-Correlation Feature (DFPCD) to solve these classification problems are explained with the following equations. The DFPCD characteristic is a parameter that can be received and measured. This changes depending on the multipath environments.

First an anechoic reference sample is selected from the SPLOS data and the corresponding correlator outputs from both carrier frequencies are stored: Sfi and S _f2 . This represents the expected correlator output under normal conditions and is used as a reference when comparing to new correlator outputs.

wobei $${\mu }_X=\frac{1}{n}{\displaystyle \sum _{i=1}^n{X}_i}$$

der arithmetische Mittelwert des Vektors X ist und $${\sigma }_x=\sqrt{{\displaystyle \sum _{i=1}^n{\left(X_i-{\mu }_x\right)}^2}}$$

seine Standardabweichung ist. Wenn X und Y eine positive lineare Korrelation haben, wird r gleich 1. Dies kann verwendet werden, um die Ähnlichkeit zweier Signale zu messen, wie dies hierin verwendet wird.Then the Pearson correlation coefficient r, a measure of the linear correlation between two vectors of the data X and Y, is defined: $$\text{r}\left(\text{X}\text{, Y}\right)=\frac{{\displaystyle {\sum }_{i=1}^n\left(X_i-u_X\right)\left(Y_i-u_y\right)}}{{\sigma }_X{\sigma }_Y}$$

in which $${\mu }_X=\frac{1}{n}{\displaystyle \sum _{i=1}^n{X}_i}$$

is the arithmetic mean of the vector X and $${\sigma }_x=\sqrt{{\displaystyle \sum _{i=1}^n{\left(X_i-{\mu }_x\right)}^2}}$$

its standard deviation is. When X and Y have a positive linear correlation, r equals 1. This can be used to measure the similarity of two signals as used herein.

Then an anechoic reference sample is selected from the SPLOS data and the corresponding correlator outputs from both carrier frequencies are stored: Sfi and S _f2 . This represents the expected correlator output under normal conditions and is used as a reference when comparing to new correlator outputs.

Then, depending on the interference between the direct signal and reflections, the received signal strength will increase or decrease the maximum correlation of the received correlator output R. This signal strength difference is represented by the scaling factor K; 3 referenced. $$\text{K}=\frac{\text{Max}\left(\mathrm{R.}\right)}{\text{Max}\left(\mathrm{S.}\right)}$$

To enable the similarity of the correlator outputs, R is scaled using the scaling factor K: $${\mathrm{R.}}_{skaliert}=\frac{1}{K}\mathrm{R.}$$

Then the similarity between the correlator outputs is given by: $$\text{r}\left(\text{S.}{\text{, R}}_{\text{scaled}}\right)$$

The parameters K and r are then combined to form an MP and an NLOS acquisition parameter D: $$\text{D.}\left(\text{S.}\text{, R}\right)=\left|\frac{K}{r}-1\right|$$

This is done because the scaling factor K is expected to increase when reflections are present, while the signal similarity r is expected to decrease. The subtraction of 1 is used for normalization to center the SPLOS samples around 0.

$${\text{DFPCD}}_{\text{f2}}=\text{D}\left({\text{S}}_{\text{f2}},{\text{ R}}_{\text{f2}}\right).$$

The DFPCD characteristic is then calculated for both frequencies: $${\text{DFPCD}}_{\text{f}1}=\text{D.}\left({\text{S.}}_{\text{f}1},{\text{R.}}_{\text{f}1}\right),$$

$${\text{DFPCD}}_{\text{f2}}=\text{D.}\left({\text{S.}}_{\text{f2}},{\text{R.}}_{\text{f2}}\right).$$

Finally, in order to be able to differentiate between MP and NLOS cases, the absolute difference between the DFPCD features is calculated on two frequencies: $${\text{DFPCD}}_{\Delta \text{f}}=\left|{\text{DFPCD}}_{f1}-{\text{DFPCD}}_{f2}\right|$$

The anechoic reference correlation shape can either be sampled on real satellite data or various other automatically generated methods can be used to compute a feature similar to DFPCD, e.g., the area under the correlation shapes.

1. 1. Wahl geeigneter Satelliten aufgrund dynamischer Parameter:
   1. a) Es kann hierzu z. B. die Fluktuation bzw. Varianz des gemessenen Signal-/Rausch-Verhältnisses genutzt werden: Ist die Fluktuation stark, kann angenommen werden, dass Reflexionen bzw. Abschattungen auftreten.
   2. b) Die Messung des Pseudorange oder die Doppler-Verschiebung variiert stark.
2. 2. Wahl geeigneter Satelliten aufgrund statischer Parameter:
   • Z. B. das Ausschließen von schwach empfangbaren Satelliten, d. h. niedriges CNo.
3. 3. Wahl geeigneter Satelliten durch Vergleich mit Inertial-Messungen und Odometer:
   • Hier kann durch Vergleich mit vertrauenswürdigen Sensoren, z. B. Gyroskop, Beschleunigungsmesser, Odometer, ein Maß bestimmt werden, inwieweit Signale eines bestimmten Satelliten plausibel sind.

The choice of suitable satellites can be done in different ways:

1. 1. Choice of suitable satellites based on dynamic parameters:
   1. a) It can for this purpose z. For example, the fluctuation or variance of the measured signal / noise ratio can be used: If the fluctuation is high, it can be assumed that reflections or shadowing occur.
   2. b) The measurement of the pseudorange or the Doppler shift varies greatly.
2. 2. Choice of suitable satellites based on static parameters:
   • For example, excluding satellites with weak receivables, ie low CNo.
3. 3. Selection of suitable satellites by comparison with inertial measurements and odometers:
   • By comparison with trustworthy sensors, e.g. B. gyroscope, accelerometer, odometer, a measure can be determined to what extent signals from a particular satellite are plausible.

The methods mentioned are relatively easy to implement, but the methods mentioned under 1) and 3) are not really real-time methods, since they are of a time-statistical nature.

1 shows a possible sequence of the method in a flowchart, whereby two alternative procedures are taken into account.

In a first step 10 sends a satellite signals in a next step 12 reflected from buildings. It is done in one step 14th the received correlation shape is stored in R and then in one step 16 R scaled to S. This is done as alternatives in one step 20th an anechoic satellite or a mathematical model (step 22nd ) is used. One of the two alternatives, i.e. step 20th or 22nd , is used to do in one step 24 make a reference correlation. This reference correlation is done in one step 26th stored in S, which in turn is when scaling in step 16 is used.

In one step 30th a scaling factor K is then determined and next to it in one step 32 a scaled correlation form R _{scaled is} created. This becomes one step 34 along with the result of the step 26th created a Pearson correlation of S and R _scaled . It will then be in one step 40 the correlation factor and the scaling factor combined. Finally it takes place in one step 42 the recording of MP and NLOS cases.

2 shows a possible scenario for carrying out the method with a recipient 50 , a first building 52 , a second building 54 , a first satellite 60 that is blocked, a second satellite 62 , in which there is an MP case, a third satellite 64 , in which there is no impairment, thus a SPLOS case, and a fourth satellite 66 who has an NLOS case.

It should be noted that for a correct function of the receiver 50 an SPLOS link, as is the case with the third satellite 64 the case is needed. MP and NLOS errors are typically caused by reflections on buildings, such as the buildings 52 and 54 , causes how these are in big cities. It should also be noted that a satellite can be completely blocked, as in this example the first satellite 60 which reduces the number of satellites available or usable.

3 shows triangular shapes in four graphs, how these are evaluated when carrying out the presented method. In a first graph 100 , on its abscissa 102 the time [µs] and on its ordinate 104 the correlation power is plotted, there is a first curve for f ₁ = L1CA for K = 0.30811 106 , whose peak 108 S _{shows L1CA} for a reference SPLOS, and a second curve 110 , whose peak 112 R _{shows L1CA} for MP. In a second graph 120 , on its abscissa 122 the time [µs] and on its ordinate 124 the correlation power is plotted, are a first curve for f ₂ = L2C for K = 1.4517 126 , whose peak 128 R _{shows L2C} for MP, and a second curve 130 , whose peak 122 S _L2C for reference SPLOS shows shown. The Correlation form measures the delay between the satellite signal and the receiver, 0 µs is the maximum correlation.

In a third graph 140 , on its abscissa 142 the time [µs] and on its ordinate 144 the correlation power is plotted, there is a first curve for f ₁ = L1CA for r = 0.99714 146 , whose peak 148 S _{shows L1CA} for a reference SPLOS, and a second curve 150 , whose peak 152 1 / K * R _L1CA for MP shows. In a fourth graph 160 , on its abscissa 162 the time [µs] and on its ordinate 164 the correlation power is plotted, there is a first curve for f ₂ = L2C for r = 0.9964 166 , whose peak 168 1 / K * R _L2C for MP shows and a second curve 170 , whose peak 172 S _L2C for reference SPLOS shows shown.

The illustration shows that the two-frequency postcorrelator difference between a reference SPLOS correlator output and an MP-influenced correlator output takes into account their difference in signal strength and asymmetry.

3 thus shows, by way of example, several forms of correlation, as may be seen at the output of the correlator. The power is plotted on the ordinate and the time shift of the received code against the internal code is plotted on the abscissa. In GPS, the shape is basically triangular and possibly distorted.

• - Messung der von den Satelliten empfangenen Leistung
  1. a) beispielhafte Ausführungsform:
     • Hierbei wird die empfangene Leistung dadurch bestimmt, dass lediglich das Maximum der resultierenden Korrelationsfunktion genommen wird, bspw. der Spitzenwert des Dreiecks, im Falle von QPSK (Quadrature Phase Shift Keying). Dies ist ein Modulationsverfahren, das u. a. mehrere GPS-Signale benutzt. Wichtig ist, dass die Korrelationsform von Binary Phase Shift Keying (BPSK) und QPSK ein Dreieck ist. $${\text{P}}_{\text{R}}=\text{max}\left(\text{R}\right)$$
       
  2. b) die empfangene Leistung kann ebenfalls bestimmt werden, indem anstelle dessen der CNo-Wert entnommen wird. Der CNo-Wert (carrier-to-noise-density ratio) gibt an, wie stark das GPS-Signal eines Satelliten ist. $${\text{P}}_{\text{R}}=\text{k*T*F*CNo}$$
     
  3. c) die empfangene Leistung kann ebenfalls bestimmt werden, indem der Bereich unter der resultierenden Korrelationsform P(τ) berechnet wird. T bezeichnet die Grenzen von R(τ), die zu berücksichtigen sind. Das Ergebnis kann von dem verwendeten Modulationsschema verschieden abhängig sein. $${\text{P}}_{\text{R}}=\frac{k}{2T}{\displaystyle \underset{-T}{\overset{+T}{\int }}R\left(\tau \right)d\tau }$$
     
  4. d) anstelle der Verwendung der unter c) gezeigten Lösung, die lediglich ein arithmetisches Mittel für ein diskretes Modell repräsentiert: $$P_R=\frac{k}{N}{\displaystyle \sum _{i=1}^N{R}_i}$$
     
     kann diese ebenfalls auch als quadratischer Mittelwert berechnet werden: $$P_R=\frac{k}{N}\sqrt{{\displaystyle \sum _{i=1}^N{R}_i^2}}$$
     
     oder als geometrisches Mittel: $$P_R=k^N\sqrt{{\displaystyle \prod _{i=1}^N{R}_i}}$$
     
• - Verwenden eines Bilds zum Vergleichen der empfangenen Signale
  1. a) beispielhafte Ausführungsform:
     • Es ist nicht klar gezeigt, wo das Bild der Korrelationsfunktion S(τ) herkommt.
  2. b) S(τ) kann eine diskrete Funktion S_i = f(τ_i) sein, die in dem Empfänger für jedes Modulationsthema gespeichert ist und die immer konstant ist.
  3. c) S(τ)kann eine diskrete Funktion S_i = f(τ_i) sein, die in dem Empfänger für jedes Modulationsthema gespeichert ist und die von dem empfangenen Signal extrahiert wird, falls bekannt ist, dass keine Reflexionen vorliegen oder ein reines empfangenes Signal angenommen wird.
• - Berechnen des Skalierungsfaktors
  • a) beispielhafte Ausführungsform:
    • Hierbei ist der Skalierungsfaktor das Verhältnis der Maxima der Werte für das empfangene Korrelationssignal bzw. des Bildkorrelationssignals. $$\text{K=}\frac{\text{max}\left(R\right)}{\text{max}\left(S\right)}$$
      
  • c) dies kann ebenfalls berechnet werden, indem die Lösungen, die unter der Überschrift „Messung der von den Satelliten empfangenen Leistung“ aufgeführt sind, anstelle der verwendeten maximalen Werte verwendet werden.
• - Korrelation
  1. a) beispielhafte Ausführungsform:
     • Hierbei wird die Korrelation durchgeführt, indem die Pearson-Gleichung verwendet wird, dies beseitigt die Offsets beider Signale und normalisiert Signalamplituden-Differenzen, indem Standardabweichungen verwendet werden.
     $$r\left(R_{skaliert},S\right)=\frac{{\displaystyle {\sum }_{i=1}^N{R}_{skaliert,i}}-{\mu }_R)\left(S-{\mu }_S\right)}{{\sigma }_{R_{skaliert}}-{\sigma }_R}$$
     
  2. b) da die Signalamplitude bereits durch die vorstehende Lösung ausgeglichen wurde, ist eine Division durch beide Standardabweichungen nicht erforderlich. Dies bedeutet einfach: $$r\left(R_{skaliert},S\right)=\frac{1}{N}{\displaystyle \sum _{i=1}^N{R}_{skaliert,i}}-{\mu }_R)\left(S_i-{\mu }_S\right)$$
     
  3. c) ein weiterer Weg, die beiden Korrelationsformen zu vergleichen, besteht darin, einfach die Quadrate der Differenzen zu addieren. $$d\left(R_{skaliert},S\right)\sqrt{\frac{1}{N}{\displaystyle \sum _{i=1}^N{\left(R_{skaliert,i}-S_i\right)}^2}}$$
     
  4. d) der einfachste Weg kann darin bestehen, lediglich das arithmetische Mittel der Beträge der Differenzen zu bilden. $$\left(R_{skaliert},S\right)=\frac{1}{N}{\displaystyle \sum _{i=1}^N\left|R_{skaliert,i}-S_i\right|}$$
     
• - Analysieren der Signalqualität der Satelliten Der beschriebene Algorithmus liefert zwei Zahlen, um die Qualität der Satelliten zu beurteilen, was insbesondere bedeuten kann, ob diese Reflexionen haben oder nicht. Die Zahlen, die unter „Messung der von den Satelliten empfangenen Leistung“ beschrieben sind, zeigen einen Wert für die empfangene Signalleistung. Durch Vergleichen der Werte bei zwei oder mehr Frequenzen kann man sagen, dass je größer der Unterschied ist, desto größer die Wahrscheinlichkeit ist, dass ein Multipfad-Fall vorliegt. Andererseits können wir unter Berücksichtigung der Korrelation unter dem Kapitel „Korrelation“, bei der die Formdifferenzen zu einer Referenz berechnet werden, sagen, dass je mehr diese von der Differenz abweichen, desto größer die Wahrscheinlichkeit des Vorliegens eines Multipfad-Falls ist.
  1. a) beispielhafte Ausführungsform:
     • Hierbei wird ein formaler Entscheidungsparameter D durch Kombinieren von K und r definiert. D kann auf die Ergebnisse von zwei oder mehr Frequenzen angewendet werden, um die DFPCD-Werte zu bestimmen.
  • b) eine Entscheidung kann ebenfalls getroffen werden, indem mathematische oder heuristische Mittel angewendet werden, um die Parameter K, r und andere verfügbare Parameter von dem GNSS-Empfang zu kombinieren, wie bspw.
    • - die Frequenz der Signalfluktuationen auf allen betrachteten Frequenzen,
    • - die Elevation von Satelliten, die in Frage stehen,
    • - die Nachverfolgungszeit eines bestimmten Satelliten,
    • - die Doppler-Überwachungen für bestimmte Satelliten und Anpassposition-Aktualisierung-Überprüfungen,
    • - die gemessene Trägerphasen und Plausibilisierungsüberprüfungen.
  • c) Zusätzlich können Ergebnisse aller vorstehend genannten Mittel mit Trägheitssensoren, wie bspw. Multifrequenz-GNSS und Koppelnavigation, kombiniert werden. Dies kann ebenso genutzt werden, um Plausibilisierungsüberprüfungen vorzunehmen.

Possible extensions with which alternative solutions are presented in the context of the invention are discussed below.

• - Measurement of the power received by the satellites
  1. a) exemplary embodiment:
     • In this case, the received power is determined in that only the maximum of the resulting correlation function is taken, for example the peak value of the triangle, in the case of QPSK (Quadrature Phase Shift Keying). This is a modulation method that uses several GPS signals, among other things. It is important that the correlation form of Binary Phase Shift Keying (BPSK) and QPSK is a triangle. $${\text{P}}_{\text{R.}}=\text{Max}\left(\text{R.}\right)$$
       
  2. b) the power received can also be determined by taking the CNo value instead. The CNo value (carrier-to-noise-density ratio) indicates how strong the GPS signal from a satellite is. $${\text{P}}_{\text{R.}}=\text{k * T * F * CNo}$$
     
  3. c) the received power can also be determined by calculating the area under the resulting correlation shape P (τ). T denotes the limits of R (τ) that must be taken into account. The result can vary depending on the modulation scheme used. $${\text{P}}_{\text{R.}}=\frac{k}{2T}{\displaystyle \underset{-T}{\overset{+T}{\int }}\mathrm{R.}\left(\tau \right)d\tau }$$
     
  4. d) instead of using the solution shown under c), which only represents an arithmetic mean for a discrete model: $$P_{\mathrm{R.}}=\frac{k}{N}{\displaystyle \sum _{i=1}^N{\mathrm{R.}}_i}$$
     
     this can also be calculated as a quadratic mean: $$P_{\mathrm{R.}}=\frac{k}{N}\sqrt{{\displaystyle \sum _{i=1}^N{\mathrm{R.}}_i^2}}$$
     
     or as a geometric mean: $$P_{\mathrm{R.}}=k^N\sqrt{{\displaystyle \prod _{i=1}^N{\mathrm{R.}}_i}}$$
     
• - Using an image to compare the received signals
  1. a) exemplary embodiment:
     • It is not clearly shown where the picture of the correlation function S (τ) comes from.
  2. b) S (τ) can be a discrete function S _i = f (τ _i ) which is stored in the receiver for each modulation theme and which is always constant.
  3. c) S (τ) can be a discrete function S _i = f (τ _i ) which is stored in the receiver for each modulation theme and which is extracted from the received signal if it is known that there are no reflections or a pure received one Signal is accepted.
• - Calculating the scaling factor
  • a) exemplary embodiment:
    • The scaling factor is the ratio of the maxima of the values for the received correlation signal or the image correlation signal. $$\text{K =}\frac{\text{Max}\left(\mathrm{R.}\right)}{\text{Max}\left(\mathrm{S.}\right)}$$
      
  • c) this can also be calculated by using the solutions given under the heading “Measuring the power received by the satellites” instead of the maximum values used.
• - correlation
  1. a) exemplary embodiment:
     • Here the correlation is performed using Pearson's equation, this removes the offsets of both signals and normalizes signal amplitude differences using standard deviations.
     $$r\left({\mathrm{R.}}_{skaliert},\mathrm{S.}\right)=\frac{{\displaystyle {\sum }_{i=1}^N{\mathrm{R.}}_{skaliert,i}}-{\mu }_{\mathrm{R.}})\left(\mathrm{S.}-{\mu }_{\mathrm{S.}}\right)}{{\sigma }_{{\mathrm{R.}}_{skaliert}}-{\sigma }_{\mathrm{R.}}}$$
     
  2. b) since the signal amplitude has already been compensated for by the above solution, division by both standard deviations is not necessary. This simply means: $$r\left({\mathrm{R.}}_{skaliert},\mathrm{S.}\right)=\frac{1}{N}{\displaystyle \sum _{i=1}^N{\mathrm{R.}}_{skaliert,i}}-{\mu }_{\mathrm{R.}})\left({\mathrm{S.}}_i-{\mu }_{\mathrm{S.}}\right)$$
     
  3. c) Another way to compare the two forms of correlation is to simply add the squares of the differences. $$d\left({\mathrm{R.}}_{skaliert},\mathrm{S.}\right)\sqrt{\frac{1}{N}{\displaystyle \sum _{i=1}^N{\left({\mathrm{R.}}_{skaliert,i}-{\mathrm{S.}}_i\right)}^2}}$$
     
  4. d) the simplest way can be to only form the arithmetic mean of the amounts of the differences. $$\left({\mathrm{R.}}_{skaliert},\mathrm{S.}\right)=\frac{1}{N}{\displaystyle \sum _{i=1}^N\left|{\mathrm{R.}}_{skaliert,i}-{\mathrm{S.}}_i\right|}$$
     
• Analyzing the signal quality of the satellites The algorithm described provides two numbers to assess the quality of the satellites, which can mean in particular whether they have reflections or not. The numbers described under “Measuring the Power Received by the Satellites” show a value for the received signal power. By comparing the values at two or more frequencies, it can be said that the greater the difference, the greater the likelihood that there is a multipath case. On the other hand, taking into account the correlation under the chapter "Correlation", in which the shape differences are calculated to a reference, we can say that the more these deviate from the difference, the greater the probability of a multipath case being present.
  1. a) exemplary embodiment:
     • Here, a formal decision parameter D is defined by combining K and r. D can be applied to the results of two or more frequencies to determine DFPCD values.
  • b) a decision can also be made by using mathematical or heuristic means to combine the parameters K, r and other available parameters from the GNSS reception, such as e.g.
    • - the frequency of the signal fluctuations on all considered frequencies,
    • - the elevation of satellites in question,
    • - the tracking time of a particular satellite,
    • - the Doppler surveys for specific satellites and matching position update checks,
    • - the measured carrier phases and plausibility checks.
  • c) In addition, results of all of the above-mentioned means can be combined with inertial sensors such as multi-frequency GNSS and dead reckoning. This can also be used to carry out plausibility checks.

The elevation of a satellite indicates how high it is in the sky, i.e. H. 0 degrees is on the horizon, 90 degrees is at the zenith. The speed of the receiver can be determined with Doppler measurements. You can then check whether the position of the receiver corresponds to the speed derived from the Doppler measurements.

Claims (9)

A method for classifying position signals transmitted by at least one satellite (60, 62, 64, 66) and received by a receiver (50), wherein both multi-frequency transmission and post-correlation analysis are performed so that a correlation form is received at two frequencies is used to classify the received signals and in this way to detect multipath errors and NLOS errors. Procedure according to Claim 1 , in which the position signals are classified into three classes, namely normal, multipath and NLOS. Procedure according to Claim 1 or 2 , in which external sensors are also used. Method according to one of the Claims 1 to 3 , which also uses wireless signals. Method according to one of the Claims 1 to 4th , in which position signals in which a multipath error has been detected are further processed to extract a pseudorange measurement. Method according to one of the Claims 1 to 5 , in which position signals for which an NLOS error was detected are discarded. Arrangement for classifying position signals which are used to carry out a method according to one of the Claims 1 to 6th is set up. Computer program with program code means which is set up to implement a method according to one of the Claims 1 to 6th execute when the computer program on a computing unit, in particular a computing unit in an arrangement according to Claim 7 , is performed. Machine-readable storage medium with a computer program stored thereon Claim 8 .

Images