RADAR SENSOR

Abstract

A radar sensor having a signal generating device which generates an outgoing signal as a radar signal that is to be emitted. The radar sensor also has a signal receiving device for receiving and processing received signals as reflected radar signals. The received signals can be processed with a prediction method in order to determine a predicted signal, which can be compared to the received signal in order to eliminate disruptions deviating therefrom.

Claims

1. A radar sensor comprising: a signal generating device, which generates an outgoing signal as a radar signal that is to be emitted; a signal receiving device for receiving and processing received signals as reflected radar signals, wherein the received signals are be processed with a prediction method in order to determine a predicted signal, with which the received signal is compared in order to reduced disruptions deviating therefrom.

2. A method for operating a radar sensor comprising the steps of:, providing the radar sensor with a signal generating device; generating an outgoing signal via the signal generating device as a radar signal that is to be emitted; providing the radar sensor with a signal receiving device; receiving and processing received signals via the signal receiving device as reflected radar signals, processing the received signals with a prediction method in order to determine a predicted signal, which is compared with the received signal in order to reduce disruptions deviating therefrom.

3. The method according to claim 2, wherein the received signal or a mixed received signal represents a predicted sinusoidal signal.

4. The method according to claim 2 wherein a linear predictive encoding is carried out.

5. The radar sensor according to claim 3 further comprising the step of detecting discontinuities in the received signal or in the mixed received signal caused by interferences through evaluation of a prediction error.

6. The method according to claim 5, wherein the prediction error is used as a test value.

7. The method according to claim 5 wherein the prediction error is calculated in a comparison of the actual signal amplitude x(k) with the predicted value x̂(k).

8. The method according to claim 1 wherein a non-causal backward prediction is used in addition to a causal forward prediction.

9. The method according to claim 8, wherein an amplitude value that is to be predicted is determined from a weighted averaging of two prediction results.

10. (canceled)

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Reference is now made more particularly to the drawings, which illustrate the best presently known mode of carrying out the invention and wherein similar reference characters indicate the same parts throughout the views.

[0019] FIG. 1 shows a diagram with an emission signal.

[0020] FIG. 2 shows a schematic depiction of a radar sensor.

[0021] FIG. 3 shows a diagram of a signal curve.

[0022] FIG. 4 shows a diagram of a signal curve.

[0023] FIG. 5 shows a diagram of a disrupted signal.

[0024] FIG. 6 shows a diagram of a signal curve after reduction of interference.

DETAILED DESCRIPTION OF THE DRAWINGS

[0025] FIG. 1 shows a diagram 1 for illustrating an outgoing signal comprised of a series of individual signals. The diagram shows the frequency f(t) of the outgoing signal as a function of time t. The individual signals f.sub.0.sup.A, f.sub.0.sup.B, f.sub.0.sup.C, f.sub.1.sup.A, f.sub.1.sup.B, f.sub.1.sup.C, etc. form a series emitted as an outgoing signal. If the radar sensor is operated with the so-called LFMSK transmission method (Linear Frequency Modulated Shift Keying), three nested individual signals A, B, C are emitted. A constant frequency is emitted thereby for a time period of ca. 25 μs (burst) in each case, which is then modified linearly for each of the three individual signals. It should be noted thereby that the frequencies of the signals f.sub.0.sup.A, f.sub.0.sup.B, f.sub.0.sup.C increase, wherein the frequencies in the next series of signals f.sub.1.sup.A, f.sub.1.sup.B, f.sub.1.sup.C increase again, wherein the frequency f.sub.1.sup.A is greater than the frequency f.sub.0.sup.A. An increasing frequency is referred to as an up-chirp, and a decreasing frequency is referred to as a down-chirp. In addition to the up-chirp and down-chirp, a mono-frequency individual signal, the so-called Doppler-chirp, is also used. The three signal types are used alternatingly thereby. In the example in FIG. 1, only up-chirps are used. Down-chirps or Doppler-chirps could also be used, however.

[0026] FIG. 2 shows a schematic depiction of a radar sensor 10, which has a transmitting antenna 11 and two receiving antennas 12, 13. The reflected received signals 14, 15 are mixed in the receivers 17 with the outgoing signal 16 of the transmitter 18, and the frequency range is transformed. This signal 21 is mixed in an HF element 19, and transferred to the digital signal processor 20 for further evaluation. The digital signal processor 20 transmits control signals 22 for controlling the radar sensor to the HF element 19. The radar sensor 10 thus has a signal generating device thereby, which includes the transmitter 16, the HF element 19 and the signal processor 20. The radar sensor 10 also has a signal receiving device thereby, which includes the receiver 17, the HF element 19 and the signal processor 20.

[0027] The received signals 14, 15 represents a superimposing of reflections from relevant targets, but also reflections from undesired targets, also referred to in radar technology as clutter, so-called disruption echoes caused by reflections on the ground, shoulder constructions and by rain. The OS-CFAR method (Ordered Statistic—Constant False Alarm Rate) is used to detect the relevant targets. Following detection, a basis signal is obtained for a relevant target, the frequency of which is obtained from the distance to the target and its relative speed. With a Doppler-chirp, the frequency is merely a function of the relative speed. The phase difference between the spectral components in the baseband between two ramps in each case is likewise obtained from the distance to and the relative speed of the relevant target. By solving a linear equation, the distance to and the relative speed of the target can be determined from the frequency and this phase difference for each measurement cycle. Furthermore, the run-time difference of the reflected signal, and thus the angle of incidence, can be determined by means of the phase difference between the two receiving antennas at the position of the frequency that is to be evaluated.

[0028] Raw target parameters can be determined from these signals, possibly using further data, which can be used for identifying objects by means of subordinate processing steps. The signal level and/or a reliability of the generated value can be used advantageously for further data.

[0029] FIGS. 3 and 4 show a signal curve 30 with a disruption. FIG. 3 shows an up-chirp in the frequency/time diagram, intersected by a CW disrupter. The CW disruption results in a pulse-like increase 31 in the amplitude of the time signal in the region of the intersection points of both signals, see FIG. 4. This increase 31, which can also extend over numerous measurement points, causes a broadband increase in the noise in the frequency range. Reflections from relevant targets are masked as a result, and can no longer be detected in the series.

[0030] The CW disruption is an example of a disruption thereby. Other disruptions could also occur. More complex disruptions could be caused, for example, by LFMCW (Linear Frequency Modulated Continuous Wave) and FCSM (Fast Chip Sequence Modulation). With these disruptions, there are increased intersections between the outgoing signal of the victim and the interference signal in the frequency curve.

[0031] As a result, disruptions in the form of an amplitude increase can be received in numerous time windows, which increase the noise level such that a detection of relevant targets without a correction of the received signal becomes difficult.

[0032] Because the receiving antennas are subject to different antenna diagrams due to manufacturing variances and asymmetrical couplings, the effects of the disruptions on the wanted signal in the two receiving antennas may also be different. In an exemplary case, only one of the two receiving antennas is disrupted. Furthermore, depending on the disruption, either just one, or all three A-, B-, and C-ramps may be affected. This may be attributed to the fact that the three ramps have different frequencies, and are temporally offset, see FIG. 1. Thus, by way of example, with a narrowband Doppler disrupter having the same Doppler frequency in the range of ramp A, the frequencies of only this ramp are strongly disrupted. The disruptions of the two other ramps are caused by the harmonics of the disrupter, and frequently cannot be detected, due to the low disruption intensity.

[0033] In order to enable an errorless detection of the reflections from relevant targets in the receiving spectrum, disruptions must be detected and eliminated.

[0034] Because it is sufficient to detect the disruption in just one of the six ramps with three ramps per receiving antenna, signals can also be corrected in which the disruption is not measured directly.

[0035] Because the time signal of each target object that is to be detected corresponds after the mixing stage to a sine function or a sinusoidal signal, as a superimposing of sine signals, the received signal, depicting the superimposing of all of the reflections of the target reflecting the outgoing signal, is composed only of harmonic oscillations. This signal shape of the harmonic oscillations can be readily predicted using a linear predictor.

[0036] The method of the linear predictive encoding LPC (Linear Predictive Coding) predicts the future curve of a signal based on the preceding signal values. A linear combination is formed from the preceding sampling values thereby.

[0037] Accordingly, for the predicted value at point k:

[00001]$\hat{x}\left(k\right)=\underset{i=1}{\overset{n}{.Math.}}.Math.a_t.Math.x\left(k-i\right)$

wherein a.sub.1 represent the prediction coefficients with i=1 to n.

[0038] These prediction coefficients are determined using a correlation matrix. For this, a correlation is formed from the preceding signal values, containing data regarding the signal characteristic. Based on this signal characteristic, the future curve of the signal can then be predicted on the basis of the last value.

[0039] Discontinuities in the received signal due to interferences can be detected by evaluating the prediction error, or the LPC prediction error. This prediction error is thus used as a test value.

[0040] The prediction error can be calculated according to the following equation, through comparison of the actual signal amplitude x(k) with the predicted value {circumflex over (x)}(k):

[00002]$e\left(k\right)=x\left(k\right)-\hat{x}\left(k\right)=x\left(k\right)-\underset{i=1}{\overset{n}{.Math.}}.Math.a_t.Math.x\left(k-i\right)$

In the case of a disruption, there is an increased prediction error due to the discontinuity, which can be determined.

[0041] With the detection through linear prediction, the entire signal curve of the examined chirp is predicted. The estimated value is then compared with the actual signal value. Using a suitable threshold value, a detection criterion can be derived in order to determine a discontinuity.

[0042] A pulse-shaped disruption in the signal is subsequently repaired using the preceding, undisrupted signal values. This repair is obtained through the forward projection of the predicted curve without the effect of the disruption.

[0043] In order to prevent amplitude jumps, a non-causal backward prediction can optionally also be used, in addition to the causal forward prediction. The amplitude value that is to be predicted is to be subsequently determined from the weighted averaging of both prediction results.

[0044] The result of the above interference reduction can be seen in FIGS. 5 and 6.

[0045] FIG. 5 shows a sinusoidal signal curve 40 having discontinuous, pulse-like rises 41 due to disruptions. After the prediction and the determination of the sinusoidal portions, the disruptions in the corrected signal 50 of FIG. 6 can no longer be detected. The reduction of the error also results in a reduction of the noise level. Masked relevant targets in the frequency range of the received signal can only be detected using the corrected signal.

LIST OF REFERENCE SYMBOLS

[0046] 1 diagram [0047] 10 radar sensor [0048] 11 transmitting antenna [0049] 12 receiving antenna [0050] 13 receiving antenna [0051] 14 signal [0052] 15 signal [0053] 16 outgoing signal [0054] 17 receiver [0055] 18 transmitter [0056] 19 HF element [0057] 20 signal processor [0058] 21 signal [0059] 22 control signal [0060] 30 signal curve [0061] 31 rise [0062] 40 signal curve [0063] 41 increase [0064] 50 corrected signal