DE102017007961A1 - Method and device for detecting a fault condition of a radar sensor - Google Patents

Abstract

The invention relates to a method and a device for detecting a fault condition of a radar sensor due to at least one interference signal in a received signal of the radar sensor (1), in which the received signal, preferably after analog processing, is sampled and subjected to digital signal processing, characterized in that from the sampled received signal and / or during the digital signal processing at least one data set or a data stream is obtained, which is supplied to a learning classifier (23) with which a presence and / or a type of the fault condition is estimated.

Description

The invention relates to a method for detecting a fault condition of a radar sensor due to at least one interference signal in a received signal of the radar sensor, in which the received signal, preferably after analog preprocessing, scanned and subjected to digital signal processing, the invention further relates to a device for detecting at least an interference condition of a radar sensor due to at least one interference signal in a reception signal of the radar sensor, with a reception signal processing unit, with an analog-to-digital converter for sampling the reception signal and with a digital signal processing unit for digital signal processing of the sampled reception signal.

Radar sensors are increasingly used in motor vehicles, especially in driver assistance systems or in the context of autonomous driving functions. But with the increasing popularity of radar sensors in the automotive industry, the likelihood of the radar sensors interfering with each other increases. Thus, a radar sensor can receive signals from other automotive radar sensors in addition to its own reflected signal on objects in the environment, d. H. it comes to mutual interference. However, interfering signals may originate not only from vehicle radars operating in the same frequency range, but also from non-automotive applications using the same technology. In addition to these unintentionally occurring mutual interference, it is also possible to introduce interference signals that are specifically generated to disrupt automotive radar systems.

In order to achieve the high availability of the sensors required for highly automated and autonomous driving, the sensors must be as resistant as possible to the described disturbing influences, i. H. suitable sensor-internal or external measures must be found with which the influence of the disturbances can be reduced or suppressed. However, the basis for this is, first of all, that a disturbing influence per se and its type is detected by the radar sensor.

In current vehicle radar sensors used Störerkenner evaluate a combination of certain influenced by electromagnetic interference sensor parameters. The basis for this is a modeling based on simple and clear sturgeon scenarios. However, the cost of such modeling increases significantly with the number of parameters considered.

Out DE 10 2015 119 482 A1 a radar sensor is known in which an interference detection is performed in transmission pauses to detect with the own radar signal interfering signals from other radar sensors. For this purpose, a local oscillator of the radar sensor is driven to generate several Suchsweeps in which the frequency of the local oscillator increases linearly. The local oscillator signal is supplied to a mixer with which the signal applied to the receiving antenna, which may contain the interfering signal, is down-mixed. Subsequently, the down-mixed signal is filtered by means of a bandpass filter, converted by means of an analog-to-digital converter into a digital signal and processed in a digital signal processor (DSP). By evaluating the digital signal for a large number of search sweeps, it is possible to detect the presence and type of the disturbance and thus classify it, for example, as a signal of a slow chirp. It is also possible in this way to draw conclusions about the bandwidth occupied by the interference sensor and its center frequency.

The object of the invention is to provide a method and a device for detecting at least one interference signal in a high-frequency received signal of a radar sensor, in particular of a vehicle, with which an improved detection of a fault condition is made possible.

The object is achieved by a method and a device having the features of the independent patent claims 1 and 8. Advantageous developments and refinements are the subject of the dependent claims.

In order to develop a method according to the preamble of independent claim 1 such that an improved detection of an interference condition is made possible, it is provided according to the invention that at least one data set or a data stream is obtained from the sampled received signal and / or during digital signal processing learning classifier is estimated with the presence and / or a type of fault condition is estimated. The learning classifier is preferably trained by machine learning, particularly supervised learning techniques. For example, the classifier may be in the form of an artificial neural network. The sampling of the received signal may be preceded by a processing of the signal received directly with at least one receiving antenna of the radar system, in particular a low-noise preamplification with a so-called LNA (low noise amplifier), filtering in one or more filter stages, frequency conversion to an intermediate frequency band or one Baseband, a gain and / or a filter in the intermediate frequency band or in the baseband.

The sampled received signal is preferably processed in a multistage digital signal processing unit, for example in a digital signal processor (DSP). The at least one data set or the data stream is obtained before, in and / or after one of the stages of the digital signal processing unit. In this way, the at least one data record or the data stream can be obtained from the high-frequency received signal.

Particularly preferably, the received signal contains at least one useful signal and / or the at least one interference signal, wherein the at least one interference signal has a frequency in a frequency range in which the at least one useful signal is to be expected. The at least one useful signal is obtained, for example, by reflection of a radio-frequency transmitted by the radar sensor via at least one transmitting antenna high-frequency transmission signal to one or more stationary or moving object. The at least one interference signal can originate, for example, from another radar system operating in the same frequency range, in particular another vehicle radar sensor, and / or a jamming transmitter.

In particular, the radar sensor and the radar system generating the interfering signals use different or identical radar technologies, in particular in each case either FMCW radar including chirp sequence or pulse radar or OFDM or pseudo noise (PN) technology.

For a training of the learning classifier, the radar sensor is particularly preferably exposed to a multiplicity of known electromagnetic scenarios and / or known interference states in which the high-frequency received signal received via at least one receiving antenna contains one or more known useful signals and / or one or more known interference signals. The respective data sets or data streams obtained from the high-frequency received signals are respectively assigned to the known electromagnetic scenarios and / or known interference states with the associated known useful signals and / or known interference signals, wherein an estimate of the presence and / or type of the interference state is based on the assignment he follows. The classifier can thus be trained in particular the ability to create associations and to the probable composition of the received signal, i. H. to conclude useful and / or disturbing signals contained therein or to make an estimate regarding the presence and / or type of Störzuständs. In particular, when training the learning classifier to the useful or interfering signals each have a time function (amplitude or power over time), a spectrum, a modulation, frequency and / or phase shift, location, speed, and / or technology used the useful or disturbing signals generating radar sources be known.

In particular, the known useful signals are emitted by a simulator and / or arise by reflection of a transmitted by the radar sensor via at least one transmitting antenna high-frequency transmission signal at known positions and / or moving in a known manner relative to the radar sensor objects. Alternatively or in combination, the known interference signals are generated by at least one radar system located at a known position and / or moving in a known manner relative to the radar sensor.

The radar sensor and the radar system generating the known interference signals preferably use different or identical radar technologies, in particular FMCW radar or pulse radar technology in each case.

In order to further develop a device according to the preamble of independent claim 8 such that an improved detection of a fault condition is made possible, according to the invention a processor and memory unit is provided on which a learning classifier is implemented for estimating a presence and / or a type of fault condition on the basis of at least one data record obtained directly or indirectly from the sampled received signal or a data stream obtained directly or indirectly from the sampled received signal.

The digital signal processing unit preferably has several stages. Furthermore, a derivation unit indirectly or directly connected to the process and storage unit is preferably provided for discharging at least one data set or data stream before, from, or after one of the stages.

The particular advantage of the method according to the invention and the device according to the invention is that the outlay for modeling disturbance scenarios, which increases significantly with the number of parameters considered, is avoided. The use of a machine learning based classifier requires only the provision of different electromagnetic training scenarios, which can be complex and blurred. By training the learning classifier with a variety of different training scenarios, the accuracy or Increase the accuracy of the classifier. In this way, a more precise and detailed detection of a fault condition per se and the nature of the fault is possible.

Further advantages, features and details emerge from the following description in which an exemplary embodiment is described with reference to the drawing. The same, similar and / or functionally identical parts are provided with the same reference numerals.

Show it:

1 a simplified block diagram of an embodiment of a device according to the invention

1 shows a radar sensor 1 in a greatly simplified representation, in particular only one reception branch and only the HF part of a transmission branch is indicated. One with the receiving antenna 3 received high-frequency radar signal, for example in a frequency range at 24 GHz, 77 GHz, 79 GHz, 122 GHz, 160 GHz, or 220 GHz, is initially prepared analog (received signal processing unit 7 ). This could include a preprocessing, not shown here, for example, a pre-amplification with a low-noise preamplifier (LNA) and / or a pre-filtering. Further, a frequency conversion to an intermediate frequency band or to a baseband by mixing in a mixer 9 with an LO signal of a local oscillator 11 be included. The converted signal can be filtered in the filter 13 and a gain with the amplifier 15 be subjected. The (preprocessed) receive signal which is still present in analog form is subsequently connected by means of an analog-to-digital converter 17 sampled, ie the received signal is now in digital form - as a sampled received signal - before, the following in a digital signal processing unit 19 (Digital Signal Processor - DSP) is processed.

• – Beobachtung der Kontur des Rauschteppichs über alle Entfernungs-, Doppler-, Winkelbereiche (nach den Entfernungs-, Doppler-, Winkel-FFT's)
• – Beobachtung des Verlaufs von Zielechos über alle Entfernungs-, Doppler-, Winkelbereiche (z. B. Verschmierung, Verbreiterung, Unterbrechung, Zerfall in Einzelziele, Auftreten eines zusätzlichen Ziels („Geisterziel”), Wegfall eines beobachteten Ziels, Verringerung der Beobachtungsreichweite eines Ziels).

The digital signal processing unit 19 can be several levels 19 ' . 19 '' . 19 ''' in which the digital signal or the digital signal sequence which is obtained after the analog-to-digital conversion is processed. The exact structure of the digital signal processing unit 19 , ie in particular type and number of stages 19 ' . 19 '' . 19 ''' , depends on the type of radar sensor. By means of a diversion unit 20 the at least one data record or the data stream, which is obtained directly or indirectly from the sampled received signal, can be passed out and the processor and memory unit 21 be fed on which the learning classifier 23 is implemented. By way of example, the sampled received signal, which represents at least one data record or the data stream, can already be used by the learning classifier via the derivation unit 20 is supplied. The at least one data set or the data stream can in particular, without a restriction of the general inventive concept, be forwarded before, in or during the following processing stages: stages in which one or more Fourier transforms (FFT), one or more multiple decimations, one or more filters, one or more threshold formations, tracking, one or more cluster formations, one or more object formations are performed. It would be conceivable, using the example of a car radio with digital beamforming in the case of reception and chirp sequence modulation method, inter alia, the following observations according to the above-mentioned processing stages:

• - observation of the contour of the noise carpet over all distance, Doppler, angle ranges (after the distance, Doppler, angle FFT's)
• - Observing the course of target echoes across all range, Doppler, angular ranges (eg smearing, broadening, interruption, decay into individual targets, occurrence of an additional target ("ghost target"), elimination of an observed target, reduction of the observation range of a target ).

Subsequently, appropriate countermeasures can be initiated according to the detected fault condition or one or more interference signals. Purely by way of example, various measures are listed, which could be used individually or in combination.

Thus, in the space / angle range, for example, adaptive interference suppression can be performed by electronic antenna pattern shaping if digital beamforming is possible in the case of reception.

In the time domain, for example, a reconstruction of the disturbed modulation time signal (eg ramps) can take place.

In the frequency domain in particular a filtering out of the known interference spectrum is possible. As further countermeasures, a coding of the radar transmission signal, a mutual understanding of successive radars (assuming a corresponding communication possibility), application of a "listen before talk" technology and / or a "detect and avoid" technology can be provided.

While the invention has been further illustrated and explained in detail by way of preferred embodiments, the invention is not limited by the disclosed examples and other variations can be made by those skilled in the art can be derived without departing from the scope of the invention. It is therefore clear that a multitude of possible variations exists. It is also to be understood that exemplified embodiments are really only examples that are not to be construed in any way as limiting the scope, applicability, or configuration of the invention. Rather, the foregoing description and description enable the skilled artisan to practice the exemplary embodiments, and those of skill in the knowledge of the disclosed inventive concept may make various changes, for example, to the function or arrangement of particular elements recited in an exemplary embodiment. without departing from the scope defined by the claims and their legal equivalents, such as further explanation in the description.

LIST OF REFERENCE NUMBERS

1

radar sensor

3

receiving antenna

5

transmitting antenna

7

Reception signal processing unit

9

mixer

11

Local oscillator

13

filter

15

amplifier

17

Analog to digital converter

19

digital signal processing unit

19 '

Stage of the digital signal processing unit

19 ''

Stage of the digital signal processing unit

19 '' '

Stage of the digital signal processing unit

20

Ausleiteinheit

21

Processor and memory unit

23

learning classifier

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102015119482 A1 [0005]

Claims (9)

Method for detecting a fault condition of a radar sensor due to at least one interference signal in a received signal of the radar sensor ( 1 ), in which the received signal, preferably after analog processing, is sampled and subjected to digital signal processing, characterized in that at least one data record or a data stream is obtained from the sampled received signal and / or during the digital signal processing , which corresponds to a learning classifier ( 23 ) is supplied, with which a presence and / or a kind of the fault condition is estimated. A method according to claim 1, characterized in that the sampled received signal in a multi-stage digital signal processing unit ( 19 ) and that the at least one data set or the data stream before, in and / or after one of the stages ( 19 ' . 19 '' . 19 ''' ) of the digital signal processing unit ( 19 ) is won. A method according to claim 1 or 2, characterized in that the received signal contains at least one useful signal and / or the at least one interference signal, wherein the at least one interference signal has a frequency in a frequency range in which the at least one useful signal is to be expected. Method according to claim 3, characterized in that the radar sensor ( 1 ) and the radar system generating the interfering signals use different or the same radar technologies, in particular in each case either FMCW radar or pulse radar technology. Method according to one of Claims 1 to 4, characterized in that training of the learning classifier ( 23 ) the radar sensor ( 1 ) is exposed to a multiplicity of known electromagnetic scenarios and / or known interference states in which the signal is transmitted via at least one receiving antenna ( 3 Received signal received in each case one or more known useful signals and / or one or more known interfering signals, that each obtained from the sampled received signals records or data streams each of the known electromagnetic scenarios and / or known interference states with the associated known useful signals and / or be assigned to known interference signals, and that on the basis of the assignment, an estimate for the presence and / or type of fault condition takes place. A method according to claim 5, characterized in that the known useful signals are emitted by a simulator and / or by reflection from the radar sensor ( 1 ) via at least one transmitting antenna ( 5 ) emitted at known positions and / or in a known manner relative to the radar sensor moving objects, and / or that the known interference of at least one located at a known position and / or in a known manner relative to the radar sensor ( 1 ) moving radar system are generated. Method according to claim 5 or 6, characterized in that the radar sensor ( 1 ) and the radar system which generates the known interference signals use different or identical radar technologies, in particular FMCW radar or pulse radar technology in each case. Device for detecting at least one fault condition of a radar sensor ( 1 ) due to at least one interfering signal in a received signal of the radar sensor ( 1 ), with a received signal conditioning unit ( 7 ), with an analog-to-digital converter ( 17 ) for scanning the received signal and with a digital signal processing unit ( 19 ) for the digital signal processing of the sampled received signal, characterized in that a processor and memory unit ( 21 ), on which a learning classifier ( 23 ) is implemented for estimating a presence and / or a type of fault condition on the basis of at least one data record obtained directly or indirectly from the sampled received signal or a data stream obtained directly or indirectly from the sampled received signal. Apparatus according to claim 8, characterized in that the digital signal processing unit ( 19 ) several stages ( 19 ' . 19 '' . 19 ''' ), and that one with the process and storage unit ( 21 ) indirectly or directly connected diversion unit ( 20 ) for deriving at least one data set or data stream before, from or after one of the stages ( 19 ' . 19 '' . 19 ''' ) is provided.

Images