GB2565824A

GB2565824A - Radar system and method for optimizing radar detection of objects

Info

Publication number: GB2565824A
Application number: GB1713624.3A
Authority: GB
Inventors: Sahyoun Walaa; Le Bars Philippe
Original assignee: Canon Inc
Current assignee: Canon Inc
Priority date: 2017-08-24
Filing date: 2017-08-24
Publication date: 2019-02-27
Also published as: GB201713624D0

Abstract

Means for optimising detection of objects by a radar system comprising a sparse array antenna comprising a plurality of antenna elements each configured to receive a signal 300 based on reflection of an emitted radar wave on at least one object. The method comprising obtaining at least one vector representing the spatial arrangement of the plurality of antenna elements; for each one of the at least one vector, determining a subset 303 of antenna elements forming a minimum redundancy sub-vector; determining a mask 306 based on the at least one sub-vector thus formed; and computing a radar image (Fig.5b) based on the mask (Fig.5a) and the vector, thereby optimizing object detection. Determining the mask may further comprise obtaining a correlation matrix 304 based on the sub-vector, computing a radar image by applying an array signal algorithm to said matrix, and comparing values of the energy of the radar image with a predetermined threshold. The threshold may be based on the minimum gap between the main and side lobes. The sparse array may be square or rectangular in shape and represented by either one or two one-dimension vectors respectively.

Description

METHOD FOR OPTIMIZING RADAR DETECTION OF OBJECTS

FIELD OF THE INVENTION

The present invention relates in general to millimeter wave radars able to reconstruct Direction of Arrival (DOA) and Angle of Arrival (AOA) images of echo signals based on reflection of an emitted radar wave on objects (also called scatterers), in particular to sparse antenna radars.

More specifically, the present invention provides a method for optimizing radar detection of objects by a radar system comprising a sparse array antenna comprising a plurality of antenna elements, as well as such a radar system.

BACKGROUND OF THE INVENTION

There exist numerous kinds of millimetre wave radar systems, for example comprising an array of antenna elements, arranged according to a certain scheme.

Angle of Arrival (AoA) techniques are specific methods of Direction of Arrival (DoA) which are known to be particularly efficient to detect objects located close to the radar and/or when the relative speed of the radar and the object is not null.

The accuracy of these methods depends on the aperture of the antenna array which is defined by the distance between the two most distant aligned antenna elements.

For instance, Figure 1a shows an example of full array antenna 101 (also called dense array antenna) comprising 20x20 antenna elements, spaced apart by a distance equal to half the wavelength of the radar signal.

In this example, the antenna elements are disposed regularly in a square so that all discrete positions are occupied by an antenna element. This distribution may be described using a one-dimension vector G, set here as follows:

G=[1 1111111111111111111], where all the positions are occupied by an antenna element, “1” meaning “busy position”, the 2D figure being the representation of the Kronecker product of vector G by itself.

Each antenna element is able to transmit a radar signal and to receive echoes due to reflection of the radar signal on objects. Since the antenna elements transmit together a same radar signal, they can be seen as one single transmission antenna. However, they may each receive different echo signals, so that they consist of a set of 20x20 small reception antennas.

This kind of antenna provides a relatively good resolution as can be seen on Figures 1b and 1c. These figures are spectrums showing the AoA elevation and azimuth obtained by applying the so-called Barlett method (which is a DoA technique) to the signals received by the full array antenna of Figure 1a, when there is a single object 102 (Figure 1b) and when there are several (here 4) objects 103a, 103b, 103c and 103d (Figure 1c).

It is recalled that in the field of antenna engineering, the main lobe is defined as the lobe containing the maximum power and roughly represents the position of a target. The side lobes are defined as the local maxima of the far field radiation pattern other than the main lobe.

As one can see on the Figures, there are several side lobes 104a, 104b and 104c around the objects 103a to 103d shown in Figure 1c. These visible side lobes result from the combination of less visible side lobes.

The two spots 103c and 103d have their main lobes overlapped. This is because the resolution at such azimuth and elevation angles of the full array antenna is not sufficient to create a proper separation between these two objects.

Also, a general drawback of full array antennas is that they require an important computation power to compensate for the redundancy of echo signals which is due to the numerous close reception antennas.

Another example of array antenna is shown in Figure 2a. In this example, a sparse array antenna 201 comprising 9x9 antenna elements arranged in a square of 20x20 positions, meaning that some positions of the square are unoccupied, is considered.

The distribution of the antenna may be described using the following onedimension vector G:

G=[1 010100101001100001 1], where only some positions are occupied by an antenna element, “0” meaning “unoccupied position” and “1” meaning “busy position”.

Thus, the number of antenna elements is smaller than in the example of Figure 1a. As a consequence, the redundancy of echo signals is reduced so as the required computation power. This kind of antenna is thus a better solution in terms of signal processing and compactness.

However, the accuracy provided by such array antennas is less good than full array antennas and as can be seen on Figures 2b and 2c, a degradation of antenna pattern is observed due to the presence of large side lobes 203a, 203b, 203c, 203d, and artefacts 205a, 205b, 205c around the main spots 202 and 204, even if the two objects 203c and 203d are represented with a better separation than with the full array antenna shown in Figure 1a.

There is thus a need to improve methods for radar detection of objects by a multi-receiver radar system, in particular comprising a sparse array antenna.

SUMMARY OF THE INVENTION

The present invention has been devised to address one or more of the foregoing concerns.

According to a first aspect of the invention, there is provided a method for optimizing radar detection of objects by a radar system comprising a sparse array antenna comprising a plurality of antenna elements each configured to receive a signal based on reflection of an emitted radar wave on at least one object, the method comprising the following steps:

obtaining at least one vector representing the spatial arrangement of the plurality of antenna elements;

for each one of the at least one vector, determining a subset of antenna elements forming a minimum redundancy sub-vector;

determining a mask based on the at least one sub-vector thus formed; and computing a radar image based on the mask and the vector, thereby optimizing object detection.

Therefore, the method of the invention makes it possible to obtain a radar image of better quality with a sparse array antenna than in the prior art.

More specifically, thanks to the mask applied to the radar image, a higher range and angular resolution with lower side lobes is achieved, without need to have a full array antenna.

Thus, a higher resolution is achieved with a smaller number of antenna elements compared to a full array antenna and a smaller number of transmission/reception modules is used to achieve better results with a lower computation power consumption.

Optional features of the invention are further defined in the dependent appended claims.

According to a second aspect of the invention, there is provided a radar system for optimizing radar detection of objects, the radar system comprising:

a sparse array antenna comprising a plurality of antenna elements each configured to receive a signal based on reflection of an emitted radar wave on at least one object;

a signal processing unit configured for:

The second aspect of the present invention has optional features and advantages similar to the first above-mentioned aspect.

Since the present invention may be implemented in software, the present invention may be embodied as computer readable code for provision to a programmable apparatus on any suitable carrier medium, and in particular a suitable tangible carrier medium or suitable transient carrier medium. A tangible carrier medium may comprise a storage medium such as a floppy disk, a CD-ROM, a hard disk drive, a magnetic tape device or a solid state memory device or the like. A transient carrier medium may include a signal such as an electrical signal, an electronic signal, an optical signal, an acoustic signal, a magnetic signal or an electromagnetic signal, e.g. a microwave or RF signal.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the invention will now be described, by way of example only, and with reference to the following drawings in which:

Figure 1a illustrates a typical example of a full array antenna having a square shape;

Figure 1b is a spectrum showing the AoA elevation and azimuth obtained by applying the so-called Barlett to the signals received by the full array antenna of Figure 1a, when a single object is detected;

Figure 1c is a spectrum showing the AoA elevation and azimuth obtained by applying the so-called Barlett to the signals received by the full array antenna of Figure 1a, when a plurality of objects are detected;

Figure 2a illustrates a typical example of sparse array antenna having a square shape;

Figure 2b is a spectrum showing the AoA elevation and azimuth obtained by applying the so-called Barlett to the signals received by the sparse array antenna of Figure 2a, when a single object is detected;

Figure 2c is a spectrum showing the AoA elevation and azimuth obtained by applying the so-called Barlett to the signals received by the sparse array antenna of Figure 2a, when a plurality of objects are detected;

Figure 3 illustrates general steps of an optimization method according to embodiments of the present invention;

Figure 4a shows a mask computed for one object detection;

Figure 4b shows the results obtained by applying the method according to embodiments of the present invention using the mask of Figure 4a;

Figure 5a shows a mask computed for multiple object detection;

Figure 5b shows the results obtained by applying the method according to embodiments of the present invention using the mask of Figure 5a;

Figure 6 illustrates an example of sparse array antenna having a rectangular shape; and

Figure 7 illustrates a radar system according to embodiments.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

In the following description, it is considered a radar system comprising a sparse array antenna comprising a plurality of antenna elements each configured to receive a signal based on reflection of an emitted radar wave on at least one object.

The invention aims at providing a sparse array antenna (which may be defined as an antenna comprising a smaller number of antenna elements than a full array antenna) which can achieve the same goal (resolution) as a full array antenna.

According to general embodiments, a method for optimizing radar detection of objects by the radar system comprises the determination of a sub-set of antenna elements from among the antenna elements of the sparse array antenna, so that the sub-set of antenna elements forms a Minimum Redundancy Antenna (MRA).

According to the article entitled “Minimum redundancy linear arrays” written by A.T. Moffett, the MRA is defined as the minimum number of antenna elements which can provide the same resolution (accuracy) as a full array antenna but with minimum redundant received signals.

A method according to embodiments also comprises determining a mask based on the determined sub-set and applying this mask to a non-filtered radar image in order to obtain an optimized radar image with a better accuracy than the non-filtered radar image.

In the following description, the spatial distribution of the antenna elements of the sparse array antenna is described using one or two one-dimension (1D) vector G made of “0” and “1” values, wherein “0” means “unoccupied position” and “1” means “busy position”.

Thus, in case of a square array antenna, the 2D spatial distribution of the antenna elements may be obtained using the Kronecker product (symbolized by “o”) of the 1D vector G: G o G.

In case of a rectangular array antenna, the 2D spatial distribution of the antenna elements may be obtained using the Kronecker product of the two 1D vectors Gh (for horizontal) and Gv (for vertical): Gh o Gv.

Figure 3 illustrates general steps of an optimization method according to embodiments of the present invention. This algorithm is typically performed by the radar system 700 shown in Figure 7.

In this example, it is assumed that the radar 700 comprises the square sparse array antenna shown in Figure 2a. It is recalled that the vector G for this sparse array antenna is the following one: [1 010100101001100001 1],

It is assumed that a radar wave is emitted by the radar 700. In practice, each antenna element of the transmission array may be activated one after the other (i.e. sequentially).

At step 300, the radar, and more particularly the antenna elements of the sparse array antenna 702 shown in Figure 7 receive a signal issued from on reflection of the emitted radar wave on at least one object, for instance objects 720 and 721 shown in Figure 7.

The signal processing unit 701 of the radar 700 collects the response of the antenna elements. This response is made of the amplitude and the phase of the received signal relative to those of the emitted signal. As the amplitude and the phase of the signal are different for each receiving antenna element, a matrix is adapted to represent the response of the whole sparse array antenna 702.

The spatial response is such that the amplitude difference between the side lobe having the highest amplitude and the main lobe is superior to a predetermined threshold. In an embodiment, it corresponds to the highest amplitude difference.

At step 301, a covariance or correlation matrix is built for the whole sparse array antenna, based on the vector G and more specifically the G o G Kronecker product. Thus, the covariance matrix is simply the sum of the products of the vector containing the received signals with its Hermitian transposed. It is recalled that in this context, the correlation matrix is equivalent to the covariance matrix. This is because the average offset is null, due to the fact that an antenna does not offset an AC signal.

At step 302, a 2D AOA spectrum is computed by applying an algorithm of estimation of the Direction of Arrival such as the so-called Barlett method. For instance, it consists in multiplying the correlation/covariance matrix obtained at step 301 with steering vectors. The obtained AOA spectrum shows the azimuth in function of the elevation of the angle of arrival. It typically shows peaks at the AOA of the energy and secondary lobes.

Examples of 2D AOA spectrums obtained using the Barlett method when the collected response corresponds to a single object and when the collected response corresponds to a plurality of objects are shown in Figures 2b (single object) and 2d (plurality of objects).

At step 303, a subset of antenna elements is determined based on the response collected at step 300. This subset forms the so-called MRA of the sparse array antenna 702. It may be represented by a sub-vector of the 1D vector G, denoted “sub G” in the following description. Advantageously, as the sub G array is integrated into the G array, there is no extra data collection to be done, the measures made for G contains the measures for sub-G.

In this example, it is assumed that the sub G vector corresponds to the underlined part of the vector G:

[1 0 1 0 1 0 0 1 0 1 00 1 1 0000 1 1], i.e. sub G = [1 0 1 0 0 1 1],

At step 304, a covariance or correlation matrix is built for the subset of antenna elements, based on the sub-vector sub G and more specifically the sub G o sub G Kronecker product. Thus, the covariance matrix is simply the sum of the products of the sub-vector containing the received signals with its Hermitian transposed.

In some embodiments, step 304 is performed after or during step 301 and comprises the extraction of the correlation/covariance sub-matrix of sub G o sub G from the correlation/covariance matrix of G o G. In the present example, positions from 8 to 14 of the vector G are selected in row and column.

At step 305, a 2D AOA spectrum is computed by multiplying the correlation/covariance matrix obtained at step 304 with steering vectors. This step is similar to step 302.

At step 306, the 2D AOA spectrum computed at step 305 is used to create a mask (i.e. a mask matrix). In practice, the AOA spectrum may be described as a AoA matrix. The energy of each point of the spectrum corresponding to an element of the AOA matrix is compared to a predetermined threshold and when it is above, the corresponding matrix element is represented by a “1” in the mask matrix while it is represented by a “0” if the energy is below the threshold.

The predetermined threshold may be set as the minimum gap between the main lobe and the side lobes amplitudes. The minimum gap may be defined as an amplitude difference between the side lobe having the highest amplitude and the main lobe.

In practice, it may be computed as the Z transform of the G matrix. For illustration purposes, if the gap is equal to 5 dB, the threshold may be defined as the maximum power of the main lobe minus 5 dB.

Alternatively, the threshold may be determined using the so-called Otsu algorithm described in the article A threshold selection method from gray-level histograms” (IEEE Trans. Sys., Man., Cyber. 9 (1): 62-66 (1979)), by applying this algorithm to the AOA spectrum computed at step 305 based on the sub-vector sub G. The Otsu algorithm is commonly used to transform image with levels of gray into a binary (black or white) image.

For instance, the mask 405 shown in Figure 4a can be obtained by applying this method to the sparse array antenna shown in Figure 2a when the collected response corresponds to a single object (Figure 2b).

As another example, the mask 505 shown in Figure 5a can be obtained by applying this method to the sparse array antenna shown in Figure 2a when the collected response corresponds to a plurality of objects (Figure 2c).

In some embodiments, steps series 301-302 and 303-306 may be performed in parallel when these series of steps are independent.

At step 307, the 2D AOA spectrum obtained at step 302 is filtered with the mask created at step 306 in order to form a filtered radar image.

The results of such masking with the mask of Figure 4a on the 2D AOA spectrum of Figure 2b (single object in the scene) is shown in Figure 4b. As we can see, there are no more secondary lobes but only a thin central spot 406 corresponding to the detected object.

Also, the results of such masking with the mask of Figure 4b on the 2D AOA spectrum of Figure 2c (plurality of objects in the scene) is shown in Figure 5b. As we can see, the secondary lobes are only a few and a good separation is achieved between central spots 506a to 506d corresponding to detected objects.

In order to judge the quality of the filtered AOA image, a ratio of the main lobe power to the entire spectrum power may be introduced. The main lobe power represents the sum of the power of all the elements in the AOA matrix where the power is higher than half of the main lobe power. This latter parameter is divided by the sum of the power of all the elements of the AOA matrix. If the power is more concentrated in the main lobe and all the rest is negligible, the ratio becomes closer to 0 dB.

The table below compares the ratio for the AOA spectrum shown in Figure 1b (single object in the scene) for the full square antenna 101 shown in Figure 1a, the non-filtered AOA spectrum shown in Figure 2b (single object in the scene) and the filtered AOA spectrum shown in Figure 4b for the sparse square antenna 202 shown in Figure 2a:

Main Lobe power/spectrum power (dB) Full array antenna -7.6
Sparse array antenna	-19.8
Sparse array antenna with Mask i	-5.3

The sparse array antenna with mask is closer to zero than the full array antenna which means that the power is more concentrated in the main lobe.

The sparse antenna without a mask shows a lower ratio because of the side lobes that appear in the non-filtered AOA spectrum shown in Figure 2b and dissipate the power.

Therefore, the mask has enhanced the ratio for the sparse array antenna of

14.5 dB while omitting the side lobes created by the sparse antenna.

As a conclusion, the sparse array antenna (here of 9x9 antenna elements) with mask correction suppresses the side lobes and presents a better quality than a full antenna array of the same size (20x20) but with less antenna elements number (55% less antenna elements occupation).

Correspondingly, the table below compares the ratio for the AOA spectrum shown in Figure 1c (plurality of objects in the scene) for the full square antenna 101 shown in Figure 1a, the non-filtered AOA spectrum shown in Figure 2c (plurality of objects in the scene) and the filtered AOA spectrum shown in Figure 5b for the sparse square antenna 202 shown in Figure 2a:

Main Lobe power/spectrum power (dB)

Full array antenna

-6.8

Sparse array antenna	-17.9
Sparse array antenna with Mask	-5.9

Compared to the results obtained for a single object, a centered object presents a higher ratio (i.e. closer to zero) than the plurality of objects because there are some side lobes after mask correction that remain, as shown in Figure 5b.

The sparse array antenna with mask is closer to zero than the full array antenna which means that the power is more concentrated in the main lobe and that filtered AOA spectrum has less side lobes and a better resolution.

Therefore, the mask has enhanced the ratio for the sparse array antenna of 12 dB while omitting the side lobes 205a, 205b and 205c shown in Figure 2c.

As a conclusion, the sparse array antenna (here of 9x9 antenna elements) with mask correction provides a AOA spectrum of better quality than a full array antenna of the same size (20x20) but with a smaller number of antenna elements (55% less occupied position).

Thanks to the present invention, a sparse array antenna may be used can to reconstruct AOA images in radar systems, because the mask correction is very efficient to improve the results.

Even though the above-described examples relate to a square array antenna, the present invention is not limited thereto.

Thus, in other embodiments, the sparse array antenna may be rectangular with two different numbers of antenna elements for the transmission 1D array and for the reception 1D array.

For illustration purposes, it is considered a sparse array antenna having 8 occupied antenna elements on 14 positions in the vertical direction (transmission array) and 12 occupied antenna elements on 24 positions in the horizontal direction (reception array) as shown in Figure 6. In this case, two 1D vectors Gh and Gv are used to represent the distribution of the sparse array antenna in the horizontal direction (Gh) and in the vertical direction (Gv):

Gh=[1 0100100011110011100001 1];

Gv=[1 0 10 10 10011011],

Steps 300 to 307 are applied in a similar way as for the square array but with small arrangements to steps 301, 303 and 304.

At step 301, the covariance or correlation matrix is built for the whole sparse array antenna, based on the vectors Gh and Gv and more specifically the Gh o Gv Kronecker product.

At step 303, two different subsets of antenna elements are determined. These subsets form the so-called MRA of the sparse array antenna and are represented by two sub-vectors of the 1D vectors Gh and Gv, denoted “sub Gh” and “sub Gv” in the following description. In this example, the sub Gh and Gv vectors correspond to the underlined parts of vector Gh and Gv:

[1 0 1 00 1 000 1 1 1 1 00 1 1 1 0000 1 1], i.e. sub Gh = [1 0 0 1 0 0 0 1 1 1];

[10 10 10 10 0 11 0 1 1], i.e. sub Gv = [1 0 1 0 0 1 1],

At step 304, the covariance or correlation matrix is built for the subsets of antenna elements, based on the sub-vectors sub Gh and Gv and more specifically the sub Gh o sub Gv Kronecker product.

The threshold used to create the mask is defined as the minimum gap existing between the main lobe and side lobes for both sub Gh and sub Gv vectors.

The same results are observed in the case of a rectangular sparse array antenna, i.e. a resulting radar image of better quality than provided by a full array antenna of the same size but with a smaller number of antenna elements.

Figure 7 illustrates a radar system according to embodiments.

In the given example, the radar 700 comprises a signal processing module

701 and a set of antenna elements 702 forming a sparse array antenna.

The sparse array antenna 702 is configured to emit an electromagnetic signal which reaches two objects 720 and 721. The reflected signals (echoes) are captured by the set of antenna elements 702 together with noise.

The radar 700 is preferably connected to a human-machine interface 730, for instance a screen. The human-machine interface 730 is connected to the signal processing module 701 and displays the relative positions of the objects 720 and 721. The human-machine interface 730 may also be configured to display video data.

While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive, the invention being not restricted to the disclosed embodiment. Other variations to the disclosed embodiment can be understood and effected by those skilled in the art in putting into practice (i.e. performing) the claimed invention, from a study of the drawings, the disclosure and the appended claims.

In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality. A single processor or other unit may fulfil the functions of several items recited in the claims. The mere fact that different features are recited in mutually different dependent claims does not indicate that a combination of these features cannot be advantageously used. Any reference signs in the claims should not be construed as limiting the scope of the invention.

Claims

1. A method for optimizing radar detection of objects by a radar system comprising a sparse array antenna comprising a plurality of antenna elements each configured to receive a signal based on reflection of an emitted radar wave on at least one object, the method comprising the following steps:

2. The method of claim 1, wherein computing a radar image comprises: generating a correlation matrix based on the at least one vector;

computing a non-filtered radar image by applying an array signal algorithm to the correlation matrix; and applying the mask to the non-filtered radar image to obtain a filtered radar image.

3. The method of claim 1, wherein determining the mask comprises: obtaining a correlation matrix based on the at least one sub-vector; computing a radar image by applying an array signal algorithm to the correlation matrix thus obtained; and comparing values of the energy of the radar image with a predetermined threshold; and computing the mask based on the result of the comparing step.

4. The method of claim 3, wherein obtaining a correlation matrix based on the at least one sub-vector comprises extracting a sub-matrix from the correlation matrix generated based on the at least one vector.

5. The method of claim 3, wherein the predetermined threshold is based on a minimum gap between a main lobe and side lobes amplitudes of the computed radar image.

6. The method of claim 2, wherein the sparse array antenna has a square shape thereby only one vector represents the spatial arrangement of the plurality of antenna elements, and the correlation matrix is obtained based on the Kronecker product of the only vector by itself.

7. The method of claim 2, wherein the sparse array antenna has a rectangular shape thereby two vectors represent the spatial arrangement of the plurality of antenna elements and the correlation matrix is obtained based on the Kronecker product of the convolution of the two vectors.

8. The method of claim 1, wherein the step of obtaining at least one vector comprises obtaining a plurality of vectors, each vector representing a spatial arrangement of the plurality of antenna elements;

the method further comprising:

for each vector of the plurality, computing a radar image comprising side lobes and a main lobe; and determining an amplitude difference between the side lobe having the highest amplitude and the main lobe; and selecting the vector from the plurality having the highest determined amplitude difference.

9. A computer-readable storage medium storing instructions of a computer program for implementing steps of the method according to any one of claims 1 to 8 when the program is loaded and executed by a programmable apparatus.

10. A radar system for optimizing radar detection of objects, the radar system comprising:

a signal processing unit configured for:

5 determining a mask based on the at least one sub-vector thus formed; and computing a radar image based on the mask and the vector, thereby optimizing object detection.

11. The radar system of claim 10, wherein the sparse array antenna has a 10 square shape and the distribution of its plurality of antenna elements is represented by a single one-dimension vector.

12. The radar system of claim 10, wherein the sparse array antenna has a rectangular shape and the distribution of its plurality of antenna elements is

15 represented by two one-dimension vectors.