Method, a computer program product, an apparatus and a frequency-modulated continuous-wave radar system

1. A method for producing an interference reduced radar image, the method comprising:

receiving a measured sequence of beat signals, each beat signal comprising a number of segments, each segment being a beat signal time sample and having an amplitude representing the beat signal time amplitude at a time of recording the beat signal time sample, wherein the i:th segment of the j:th beat signal is identifiable as a segment i,j with amplitude A i,j , wherein i=1, . . . , a, j=1, b, wherein a and b are integers greater than or equal to two;

identifying one or more interference segments within the measured sequence of beat signals, an interference segment being a segment subject to an interference;

doctoring the measured sequence of beat signals by creating a doctored representation of segments of the measured sequence of beat signals in the form of a matrix Y with elements y i,j :

Y

=

[

y

1

,

1

…

y

1

,

b

⋮

⋱

⋮

y

a

,

1

…

y

a

,

b

]

,

wherein y i,j =m i,j ×A i,j , wherein m i,j is a masking value, the masking value being dependent on whether segment i,j is identified as an interference segment or not, the masking value being set to a lower value if segment i,j is an interference segment compared to if it is not an interference segment;

creating a doctoring mask representing the doctoring of the segments of the measured sequence of beat signals, the doctoring mask being a matrix M with elements m i,j :

M

=

[

m

1

,

1

…

m

1

,

b

⋮

⋱

⋮

m

a

,

1

…

m

a

,

b

]

;

and

estimating a reconstructed range-doppler image {circumflex over (X)} from Y and M, wherein the reconstructed range-doppler image is a range-doppler image of Y which is deconvolved using M, a range-doppler image being a time-frequency transform of a representation of a sequence of beat signals;

such that doctoring the measured sequence of beat signals at least partially removes interference effects and estimating a reconstructed range-doppler image at least partially removes image artefacts created by the doctoring.

2. The method according to claim 1 , wherein at least one of the one or more interference segments is identified by comparing at least one of: the segment amplitude; the difference between the segment amplitude and a reference segment amplitude; and, a derivative of the difference between the segment amplitude and a reference segment amplitude; to one or more thresholds.

3. The method according to claim 1 , wherein the step of estimating a reconstructed doppler-range image comprises:

finding a first preliminary estimate of a reconstructed range-doppler image from Y and M;

finding a second preliminary estimate of a reconstructed range-doppler image from Y and M;

determining the estimated reconstructed doppler-range image {circumflex over (X)} as if M⊙X 1 is a better least square approximation to Y than M⊙X 2 , wherein ⊙ refers to a Hadamard product, wherein X 1 is an inverse Fourier transform of , and wherein X 2 is an inverse Fourier transform of .

4. The method according to claim 1 , wherein the reconstructed range-doppler image {circumflex over (X)} is a range-doppler image of Y which is deconvolved with M using a Wiener deconvolution.

5. The method according to claim 1 , further comprising setting a masking value m i,j to a minimum value if segment i,j is identified as an interference segment, wherein the minimum value is a minimum value of the matrix M.

6. The method according to claim 5 , wherein the minimum value is 0.

7. The method according to claim 5 , wherein all masking values corresponding to segments which are identified as interference segments are set to the same minimum value.

8. The method according to claim 5 , further comprising setting a masking value m i,j to a maximum value if segment i,j is not identified as an interference segment, wherein the maximum value is a maximum value of the matrix M.

9. The method according to claim 5 , wherein the masking values are configured to form at least one window function in the matrix M, wherein the window function comprises either:

the elements in a row i of the matrix M wherein one of the elements in the row is an element i,j corresponding to a segment i,j which is identified as an interference segment, the window function being configured to have element values continuously increasing along the row from a minimum value at element i,j to a maximum value at an element corresponding to a segment which is not is identified as an interference segment; or

the elements in a column j of the matrix M wherein one of the elements in the column is an element i,j corresponding to a segment i,j which is identified as an interference segment, the window function being configured to have element values continuously increasing along the column from a minimum value at element i,j to a maximum value at an element corresponding to a segment which is not is identified as an interference segment.

10. The method according to claim 9 , wherein the window function comprises one of: a Hann window, a Blackman window, a Hamming window, a Bartlett window, or a flat top window.

11. The method according to claim 1 , wherein the step of estimating a reconstructed range-doppler image {circumflex over (X)} from Y and M comprises:

Fourier transforming Y to form Ŷ;

Fourier transforming |Ŷ| to form Ŷ, wherein |Ŷ| is the elementwise absolute value of Ŷ;

Fourier transforming M to form {circumflex over (M)};

Fourier transforming |{circumflex over (M)}| to form {circumflex over ({circumflex over (M)})}, wherein |{circumflex over (M)}| is the elementwise absolute value of {circumflex over (M)};

creating an inverse mask G, G being a matrix with elements g u,v , wherein each element g u,v is based on element {circumflex over ({circumflex over (m)})} u,v of the matrix {circumflex over ({circumflex over (M)})} and an estimated signal to noise ratio of {circumflex over (X)} such that g u,v ×{circumflex over ({circumflex over (m)})} u,v forms a value which is smaller for a small estimated signal to noise ratio than for a large estimated signal to noise ratio; and,

setting {circumflex over (X)} to be the inverse Fourier transform of G⊙{circumflex over (Ŷ)}, wherein ⊙ refers to a Hadamard product.

12. The method according to claim 11 , wherein the g u,v elements of G are set to

g u,v ={circumflex over ({circumflex over (m)})} u,v */(| {circumflex over ({circumflex over (m)})} u,v | 2 +R ( u,v )),

wherein {circumflex over ({circumflex over (m)})} u,v * denote the complex conjugate of {circumflex over ({circumflex over (m)})} u,v , |{circumflex over ({circumflex over (m)})} u,v | denote the absolute value of {circumflex over ({circumflex over (m)})} u,v , and R(u,v) is a number which is inversely proportional to the square of the estimated signal to noise ratio of {circumflex over (X)} at element u,v.

13. A non-transitory computer-readable medium storing computer-readable instructions which, when executed on a processing unit, will cause the processing unit to perform a method for producing an interference reduced radar image, the method comprising:

receiving a measured sequence of beat signals, each beat signal comprising a number of segments, each segment being a beat signal time sample and having an amplitude representing the beat signal time amplitude at a time of recording the beat signal time sample, wherein the i:th segment of the j:th beat signal is identifiable as a segment i,j with amplitude A i,j , wherein i=1, . . . , a, j=1, b; wherein a and b are integers greater than or equal to two;

identifying one or more interference segments within the measured sequence of beat signals, an interference segment being a segment subject to an interference;

doctoring the measured sequence of beat signals by creating a doctored representation segments of the measured sequence of beat signals in the form of a matrix Y with elements y i,j :

Y

=

[

y

1

,

1

…

y

1

,

b

⋮

⋱

⋮

y

a

,

1

…

y

a

,

b

]

,

wherein y i,j =m i,j ×A i,j , wherein m i,j is a masking value, the masking value being dependent on whether segment i,j is identified as an interference segment or not, the masking value being set to a lower value if segment i,j is an interference segment compared to if it is not an interference segment;

creating a doctoring mask representing the doctoring of the measured sequence of beat signals, the doctoring mask being a matrix M with elements m i,j :

M

=

[

m

1

,

1

…

m

1

,

b

⋮

⋱

⋮

m

a

,

1

…

m

a

,

b

]

;

and

estimating a reconstructed range-doppler image {circumflex over (X)} from Y and M, wherein the reconstructed range-doppler image is a range-doppler image of Y which is deconvolved using M, a range-doppler image being a time-frequency transform of a representation of a sequence of beat signals;

such that doctoring the measured sequence of beat signals at least partially removes interference effects and estimating a reconstructed range-doppler image at least partially removes image artefacts created by the doctoring.

14. An apparatus for producing an interference reduced radar image, the apparatus comprising a control circuit configured to:

receive a measured sequence of beat signals, each beat signal comprising a number of segments, each segment being a beat signal time sample and having an amplitude representing the beat signal time amplitude at a time of recording the beat signal time sample, wherein the i:th segment of the j:th beat signal is identifiable as a segment i,j with amplitude A i,j , wherein i=1, . . . , a, j=1, b; wherein a and b are integers greater than or equal to two;

identify at least one interference segment within the measured sequence of beat signals, the interference segment being a segment subject to an interference;

doctor the measured sequence of beat signals by creating a doctored representation of segments of the measured sequence of beat signals in the form of a matrix Y with elements y i,j :

Y

=

[

y

1

,

1

…

y

1

,

b

⋮

⋱

⋮

y

a

,

1

…

y

a

,

b

]

,

wherein y i,j =m i,j ×A i,j , wherein m i,j is a masking value, the masking value being dependent on whether segment i,j is identified as an interference segment or not, the masking value being set to a lower value if segment i,j is an interference segment compared to if it is not an interference segment;

create a doctoring mask representing the doctoring of the measured sequence of beat signals, the doctoring mask being a matrix M with elements m i,j :

M

=

[

m

1

,

1

…

m

1

,

b

⋮

⋱

⋮

m

a

,

1

…

m

a

,

b

]

;

estimate a reconstructed range-doppler image {circumflex over (X)} from Y and M, wherein the reconstructed range-doppler image is a range-doppler image of Y which is deconvolved using M, a range-doppler image being a time-frequency transform of a representation of a sequence of beat signals which transforms the representation of the sequence of beat signals from a time domain to a frequency domain;

whereby doctoring the measured sequence of beat signals at least partially removes interference effects and estimating a reconstructed range-doppler image at least partially removes image artefacts created by the doctoring.

15. A frequency-modulated continuous-wave, FMCW, radar system, comprising:

a radar unit;

an apparatus for producing an interference reduced radar image, the apparatus comprising:

a control circuit configured to:

receive a measured sequence of beat signals, each beat signal comprising a number of segments, each segment being a beat signal time sample and having an amplitude representing the beat signal time amplitude at a time of recording the beat signal time sample, wherein the i:th segment of the j:th beat signal is identifiable as a segment i,j with amplitude A i,j , wherein i=1, . . . , a, j=1, b; wherein a and b are integers greater than or equal to two;

identify at least one interference segment within the measured sequence of beat signals, the interference segment being a segment subject to an interference;

doctor the measured sequence of beat signals by creating a doctored representation of segments of the measured sequence of beat signals in the form of a matrix Y with elements y i,j :

Y

=

[

y

1

,

1

…

y

1

,

b

⋮

⋱

⋮

y

a

,

1

…

y

a

,

b

]

,

wherein y i,j =m i,j ×A i,j , wherein m i,j is a masking value, the masking value being dependent on whether segment i,j is identified as an interference segment or not, the masking value being set to a lower value if segment i,j is an interference segment compared to if it is not an interference segment;

create a doctoring mask representing the doctoring of the measured sequence of beat signals, the doctoring mask being a matrix M with elements m i,j :

M

=

[

m

1

,

1

…

m

1

,

b

⋮

⋱

⋮

m

a

,

1

…

m

a

,

b

]

;

estimate a reconstructed range-doppler image {circumflex over (X)} from Y and M, wherein the reconstructed range-doppler image is a range-doppler image of Y which is deconvolved using M, a range-doppler image being a time-frequency transform of a representation of a sequence of beat signals which transforms the representation of the sequence of beat signals from a time domain to a frequency domain;

whereby doctoring the measured sequence of beat signals at least partially removes interference effects and estimating a reconstructed range-doppler image at least partially removes image artefacts created by the doctoring; and

wherein the radar unit comprises:

a transmit antenna configured to transmit a first sequence of signals;

a receive antenna configured to receive a second sequence of signals in response to the first sequence of signals transmitted by the transmit antenna; and

a mixer configured to mix the first sequence of signals and the second sequence of signals to generate a sequence of beat signals;

the radar unit being configured to deliver the sequence of beat signals to the apparatus as a measured sequence of beat signals, such that the apparatus produces the interference reduced radar image.