Automotive radar with hardware accelerated target detection capability

1. A radar system, comprising:

a radio-frequency (RF) transmitter unit and first plurality of transmit antennas which are connected to transmit MIMO radar signals;

an RF receiver unit and a first plurality of receive antennas which are connected to generate digital output signals from MIMO radar signal returns received in response to the MIMO radar signals; and

a radar control processing unit connected and configured to control the RF transmitter unit and the RF receiver unit;

wherein the radar control processing unit is configured to generate, from the digital output signals, a target response signal in at least a first dimension; and

wherein the radar control processing unit is further configured to perform cell-averaging constant false alarm rate (CFAR) target detection by convolving the target response signal with a weighted kernel window signal to generate an output signal having a sign that indicates a target detection decision.

2. The radar system of claim 1 , wherein the target response signal and the weighted kernel window signal are each a one-dimensional signal, a two-dimensional signal, or a three-dimensional signal.

3. The radar system of claim 1 , wherein the radar control processing unit is configured to perform a one-dimensional convolution, a two-dimensional convolution, or a three-dimensional convolution of the target response signal with the weighted kernel window signal.

4. The radar system of claim 1 , wherein the radar control processing unit is configured to convolve the target response signal with the weighted kernel window signal in a frequency domain using a Fast Fourier Transform hardware accelerator and an Inverse Fast Fourier Transform hardware accelerator.

5. The radar system of claim 1 , wherein the radar control processing unit is configured to construct the target response signal as a target response magnitude vector x=[x 1 , x 2 , . . . x N ] of 1D target responses.

6. The radar system of claim 5 , wherein the radar control processing unit is configured to convolve the target response signal with the weighted kernel window signal by:

adding head and tail padding to the target response magnitude vector x to generate a padded target response magnitude vector;

computing or retrieving a weighted kernel window vector s comprising W weighting elements [s 1 , s 2 , . . . s W ], where W<N;

computing a convolution kernel window vector ( ) by reversing an index order of elements in the weighted kernel window vector s to generate an order-reversed window weight vector and computing complex conjugate values for the order-reversed window weight vector;

adding zero padding to the convolution kernel window vector ( ) to generate a padded convolutional kernel window vector having a length to match the padded target response magnitude vector;

computing a discrete target response spectrum {tilde over (X)} by performing a fast Fourier transform of the padded target response magnitude vector;

computing a discrete window spectrum {tilde over (S)} by performing a fast Fourier transform of the padded convolutional kernel window vector;

computing an element-wise multiplication product of the discrete target response spectrum {tilde over (X)} and the discrete window spectrum {tilde over (S)} to generate a spectral product {tilde over (Y)}={tilde over (X)}{tilde over (S)}; and

applying an inverse fast Fourier transform to convert the spectral product back to generate a padded detection decision metric output vector {tilde over (y)}=IFFT{{tilde over (Y)}} which contains a final detection decision metric output vector y which is padded with head and tail paddings.

7. The radar system of claim 6 , wherein the W weighting elements [s 1 , s 2 , . . . s W ] comprises:

a CUT element having a weight of −1;

a plurality of guard elements having a weight of 0 and disposed on each side of the CUT element; and

a plurality of averaging elements having a predetermined scaling weight and disposed on each side of the plurality of guard elements.

8. The radar system of claim 6 , wherein the W weighting elements [s 1 , s 2 , . . . s W ] comprise a plurality of weighted averaging elements based on a data training process.

9. The radar system of claim 6 , wherein the radar control processing unit is configured to generate the output signal by declaring a target detection for each negative-value element of the final detection decision metric output vector y.

10. A method for operating a radar system, comprising:

transmitting MIMO radar signals from transit antennas in a plurality of transmit and receive antennas;

receiving MIMO radar return signals at receive antennas in the plurality of transmit and receive antennas;

generating, at a plurality of receive channels, digital output signals from the MIMO radar return signals; and

processing the digital output signals to perform cell-averaging constant false alarm rate (CFAR) target detection by generating, from the digital output signals, a target response signal in at least a first dimension and by convolving the target response signal with a weighted kernel window signal to generate an output signal having a sign that indicates a target detection decision.

11. The system of claim 10 , wherein the target response signal and the weighted kernel window signal are each a one-dimensional target response signal, a two-dimensional target response signal, or a three-dimensional target response signal.

12. The system of claim 10 , wherein convolving the target response signal with the weighted kernel window signal comprises performing a one-dimensional convolution, a two-dimensional convolution, or a three-dimensional convolution of the target response signal with the weighted kernel window signal.

13. The system of claim 10 , wherein processing the digital output signals comprises convolving the target response signal with the weighted kernel window signal in a frequency domain using a Fast Fourier Transform hardware accelerator and an Inverse Fast Fourier Transform hardware accelerator.

14. The method system of claim 10 , wherein convolving the target response signal with the weighted kernel window signal comprises:

computing a discrete target response spectrum by applying a fast Fourier transform to at least the target response signal;

retrieving a discrete window spectrum by that is computed by applying a fast Fourier transform to a zero-padded convolutional kernel window signal generated from the weighted kernel window signal;

performing an element-wise product of the discrete target response spectrum and the discrete window spectrum to compute a spectral product; and

applying an inverse fast Fourier transform to the spectral product to generate a padded detection decision metric output signal which contains a final detection decision signal.

15. The method of claim 14 , where the target response signal comprises a target response magnitude vector x=[x 1 , x 2 , . . . x N ] of 1D target responses, and where computing the discrete target response spectrum comprises applying a fast Fourier transform to a padded target response signal comprising a head padding, the target response signal, and a tail padding.

16. The method of claim 15 ,

where the weighted kernel window signal comprises W<N weighting elements [s 1 , s 2 , . . . s W ]; and

where the zero-padded convolutional kernel window signal is generated from the weighted kernel window signal by:

computing a convolution kernel window vector ( ) by reversing an index order of elements in the weighted kernel window vector s to generate an order-reversed window weight vector and computing complex conjugate values for the order-reversed window weight vector, and

adding zero padding to the convolution kernel window vector ( ) to generate the zero-padded convolutional kernel window signal having a length to match the padded target response signal.

17. The method claim 16 , wherein the W<N weighting elements [s 1 , s 2 , . . . s W ] comprises:

a CUT element having a weight of −1;

a plurality of guard elements having a weight of 0 and disposed on each side of the CUT element; and

a plurality of averaging elements having a predetermined scaling weight and disposed on each side of the plurality of guard elements.

18. A radar apparatus, comprising:

a transmitter configured to transmit a first transmit signal and a second transmit signal;

a receiver configured to produce digital output signals from first and second transmit signal returns received in response to the first and second transmit signals; and

a digital signal processor configured to:

generate, from the digital output signals, a target response signal in at least a first dimension; and

perform cell-averaging constant false alarm rate (CFAR) target detection by convolving the target response signal with a weighted kernel window signal to generate an output signal having a sign that indicates a target detection decision, thereby identifying one or more targets in the first and second transmit signal returns.

19. The radar apparatus of claim 18 , wherein the target response signal and the weighted kernel window signal are each a one-dimensional signal, a two-dimensional signal, or a three-dimensional signal.

20. The radar apparatus of claim 18 , wherein the digital signal processor is configured to perform a one-dimensional convolution, a two-dimensional convolution, or a three-dimensional convolution of the target response signal with the weighted kernel window signal.

21. The radar apparatus of claim 18 , wherein the digital signal processor is configured to convolve the target response signal with the weighted kernel window signal in a frequency domain using a Fast Fourier Transform hardware accelerator and an Inverse Fast Fourier Transform hardware accelerator.