Radar system and method for transmitting data in a radar system
According to various embodiments, a radar system is described including a first radar processing device and a second radar processing device, wherein the first radar processing device is configured to generate radar data and to transmit the radar data partially to the second radar processing device for further processing, wherein the first radar processing device is configured to omit parts of the radar data from the transmission and wherein the second radar processing device is configured to reconstruct the omitted parts using a machine learning model trained to supplement radar data with additional radar data and is configured to further process the transmitted parts of the radar data in combination with the additional radar data.
1. A radar system, comprising:
a first radar processing device and a second radar processing device,
wherein the first radar processing device is configured to generate radar data and to transmit the radar data partially to the second radar processing device for further processing, wherein the first radar processing device is configured to omit, according to a mask, parts of the radar data from the transmission, wherein the first radar processing device is configured to signal the mask to the second radar processing device; and
wherein the second radar processing device is configured to reconstruct the omitted parts using a machine learning model trained to supplement radar data with additional radar data and is configured to further process the transmitted parts of the radar data in combination with the additional radar data, wherein the machine learning model has been trained with training data that comprises data samples for a sequence of sampling times, wherein the data samples have been set to zero for at least a portion of the sampling times in the sequence.
2. The radar system of claim 1 , wherein the radar data are radar samples, the first radar processing device is configured to omit radar samples for certain sampling times and the second radar processing device is configured to reconstruct the omitted radar samples and to perform a range Fourier transformation using the reconstructed radar samples.
3. The radar system of claim 2 , wherein the first radar processing device is configured to omit radar samples for certain combinations of sampling time, radar channel and chirp.
4. The radar system of claim 2 , wherein the second radar processing device is configured to reconstruct the omitted radar samples by extrapolating and/or interpolating transmitted radar samples for other sampling times, radar channels and/or chirps transmitted from the first radar processing device to the second radar processing device.
5. The radar system of claim 1 , wherein the radar data are range Fourier transform results, the first radar processing device is configured to omit range Fourier transform results for certain chirps and the second radar processing device is configured to reconstruct the omitted range Fourier transform results to perform a Doppler Fourier transformation using the reconstructed range Fourier transform results.
6. The radar system of claim 5 , wherein the first radar processing device is configured to omit range Fourier transform results for certain combinations of range bin, radar channel and chirp.
7. The radar system of claim 5 , wherein the second radar processing device is configured to reconstruct the omitted range Fourier transform results by extrapolating and/or interpolating range Fourier transform results for other range bins, radar channels and/or chirps from the first radar processing device to the second radar processing device.
8. The radar system of claim 1 , wherein the radar data are Doppler Fourier transform results, the first radar processing device is configured to omit Doppler Fourier transform results for certain radar channels and the second radar processing device is configured to reconstruct the omitted Doppler Fourier transform results to perform direction of arrival determination using the reconstructed Doppler Fourier transform results.
9. The radar system of claim 8 , wherein the first radar processing device is configured to omit Doppler Fourier transform results for certain combinations of range bin, radar channel and Doppler bin.
10. The radar system of claim 9 , wherein the second radar processing device is configured to reconstruct the omitted Doppler Fourier transform results by extrapolating and/or interpolating Doppler Fourier transform results for other range bins, radar channels and/or Doppler bins that are different from the certain combinations of range bin, radar channel and Doppler bin used to omit Doppler Fourier transform results.
11. The radar system of claim 1 , wherein the first radar processing device is configured to discard the parts of the radar data.
12. The radar system of claim 1 , wherein the first radar processing device is configured to determine a compression rate and determine the mask according to the determined compression rate.
13. The radar system of claim 1 , wherein the first radar processing device is a Monolithic Microwave Integrated Circuit and the second radar processing device is a microcontroller.
14. The radar system of claim 1 , wherein the first radar processing device is a radar sensor device of a vehicle and wherein the second radar processing device is a processing device of an Advanced Driver Assistance System of the vehicle.
15. The radar system of claim 1 , comprising a training device configured to train the machine learning model with the training data.
16. The radar system of claim 1 , wherein the machine learning model is a neural network.
17. The radar system of claim 1 , wherein the machine learning model comprises at least one of a multi-layer perceptron, an autoencoder and a UNet.
18. A method for transmitting data in a radar system, the method comprising:
generating radar data by a first radar processing device;
transmitting the radar data partially from the first radar processing device to a second radar processing device for further processing, wherein the first radar processing device omits, according to a mask, parts of the radar data from the transmission and signals the mask to the second radar processing device;
reconstructing the omitted parts, by the second radar processing device, using a machine learning model trained to supplement radar data with additional radar data, wherein the machine learning model has been trained with training data that comprises data samples for a sequence of sampling times, wherein the data samples have been set to zero for at least a portion of the sampling times in the sequence; and
further processing the transmitted parts of the radar data in combination with the additional radar data by the second radar processing device.
19. The method of claim 18 , further comprising training the machine learning model.