Hi, I'm supposed to analize a received signal on a Doppler continuous wave radar for an university practice. I only have a file with samples taken at 20KHz. The only information I know about the radar is its center frequency (24GHz) and trasmitted signal power. The recieved signal values that I have are the baseband outputs of the front-end that have a DC offset of 2.6Volts and seem to be a sinusoidal wave. I'm new on radar comunications but not on signal proccesing and I'm having problems...I'm trying to perform the FFT to get the doppler frequency offset but I'm not sure about what I'm doing because I haven't found it. To obtaining such frequencial resolution I've taken 200 samples of the received signal and I've perform the 2048-FFT. Could someone help me?? I know I'm making a "very simple" mistake but I'm not being able to find it.... Thanks!

# FFT and number of samples..A "simple" doubt?

Started by ●June 24, 2008

Reply by ●June 24, 20082008-06-24

On 24 Jun, 12:43, "night" <sandracorr...@gmail.com> wrote:> �I'm new on radar comunications but not on signal proccesing and I'm > having problems...I'm trying to perform the FFT to get the doppler > frequency offset but I'm not sure about what I'm doing because I haven't > found it. > > � � To obtaining such frequencial resolution I've taken 200 samples of the > received signal and I've perform the 2048-FFT.How many samples do you have? As a concept check, use all available data and compare the spectrum with the spectrum of the baseband reference data. Apart from that, I think the usual way to estimate Doppler shift is to modulate the baseband reference up or down in frequency, and use a correlator to do the comparision in time domain. Use several correlators, one for a Doppler interval, and search for the correlator with highest output magnitude to find the Doppler shift. Rune

Reply by ●June 24, 20082008-06-24

I've aprox. 13secs of samples at 20KHz so I have a great number of samples but I took only a few because I wasn't able to see anything performing the FFT over a lot of samples... So, I should generate a reference baseband signal and correlate it with the received signal (I know how to do it but...at what frequency u think I should modulate it?I only know that the trasmitter uses one carrier and it doesn't use frequency modulation signal transmission) I'm not finding documentation about continuos wave radar (without frequency modulation) so thanks for your help!!!!

Reply by ●June 24, 20082008-06-24

On 24 Jun, 13:09, "night" <sandracorr...@gmail.com> wrote:> I've aprox. 13secs of samples at 20KHz so I have a great number of samples > but I took only a few because I wasn't able to see anything performing the > FFT over a lot of samples... > > So, I should generate a reference baseband signal �and correlate it with > the received signal (I know how to do it but...at what frequency u think I > should modulate it?I only know that the trasmitter uses one carrier and it > doesn't use frequency modulation signal transmission)You need to read up on Doppler. A Doppler shift is conceptually the same as an AM-modulation. This modulation happens because of the interaction between target and pulse, not because of anything happening in the radar. You need knowledge about the baseband signal in order to estimate Doppler shift, since you need to compare the recieved signal with AM-modulated versions of the reference signal. Rune

Reply by ●June 24, 20082008-06-24

>On 24 Jun, 13:09, "night" <sandracorr...@gmail.com> wrote: >> I've aprox. 13secs of samples at 20KHz so I have a great number ofsample=>s >> but I took only a few because I wasn't able to see anything performingth=>e >> FFT over a lot of samples... >> >> So, I should generate a reference baseband signal =A0and correlate itwit=>h >> the received signal (I know how to do it but...at what frequency uthink =>I >> should modulate it?I only know that the trasmitter uses one carrier andi=>t >> doesn't use frequency modulation signal transmission) > >You need to read up on Doppler. A Doppler shift is >conceptually the same as an AM-modulation. This >modulation happens because of the interaction between >target and pulse, not because of anything happening in >the radar. > >You need knowledge about the baseband signal in order >to estimate Doppler shift, since you need to compare >the recieved signal with AM-modulated versions of >the reference signal. > >RuneThank u...I think I've explained me in a wrong way talking about the frequency..anyway u've told me what I needed.. I should look for getting my baseband reference signal first of all...:)>

Reply by ●June 24, 20082008-06-24

It seems to me that Doppler shift is *not* like AM modulation. It's more like time compression / expansion which for a sinusoid appears like a frequency shift. Anyway, this is CW radar which normally transmits a fixed-amplitude, sawtooth in frequency waveform. Then, in the receiver, you see leakage from the transmitter and you see the received signal. At any instant there are two frequencies which are separated by the time delay to the target and back and the rate of change of frequency in the sawtooth / frequency sweep. If things are just right, re amplitudes then you may be able to see a beat frequency in the received signal amplitude. The beat frequency would be the difference frequency which is what you want to determine range / distance / delay. So now the task would be to design the processing that will determine the (relatively stable) frequency difference between transmit and receive signals. Fred