Forums

FMCW radar signal processing independent of sweep time

Started by Nicholas Kinar October 11, 2006
I am working on an application which involves a Frequency-Modulated 
Continuous Wave (FMCW) system.  I understand that the frequency-swept wave 
is sent out from an antenna toward a reflector, and the received reflection 
is then homodyned (mixed) with the original wave.  (I understand that this 
mixing is a form of multiplying both signals in the time domain.)  This 
produces a beat frequency f_b.



However, by digital signal processing, is it possible to determine the 
distance to the reflector without homodyning the two signals?  (This is for 
lower frequencies of electromagnetic waves, whereby the received wave can be 
converted by an ADC, and then subsequently homodyned by digital signal 
processing.)



Is it possible to determine the distance to the reflector without being 
dependent on the duration T of the frequency-swept wave?



Thank you for your time and assistance!




Nicholas

> However, by digital signal processing, is it possible to determine the > distance to the reflector without homodyning the two signals? (This is for > lower frequencies of electromagnetic waves, whereby the received wave can be > converted by an ADC, and then subsequently homodyned by digital signal > processing.)
Yes, providing you accurately know the frequency of the transmitted signal over the sweep time (easy if you are using a DAC, harder if you are controlling a free-running VCO with a control voltage). The disadvantage is that the receiver bandwidth must equal the transmitter bandwidth. In comparison, for normal FMCW, at least when you are considering fairly short range operation, the beat frequencies are small, so the receiver bandwidth can be small. Note high bandwidth ADCs tend to have low bit depth (ie low dynamic range), which may be a problem depending on the maximum range, clutter, target RCS, etc.
> Is it possible to determine the distance to the reflector without being > dependent on the duration T of the frequency-swept wave?
I think you will find that the range resolution (accuracy) is dependent on the swept bandwidth B during time the transmitter "on time" T. Therefore you could reduce T by increasing the sweep rate B/T. Note however that T must be long enough to deal with your maximum range, and also that a very short T makes the duration of each received signal (or beat signal) short, which reduces the accuracy to which you can determine its frequency - whether this is a problem depends on your system parameters. Tom
Thank you ever so much for this reply, Tom; it is very much appreciated. 
Concerning the DAC, I will incorporate one into my design.  I had initially 
thought of using a VCO for this purpose, but it is easier to know the 
frequency at a certain time if I use a DAC.  I will also think of increasing 
the sweep rate.

Once again, Tom, thank you for replying to my post.  Your assistance is 
greatly appreciated, and I thank you for your suggestions.

Nicholas

"Tom" wrote:

Nicholas
> >> However, by digital signal processing, is it possible to determine the >> distance to the reflector without homodyning the two signals? (This is >> for >> lower frequencies of electromagnetic waves, whereby the received wave can >> be >> converted by an ADC, and then subsequently homodyned by digital signal >> processing.) > > Yes, providing you accurately know the frequency of the transmitted > signal over the sweep time (easy if you are using a DAC, harder if you > are controlling a free-running VCO with a control voltage). > > The disadvantage is that the receiver bandwidth must equal the > transmitter bandwidth. In comparison, for normal FMCW, at least when > you are considering fairly short range operation, the beat frequencies > are small, so the receiver bandwidth can be small. > > Note high bandwidth ADCs tend to have low bit depth (ie low dynamic > range), which may be a problem depending on the maximum range, clutter, > target RCS, etc. > >> Is it possible to determine the distance to the reflector without being >> dependent on the duration T of the frequency-swept wave? > > I think you will find that the range resolution (accuracy) is dependent > on the swept bandwidth B during time the transmitter "on time" T. > Therefore you could reduce T by increasing the sweep rate B/T. > Note however that T must be long enough to deal with your maximum > range, and also that a very short T makes the duration of each received > signal (or beat signal) short, which reduces the accuracy to which you > can determine its frequency - whether this is a problem depends on your > system parameters. > > Tom >