FMCW CFAR Radar Signal Processing

[Johnny Radar]

Below is a link to a single power versus range spectrum at a fixed
bearing from an FMCW radar taken in ground clutter.


The values returned are unitless, and are simply plotted in the link.
The output of running a CA-CFAR detector on this data is also shown,
with a P_fa of 1%
2*N = 35 (upper and lower window width).

We are not completely sure what the units returned are (mW,dB,dBm) and
just directly run the CFAR algorithm on it. However, if the data is
indeed a non-linear scale (dB), should all the data be firstly
linearized through 10^{dB/10}, before running the CFAR ?
Thanks in advance,
John.

[Johnny Radar]

Sorry, forgot the links,

FMCW Radar spectrum: Loading Image...
CA-CFAR output: Loading Image...

[One Usenet Poster]

Post by Johnny Radar
Below is a link to a single power versus range spectrum at a fixed
bearing from an FMCW radar taken in ground clutter.
The values returned are unitless, and are simply plotted in the link.
The output of running a CA-CFAR detector on this data is also shown,
with a P_fa of 1%
2*N = 35 (upper and lower window width).
We are not completely sure what the units returned are (mW,dB,dBm) and
just directly run the CFAR algorithm on it. However, if the data is
indeed a non-linear scale (dB), should all the data be firstly
linearized through 10^{dB/10}, before running the CFAR ?
Thanks in advance,
John.

When the units are logarithmic, i.e. dB, this form of CFAR is called
"logarithmic CFAR" or "log CFAR". When the units are linear, it is
normally called "linear CFAR". There are pros and cons to both
techniques. You can find several papers on this topic on the IEEE
Xplore Web site.

Mike