You are on page 1of 11

ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 1

Laboratory report No.3 FMCW radar


Constantinescu Mihai Marius 4122992
m.m.constantinescu@student.tudelft.nl
Abstract

In the first measurement we acquire the radar output signal (beat signal) resulting from the
backscattering of a large metal plate for both settings of the radar bandwidth (1GHz with 23.4 ms
sweep time and 500MHz with a 11.7 ms sweep time). The recorded signal must be processed in
MATLAB to produce proper range profiles. In the second assignment we make measurements for
three reflector setups: two reflectors are separated in radar range by 20cm, 50cm and 1m. We acquire
the beat signal in these three cases with a 500MHz bandwidth and 11.7 ms sweep time.

1 S TUDY AND DRAW THE MEASUREMENT SET- UP. T HE DRAWING SHOULD SHOW
THE TARGETS , THE RADAR AND THE DISTANCES BETWEEN THEM .
Figure 1 presents the measurement set-up, containing our radar with the antenna dish and the
mixer (switches from 500MHz to 1 GHz and 11,7 ms / 23.4 ms sweep time) connected to the
sound card of the PC for further processing. Our targets are a large metal plate for the first case
and two smaller plates positioned at 20cm, 50cm and 100cm distance one from each other, in
order to show the dependence of resolution on the bandwidth and position.

Fig. 1. Measurement Scheme


ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 2

Fig. 2. Beat Signal

2 O BTAIN FROM THE FIRST TWO MEASUREMENTS ( ON THE METAL PLATE ) THE
RANGE PROFILES BY MEANS OF FFT OF THE ACQUIRED BEAT SIGNALS AND PLOT
THEM . I NDICATE HOW THE FREQUENCY SCALE CAN BE CONVERTED TO DISTANCE
AND ADJUST THE PLOT TO GIVE THE DISTANCE SCALE . W HAT IS THE DISTANCE OF
THE FLAT PLATE MEASURED FROM THE RADAR ? H OW DOES IT COMPARE TO THE
REAL DISTANCE ?
First we aquire a beat signal, selected among the multiple beat signals (figure 2 presenst the
beat signal and figure 3 the signal after removing the triggering pulse and the DC component)
and then processing is done on that signal by taking FFT of it (for better performances we are
using 8192 samples so we have to add 0 padding to the signal). Figure 4 shows the result of
applying fft, the frequency profile and we observe our metal plate around 1200 Hz (both cases
500 MHz, 1 GHz).
The range and the frequency for a FMCW radar are related with the equation (1):
cTs fb
R= (1)
2B
where R represents the range, c is the speed of light, Ts sweep time (11,7 ms / 23.4 ms), fb beat
frequency and B the bandwidth (500MHz / 1 GHz). Using the above formula, the range profiles
for the plate can be plotted and are figure 5 shows the results. We can observe the distance of
the flat metal plate measured from the radar which is 4.4 m for 1GHz and 4.3 m for 500MHz
(compared with the first measurement (1GHZ), this result is more accurate towards the practical
distance), result that is similar with the real distance measured during the experiment( 4m). We
can also observe a second peak at 10 m representink the wall of the room.
Figure 6 compares the results of frequency and range profiles for the 2 situations (500MHz/1GHz)
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 3

Fig. 3. Required Beat Signal (without triggering pulse)

Fig. 4. Frequency profile (magnitude)

3 I NVESTIGATE THE USE OF WINDOWING ON THE BEATSIGNAL . TAKE THE PLOT


FROM 1.2 AND ADD A PROFILE AFTER APPLYING A WINDOW, E . G . THE H ANNING
WINDOW. W HAT IS THE EFFECT ?
Before doing FFT of the beat signals, we are adding a window to it, in order to smooth it.
For investigating the effect of windowing on the beatsignal we are applying a Hanning and a
Chebysev window and the plots below present the results. Figure 7 and 8 present the Range
profile using Hanning window (Magnitude and Power) and figure 9 presents the range profile
using a Chebysev window while the last two figures (figures 9, 10) present a comparison of
the results of windows (1GHz and 500 MHz case). We can observe a couple of differences
(more evident in the dB plot) like peaks get larger or ”valleys” get deeper (less riples), less
noise (these modified signals have lower noise level). Table 1 also presents the pulse widths for
different windows.
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 4

Fig. 5. Range profile (magnitude)

Fig. 6. Range / Frequency profiles (Power (dB)) blue=1GHz / green=500Mhz

4 D ISCUSS THE COMPLETE SEQUENCE OF PROCESSING STEPS YOU HAVE AP -


PLIED IN PRODUCING THE RANGE PROFILES .
As presented before we first aquired a beat signal, selected among the multiple beat signals
(after removing the triggering pulse) and then processing is done on that signal by taking FFT
of it (for better performances we are using 8192 samples so we have to add 0 padding to the
signal). Using equation (1), from which range can be derived using the frequency we finnaly
are able to get the range profiles. So in a nutshell:
• Import the beat signals from FMCW Radar
• Select a sweep period
• Add a window to the data (optional)
• 0 padding at the end of the beat signals
• Fourier Transform of the time domain signals to frequency domain
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 5

Fig. 7. Range profile using Hanning window (magnitude)

Fig. 8. Frequency / Range profile using Hanning window (Power(dB)) blue=1GHz / green=500Mhz

• Convert beat frequency to range


• Estimate distance to target

5 S TUDY THE EFFECT OF RANGE RESOLUTION . M AKE THREE RANGE PROFILE


PLOTS FROM THE SECOND ASSIGNMENT, WITH TWO CURVES PER PLOT WITH AND
WITHOUT WEIGHTING FUNCTION . W HAT CAN YOU SAY OF THE RADAR RESOLUTION ?
H OW IS IT ILLUSTRATED IN THE RANGE PROFILES ? W HAT IS AN IMPACT OF WINDOW-
ING ON THE RANGE RESOLUTION ?
The same procedure is applied for the case of the two plates (figure 11 presents the signals
of plates at 20, 50 100 cm, in the final form without triggering pulse). Figure 12 contains the
frequency profiles (with and without (hanning) window) and figures 13, 14 and 15 present as
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 6

Fig. 9. Frequency / Range profile using Chebyshev window (Power(dB)) blue=1GHz /


green=500Mhz

Fig. 10. Window Comparison (500MHz))

asked three range profile plots from the second assignment, with two curves per plot with and
without weighting function.
It can be noticed from the figures that the targets were detected in all the cases, but because of
lack of resolution it was difficult to detect two target plates (if the distance between two targets is
only 20 cm, the FMCW radar is not able to distinguish them, because it’s resolution is 30cm from
the resolution equation (2); even the targets are 50 cm apart, they still seem to be together due to
noise or maybe the applied window is not suitable). In the last case of plate separation 100 cm-s
(at frequency of 500MHz it is clearly evident that there are two target plates. We can also see
the performarces of applying a window (in this case Hanning) and we can clearly distinguish
better our targets (separate them) compared with the initial case. We further investigate the use
of a window and also aply Chebysev and Kaiser windows and then compare the results (figures
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 7

Fig. 11. Window Comparison (1 GHZ)

Fig. 12. 2 Plates beat signal at 20, 50, 100 cm separation)

16, 17, 18) - we can say that the best results are performed by the hanning window (in the 100
cm case we can clearly distinguish between the 2 plates). For the Hanning case we can also
compute the difference of the two peaks 4.2 m - 3.2 m and verify in this way the distance it’s
really 100 cm.
Theoretically the resolution of the radar is given by equation (2):
c
R= (2)
2B
where R is the resolution, c speed of light, B bandwidth so it can be clearly seen that with higher
bandwidth results a better resolution (1 GHz better than 0.5GHz). The experiment results follow
in general the theory expectations but the difference is not that huge.
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 8

Fig. 13. Frequency profile (without and with Hanning window)

Fig. 14. Range profile (with and without weighting function)

6 S TUDY THE SYSTEM IMPULSE RESPONSE IN RANGE BASED ON THE PLATE MEA -
SUREMENT OF ASSIGNMENT 1.2. W HAT CAN BE SAID OF THE IMPULSE WIDTH IN
COMPARISON TO THE RANGE RESOLUTION AS DETERMINED IN 1.5? W HAT SHOULD
THE RELATION BE IN THEORY ? D O THEORY AND MEASUREMENT FIT ? W HAT IS THE
EFFECT OF THE BANDWIDTH ON THE RANGE RESOLUTION ?
As presented at previous point in theory the resolution of the radar is given by equation (2):
c
R= (3)
2B
where R is the resolution, c speed of light, B bandwidth so it can be clearly seen that with
higher bandwidth results a better resolution (1 GHz better than 0.5GHz - 15 cm and 30 cm).
The experiment results follow in general the theory expectations but the difference is not that
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 9

Fig. 15. Range profile (with and without weighting function)

Fig. 16. Range profile (with and without weighting function)

huge. From our results it can be seen that for higher frequency the pulse width reduces, which
should be the case ideally because with high frequency the resolution increases and thus the
pulse width should reduce. Table 1 presents the pulse width (3 Db beamwidth) for all of the
windows in the 1GHz and 500MHz cases. Compared to 0.15 m and 0.3 m, the experimental
resolutions are bigger than the teoretical value. The reason may be the utilization of windows,
that will widen the resolution.
ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 10

Fig. 17. Window comparison 100cm

Fig. 18. Window comparison 50cm


ET4169 MICROWAVE RADAR AND REMOTE SENSING (2010-2011 Q3) 11

Fig. 19. Window comparison 20cm

Fig. 20. Table 1: Pulse Width

You might also like