IEEE Aerospace and Electronic Systems Magazine - June 2020 - 55

Horne et al.
Figure 6(g) illustrates the estimated SINR for each
CPI over the course of the experiment. The estimated
SINR is calculated as

Table 1.

Radar Operating Parameters
Parameter

Value

SINR ¼
Experiment center
frequency
Tx/Rx bandwidth
Max. Tx power
Antenna gain

94.28 GHz

where

s is the transmit waveform for the CPI;

33 dB

Antenna beamwidth (El)

NP is the number of PRIs in the CPI;

25 dBm

Sample type

r is the received signal collected from each PRI;
P
2
X ¼ N1
2NP ks ? rPU k is the mean of the estiP
mated interference plus noise rPU measured during
the environment sensing phase, cross correlated
with the current transmit waveform s.

3 GHz
12 bit, Real
800 MHz

Operating band
Transmit pulse length
PRF

300 À 1300 MHz
3:84 Á 10À7 s (1152
samples)
10 kHz

Pulses per CPI

128

Listen interval

5 CPIs

Target interrogation
interval

5 CPIs

Experiment length

200 CPIs

Figure 6(d) shows the Fourier domain versions of the
designed TMI waveform sTMI along with the target
response and interference.
Figure 6(e) shows the sum of the squared estimation error
between the estimated interference and the known interference from the interference definition table. The interference
estimate can only be calculated during the environmental
sensing phase, the error in that estimate approaching zero.
During other phases, the interference estimate error increases
as the actual interference evolves away from the estimated
value as
X
IErr ¼
j IPU À ^IPU j2
(16)
2NF

where NF is the number of Fourier bins, IPU and ^IPU are
defined by (14) and (15), respectively.
Figure 6(f) shows the estimated spectral occupancy of
^ as estimated by the environment sensthe interference O,
ing phase processing
P

~
2NF q
^
O ¼ 100 Â 1 À
(17)
NF
where Q is the spectral mask from the "PU Environment
Estimation" section.
JUNE 2020

(18)

1 GHz

Antenna beamwidth (Az)

Sample frequency

1 X ks ? rk2 À X
NP 2N
X

It can be seen in Figure 6(f)-(g) that the values periodically return to zero. These correspond to the time region
of the environmental sensing phase, where the values are
not available.
Figure 6(e)-(g) is constructed over the full course of
the experiment, so are only fully populated at completion.
The example plots are snapshots from the end of each
experiment.
Finally, Figure 6(h) provides a comparison of the resolution and sidelobe performance between the ERA processed
full-band LFM waveform and the designed waveform, by
way of their respective autocorrelation sequences.

RESULTS
The following results are based on the radar operating characteristics defined in Table 1 and the characteristics of the
interference and target defined in Table 2. All parameters
apply to both simulation and HITL experiments.
To facilitate the comparison of the FFAST algorithm
with nonadaptive and PU avoidance adaptive techniques,
three experimental configuration are considered, which
employ different waveforms in the exploitation cycle,
namely, a fixed, full bandwidth LFM like waveform, a
notched waveform designed to avoid the PU transmissions,
and the FFAST algorithm incorporating TMI. The nonadaptive transmission option is created by setting the spectral
mask such that the full bandwidth is available to the radar
transmission. The resulting frequency domain representation of the waveform, along with that for the original LFM
are shown Figure 7. It can be seen that the spectral containment at the band edges is sharper for the designed waveform
than for the pure LFM. The PU avoidance transmission configuration allows the spectral mask estimated by the MDL to
restrict the transmission in areas of PU occupancy, but does
not implement TMI. Finally, the FFAST transmission
experiment implements the full algorithm.

IEEE A&E SYSTEMS MAGAZINE

IEEE Aerospace and Electronic Systems Magazine - June 2020

Table of Contents for the Digital Edition of IEEE Aerospace and Electronic Systems Magazine - June 2020