Aerospace and Electronic Systems Magazine December 2017 - 62

MIMO Methods Applied in Over-the-Horizon Radar
ing modes are rejected. The rejection performance of the two
examples is comparable and we stress that isolating these two
modes individually on the basis of differences in elevation angle
via transmit beamforming has been undertaken simultaneously
on the same dataset recorded by a single receiver and that this
cannot be achieved without the use of multiple waveform methods. See [26] for details and examples of our previous use of this
algorithm in the MIMO radar context. While the remaining two
modes preservation cases are not shown here we have been able
to successively preserve each mode in turn and reject all other
modes to the noise floor.

RANGE DEPENDENT BEAMFORMING
Geometric consideration of the propagation paths employed in
OTHR mean that mode-selection/rejection beamformer solutions generated at one range may be ineffective for mode-rejection at different ranges. This is demonstrated in the following
example in Figures 8a-c. The receiver data has been measured
simultaneously at Elliott (1,514 km) and Tindal (1,864 km) and
a single waveform channel showing the mode structure measured at Tindal (range 1,864 km) is shown in Figure 8a. The
modes here (from minimum range to maximum range) have
been identified as 1F2l, 1F2h-o, and 1F2h-x. This range is too long
to allow for 1E propagation. We have processed the received
data as follows. Firstly, in Figure 8b, we have computed and applied a mode-selection/rejection beamformer to retain only the
strongest mode at Tindal. Note that one mode remains and that
the remaining two modes (location marked with white circles)
have been rejected to the noise floor of the Doppler-range map.
However, in Figure 8c we have applied this Tindal data generated beamformer to the shorter range Elliott data. From Figure
8c it is clear that mode rejection for the rejected-at-Tindal mode
is now poor.

AZIMUTHAL DEPENDENCE
In the previous section we presented results demonstrating that
a MVDR mode-selective beamformer solution computed for one
range is ineffective at different ranges. We have also investigated
the azimuthal dependency of the MVDR mode-selective beamformer solution. As with the range varying case, we are able to
show that the mode-selective beamformer solution computed at
one range and azimuth is ineffective when applied to data collected
at comparable range but different azimuth.
Table 1 lists the locations of down-range receiver sites used
in the experiment. The three sites at Mt. Everard, Hermannsburg,
and Kings Canyon are of similar range but different azimuth. In
fact, this was the key criteria in selecting the sites. We have investigated all combinations of the MVDR mode-selective beamformer solution computed using data at one location and applied
at that location and also applied to the remaining two locations. In
all cases the mode-selective beamformer solution determined at a
given location did not prove to be effective when applied to data
received at the remaining locations, yet was highly effective as a
mode selection beamformer for data at the same location.
60

Figure 9.

(a) Single Channel Response showing the mode structure measured at
Kings Canyon (range 867 km azimuth -30.1°). (b) and (c) A series of
range-Doppler maps show the effect of the adaptive transmitter processing for different azimuths and similar range.

We present one example showing the azimuthal dependency of
the mode-selective beamformer solution in Figure 9. These results
are typical of all the cases examined.

IEEE A&E SYSTEMS MAGAZINE

DECEMBER 2017

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine December 2017