Aerospace and Electronic Systems Magazine December 2017 - 28

Design of a Network of Skywave Over-the-Horizon Radars

Figure 8.

The temporal performance system gain equivalence for the radar network from 0 to 30 dB values of system gain.

diagram. For simplicity, we constrain the radar systems to have
equivalent design and additional marginal system gain value. Obviously, a more global optimisation process may be conducted at
great computational cost.
We incorporate the environmental conditions by taking the
average of the probabilities of equivalent performance across season and solar activity level. Using these probabilities, we peruse
the answer to the question: what is the minimum extra sensitivity
required of the radar design to achieve a significant performance
improvement? The answer is the "equivalent performance gain"
for each input sensitivity or additional marginal system gain. We
have displayed the equivalent performance gain value against
the additional marginal system gain for the temporal and spatial
performances. Figure 8 demonstrates the equivalent performance
gains for each flight temporally, while Figure 9 shows the spatial
equivalent performance gains.
These equivalent performance gain value curves lose the indication of absolute performance. To assess the key points of the
sensitivity to performance trade-off, we must look at the mostly-atleast hours of coverage performance tables, such as Table 1.
We identify in Figure 8 that the 0 and 3 dB additional marginal
system gain values are mostly equivalent, except for the SYD circling flight. Because 0 dB is the base-level radar design, we find
that this is what we characterise as a performance cliff. Any radar
designs with sensitivity levels near these values will produce unacceptable performance.
All the flights have their performance improved with additional marginal system gains of 3 dB and above, except for the
SYD circling flight. We characterise the SYD circling flight as a
performance plateau for values of 3 dB and higher, as there is no
improvement in performance for the increased sensitivity levels.
Note that the spatial equivalent performance curves in Figure 9 all
reach a performance plateau quicker than the temporal curves in
Figure 8 because the spatial locations of performance clearly split
into areas of good and poor performance due to the geometry of
the radars, the target geometry, and the utilisation of propagation
via adequate frequency agility. Sensitivity quickly loses impact as
26

Figure 9.

The spatial performance system gain equivalence for the radar network
from 0 to 30 dB values of system gain.

these issues take dominant effect. Although, by contrast, the hours
of performance have a smooth transition from no performance to
good performance, as scaled by radar sensitivity, conditioned on
the geometry issues not dominating performance outcomes.
Another consideration in the choice of radar design is the stability of performance. If the radar network is to overcome environmental variability, we must increase the system design above the
minimum requirement, as the day-to-day propagation power may
decrease or the noise may increase with respect to the monthly median environment. ITU suggests the temporal noise variation has
values in the range of 5.3 to 10.6 dB [10] for residential locales.
Using a temporal standard deviation of noise of 4 and 8 dB, we
observe that for 6 to 11 days a month, we expect performance to
be worse than a −3 dB fluctuation. This means that if we design
our radar systems to a sensitivity level of an additional marginal
system gain of 6 dB, we expect to fail the tracking missions 6 to
11 days a month, just based on environmental fluctuations, as the
performance cliff is located at an additional marginal gain of 3 dB,
resulting in the evaluated margin of a 3 dB sensitivity buffer.
We identify 6 dB as being the minimum additional system gain
to achieve reasonable network performance for all but the BNE
to SYD flight. This could be reasonably realised through the doubling of two of either the transmit array, receive array, or transmitter power of the reference design. This achieves most of the day
performance for low solar activity and most of the day coverage
for medium solar activity, as shown in Table 1. This system gain
also maximises the spatial performance of the radar network, if we
abandon the BNE to SYD flight mission. We expect the environmental variation to significantly reduce performance for 6 to 11
days a month.

NETWORK PERFORMANCE ASSESSMENT
We have identified that our radar network design is not suitable
for achieving all the flight missions. Now, we may investigate the
marginal utility each radar provides to the success of the missions

IEEE A&E SYSTEMS MAGAZINE

DECEMBER 2017

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