Aerospace and Electronic Systems - October 2018 - 52

Daum
the SISO or SIMO radar designed by smart experienced antenna
designers and radar system engineers or not? Was the design of the
SISO or SIMO radar optimized to compete with the MIMO radar,
or was it a stupid SISO or SIMO radar design, as we have seen
so often in academic papers? Was the cost of the MIMO radar the
same as the SISO or SIMO radar? After some digging, I guess that
the actual source of these figures is [4], but that paper does not
answer these questions either. Also, the analysis in Figures 3.10 to
3.13 is flawed, and it is definitely not a fair apples-to-apples comparison, because it assumes the same SNR for the MIMO and SISO
or SIMO radars, whereas MIMO suffers an irretrievable SNR loss
of a factor of N in track, where the SISO or SIMO radars have a
transmit antenna gain that is a factor of N higher than the MIMO
radar in track. Longer coherent integration time obviously improves
the suppression of clutter for GMTI applications, and MIMO radar allows continuous integration of the targets and clutter owing to the omnidirectional transmit antenna pattern, but SISO and
SIMO radars can also achieve long coherent integration times using
pulse-Doppler waveforms; this elementary fact seems to have been
ignored in the design of the strawman SISO or SIMO radars by
MIMO radar researchers, as in [7]. Likewise, better suppression of
clutter for GMTI can be achieved by using a larger receive aperture
(to narrow the receive beam) for SISO or SIMO or MIMO, or a
larger transmit aperture for MIMO radar, as in [7]. But why increase
the transmit antenna size rather than the receive antenna size? It is
generally less expensive to use a bigger receive antenna than a bigger transmit antenna, owing to the cost of power and cooling and
high power microwave components. I discussed exactly such issues
with Dan Bliss (the author of [7] and a MIMO radar expert at MIT
Lincoln Lab) at the Orlando airport after the IEEE Conference on
waveform diversity in 2009, and Dan agreed that these were valid
questions. In summary, the analysis of the MIMO vs. non-MIMO
for GMTI applications given in chapter 3 of this book is not a fair
apples and apples comparison; see [5], [6] and [16] for details.
There are many important issues and topics that are not mentioned or not covered thoroughly in this book, including: (1) SIMO
phased array radars (i.e., single transmit waveform but multiple receive channels to form multiple simultaneous receive beams); (2)
correlated MIMO radar, which does not transmit orthogonal waveforms from distinct antenna elements, but rather transits correlated
waveforms between distinct antenna elements, with the purpose of
focusing the transmit energy in space (rather than forming an omnidirectional pattern with zero transmit antenna gain) in order to
mitigate the loss of SNR (see [2] for details); (3) the useful area of
range-Doppler space is reduced by a factor of N for MIMO radars
with N degrees of freedom for any transmitted waveforms (see [1]
for more details); (4) on page 50 the authors correctly note that
the signal-to-noise ratio (SNR) for MIMO is a factor of N smaller
than for the corresponding SIMO or SISO radar, but they then go
on to say that this can be fixed by exploiting the longer time that
the targets are illuminated by the fatter MIMO transmit beams;
but this does not account for real world effects, as pointed out in
[5], including the limited coherent integration time due to the lack
of coherence of both the target and the propagation medium (i.e.,
troposphere or ionosphere); moreover, this attempt to compensate
for the loss of SNR only makes sense for search, whereas for track,
52

radar energy that is transmitted where there are no targets is irrevocably lost, to the severe detriment in MIMO radar performance;
this is the main reason that hard boiled engineers do not use MIMO
radars; (5) search performance for MIMO radar is clearly inferior
to SIMO or SISO radars, owing to scan loss for any planar or linear array, despite the longer integration time allowed by MIMO
radars (see [6] for a detailed analysis) assuming radar front end
noise but no clutter or jamming; (6) the cost and risk of designing
a MIMO radar is generally much higher than the corresponding
SISO or SIMO radar, owing to the tight coupling between antenna
design and waveform design for MIMO radar as noted in [5], larger system complexity of MIMO radars, difficulty of calibrating all
the MIMO receive channels, novelty of MIMO radars, lack of real
world experience with MIMO radars, etc.
Nevertheless, there are certain niche applications where MIMO
radar actually might make sense, including HF over the horizon
radars, where we can adaptively null spread Doppler clutter, as explained in the superb paper by Gordon Frazer [11]; unfortunately,
this book only spends one page on this important application of
MIMO radar. GMTI radar is another important niche application of
MIMO, and the bulk of this book is focused almost exclusively on
that single application. This book gives many nuts and bolts technical details about how to make MIMO radar work for GMTI applications in the real world. The authors do a good job explaining
this topic, using intuition and nice simple easy to grasp analysis.
GMTI is a good application of MIMO because the target speeds
are very low and the range interval is small, because the targets
are on the ground (by definition). That is, the slow moving ground
targets are in a very small area of range-Doppler space. This means
that MIMO does not suffer very much from the reduction in usable range-Doppler space (by a factor of N compared with SIMO
or SISO); see [1] for details. Second, GMTI radar system performance is generally limited by clutter and jamming rather than
thermal noise, and hence MIMO does not suffer very much from
the reduction in SNR by a factor of N relative to SISO or SIMO
[5]. Thirdly, GMTI radars typically operate in a track-while-scan
mode rather than a track mode. It is curious that this book does not
explain why GMTI is a good application for MIMO in contrast to
many other potential bad applications of MIMO. The theoretical
result proved in [1] is fundamental for MIMO radar performance;
it is like Ohm's law in electrical engineering. It is curious that the
results in [1] are not mentioned in this book.
Recall the quote at the beginning of this review (from page 11
of the book) which blamed the slow adoption of MIMO radars in
the real world (in contrast with paper radars in academic papers) on
"a cloud of controversy." As an alternative explanation of the lack
of real world applications of MIMO radar, is it possible that smart
experienced hard boiled antenna designers and radar system engineers have actually done a quantitative cost and risk tradeoff between
MIMO radar vs. boring old phased array radars, and concluded that
MIMO radar is too risky and too expensive relative to properly designed boring old SISO and SIMO radars? There is no such cost analysis or credible cost model mentioned in this book. Without a credible
cost model it is impossible to decide whether MIMO radar is better
than boring old phased array radars. On the other hand, rumor has it
that MIMO radar has indeed been adopted for automobiles in the real

IEEE A&E SYSTEMS MAGAZINE

OCTOBER 2018
Aerospace and Electronic Systems - October 2018

Table of Contents for the Digital Edition of Aerospace and Electronic Systems - October 2018
