IEEE - Aerospace and Electronic Systems - April 2022 - 11

Kellett et al.
it is likely that any non-COTS available functions will
have to be implemented on FPGAs instead ofASICs.
FPGAs provide flexibility through their reprogrammFigure
4.
Illustration ofan undersampled signal received at baseband.
ELEMENT LEVEL DIGITIZATION
Traditionally radars have converted from analog signals to
digital samples at the subarray level, with individual elements
being controlled on transmit and receive by analog
phase shifters. When receiving, information is lost upon
the analog combination of signals at the subarray level.
This is because each subarray is programmed with one
receive pattern prior to the reception. While digital beamforming
is possible using the data at subarray levels, the
angular limits are set by the subarray beam width and also
by grating lobe levels that grow dramatically with scan off.
Introducing element level digitization theoretically
allows the radar to create multiple different arbitrary receive
patterns, each suited to a different task and simultaneous in
operation. This is not beyond the architecture of current
MFRs, however, they require additional analog connections
where each element is a member ofmultiple subarrays. Each
connection must include analog components such as phase
shifters that add to the complexity ofthe overall design.
The ability to form multiple, extremely high gain
receive patterns could allow for broadening of the transmit
beam. In broadening the transmit beam more targets can
be illuminated at once, thus allowing the beam to dwell at
one point in space for longer increasing the Doppler resolution
of the radar. Broadening the transmit beam does,
however, increase the amount of clutter illuminated at any
one time although this is mitigated by the narrow receive
beam widths.
The ability to form multiple arbitrary receive patterns
may be also be limited by the computational power
required to process the data from the thousands of elements
typically found on naval MFRs. For example
MUSIC, a super-resolution direction finding algorithm
used in ES and some radars has complexity Oðn2Þ, where
n is the number of receivers, be they at the sub-array level
or at the element level [19].
It is recognized that much of the signal processing
close to the ADC is likely to occur on application-specific
integrated circuits (ASICs) or field programmable gate
arrays (FPGAs), not general-purpose processing units
such as GPUs. Due to ASICs' high fixed, nonrecurring
engineering costs and the relatively low number of units
required for MFRFS (less than a few million) production,
APRIL 2022
ability that would enable further improvements to the hypothetical
MFRFS after deployment. FPGAs can also be
reprogrammed at runtime to perform different tasks;
although the timescales for this (greater than 100 ms) using
traditional architectures may be too long for radar applications.
Some novel architectures and techniques may support
reprogramming times much quicker than this [24].
FPGAs have seen a reduction in cost for a given performance.
Architecture changes and specialist logic cells
make performance metrics such as gate count ineffectual
when analyzing long-term trends, however, Trimberger
[25] and Kuon et al. [26] provide a history of the development
ofthe FPGA. In summary, based on Xilinx data, there
has been a 100 fold increase in speed and a 1000 fold reduction
in power for each logic cell, and a 1000 fold increase in
the number oflogic cells on a FPGA since 1988 [25].
It is also debatable on how many analog RF components
can be removed by moving to an element level digitization
architecture, as some components such as RF
limiters and circulators will have to be moved to element
level instead of subarray level. However, it should be
noted that element level digitization relaxes some component
requirements, such as dynamic range on the ADC.
This is because the received signal power relative to the
noise floor is lower for one element than the coherently
combined powers ofmany elements.
The benefits from element-level transmit digitization
are less clear. Unlike in the receive case, there can only be
one transmit pattern at a time, although it can still vary
across frequencies. This is achieved by adding different
phase shifts to different frequency components prior to
digital to analog conversion.
Element-level transmit digitization avoids the problem
of squint caused by analog phase beamforming. Further, it
could enable the MFRFS to transmit both radar pulses and
communications or EA at the same time without impairing
the beamforming functionality of any one function
(assuming the functions are on different frequencies). It
should be noted, however, that operating multiple transmit
functions at the same time will require total transmit
power to be split, thus reducing the sensitivity ofthe radar.
MIMORADAR
Multiple-input multiple-output (MIMO) radar has been a
topic of significant research in the past decade. Proponents
claim they can provide finer spatial resolution for the same
number of elements, in addition to an increased maximum
number of identifiable targets and higher sensitivity to
slow-moving targets [21]. On the other hand, it should be
noted that this can come at the cost of sensitivity in general
due to reduced directivity on transmit.
IEEE A&E SYSTEMS MAGAZINE
11
IEEE - Aerospace and Electronic Systems - April 2022

Table of Contents for the Digital Edition of IEEE - Aerospace and Electronic Systems - April 2022
