IEEE Aerospace and Electronic Systems Magazine - January 2021 - 5

Credit: Image licensed by Ingram Publishing

exploit antenna gains (in contrast to LTE, these tasks are
carried out using an omnidirectional antenna pattern).
To be able to carry out the indicated tasks, channel
state information (CSI) is needed. Much research today
focuses on improving CSI estimation efficiency. As
pointed out in [9], in addition to channel gain, channel
matrix contains information about DOA; so being able to
retrieve knowledge about transmitter location efficiently
will reduce the CSI estimation effort.
A variety of different methods for estimating node
positions exists in both theory and practice. Among others,
TDOA-based techniques must be mentioned, which rely
on spatially distributed receiver networks, where the feature, which transmitted the signal arrives in different times
to receivers, is exploited. Practical TDOA networks rely
on the fact that the user is within a coverage of at least
four base stations. This means that in complex propagation
environments with a high number of obstacles, the
receiver network needs to be dense, leading to high lifecycle cost. In the 5G scenario, where potentially both
source and sink need to be aware of each other's locations [8], TDOA-based systems will need extra synchronization between network nodes, adding more complexity
into the system.
So called " hybrid frameworks " have been proposed,
for example, in [10] and [11], where localization uses different methods (e.g., array-based DOA estimation and
TDOA-based systems) simultaneously. Such systems can
achieve superior position estimation accuracy when compared with using the methods individually. We see the
algorithm proposed in this article as a complement to the
described frameworks. Indeed, an effective and efficient
DOA estimation algorithm will only improve overall
hybrid system performance.
A multitude of array-based DOA estimation algorithms also exist. Based on the work in [12], for example,
these can be categorized as follows:
beamforming methods [e.g., maximum variance distortionless response (MVDR) beamformer];
subspace methods (e.g., MUSIC);
parametric methods [e.g., maximum likelihood
(ML) estimator].
JANUARY 2021

It is argued in [12] that one of the disadvantages of
beamforming methods is shown to be its poor resolution
performance. Whereas for subspace methods, such as
MUSIC, the limiting factor is handling the coherence of
received signals. Multipath propagation effect will
result in non-orthogonal eigenvectors of the array
covariance matrix, which in return will cause significant
decline of algorithm performance. Another weak point
of subspace methods is that it transforms a multidimensional problem into one dimension-meaning, subspace
methods are capable of estimating only one parameter at
a time (DOA in the given case). In the MU-MIMO case,
as discussed earlier, full CSI in needed, meaning other
methods need to be employed in parallel, speaking in
favor of parametric methods. With the ML estimator as
a representative example of the class of estimators, the
threshold phenomenon, as described in [12] and [13],
might exhibit a major problem in practical situations.
The behavior of the threshold phenomenon results in
sharp degradation of the estimator in the case of exceeding a certain parameter-such as SNR or resolution
of targets.
In order to overcome the above-mentioned issues,
arrays with a high number of antenna elements could
be used. As discussed earlier, using mmW provides
good opportunities for that, and a short wavelength
allows us to pack antenna elements close to each while
restraining the physical size of the array. In the case
of ADS-B, with working frequency 1090 MHz,
increasing the number of antenna elements to a comparable level with mmW systems is not viable due to the
consequent physical size of an array. As discussed, for
example, in [7], [8], [14]-[20] even in case mmW systems, implementing classical array processing algorithms is not a straightforward task. The problem here
is twofold: computational complexity and cost- and
power-consumption efficiency.
The computational complexity increases rapidly
with increasing number of antenna elements. In [8],
[18]-[20], different ways for reducing the complexity
are proposed. They start with having a set of
" predefined beams, " and the transmitter is said to be in
the direction of the beam, which has the highest SNR.

IEEE A&E SYSTEMS MAGAZINE

IEEE Aerospace and Electronic Systems Magazine - January 2021

Table of Contents for the Digital Edition of IEEE Aerospace and Electronic Systems Magazine - January 2021