Signal Processing - September 2017 - 50

To the best of our knowledge, we are not aware of any
and tracking of GNSS signals. The main approaches in each
research paper that addresses jointly or systematically all
class are compared in Table 4. Some algorithms such as
of these tracking performance metrics. Usually, when an
code smoothing, Astrium, DET, and DPE are not applicable
unambiguous algorithm is presented in the literature, only
in acquisition, as they rely on the outputs of three tracking
one (or at best, two) of these metrics are looked at, such as
loops (code, carrier, and subcarrier), and acquisition should
code-tracking error variance and multipath performance. It
be completed before the tracking starts.
is usually understood that, by removing or diminishing the
sidelobes and keeping only a narrow main lobe, the false
Concluding remarks and further directions
lock probability decreases, but analyses of the exact false
Huge efforts are being carried out worldwide toward the
lock probabilities and mean time to lose lock under various
modernization of GNSS. One underlying characteristic of
unambiguous algorithms are still missmost of the modernized GNSS signals is
ing in the literature. Due to these many
the use of split-spectrum BOC modulaA major challenge when
metrics that have to be considered in
tions to achieve higher positioning accuradesigning a GNSS receiver cy and less intersystem interference. A
tracking, the conclusions regarding the
best algorithms in the code-tracking
for BOC-modulated signals major challenge when designing a GNSS
part are harder to reach than the conreceiver for BOC-modulated signals is how
is how to mitigate the
clusions regarding the acquisition. The
to mitigate the ambiguities created by the
ambiguities created by
section "Comparative Summary" recaps
oscillatory nature of the correlation curve.
the oscillatory nature
the advantages and disadvantages of the
These ambiguities affect both the acquisiof the correlation curve.
main unambiguous algorithms reported
tion and the tracking stages of a GNSS
so far in the literature by looking at the
receiver and there has been significant
different performance metrics in acquisition and tracking.
effort in the research community to overcome the ambiguiAs a general rule, the narrow and hybrid main lobe techniques
ties-related challenges. The ambiguity removal part can be
offer a better performance than the wide main lobe techbuilt upon any GNSS basic acquisition or tracking strucniques in terms of tracking error variance, multipath mitigature, as shown in Figure 2. The basic tracking structures
tion, and false lock threat mitigation.
reported so far in the literature were summarized, and we
discussed the tradeoff between achieving good multipath
Complexity considerations
mitigation versus having a good noise robustness. Focus
A good measure of the complexity of the unambiguous
was given on the BOC-specific challenges and solutions.
approach is the number and complexity of the filters (see
Those solutions were divided into three main classes: wide,
Figure 4) involved in the unambiguous processing. If we
narrow, and hybrid main lobe processing. We also divided
follow the division shown in Figure 5, typically the
the ambiguity mitigating solutions into three additional
approaches involving both pre- and postcor relation
classes according to the processing steps involved in
processing are more complex than the rest. Complexity
removing the ambiguities, and we discussed how the comanalysis is also hard to find in existing literature of unamplexity of the unambiguous part is affected by the processbiguous approaches. Another complexity metric can be the
ing class. We have shown that a wide main lobe correlation
simulation time to r un the acquisition or t rack ing
is good in the acquisition, as it allows the use of a higher
structures under identical parameters, but with different
time-bin step and thus a faster acquisition. On the other
algorithms. One partial analysis we have done for the
hand, a narrow main lobe correlation better preserves the
a c qu isit ion pa r t w it h t wo GNSS mo du lat ions -
ability to cope with multipath and can remove the threat of
BOC c (10, 5) and BOC c (15, 2.5) -showed that, on average,
the false locks if there is no additional sidelobe on which
compared with full BOC, SLC takes 15 times longer, BF
to lock. Thus, there is an inherent tradeoff between the
and MH take 3.2 times longer, UAL takes 2.8 times longer,
unambiguous acquisition and unambiguous tracking. This
PUDLL takes 2.7 times longer, GRASS takes 1.85 times
means that, from a GNSS receiver design perspective, the
longer, and Benedetto and SCM algorithms take only 1.1
two receiver stages of acquisition and tracking are better
times longer. While these values depend on the signal and
to be designed in a disjoint manner when dealing with
receiver parameters, such as integration times, modulation
the ambiguities.
types, channel type, and so on, they give a very good estiFor example, if the detection performance is the desired
mate of the relative order of one algorithm with respect to
metric in the acquisition stage, then dual-sideband BF unamanother in terms of complexity.
biguous approaches give the best performance. If the low
complexity of the acquisition approach is the desired metComparative summary
ric, then then single and dual sideband UAL unambiguous
The three main classes introduced in the section "Principal
approaches are the best. In the tracking approach, there are
Dichotomy of Unambiguous Solutions: Wide Main Lobe
even more metrics to consider, such as speed, complexity,
Versus Narrow Main Lobe" cover the full spectrum of techaccuracy, multipath and noise robustness, mean time to lose
niques currently existing for the unambiguous acquisition
lock, etc. According to our studies and the results reported
50

IEEE SIGNAL PROCESSING MAGAZINE

September 2017

Table of Contents for the Digital Edition of Signal Processing - September 2017