Signal Processing - September 2017 - 75

Note that, in this process, the receiver
and the PRN sequence of each satellite in
DPE receivers are able
also needs to estimate the receiver clock
view, in an individual way. On the other
to enhance tracking
bias dt that represents the offset between
hand, DPE defines a set of candidate posiand positioning under
the receiver time and the GNSS time.
tions, determines the associated time delays
challenging scenarios
This positioning approach has established
with the positions, and computes the energy
such as multipath
itself as the de facto technique for GNSS
found at the different correlation outputs, in
propagation or weak
receivers. This is due to its modularity,
a joint manner. In a way, the local replica
reuse of well-known receiver blocks, and,
is generated as a result of jointly combinsignal conditions.
importantly, notable performance over
ing all signals. Then the tentative position
the years. The two-steps technique will
that jointly maximizes correlation with all
be referred to as the conventional approach throughout the
considered satellites is selected through optimization of a cost
remainder of the article
function. The cost function appears as a result of deriving the
MLE of the PVT parameters directly from the received signal
DPE
(1), yielding to the maximization of the addition of CAFs over
The proposal of new techniques is blooming due to the advancthe M visible satellites [22], [23]:
es in digital signal processing devices, which allow increased
M
computational complexity at faster rates [18]. Recently, fundat
(5)
c
=
argmax
K i (x i (c), fd i (c), N c, N nc) 3,
/
)
mental modifications to the conventional receiver architecture
i= 1
in Figure 1(a) were proposed. Here we refer to those approaches that not only substitute certain parts of the receiver by more
sophisticated algorithms, but those that entail an essential
where the function K i (x i, fd i, N c, N nc) is defined analogousmodification of the receiver's operation. Figure 1(b) shows an
ly as in (2). This maximization contrasts to the one obtained
advanced architecture for a receiver that integrates code/carin the conventional two-steps procedures, where each CAF
rier tracking loops and the navigation solution in a single step.
is optimized independently as for (2). More on intuitive reaThis is the basis of the so-called DPE concept introduced in
soning, adding up the energy of all satellites' contributions
[13] in the context of GNSS receivers. Figure 1(b) shows that
should increase the performance of a DPE-based solution
the acquisition mode in DPE involves an initial PVT solution
when compared to two-steps solutions where each satellite is
obtained from a conventional two-steps receiver operation,
treated independently.
which is used to initialize DPE's algorithm implemented in
The main features of this single-step approach are as
tracking mode. DPE is a completely different approach to
follows:
the receiver design, where PVT solution is estimated in one
■ More information is used in the synchronization process
step. Rather than estimating a set of observables (through
thanks to the (geometrical) coupling among channels,
time delay and Doppler shifts) to infer the associated PVT,
exploiting the inherent redundancy of the system (for
a DPE-enabled receiver directly estimates PVT from the
instance, considering information of all of the constellareceived signal x (t) . The method was first introduced in the
tions when synchronizing one satellite); the key idea is to
context of localization of narrowband radio-frequency transensure that all time delays (and Doppler shifts) are referred
mitters [19] and for multiple radio signals [20]. In the context
to the same receiver location, a fact not used in the twoof self-localization, [13] presented the approach for GNSS
steps approach.
receivers, and related works exist under the collective detec■ M two-dimensional (i.e., delay and Doppler) optimization literature [21].
tion problems must be solved to estimate each pair
The key idea is to realize that time delays and Doppler
" x i, fd i ,iM= 1 _ " y i ,iM= 1 in the two-steps approach, which is
shifts of all satellites are intimately related to one another
transformed into a single n c -dimensional problem in
through the receiver motion parameters (which we gather in
DPE's framework.
a vector c ! R n ). The fact that all those signals are received
■ Use of prior/side information appears naturally since estiat the same location and at the same time instant is crucial.
mated parameters are related to the receiver's motion for
Inspecting (3) and (4), we can readily identify that x _ x ^ch
which a number of sensors are available; in contrast, addand fd _ fd ^ch, respectively. Therefore, given a value for
ing side information on the evolution of x i or fd i is not
c , it implies a value for all satellites' time delays and Dopintuitive [24].
pler shifts. One of the simplest configurations for c could be
As a consequence, receivers based on the DPE approach are
<
< <
c = 6p , v , dt@ , encompassing PVT parameters.
able to enhance tracking of satellites and deliver PVT under
Intuitively, one can see the DPE approach as the inverse
challenging scenarios such as multipath propagation or weak
process of the conventional approach in what concerns the
signal and fading conditions. Contrary to two-steps architecrelation between the user position and the time delays. Contures, the DPE approach has the ability to use (and extract
sider some observations with the carrier wiped off. On the
useful information from) weak signals and cope with signal
one hand, the conventional approach estimates the different
blockages, allowing fast recovery and even operating with low
time delays by maximizing the correlation between the signal
satellite coverage [25].
c

IEEE SIGNAL PROCESSING MAGAZINE

September 2017

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