Signal Processing - September 2017 - 121

postprocessing. The GPS signal was processed by the GRID
SDR [32], and the LTE signals were processed by the LTE
SDR. Figure 7(a) shows the experimental hardware and software setup. The environment layout, eNodeBs locations, and
the estimated receiver trajectory from GPS and LTE signals
are shown in Figure 7(b).
Figure 7(b) shows that the trajectory estimated using only
LTE signals from two eNodeBs follows closely the navigation
solution obtained by GPS. Poor geometric configuration of the
eNodeBs is one source of error contributing to the difference
between the LTE and GPS navigation solutions. By increasing
the number of eNodeBs that are capable of being heard, e.g.,
by listening to other frequency bands and obtaining the LTE
signals from other network providers, the geometric configuration could be improved. Another source of error is due to
the mismatch by the velocity random walk dynamical model
assumed by the EKF and the true dynamics of the vehicle.
Such mismatch could be alleviated by using an INS to propagate the vehicle's states [45].

Framework and Experimental Results" sections in sections
"CDMA" and "LTE," respectively, since the BTSs (eNodeBs)
geometrical configuration, C/N 0, and transmitters' oscillator
stability, among other parameters, are different. Nevertheless, Table 1 compares the main characteristics of 1) GPS C/A
code, 2) CDMA pilot signal, and 3) three LTE reference signals (PSS, SSS, and CRS). Note that there are 63 possible PN
sequences for the C/A code defined by the latest GPS Interface Specification (Interface Control Document) [46]. Table 1
shows that PSS and SSS have the worst ranging precision,
which is due to their lower bandwidth. The CRS will offer
the best ranging precision and will be robust to multipath due
to its higher bandwidth. However, the CRS is scattered in the
bandwidth, and it is not feasible to exploit conventional DLLs
to track this signal. Therefore, the design of a computationally efficient receiver for navigating with the CRS remains
a challenge.

Multisignal navigation: GPS and cellular
In TOA-based radio navigation, the quality of the receiver's
navigation solution is determined by both the pseudorange
measurement noise statistics and the spatial geometry of the
transmitters. GNSS position solutions suffer from a relatively

LTE versus CDMA
It is difficult to fairly compare the experimental results
obtained by the CDMA and LTE signals in the "Navigation

Storage

GRID GPS SDR

USRP
RIO
NI-2930
USRP
LabVIEW-Based
LTE SDR

LTE
Antennas

MATLABBased
Estimator

GPS
Antenna
(a)

GPS Nav. Solution
LTE Nav. Solution
Total Trajectory: 2 Km
RMS Error: 9.32 m
Stand. Dev.: 4.36 m
Max Error: 33.47 m

200 m
(b)

FIGURE 7. (a) The experimental setup. (b) Locations of eNodeBs and estimated receiver trajectory using GPS and LTE signals. Map data courtesy of
Google Earth.
IEEE SIGNAL PROCESSING MAGAZINE

September 2017

121

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