Aerospace and Electronic Systems Magazine August 2017 - 46

Feature Article:

DOI. No. 10.1109/MAES.2017.160176

High Sensitivity Acquisition of GNSS Signals With
Secondary Code on FPGAs
Jérôme Leclère, École de Technologie Supérieure (ÉTS), Montreal, Canada
Cyril Botteron, Pierre-André Farine, École Polytechnique Fédérale de Lausanne
(EPFL), Lausanne, Switzerland

INTRODUCTION
The modern global navigation satellite systems (GNSS) signals,
such as the Global Positioning System (GPS) L5 and L1C, and
Galileo E5 and E1, have brought several innovations: the introduction of a pilot channel that does not contain any data to allow very
long coherent integrations; the introduction of a secondary code
to offer better cross-correlations, to facilitate the synchronization
with the data, and to help interference mitigation; the introduction
of new modulations to reduce the impact of multipath; and the use
of higher chipping rates to have better accuracy and interference
mitigation.
Although having a secondary code brings some advantages,
it also presents some drawbacks. Indeed, with the modern GNSS
signals, there is now a potential sign transition (i.e., a carrier phase
shift of 180°) between each period of the primary code, unlike the
GPS L1 C/A signal that has a potential sign transition each 20 code
periods only. These sign transitions are one of the limitations of
the coherent integration time, and thus of the receiver sensitivity
[1]-[4]. Therefore, to use a long coherent integration time and get
high sensitivity, the delay of the secondary code must be estimated.
There have been several proposals to address this problem. In
[5], [6], it was proposed to synchronize with the primary code first,
and then synchronize with the secondary code. However, this implies the ability to detect the signals using only one period of the
primary code, which is not the case in the high sensitivity context.
In [7], [8], it was proposed to extend the coherent integration time
by estimating the possible combinations of several secondary code
chips, and using this to determine the secondary code delay [9], but
these methods are still not adapted to the high sensitivity context.

Authors' current addresses: J. Leclère, ÉTS LASSENA (Electrical Engineering Department), 1100 Notre-Dame West Street,
Montreal, QC H3C 1K3, Canada, E-mail: (jerome.leclere@
lassena.etsmtl.ca). C. Botteron, P.-A. Farine, EPFL STI IMT
ESPLAB, Rue de la Maladière 71b, CP 526, CH-2002 Neuchâtel 2, Switzerland.
Manuscript received July 30, 2016, revised January 2, 2017,
February 18, 2017, March 6, 2017, March 17, 2017, and ready
for publication March 22, 2017.
Review handled by A. Dempster.
0885/8985/17/$26.00 © 2017 IEEE
46

To get high sensitivity, the coherent integration time should be at
least one period of the secondary code, or a multiple of it. In [10],
it was proposed to determine the primary code delay with a serial
search and the secondary code delay with a fast Fourier transform
(FFT) based correlation; however, the serial search is too timeconsuming for a realistic implementation. In [11], the authors
proposed to perform an FFT-based correlation over one period of
the secondary code with the L5 signal; nevertheless, this requires
very large FFTs (length greater than 218), which are not compatible
with a hardware implementation. Finally, [12] proposed to perform
FFT-based correlations over one period of the primary code (doubling the length to manage the sign transition), and to combine the
results according to the secondary code chips.
In this article, we will focus on this last method. More specifically, we will compare different hardware implementations of
this method. Indeed, the combinations can be performed before or
after the correlations with the local primary code; they can be computed sequentially or in parallel; and the output can be computed
in different orders (checking all the primary code delays for one
secondary code delay, or checking all the secondary code delays
for one primary code delay). The objective is therefore to identify the most efficient implementations. Note that these different
implementations are not approximations; they all provide the same
output and thus the same performance in terms of sensitivity. We
will also present a method that approximately halves the number of
operations related to the secondary code correlation, still without
impacting the sensitivity, and see how it can reduce the processing
time with the hardware implementations.

ACQUISITION OF GNSS SIGNALS
SIGNAL DEFINITION
The signal received by a GNSS receiver is the combination of several GNSS signals coming from U different satellites, plus a noise
term. Thus, after the front-end, the discrete baseband signal can be
written as
U

sb (nTS ) =  sbu (nTS ) + ηb (nTS ),

(1)

u =1

where sbu(nTS ) is the discrete baseband signal from satellite u, n is
the discrete time index, TS is the sampling period equal to 1/fS with

IEEE A&E SYSTEMS MAGAZINE

AUGUST 2017

http://jerome.leclere@lassena.etsmtl.ca

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine August 2017