Aerospace and Electronic Systems Magazine August 2017 - 10

Feature Article:

DOI. No. 10.1109/MAES.2017.160169

Flight Data Assessment of Tightly Coupled PPP/INS
Using Real-Time Products
Ryan M.Watson, Jason N. Gross, West Virginia University, Morgantown, WV, USA
Yoaz Bar-Sever, William I. Bertiger, Bruce J. Haines, Jet Propulsion Laboratory,
Pasadena, CA, USA

INTRODUCTION
Airborne geodetic techniques are superior to their terrestrial counterparts with respect to both economy and efficiency [1]. In addition, airborne geodesy allows mapping of remote areas that would
otherwise be inaccessible. A cornerstone for most airborne geodetic
measurements is the accurate determination of the aircraft position
and orientation. Therefore, airborne geodesy was not widely used
until the advent of global navigation satellite systems (GNSSs).
Now, with precise GNSS positioning techniques, airborne geodesy
is booming within several domains, including solid Earth monitoring (e.g., crustal deformation) [2]-[4], fluid Earth monitoring (e.g.,
ice sheet or sea-level monitoring) [5]-[7], and geoid determination
[8], [9]. Despite the success of these airborne geodetic methods,
the increased availability and reliability of accurate aircraft positioning remains an important enabling technology in support of
future scientific endeavors.
Precise GNSS processing techniques can be broadly put into
two categories: (1) single-receiver processing with undifferenced
observations (i.e., precise point positioning (PPP)) and (2) carrierphase differential GNSS (CP-DGNSS) processing (e.g., real-time
kinematic (RTK) or network RTK (NRTK)). CP-DGNSS processing strategies use additional static GNSS reference receivers to
mitigate correlated error sources through cancellation by data differencing (e.g., atmospheric delays, ephemeris errors, and clock
biases) [10], whereas PPP techniques rely on global correctors for
the GNSS orbit and clocks, along with models and dual-frequency
data, to mitigate these errors [11], [12].
The most common CP-DGNSS configuration is RTK, which
consists of a single static GNSS reference receiver at a well-known

Authors' current addresses: R. Watson, J. N. Gross, Mechanical and Aerospace Engineering, West Virginia University,
395 Evansdale Drive, Morgantown, WV 26506, USA, E-mail:
(jason.gross@mail.wvu.edu). Y. Bar-Sever, W. I. Bertiger, B. J.
Haines, Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 81101, USA.
Manuscript received August 10, 2016, revised October 27,
2016, November 14, 2016, and ready for publication November 30, 2016.
Review handled by M. Braasch.
0885/8985/17/$26.00 © 2017 IEEE
10

location transmitting data to the roving platform so that doubledifferenced observables can be formed. This configuration works
well for short-baseline separation between mobile and reference
receiver locations and can readily produce centimeter-level positioning errors for airborne kinematic applications [13]-[15]; however, in this configuration, it is well known that the positioning
errors grow in proportion to the distance between the roving and
the reference receivers [16]. As such, RTK with successful integer
ambiguity resolution is only feasible for a roughly 10-km radius
around the reference receiver [10].
The maximum separation distance between the reference and
the roving receivers is greatly extended by using a network of
static GNSS receivers, i.e., NRTK [17]. The NRTK approach
allows separations of approximately 100 km. However, it has
been shown that NRTK density spacing of less than 18 km is
necessary to reduce network-side GNSS error sources to a level
at which the roving receiver's multipath errors are the dominant
error source [18]. Unfortunately, even the extended range of
NRTK remains problematic for airborne sensing applications
that can easily span hundreds of kilometers and often carry out
missions in remote locations that do not have a dense GNSS
reference network.
Another configuration, known as PPP-RTK, has been developed to leverage a network of static GNSS stations and extend the
maximum baseline separation of RTK [19]. The PPP-RTK method
resolves the carrier-phase ambiguities for the network and provides
that information in addition to precise orbit and clock information
used by traditional PPP to accurately determine the platform's position [20]. However, this positioning technique is subject to the
same baseline limitation as NRTK.
To fully overcome the limitation imposed by requiring proximity to GNSS reference stations, the PPP processing strategy is
the most promising precise GNSS processing technique for many
specialized airborne geodetic applications. Several studies have
shown that the accuracy of PPP with respect to CP-DGNSS is
comparable. For example, for static positioning, Colombo and Evans show a 10-cm agreement after PPP filter convergence [21]. In
addition, for kinematic applications, Honda et al. show decimeter-level positioning error agreement between RTK and PPP [22].
These studies show that PPP and RTK produce similar positioning
accuracy after an initial convergence period of the PPP filter. This
point was elaborated on in [23], where it is concluded that addi-

IEEE A&E SYSTEMS MAGAZINE

AUGUST 2017

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