Ashtech | Z-sensor | GPS and GPS+GLONASS RTK

Dr. Frank van Diggelen
Ashtech Inc. Sunnyvale, California, USA
ION-GPS, September 1997 Ò New Product DescriptionsÓ
This paper presents not only two new RTK
products, but also a major breakthrough in global
positioning technology, taking RTK where it has
never gone before.
The first new product is the Z-Sensor, a single board
implementation of AshtechÕs reknowned Z12
The second product is the GG-RTK receiver, the
worldÕs first GPS+GLONASS RTK product.
RTK has represented the peak of GPS performance
for several years, but there have been severe
limitations - with fewer than 5 satellites in view
RTK does not work at all, or works so slowly as to
be almost no better than DGPS in many
applications. Now, by combining GPS and GLONASS
in an RTK product, you can do RTK, for the first
time ever, in places such as open pit mines, urban
canyon, river valleys, etc., where GPS-only RTK
simply will not work.
The paper is organized as follows:
Background & RTK
Introduces RTK. Readers familiar with the
concepts may want to skip directly to The
The Products
The most significant product specifications.
RTK Performance
Presents the parameters by which we will judge
RTK performance.
Field Test Results
Presents real-world results for both products.
How much benefit do we get from the current,
partial, GLONASS constellation? More than
youÕd think.
Reliability of GPS+GLONASS
Addresses the ever popular topic of GLONASS
reliability, and shows how the GPS+GLONASS
receiver has been architectured so that, no
matter how GLONASS system performance
compares to GPS system performance, the GGRTK receiver always delivers better performance
and reliability than a GPS-only receiver.
T o obtain precise position from a GPS receiver, we use
techniques called Ò Differential GPSÓ. T his involves
two GPS receivers. One is stationary, at a known
point, we call this the Ò BaseÓ receiver. T he Base
receiver transmits data over a radio link. A second GPS
receiver, possibly moving, at an unknown point,
calculates precise position by using the signals it
receives from the satellites, and the data it receives via
radio from the Base. Differential GPS usually gives
about one meter accuracy. RTK is a special form of
Differential GPS that gives about one-hundred times
greater accuracy.
T he GPS system uses a coded signal from which a
receiver derives distance and thus position. T he code is
a string of bits, like the ones and zeros in a computer.
T he receiver sees this code as tick-marks on a giant
tape-measure, every transition from one to zero or
back appears as a tick on the Ò tape measureÓ. T he C/A
code has ticks at about 300 meter spacing. T he
encoded information gives the equivalent of the
numbers on a tape-measure, and the receiver uses these
codes to measure position to meter-level accuracy.
T he military, encrypted, P-code gives even better
accuracy, roughly twice as good as the best C/A code
receiver. But the microwave carrier, which is there
ostensibly only to carry the coded signals, provides the
best Ò tape-measureÓ of all, with tick-marks at about
20cm spacing. T he receiver can measure these signals
to centimeter precision. T he trouble is, the carrier
provides the equivalent of a very precisely graduated
tape-measure with no numbers on it. If the receiver
software could use the code to derive the numbers on
this carrier Ò tape-measureÓ, it would provide GPS
accuracy of centimeters. T his is exactly what RTK
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
GPS - a tape measure
from space
Tape measure from space
Code gives tick-marks
~every 300m, with
labels for each tick
Carrier gives tick marks ~every 20 centimeters
but with no labels
T he GPS satellites provide the equivalent of a tape measure from space. The tape has labeled tick-marks at ~300m
intervals (the code), as well as unlabelled tick-marks at ~20cm intervals (the carrier). A receiver can measure the code
to 1m precision, and the carrier to one centimeter precision. A receiver that can compute the ÔlabelsÕ on the carrier
can then deliver centimeter position accuracy. This is what RT K receivers do.
What about GLONASS? GLONASS is the Russian GPS, and is almost identical in operation to the United States GPS.
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
RT K stands for Ò Real T ime KinematicÓ, but it means
real-time-centimeters. A brief dig through the layers of
GPS history uncovers the origins of the terminology.
T he first use of GPS for centimeter position relied on
static receivers which collected only carrier data, hours
of data, to be processed later on desktop computers.
T his was called ÒstaticÓ surveying. T he technique
evolved till minutes, not hours, of data was enough,
Ò rapid-staticÓ. Later techniques required only one
static initialization of the receiver. From then on, as
long as the receiver maintained phase-lock, it could be
moved and the data would still yield centimeter
accuracy when processed back in the office. T his was
Ò kinematicÓ surveying. With faster and smaller
computers, the desktop processing moved into the GPS
receiver itself, providing results in the field, in realtime. Hence ÒReal T ime KinematicÓ.
T here used to be a distinction between RT K with static
initialization and RT K on-the-fly, but any modern
RT K receiver that cannot do on-the-fly initialization
is not worth bothering about.
Both products are available in two standard
configurations: Base and Remote. Remote units have
RT K capability, standard. Base units can provide
Differential corrections and RT K data, at the same
time, so you can operate DGPS and RTK remote units
simultaneously with the same base station.
Both products conform to the RT CM standard for
Differential and RT K data. So any other receiver
(from any manufacturer) that also conforms to the
RT CM standard can be used compatibly as a Base or
Remote unit with an Ashtech Remote or Base,
respectively, as long as it tracks the correct signal (dual
frequency GPS, compatible with Z, or single frequency
GPS+GLONASS, compatible with GG-RT K).
T he Z Sensor also supports a more concise RT K
format, as an alternative to RT CM, to reduce
bandwidth requirements for RT K. The GG-RT K
receiver achieves low bandwidth requirements, using
the standard RT CM format, thanks to the low drift
rate of GLONASS errors, and the fact that it is a single
frequency receiver.
Both receivers provide high rate RT K updates with no
extrapolation of old positions. Every position is based
on a fresh set of range measurements at the Remote
Z Sensor
Specification hilite s:
12-channel, all-in-view, full wave length code
and carrier phase on L1 and L2
Z-tracking (all observables, even with A-S on)
Horizontal accuracy (1s) 1cm.
Vertical accuracy (1s) 2cm.
10 Hz RTK position output
Less than 30ms latency
RTK on-the-fly initialization: > 99.9%
RTCM messages: 1, 2, 3, 6, 16, 18, 19, 22
Power Consumption 7.5W
Form Factors:
Z-Surveyor: Injected molded plastic housing,
including display, integral battery, removeable PCMemory Card, and optional internal radio.
Z-Surveyor FX: Metal waterproof housing including
display, internal PC-Memory Card, and optional
internal radio.
Z-Sensor: Metal housing, no display, no memory,
optional internal radio.
OEM Boards: The Z-Sensor is based on a singleboard design. The boards are available for OEM
purchase. Contact Ashtech for details.
For full specifications see product data sheets.
Specification hilite s:
· 12 channels GPS, L1 code and carrier
· 12 channels GLONASS, L1 code and carrier
· Horizontal accuracy (1s) 1cm.
· Vertical accuracy (1s) 1cm.
· 5 Hz RTK position output
· Less than 100ms latency
· RTK on-the-fly initialization: > 99.9%
· RTCM: 1, 2, 3, 9, 16, 18, 19, 22, 31, 32, 34
· Power Consumption 2.6W (Sensor) 1.8W
Form Factors:
GG-Surveyor: Metal housing, no display, internal
PC-Memory Card, optional internal radio.
GG-RTK Sensor: Metal housing, no display, no
memory, optional internal radio.
GG RTK OEM Board: Eurocard format.
Any existing GG24 receiver can be upgraded to a GGRT K with no hardware changes.
For full specifications see product data sheets.
RT K is a special form of Differential GPS. It is
differes from conventional DGPS in three major
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
1. T ypical accuracy is one hundred times as good as
2. T here is an initialization period following poweron. T his initialization calculates the integer
number of carrier phase wavelengths. This is
known as Ò fixing the integersÓ.
3. T here is a non-zero probability that the
initialization will be wrong.
mistake. At this time the receiver will return to float
mode, and then fix the integers correctly. T he satellite
geometry changes when new satellites come into view
or, if this does not happen in time, when the satellites
move enough in the sky (usually 2 - 10 minutes is
In assessing RT K performance we thus address all three
issues: reliability, speed and accuracy.
We present results achieved by the two new
products, Z-Sensor and GG-RTK. T he initialization
results are summarized in the next two figures. T hese
figures represent the results of, literally, thousand of
tests. For each test the receiver is allowed to fix
integers, and the time it takes to do this is recorded,
then the receiver is reset. T his gives us a measure of
the integer-fixing phase of RT K initialization. T ests
were done both for static and dynamic cases, the time
to initialize is the same whether the receiver is static or
RELIABILIT Y - T he best RT K receivers offer
greater than 99.9% reliability. To put this in context:
suppose you turned on your RT K receiver once per
day, the receiver fixed integers and from then on
maintained lock on at least 4 satellites all day, then
your receiver would fix integers incorrectly about once
every three years.
Both the Z-Sensor and GG-RTK receivers provide
greater than 99.9% reliability, as well as providing the
user with control of the reliability. T he user may
choose from three Ò formal reliabilityÓ settings,
corresponding to probabilities: 95%, 99% and 99.9%.
T he receiver guarantees that the achieved reliability is
greater than the formal reliability setting. T he greater
the reliability the slower the initialization.
SPEED RT K initialization is split into two stages, the
acquisition phase (when the satellites signal is acquired)
and the integer fixing stage (when the integer numbers
of wavelengths are computed). T he GG-RT K receiver
is always faster than a dual-frequency receiver for the
acquisition phase (because it only needs to acquire the
single-frequency C/A code, which is easy). On short
baselines (<1km) the GG-RT K receiver is also faster
than dual-frequency receivers for the integer-fixing
stage. On medium and long baselines the Z receiver is
faster than GG-RT K for the integer fixing stage.
T ypical times to fix integers are 30 seconds through 2
ACCURACY - Once the integers are fixed
correctly, RTK accuracy is at the centimeter-level.
During the integer fixing phase, while the integers are
being fixed they are modeled as real numbers (or
floating-point numbers), and the position is referred to
as a Ò floatÓ solution. Float solutions have accuracy
ranging from DGPS levels (meter level) to decimeter
level, depending on how long the receiver has been
tracking the signals.
If the integers are fixed incorrectly, the position will
have float-solution accuracy, but the statistical
indicators available in the field will make it look like it
has centimeter accuracy. T his will persist until the
satellite geometry changes, and the receiver realizes its
T he first plot shows the results for the Z receiver. T he
results are organized into three baselines: short (<1km),
medium (3 to 7 km) and long (19 km). For the long
baseline tests, data was logged to a PC and processed on
the PC to give an approximation of what we expect to
see in real time. PC-processed data is shown as a dotted
line on the plots. Real time data collected so far agrees
closely with the data processed on the PC, but, where
we have not collected enough real-time data to give
meaningful statistical results (i.e. thousands of tests) we
show only the PC-processed results.
Time since reset (seconds)
Z Re ce ive r, RTK intializ ation, inte ge r-fixing
phase . Short, Me dium and Long base lines
T he second plot shows results for the GG-RTK
receiver, collected and displayed in a similar way. The
Z Receiver results are shown, in light grey, on this plot
for reference.
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
Once the solution is fixed, horizontal accuracy is 1cm
1s with degradation of 1 part per million on long
baselines. A GPS paper wouldnÕt be complete without a
scatter plot showing accuracy, and so here it is:
Time since reset (seconds)
GG-RTK Re ce ive r, RTK intializ ation, int.-fixing
phase . Short, Me dium and Long base lines
T he above data was all collected with the default
reliability setting, which is: formal reliability = 99%. In
both cases (Z and GG-RTK) the achieved reliability
exceeded the formal reliability.
T he results show two things very clearly:
1. For short baselines GG-RT K is much faster than
dual-frequency GPS-only RT K, with integer-fixing
initialization occuring within 1.5 seconds in 50%
of tests.
2. For longer baselines dual-frequency GPS-only RT K
is faster than GG-RT K.
T his plot was obtained from 12 hours of GG-RT K
position data, with a short baseline. In this example the
accuracy is well better than the specified accuracy. T he
horizontal rms accuracy of these RT K positions is
0.5cm. T he worst case error is 1.3cm
What about floating point accuracy? T he following
plot shows the typical behavior seen when the GGRT K receiver does take a long time to fix integers. In
this example the receiver took 11 minutes to fix
integers on a 7km baseline, but the floating point
accuracy converged to 20 centimeters within a few
minutes - ideal performance for applications requiring
10-20 cm aaccuracy levels, such as guidance and
machine control.
Time since reset (minutes)
Float solution convergence of GG-RTK
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
T he current satellite constellations provide 25
healthy GPS satellites, and 40 healthy GPS+GLONASS
satellites. For RTK initialization, 5 satellites are
required. T his is no problem if the whole sky is visible,
and, as already shown, dual frequency GPS-only RT K
performance is similar to single-frequency GG-RTK
when the whole sky is visible. However, there are
occassions when there is no comparison between a
GPS-only system, and a GPS+GLONASS system, and
this is when large parts of the sky are blocked, such as
in an open-pit mine, urban canyon, or river valley. To
demonstrate this we did a simulation with a 30° mask
angle (this is typical for an open pit mine). In this
environment, 5 or more GPS sats are available only 6
hours per day, and the 6 hours is fragmented throug the
day, making RT K a practical impossibility in these
environments - if you only have GPS. But, with the
combined GPS+GLONASS constellation the availability
of 5 or more satellites improves by 300% to 18 hours
per day.
T he following plot shows todayÕs satellite availability,
comparing what you get with 25 satellites, to what you
get with 40 satellites. T he plot shows what percentage
of the time the indicated number of satellites are
visible. T his plot was generated by doing an 8-day
simulation, with a 10 degree mask angle.
% Availability
T his paper ends with the bottom line - cost.
Both new receivers are available at significant cost
reduction over competing products. T he Z receiver has
been cost reduced by integrating what used to be on 5
circuit boards into one single board. T he GG-RT K
receiver is even less costly for a surprising reason adding GLONASS to RT K reduces cost. HereÕs how:
T he second GPS frequency is encrypted. T his means
that dual frequency GPS systems for civilian use have
to perform significant extra processing to extract the
observables from the encrypted signal. T his makes the
receivers more complex and therfor more costly. By
using GLONASS instead of the second GPS frequency
to provide the extra observables required for RTK, the
cost of RT K systems has been significantly reduced.
For GG-RTK we automatically enable RAIM
(standard) whenever you set up a base station. If the
base station sees errors on the order of 100m, then
RTK (or Differential) , simply removes the errors as
part of the normal operation. If the base station sees
an unexpected error (e.g. worse than 1km) on any
satellite, then it knows something is wrong, and it
immediately removes that satellite from the set of
broadcast data, and then the remote station stops
using the satellite too.
So, what many think is the BIG issue for
GPS+GLONASS, satellite reliability, is a non-issue
for RTK or Differential, if youÕre using an Ashtech
base station.
The remote unit operates its own RAIM algorithm
to detect and repair cycle slips. The remote receiver
also weights measurements appropriately, so that the
combined GPS+GLONASS position is always at least
as accurate than either system alone.
³ 12
³ 11
³ 10
# Satellites
Satellite availability, Kansas City, September
What about the performance of the satellites
themselves? There have been documented cases, for
both GPS and GLONASS, of the satellite clocks
generating errors of thousands of kilometers in
stand-alone positions for many minutes before the
respective system control set the satellite to
T here are now two options for RT K: dualfrequency-single-system and single-frequency-dualsystem. Dual frequency systems have advantages on
longer baselines, and present users with a costperformance trade-off for anything but short baselines.
T he trade-off becomes noticeable at baselines of
around 5km. For shorter baselines there is no trade-off,
single-frequency-dual-systems not only costs less, but
perform better than dual-frequency GPS-only RT K.
Ashtech is a registered trademark, and GPS+GLONASS,
and GG-RT K are trademarks of Ashtech Inc.
Presented at ION-GPS 97, Kansas City, MO. ÒNew Product DescriptionsÓ Session.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF